ceramic armor materials by design

CeramicArmorMaterials by Design

This page intentionally left blank

Ceramic ArmorMaterials by Design

Volume 134

CeramicTransactions

Published byThe American Ceramic Society735 Ceramic PlaceWesterville, Ohio 43081www.ceramics.org

Edited by

James W. McCauleyU.S. Army Research Laboratory

Andrew Crowson U.S. Army Research Laboratory

William A. Gooch, Jr.U.S. Army Research Laboratory

A.M. RajendranU.S. Army Research Laboratory

Stephan J. BlessThe University of Texas at Austin

Kathryn V. LoganGeorgia Institute of Technology

Michael NormandiaU.S. Army Research Laboratory

Steven WaxU.S. Defense Advanced Research Projects Agency

Proceedings of the Ceramic Armor Materials by Design Symposium held at the Pac RimIV International Conference on Advanced Ceramics and Glass, November 4–8, 2001 inWailea, Maui, Hawaii.

Proceedings of the Ceramic Armor Materials by Design Symposium held at the Pac Rim IVInternational Conference on Advanced Ceramics and Glass, November 4–8, 2001 in Wailea, Maui,Hawaii.

Copyright 2002,The American Ceramic Society. All rights reserved.

Statements of fact and opinion are the responsibility of the authors alone and do not imply anopinion on the part of the officers, staff, or members of The American Ceramic Society.TheAmerican Ceramic Society assumes no responsibility for the statements and opinions advancedby the contributors to its publications or by the speakers at its programs. Registered namesand trademarks, etc., used in this publication, even without specific indication thereof, are not tobe considered unprotected by the law.

No part of this book may be reproduced, stored in a retrieval system, or transmitted in anyform or by any means, electronic, mechanical, photocopying, microfilming, recording, or other-wise, without prior written permission from the publisher.

Authorization to photocopy for internal or personal use beyond the limits of Sections 107 and108 of the U.S. Copyright Law is granted by the American Ceramic Society, provided that theappropriate fee is paid directly to the Copyright Clearance Center, Inc., 222 Rosewood Drive,Danvers, MA 01923 USA, www.copyright.com. Prior to photocopying items for educationalclassroom use, please contact Copyright Clearance Center, Inc.

This consent does not extend to copying items for general distribution or for advertising or promotional purposes or to republishing items in whole or in part in any work in any format.

Please direct republication or special copying permission requests to the Senior Director,Publications,The American Ceramic Society, PO Box 6136,Westerville, Ohio 43086-6136, USA.

COVER PHOTO:“Post-test photograph of impact of AP bullet against ceramic/aluminum tar-get: (a) front view of ceramic element;” is courtesy of Charles E. Anderson Jr., and appears asfigure 4a in his paper “Developing an Ultra-Lightweight Armor Concept,” which begins on page485.

For information on ordering titles published by The American Ceramic Society, or to request apublications catalog, please call 614-794-5890.

Printed in the United States of America.

4 3 2 1–05 04 03 02

ISSN 1042-1122 ISBN 1-57498-148-X

v

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi

CERAMIC ARMOR DEVELOPMENTAn Overview of Ceramic Armor Applications . . . . . . . . . . . . . . . 3

William A. Gooch Jr., U.S. Army Research Laboratory

Armor Ceramics Under High-Velocity Impact of a Medium-Caliber Long-Rod Penetrator. . . . . . . . . . . . . . . . . . . . 23

Hans-Jürgen Ernst,Volker Wiesner, and Thomas Wolf,French-German Research Institute of Saint-Louis (ISL)

Practical Issues in Ceramic Armor Design . . . . . . . . . . . . . . . . . 33Bryn James, Defense Science and Technology Laboratories

Ballistic Development of Tungsten Carbide Ceramics for Armor Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Pierre-François Peron, Etablissement Technique de Bourges

Ballistic Development of U.S. High Density Tungsten Carbide Ceramics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

William A. Gooch and Matthew S. Burkins, U.S. Army Research Laboratory; Richard Palicka, Cercom Incorporated

Initial Tests on Ceramics in Composite Armor . . . . . . . . . . . . . 63W. Lanz, RUAG Land Systems

Structure and Properties of Shock-Resistant Ceramics Developed at the Institute for Problems in Materials Science . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

B.A. Galanov, O.N. Grigoriev, S.M. Ivanov, and V.V. Kartuzov, National Academy of Sciences of Ukraine

Ceramic Armor with Submicron Alumina Against Armor Piercing Projectiles. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

E. Strassburger and B. Lexow, Fraunhofer-Institut für Kurzzeitdynamik Ernst-Mach-Institut (EMI); A. Krell, Fraunhofer-Institut für Keramische Technologien und Sinterwerkstoffe

Contents

vi

Armor Alumina Ceramics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91Eugene Medvedovski, Ceramic Protection Corporation

Ballistic Performance of Alumina Ceramic Armors. . . . . . . . . . 103Murat Vural and Zeki Erim, Istanbul Technical University;B.A. Konduk and A.H. Ucisik, Bogazici University

PENETRATION AND BALLISTIC TESTINGAn Overview of Ballistic Testing Methods of Ceramic Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

Michael J. Normandia and William A. Gooch, U.S. Army Research Laboratory

Theory and Experimental Test Methods for Evaluating Ceramic Armor Components . . . . . . . . . . . . . . . . . 139

Marc A. Adams, Jet Propulsion Laboratory

Long Rod Penetration of Ceramics . . . . . . . . . . . . . . . . . . . . . 151D.L. Orphal, International Research Associates

Depth of Penetration Testing . . . . . . . . . . . . . . . . . . . . . . . . . . 165Bryn James, Defense Science and Technology Laboratories

Transition Between Interface Defeat and Penetration for a Given Combination of Projectile and Ceramic Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173

Patrik Lundberg, René Renström, and Lars Westerling, Swedish Defense Research Agency, FOI

SHOCK AND HIGH STRAIN RATE DYNAMICDynamic Fracture of Ceramics and CMC . . . . . . . . . . . . . . . . 185

Albert S. Kobayashi, University of Washington

Compressive Fracture of Brittle Solids Under Shock-Wave Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197

G. I. Kanel, Institute for High Energy Densities; S. J. Bless, The University of Texas at Austin

Recent Developments in Split Hopkinson Pressure Bar Testing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217

W. Chen and B. Song, The University of Arizona; D. J. Frew and M. J. Forrestal, Sandia National Laboratories

vii

Using Bar Impact to Determine Dynamic Properties of Ceramics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225

Stephan J. Bless, The University of Texas at Austin

Shock Compression and Release Properties of Coors AD995 Alumina. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233

William D. Reinhart and Lalit C. Chhabildas, Sandia National Laboratories; Dennis E. Grady, Applied Research Associates; andTsutomu Mashimo, Kumamoto University

Compressibility and Shear Strength of Titanium Diboride Under Plane Shock Wave Loading . . . . . . . . . . . . . . . . . . . . . . 249

D. P. Dandekar and E. J. Rapacki, U.S. Army Research Laboratory

Dynamic Indentation Damage of Ceramics . . . . . . . . . . . . . . . 261Do Kyung Kim, Chul-Seung Lee, and Young-Gu Kim, Korea Advanced Institute of Science and Technology; Chang Wook Kim and Soon Nam Chang, Agency for Defense Development

Taylor-Impact Experiments for Brittle Ceramic Materials. . . . . 269L. C. Chhabildas and W. D. Reinhart, Sandia National Laboratories;D. P. Dandekar, U.S. Army Research Laboratory

ANALYTICAL AND COMPUTATIONAL MODELINGHistorical Perspective on Ceramic Materials Damage Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281

A.M. Rajendran, U.S. Army Research Laboratory

A Comparison of Ceramic Material Models . . . . . . . . . . . . . . 299Douglas W.Templeton, U. S. Army Tank Automotive Research,Development, and Engineering Center;Timothy J. Holmquist,Network Computing Services Inc./Army HPC Research Center;Hubert W. Meyer Jr., David J. Grove, and Brian Leavy, U.S. Army Research Laboratory

Modeling Ceramic Dwell and Interface Defeat . . . . . . . . . . . . 309Timothy J. Holmquist and Gordon R. Johnson, Network CS/Army High Performance Computing Research Center

3D Finite Element Analysis of Impact Damage in Metallic and Ceramic Targets . . . . . . . . . . . . . . . . . . . . . . . . . . 317

Fenghua Zhou and Jean-Francois Molinari, Johns Hopkins University

viii

A Numerical Investigation of Microcracking Diffusion in Sandwiched Glass Plates . . . . . . . . . . . . . . . . . . . . 329

Z. Chen and L. Shen, University of Missouri-Columbia; G.I. Kanel and S.V. Razorenov, Russian Academy of Sciences

Analytic Model for Penetration of Thick Ceramic Targets . . . . 337James D.Walker, Southwest Research Institute

Grain Level Analysis of Ceramic Microstructures Subjected to Impact Loading . . . . . . . . . . . . . . . . . . . . . . . . . . 349

Pablo D. Zavattieri and Horacio D. Espinosa, Northwestern University

Analysis and Modeling of Ceramic Armor Penetration. . . . . . . 361S.J. Cimpoeru and R.L.Woodward, DSTO Aeronautical and Maritime Research Laboratory

Overview of the Rajendran-Grove Ceramic Failure Model . . . 371D. J. Grove and A. M. Rajendran, U. S. Army Research Laboratory

DAMAGE EVOLUTION AND MICROMECHANISMSFailure Phenomenology of Confined Ceramic Targets and Impacting Rods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385

Donald A. Shockey and A.H. Marchand, SRI International;S.R. Skaggs, G.E. Cort, M.W. Burkett, and R. Parker, Los Alamos National Laboratory

Micro-Mechanisms of Compression Failure . . . . . . . . . . . . . . . 403Sia Nemat-Nasser and Sai Sarva, University of California, San Diego

Damage Mitigation in Ceramics: Historical Developments and Future Directions in Army Research . . . . . . . . . . . . . . . . . 421

D.M. Stepp, U.S. Army Research Office

Indentation Damage Behavior of Armor Ceramics. . . . . . . . . . 429Do Kyung Kim and Chul-Seung Lee, Korea Advanced Institute of Science and Technology; Chang Wook Kim and Soon Nam Chang, Agency for Defense Development

Progress in the 3-D Visualization of Interior Ballistic Damage in Armor Ceramics . . . . . . . . . . . . . . . . . . . . . . . . . . 441

Joseph M.Wells, Nevin L. Rupert, and William H. Green, U.S. Army Research Laboratory

ix

PROCESSING AND MANUFACTURINGAn Assessment of Low Cost Manufacturing Technology for Advanced Structural Ceramics and Its Impact on Ceramic Armor . . . . . . . . . . . . . . . . . . . . . . 451

Richard E.Tressler, The Pennsylvania State University

High-Purity Submicron �-Al2O3 Armor Ceramics Design, Manufacture, and Ballistic Performance . . . . . . . . . . . . 463

Andreas Krell, Fraunhofer Institut für Keramische Technologien und Sinterwerkstoffe(IKTS); Elmar Strassburger, Fraunhofer Institut für Kurzzeitdynamik (EMI)

Solid Freeform Fabrication of Advanced Armor Concepts: Opportunities for Design and Manufacture . . . . . . . 473

R.C. McCuiston, S.C. Danforth, M.J. Matthewson, and D.E. Niesz,Rutgers,The State University of New Jersey

ULTRA-LIGHTWEIGHT AND NOVEL CONCEPTSDeveloping an Ultra-Lightweight Armor Concept . . . . . . . . . . 485

Charles E. Anderson Jr., Southwest Research Institute

Ceramics That Exhibit a Threshold Strength . . . . . . . . . . . . . . 499F. F. Lange, M.P. Rao, K. Hbaieb, and R.M. McMeeking,University of California at Santa Barbara

Novel Ideas in Multi-Functional Ceramic Armor Design . . . . . 511Sia Nemat-Nasser, Sai Sarva, Jon B. Isaacs, and David W. Lischer,University of California, San Diego

A New Family of Reaction Bonded Ceramics for Armor Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 527

M. K. Aghajanian, B. N. Morgan, J. R. Singh, M Cubed Technologies, Inc.;J. Mears and R. A.Wolffe, Simula Safety Systems, Inc.

Flexible Ceramic Coated Fiber Fabrics for Lightweight Protection Systems . . . . . . . . . . . . . . . . . . . . . . . . 541

Konstantin von Niessen and Rainer Gadow, University of Stuttgart

Improved Performance of Alumina Ceramics with Carbon Nanotube Reinforcement . . . . . . . . . . . . . . . . . . . . . . 551

Michael Sennett, Natick Soldier Center; Sekyung Chang,Robert H. Doremus, Richard W. Siegel, Pulickel M. Ajayan, and Linda S. Schadler, Rensselaer Polytechnic Institute

x

Recent Progress on the Influence of Microstructure and Mechanical Properties on Ballistic Performance . . . . . . . . 557

J.C. LaSalvia, U.S. Army Research Laboratory

Transparent ArmorTransparent Armor Materials: Needs and Requirements . . . . . 573

Parimal J. Patel and Gary A. Gilde, U.S. Army Research Laboratory

Microwave Reactive Sintering to Fully Transparent Aluminum Oxynitride (AlON) Ceramics . . . . . . . . . . . . . . . . . 587

Dinesh Agrawal, Jiping Cheng, and Rustum Roy, The Pennsylvania State University

An Investigation of the Transmission Properties and Ballistic Performance of Hot Pressed Spinel. . . . . . . . . . . . . . . 595

Mark C.L. Patterson, Technology Assessment & Transfer Inc.; Don W. Roy,Independent; and Gary Gilde, U.S. Army Research Laboratory

Microstructure and Macrostructure EffectsThe Effect of Microstructure on the Dynamic Behavior of Composite Alumina/Titanium Diboride . . . . . . . . . . . . . . . . 611

Kathryn V. Logan, Georgia Institute of Technology

Phase Equilibrium Studies in Al2O3-TiB2 . . . . . . . . . . . . . . . . . . 623Isabel K. Lloyd, University of Maryland; Kevin J. Doherty and Gary A. Gilde, U.S. Army Research Laboratory

Microstructure Development of Aluminum Oxide/TitaniumDiboride Composites for Penetration Resistance . . . . . . . . . . 629

J.W. Adams, G.A. Gilde, and M. Burkins, U.S. Army Research Laboratory; L. Prokurat Franks, U.S. Army Tank-Automotive and Armaments Command

The Effect of Metal-Ceramic Bonding on Ballistic Impact. . . . . 635Kevin J. Doherty, U.S. Army Research Laboratory

Aspects of Geometry Affecting the Ballistic Performance of Ceramic Targets. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 643

I. M. Pickup, A. K. Barker, R. Chenari, and B. J. James, Defense Science and Technology Laboratories;V. Hohler, K.Weber, and R.Tham,Faunhofer-Institut fur Kurzzeitdynamik (EMI)

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 651

xi

This volume contains the proceedings of the “Ceramic Armor Materials byDesign” symposium held at the Pac Rim IV International Conference onAdvanced Ceramics and Glasses held November 4–8, 2001 in Wailea, Maui,Hawaii.

In 1998, the Army formally approved a new basic research StrategicResearch Objective (SRO)—“Armor Materials by Design”. This actionresulted from a critical assessment of the survivability requirements offuture lightweight weapon systems, as well as the emerging materials andmechanics science and engineering that could be brought to bear on thisproblem. It was concluded that there was a critical need for an integrated,multi-disciplinary basic research program that would result in the capabilityto actually design materials for passive, kinetic energy, armor applications.

Since some high performance structural ceramics have been shown to haveoutstanding armor properties at relatively low weight, the symposium wasorganized to address the ceramic armor aspects of the SRO. Researchersfrom around the world working in private industry, academia, and govern-ment organizations on passive transparent and opaque ceramic armorwere invited to participate in this special program.

It was the goal of the symposium to connect ballistic performance tomacro, micro, and crystallographic mechanisms of damage evolution as wellas static and high strain rate mechanical properties and to assess the cur-rent status of computer codes to model and simulate the ballistic per-formance of these materials against kinetic energy projectiles. Currentstate-of-the-art research and development, as well as some historical con-tent, was incorporated into an integrated program.

Most of the credit for this symposium goes to the organizing committeeconsisting of William A. Gooch Jr. and Michael Normandia, U.S. Army

Preface

Research Laboratory, Andrew Crowson, A. M. Rajendran, and David Stepp,Army Research Office of the Army Research Laboratory, Stephan J. Bless,University of Texas, and Steven Wax, Defense Advanced Research ProjectsAgency.

The symposium was co-sponsored by Steven Wax of the U. S. DefenseAdvanced Research Projects Agency, Andrew Crowson, A. M. Rajendranand David Stepp of the Army Research Office of the Army ResearchLaboratory, and William A. Gooch Jr. and James W. McCauley of the U. S.Army Research Laboratory.This support was critical to the success of thesymposium.

Finally, thanks also go to Ms. Susan J. Burns, Battelle, Research Triangle Park,NC for her tremendous help with assembling this book.

James W. McCauley, Chair, Organizing Committee

Editors

James W. McCauley

Andrew Crowson

William A. Gooch, Jr.

M. Rajendran

Stephan J. Bless

Kathryn V. Logan

Michael Normandia

Steven Wax

Ceramic Armor Development

AN OVERVIEW OF CERAMIC ARMOR APPLICATIONS

William A. Gooch Jr.

U.S. Army Research Laboratory

Weapons and Materials Research Directorate

Aberdeen Proving Ground, MD 21005-5066

ABSTRACT

The increasing capability of modern anti-armor threats and the need to field

lower weight combat vehicles, capable of engaging an opponent with little

preparation, have intensified the need for highly effective passive armor systems.

Ceramic armor technology offers significant advantages for meeting future

protection requirements, particularly for the U.S. Army’s Future Combat System.

The investigation and application of ceramics against small arms threats has a

long history, dating back to the early 1960s and the ballistic performance of

ceramic armors for personnel protection is very high; the principles governing

these defeat mechanisms and the design parameters against such threats are now

generally understood. However, achieving similar ceramic performance versus

larger caliber, kinetic energy penetrator threats have long presented a difficult

challenge. This paper presents an overview and discussion of the ballistic

requirements, ceramic design factors and a chronology of significant U.S.

developments and applications of ceramics for armor.

INTRODUCTION

The application of ceramics for armor continues to be primarily used in

lightweight armor systems for protection against small arms and machine gun

threats. The design of these systems is typically based upon the mechanical

properties of the ceramic to fracture the penetrator and the ability of a rear

compliant layer to catch the projectile debris and the damaged ceramic material.

For defeat of these low-velocity, short projectiles, the fracture mechanism occurs

very early in the process with the majority of the interaction time dedicated to

energy conversion of the kinetic energy of the debris into deformation and

delamination of compliant backing. For medium caliber and heavy armor

applications, where the dominant threat is modern, high velocity, heavy metal

eroding projectiles, the defeat mechanisms are much more complicated and of

longer time duration. For the past three decades, a wide variety of research

Ceramic Armor Materials by Design 3

To the extent authorized under the laws of the United States of America, all copyright interests in this publication are the propertyof The American Ceramic Society. Any duplication, reproduction, or republication of this publication or any part thereof, withoutthe express written consent of The American Ceramic Society or fee paid to the Copyright Clearance Center, is prohibited.

programs, both domestic and foreign, have focused on developing improved

ceramic armor systems for the defeat of these threats. This paper presents an

overview and discussion of the ballistic requirements, ceramic design factors and

a chronology of significant developments and applications of ceramics for armor,

with emphasis on research conducted on ceramic armors at the U.S. Army

Research Laboratory (ARL).

TERMINAL BALLISTIC EFFECTS

A review of the difference in terminal ballistic effects observed during the

interaction of different classes and caliber’s of kinetic energy (KE) projectiles is

important to understand the required defeat mechanisms and armor designs. The

delineation between the threat projectiles is primary related to the caliber, velocity

and energy available, but is not exact and some projectiles cross over into the two

categories discussed below. While the penetrator/target interactions for these two

categories involve similar processes, defeat of the higher performance, long rod

threats require different emphasis in the armor design parameters to be successful

and the progress has been much slower.

Small Arms/Heavy Machine Gun Defeat

Historically, ceramic composite armor systems were designed to defeat

armorpiercing (AP), kinetic energy projectiles, mainly in the small arms and

heavy machine gun category. These AP projectiles are purely inertial rounds,

most commonly made of hard steel (HRc 60-64), of moderate density (7.85

g/cm3) with a few select rounds employing even harder tungsten carbide (WC)

cores at higher densities (13.5-15.0 g/cm3). The hard core is generally encased in

a thin jacket of a more ductile metal for interior ballistic or aerodynamic

considerations, but penetration performance of the bullet is controlled by the core

properties. Such projectiles typically have a length to diameter (L/D) ratio in the

range of 3:1 to 5:1 with moderate muzzle velocities of less than 1 km/s. The

generally accepted high-end caliber is 14.5-mm, typified by the Soviet KPV

family of heavy machine guns. Some saboted, light armor-piercing (SLAP)

rounds have velocities up to 1.3 km/s but with reduced core weight. Overall, these

projectiles tend to produce a total KE on the order of 103 - 10

4 J.

Early Research [1–4] discovered that the perforation of ceramic armor

systems occurred in three general stages: 1. shattering; 2. erosion; and 3. catching.

During the shattering phase, the penetrator fractures and breaks on the surface of

the ceramic plate; the high compressive strength of the ceramic overmatches the

loading produced by the penetrator impact, and the penetrator material flows and

shatters. This initial stage is followed by a period of damage accumulation in the

ceramic material initiated by tensile wave reflections, and bending of the ceramic

4 Ceramic Armor Materials by Design

tile and backing plate. During the second stage of ceramic armor penetration, the

ceramic material is cracking, but the ceramic material can still contribute to defeat

of the penetrator core through erosion mechanisms. In the final catching phase,

the ceramic has lost considerable strength, but ceramic and backing combine to

reduce the velocity through momentum transfer mechanisms.

The defeat mechanism for hard-core AP projectiles is primarily stages 1 and 3

with projectile fracture upon impact against an armor plate having sufficient

hardness and/or high obliquity. The shattering and subsequent dispersion of the

fragments result in a dissipation of the kinetic energy of the core over a larger

area than if intact, thereby achieving defeat of the round with a reduced amount of

armor plate. Monolithic ceramic plates were best suited to produce the shattering

phenomena due to their high hardness and low densities. However, ceramic armor

requires a backup component to support the ceramic and delay failure during the

initial impact/shattering interaction; the backup component then serves to absorb

the residual projectile fragments and comminuted ceramic particles (Phase 3). The

state of the art in protection against small arms threats is typified in lightweight,

two-component ceramic faced composite armors designed for use in breast plates

for personnel body armor, armored helicopter seats and appliques to metal or

composite based vehicle structures.

Heavy Metal Long Rod KE Projectile Defeat

The mechanism for defeat of long rod penetrators (LRP) is more complex than

for the conventional AP projectiles described above. These penetrators are

commonly made of high strength, high density materials, such as tungsten

sintered alloy or depleted uranium, having densities near 18 g/cm3 with moderate

hardness, good toughness and ductility; hence, the projectiles are not susceptible

to shattering as hard core, relatively brittle, AP projectiles. This category includes

APDS and armor-piercing, discarding-sabot, fin stabilized (APDSFS) projectiles,

in calibers from 20-mm up to >140-mm. These LRPs are designed with a high

L/D ratio (currently fielded examples exceed 30:1) and the high density core

material coupled with relatively high muzzle velocity (1.3 - >1.6 km/s), yields KE

in excess of 106 J, creating a high energy density per unit area of target impacted

than with a corresponding hard core AP round. These factors, when combined

with the greater projectile length and reduced propensity for fracture, makes the

LRP a much more effective penetrator. Even if the frontal portion of the LRP can

be effectively damaged, a substantial portion of the rod remains to continue the

armor penetration process. Thus, the conditions that allow a simple ceramic


composite to function effectively for small arms defeat do not apply when the

armor is impacted by a LRP. The primary defeat mechanism is erosion (Phase 2)

and the effectiveness is relatively low for simple ceramic armor systems.

SIGNIFICANT DEVELOPMENTS IN CERAMIC ARMOR TECHNOLOGY

Figure 1. Abex-Norton

Ceramic/ Metal Composite

Attempts to increase performance of ceramic tiles continued during the 1980’s

to present, as penetrator threats evolved. Researchers realized that increased

efficiency of ceramics might be possible by lengthening the duration of the

shattering stage of the penetration process, and/or by increasing the efficiency of

the erosion process of the comminuted ceramic material. These researchers found

that modest lateral confinement allowed constraint

of the broken ceramic pieces, thereby enhancing the

erosion phase of the ceramic penetration process.

This confinement could be obtained by casting, as

seen in the then very efficient armor developed in

1984-86 by ABEX-NORTON where silicon carbide

tiles were inserted into very accurately cast

aluminum matrices [5](Figure 1). Additional

examples include test geometry’s proposed by

Woolsey and others [6,7] to provide a stiff and

substantial confinement of the ceramic tiles in

depth of penetration (DOP) configurations.

The most significant observations during this period, however, were in 1987

by Hauver et al [8] who examined test geometry’s that delayed the generation of

damage in the ceramic tile, thereby increasing the duration of the shattering phase

of the penetrator defeat process. As penetrator threats increased in length and L/D

ratio, Hauver realized that the shattering stage duration was critical to the overall

efficiency of the ceramic defeat process; he demonstrated ceramic tile

confinement geometry’s that substantially

increased the shattering/erosion phase of the

penetrator defeat process to completely erode

the penetrator (Dwell). These experiments

employed compressive confinement of the

ceramic tile (heat shrink of the metal

surround), in combination with techniques to

delay tensile wave and bending damage to the

ceramic. The ceramic performance was

enhanced through control of system geometry

to minimize damage and increase the

shattering stage of penetrator defeat.Figure 2. Hauver’s Observation

of Ceramic Dwell in Laboratory


However, the overall mass and space efficiencies of these laboratory packages

were low, due to the considerable confinement materials employed in the

geometry.

In 1994, research lead to the demonstration of a set of medium caliber and

full-scale armor targets that incorporated existing ceramic defeat knowledge into

an armor technology known as tandem ceramic armor (TCA) [9]. TCA

determines the optimum performance of a specific cross-sectional ceramic armor

design and then repeats the designs in multiple, shock-isolated sections; the

performance is thus additive (Figure 3). Laboratory targets, utilizing conventional

laminated ceramic-metal technology, demonstrated system designs that produced

the state of the art for KE performance. A limiting factor, however, was the space

requirements that grew as the penetrator performance increased.

TANDEM ARMOR SYSTEM

1. CERAMIC TILE

2. CONFINEMENT FRAME

3. POLYMERIC ADHESIVE

4. SUPPORTING PLATE

METAL/COMPOSITE)

5. THIN GRP SECTION

(OPTIONAL)

6. HONEYCOMB/ISOLATION

MATERIAL

7. VEHICLE HULL

Figure 3. Tandem Ceramic Armor Concept

The latest efforts to generate increased efficiency in ceramic armors are to

enhance both the erosion and “dwell” mechanisms of ceramic armor for

penetrator defeat. The development of hot-isostaticpress (HIP) processing of

ceramics with metal surrounds (Figure 4) has demonstrated dwell on the ceramic

front surface of laboratory scale

threats at efficient armor system

areal densities [10]. This HIP

processing forms a macro

composite through the

generation of residual

compressive stresses (mismatch

of thermal expansion

coefficients of the ceramic tiles

and metal confining plates) in Figure 4. Hot-Isostatic Pressed Metal

Encapsulated Ceramic


the ceramic tile during cool down of the HIP assembly from the pressing

temperature. The macrocomposite is then able to withstand the large ballistic

bending loads during round impact, so that the ceramic tile resists fracture and

retains a high compressive strength. The macro-composite formed by HIP

processing also keeps the broken ceramic pieces confined during the second

erosive phase of the ceramic armor defeat process, should it occur, thus

maintaining a high erosive efficiency.

CERAMIC ARMOR DESIGN REQUIREMENTS

As with all armor systems, many design factors and production decisions

influence the effectiveness of ceramic armors. These processes have to be

understood and controlled to maintain performance.

Ceramic Type

The technical ceramics available for armor are numerous, but generally are

divided into the lower cost sintered and the higher cost hot-pressed ceramics. The

higher cost ceramics are justified when the lowest areal weight system is the main

requirement with the prime ceramics being boron carbide for body armor and

airborne platforms and silicon carbide for ground vehicles. The high density

tungsten carbide ceramic has specific applications where space is a limiting factor

[11]. The lower cost sintered 99.5% aluminum oxide or silicon carbide or the

reaction-bonded ceramics can be used were weight is not the driving requirement.

However, for many armor applications, ultra-high hard steels, titanium or

laminates of these materials with aluminum/composite backings are very

competitive in performance with significant engineering advantages. Table 1 lists

some of the primary producers of ceramics used today in armor applications.

Table 1. Current Ceramic Armor Producers

Ceramic Type Producer/Type

Sintered Coors CAP3 99.5% Alumina

Morgan Matroc (UK) Alumina

ETEC Alumina

Ceradyne Sintered SiC

Pure Carbon SiC

Reaction-Bonded M-Cubed (Simula) SiC

MC2 (Australia) SiC, B4C

Metal Matrix Composite Lanxide Dimox AS109

Lanxide Dimox-HT

Hot Pressed Cercom B4C, TiB2, WC

Ceradyne B4C, TiB2, SiC

Saint-Gobain B4C


Bonding/Impedance Effects: The use of hard face ceramic materials, bonded

unto metal and composite backings, is typified in the classic work by Wilkens et

al [12,13,14] on understanding the fundamental penetration mechanics that occur

during the interaction of a hardcore steel projectile with a hard face aluminum

oxide ceramic on aluminum. The primary applications involve bonding the

ceramic to the metal or composite backing with lowdensity, low-impedance, and

low shear strength adhesives. The unfavorable impedance effects and induced

tensile failure across these boundaries and at the lateral boundaries of the ceramic

are well documented by Hauver [15,16,17] for eroding long rod penetrators.

A less understood, but equally important effect from the use of a low-shear

strength adhesive has been documented by a number of researchers. In 1993,

Furlong et al [18] presented an exact solution for the transmission of spherical

waves across planar surfaces; the coefficients of reflection and refraction were

shown to depend not only on the acoustic impedance’s of the media, but also on

the boundary conditions at the interface, the wave face curvature, and the source

frequency. Three types of boundaries can exist: 1) free, as with a free standing

ceramic plate; 2) no shear-coupling, as with two unbonded or lightly bonded

plates or 3) shear-coupling, where good adhesion or coupling exists, allowing the

transmission of transverse motion and stress. The latter shear-coupled or no slip

condition provides the best interface for a ceramic/metal armor design. Similar

investigations were conducted by Alme [19]. Leighton et al [20] discussed the

increased ballistic performance of laminated ceramic-titanium composites that

resulted from increased interlayer bond strength (strong, shear-coupled

metallurgical bonds). These effects are inherent in functionally gradient materials

(FGM) composites as observed by Gooch et al [21] where metal layers transition

into the ceramic layers without interfaces.

Adhesive Thickness and Uniformity: In simple bonded ceramic-metal

laminates, an important factor to eliminate variability in ballistic performance is

to maintain a uniform adhesive bonding layer at the minimum thickness. Burkins

[22] modified the standard DOP test configuration by examining the ballistic

results of a set of Taguchi experiments where the rear ceramic/metal bond

thickness and lateral side confinement bond thickness were varied. The least

variance occurred with a minimum bond thickness for the side and rear. For DOP

tests, the maximum bond thickness allowed for the rear and sides is 0.127-mm

(0.005 in). The uniformity of the bond thickness is maintained by placing spacers

in the adhesive.


Confinement and Stiffness: The design of efficient ceramic systems begins by

considering the mechanisms by which a ceramic tile fails during loading and

designing the armor system to reduce the stresses contributing to early failure of

the ceramic tile. Consideration of the ballistic event with emphasis on penetrator

interface defeat on the ceramic front surface (Figure 5) lead Horwath [23] to

determine two primary areas of concern: (1) the compressional loading of the

ceramic directly under the penetrator rod, and (2) the maximum flexure of the

ceramic plate and tensile stress/strain at the ceramic plate rear surface. These two

factors are heavily influenced by the side and rear confinement thickness and

materials.

Compressive Loading Under

Projectile

Deflection and Back Surface

Stress of Ceramic tile Under

Ballistic Load

Figure 5. Primary Areas of Ceramic Tile

Multi-hit Requirements, Edge and Joint Impacts: As with all armor systems,

the requirement to provide full protection against multiple impacts is still valid for

ceramic armor designs. This requirement significantly impacts the design and tile

size of ceramic designs. Table 2 lists the U.S. Army minimum impact spacing

requirements for metals, metal laminates and ceramic laminates.

Table 2. Multi-hit Impact Requirements for Vehicular Armor

WEAPON CALIBER METALS AND METAL

LAMINATES* (mm)

CERAMIC

LAMINATES* (mm)

7.62-mm 27 54

12.7-mm 45 90

14.5-mm 51 102

20-mm 70 140

23- to 25-mm 75 145

30- to 50-mm 152 76 (105)** 152

* Minimum spacing between impacts measured from center to center of impacts

in vertical plane

** Full bore AP bullets only


The requirement to impact edges or joints and maintain the same ballistic

performance as center tile impacts is a major design requirement for ceramic

armor systems. Generally, ceramic design is driven by the edge or joint protection

with the center providing greater protection. In some ceramic designs, the edges

of the ceramic tiles are raised to increase the thickness to equalize the protection

across the ceramic tile face. These factors result in increased areal weight for the

design.

APPLICATIONS OF CERAMIC ARMOR FOR COMBAT VEHICLES

The application of ceramics as the main protection technology has made

major advances in the last decade and represents the accepted technology, in use

today, for small arms and heavy machine gun protection, primarily as a ceramic

laminate applique over metal structural base armors and a few, newer composite

based systems. A few systems were designed against 30-mm APDS, but few

armored systems have been designed against larger threats. The following

paragraphs describe some of the military armor applications in use or

development today. These are representative, but not inclusive, of the myriad

examples of ceramic armors under development worldwide. The information was

provided by the fabricators and producers of the ceramics and products.

Armorworks

Armorworks Incorporated of Phoenix, AZ fabricates a wide range of ceramic

composite products. Shown in Figure 6 is armored kit for an AH-60H helicopter

floor. This armor system is an aluminum oxide based armor system that provides

7.62-mm APM2 protection at muzzle velocity. The armor kit consists of five

panels, two of which are removable in flight (cargo hook access) and are nested in

aft panel. The armor kit mounts on top of the floor panels using exiting fastener

points on the floor with coverage of about 5.1-m2 (55-ft

2). The armor panels

passed MIL-STD-810E environmental

testing including high and low

temperature, solar radiation, sand and

dust, salt fog, high pressure wash,

humidity, fungus, vibration-resonance

and vibration-endurance tests. The tile

and backing are bonded; the gross

panel shape is then fabricated by

cutting and grinding and diamond

saws and cores drills the holes to the

final panel configuration.

Courtesy Amorworks

Figure 6. Armorworks AH-60H Floor

Armor


Ceradyne Incorporated

Ceradyne Incorporated of Costa Mesa, CA develops and produces a wide

range of advanced ceramics for many applications including ballistic grades such

as hot-pressed boron carbide,

silicon carbide and titanium

diboride, pressureless sintered

silicon carbide and reaction-bonded

and sintered silicon nitride.

Ceradyne has a long history of

armor development beginning in the

1960’s with the first applications of

boron carbide for combat helicopter

protection. Today, Ceradyne

designs, develops and manufactures

ceramic armor such as the ceramic

breast plates and Cobra helicopter

bucket seat of Figure 7.

Photos Courtesy of Ceradyne

Figure 7. Ceradyne Body Armor and

Helicopter Seat

Cercom Incorporated

Cercom Incorporated of Vista, CA has been a prime producer of a wide range

of commercial and ballistic grades of ceramics since 1985. Using their pressure-

assisted densification (PAD) process, Cercom has hot-pressed large quantities of

aluminum nitride, boron carbide,

silicon carbide, silicon nitride,

titanium diboride and tungsten

carbide ballistic ceramics for the

U.S. Army. Figure 8 shows Cercom

ceramic tiles on the European Tiger

helicopter seat and two different

types of Cercom hotpressed boron

carbide body armor inserts, a

single-piece, compound curvature

plate that is used in the U.S. Army

Small Arms Protective Insert

(SAPI) vest and two examples of

multiple tiles fabricated into single

protective inserts.

Photos Courtesy of Cercom

Figure 8. Cercom Ceramic Tiles for (L)

Tiger Helicopter and (R) body armor

inserts


German Ingenieurbüro Deisenroth

The German Company Ingenieurbüro Deisenroth (IBD) of Lohmar, Germany

has established itself as a world leader in the variety and quantity of vehicles

incorporating the MEXASTM ceramic/metal/composite design; MEXASTM stands

for Modular, EXpandable Armor Systems and is composed of layered appliques

that can be added to a basic vehicle structure to give the desired protection. While

not conceptionally different from other appliques, the early use and continued

application of this design is noteworthy and at least 39 different vehicles in ten

countries utilize MEXASTM, including Austria, Switzerland, Germany, Canada,

U.S., France, Italy, Finland, Sweden and Norway. Figure 9 provides a collage of

the different vehicle types utilizing the MEXASTM system, from engineer vehicles,

tactical trucks, and numerous wheeled and tracked combat vehicles.

Figure 9. Examples of tactical and combat vehicles

Photos courtesy of NDHQ Canada and Ingenieurbüro Deisenroth

that mount IBD MEXASTM armor

The design of the MEXASTM system is shown in Figure 10 where a second

layer of protection is being placed over the first. The vehicle structure provides

the base protection and this system could be configured against three different

missions or selective uparmoring of the vehicle. The panels are mounted by

threaded attachment studs that accept special recessed fasteners.


Photos courtesy of Ingenieurbüro Deisenroth

Figure 10. Multiple layer concept of IBD MEXASTM armor

Detroit Diesel General Motors of Canada

The Canadian National Defense Forces have been very active in providing

increased protection for a wide range of tactical and support Canadian equipment.

This requirement is driven by the deployment of their forces in a number of

peace-keeping operations and the threat of increased small arms threats. Shown in

Figure 11 is the Canadian LAV III Armored Personnel Carrier (APC) that has

protection against small arms AP threats. The ceramic MEXASTM composite

armor is fabricated by the Canadian company DEW Engineering and

Development Limited of Miramichi, New Brunswick, Canada under license to

IBD. The characteristic mounting hardware of IBD armor is readily visible in the

LAV III glacis area. DEW has supplied over 750 kits to the Canadian Defense

Forces.

Photos Courtesy of Program Manager Brigade Combat Team

Figure 11. The MEXASTM

armor panels mounted on the Canadian LAV III APC


General Motors Defense Systems

The U.S. Army has initiated a major development program to transform the

existing family of heavy vehicles to a lighter, more agile and deployable force.

The Future Combat System (FCS) is planned for fielding by 2015. As part of the

transformation, a contract to purchase an interim family of light vehicles under

the Interim Brigade Combat Team has been awarded to GM GDLS Defense

Group L.L.C. of Sterling Heights, MI [24]. Among the many variants is the

Infantry Combat Vehicle (ICV) shown in Figure 12. Based on the LAV III chassis

and hull, the ICV mounts a version of the MEXASTM system of IBD. The

similarities in the design and mounting are visible.

Photos Courtesy of Program Manager Brigade Combat Team

Figure 12. The ICV of the Interim Brigade Combat Team utilizes

the IBD MEXASTM

applique

The GM GDLS contract indicates the ICV is to have overhead and all around

protection for the squad and crew from 152-mm Artillery high explosive airburst

at an undisclosed distance from and above the vehicle. The ICV shall also provide

integral 360 and overhead squad and crew protection from 7.62-mm AP threats

and 360 squad and crew protection from 14.5-mm AP ammunition, both fired

from undisclosed impact conditions. The ICV shall also provide the capability to

mount add on armor packages to protect against hand held shaped charge

warheads up to and including the RPG-7.

Textron Marine and Land Systems

Textron Marine and Land Systems of New Orleans, LA is the prime fabricator

for two interesting applications of ceramic composites, the U.S. Army Armored

Security Vehicle (ASV) and the Marine Corp Landing Craft, Air Cushion

(LCAC) vehicle (Figure 13). On the initial vehicle procurement, the ceramic

composite armor kit on the ASV was produced by Simula Safety Systems of

Phoenix, AZ, based on a MEXAS. license from IBD. Textron is currently

working on a new composite armor design. The ASV offers front, rear and side


protection from 0.50-caliber armor-piercing ammunition. The LCAC is a high-

speed, over-the-beach fully amphibious, landing craft capable of carrying a 60-75

ton payload. Critical areas of the vehicle including the turbine housings are also

protected with a Simula-developed, aluminum oxide composite.

Photos Courtesy Textron

Figure 13. The Textron ASV and LCAC vehicles both mount composite armors

General Dynamics Land Systems

General Dynamics Land Systems Division (GDLS), Sterling Heights, MI has

licensed and acquired an advanced, lightweight armor technology, named

SURMAX™ Armor. This armor technology is used on the sides and rear of the

hull and the sides of the turret of the U.S. Marine Corps' Advanced Amphibious

Assault Vehicle (AAAV) to protect the vehicle from 14.5-mm AP threats and

artillery fragments (Figure 14).

Figure 14. Marine Corp AAAV and SURMAX™ being mounted

on AAAV spaceframe

Photos Courtesy General Dynamics Land Systems

SURMAX™ consists of a ceramic composite front panel attached to an armor

backing. The backing can be a composite material (such as Kevlar or S-glass) or

the structure of a vehicle (such as aluminum, steel, or titanium). The combination

of the front panel and the backing are used to stop the penetrator and the

application of SURMAX™ on the spaceframe structure of the AAAV is shown in

Figure 14. SURMAX™ is also used on the U.S. Army's wheeled Armored


Ground Mobility System (AGMS) and the flexible panels can be fit to curved

surfaces such as wheel wells (Figure 15).

Photos Courtesy General Dynamics Land Systems

Figure 15. SURMAX™ on AGMS with curved panels in wheel wells

Extensive testing of AAAV armor at tight multi-hit distances was required for

engineering development and

Government validation tests. The

AAAV SURMAX™ side armor panel

on the left of Figure 16 is a typical

validation target, after four partial

penetrations, three with 14.5-mm AP

and one with a simulated artillery

fragment. Shots 1 and 2 are located

101-mm (4”) apart. Shown on the right

is a multi-hit test panel with ten 0.50-

caliber AP impacts for the AGMS, all

partial penetrations with the tenth shot

at a distance of 76-mm (3”) from one

previous shot.

Photos Courtesy General Land Systems

Figure 16. Ballistic Multi-hit

Qualification Tests for the AAAV and

the AGMS

Simula Incorporated

Simula Incorporated, Phoenix, AZ has been a designer and fabricator of

ceramic composite armor components since 1970 for a wide range of products

from helicopter seats, aircraft armors, body armors and ceramic composite

appliques for a range of vehicles. Figure 17 shows some of Simula’s ceramic

components. On the left is an AH64 helicopter seat that is fabricated from hot-

pressed Cercom boron carbide on Kevlar backing; the middle picture shows the

Interceptor body armor vest with hot pressed ceramic tile inserts; and the photo on

the right shows one-piece sintered silicon carbide plates made by M-Cubed of

Monroe, CT which also can be used in the Interceptor vest.


Figure 17. Simula products: (L) AH64 Helicopter seat. (M)

Ceramic Inserts for the Interceptor Body Armor vest. (R)

Sintered silicon carbide one-piece inserts for body armor

Photos Courtesy Simula

United Defense Limited Partnership

The Ground Systems Division of United Defense Limited Partnership

(UDLP), headquartered in York, PA is one of the largest ground vehicle

fabricator’s in the U.S. The use of ceramic composite materials and structures has

been in development for many years and UDLP has progressed through three

generations of systems. The first generation development was the Composite

Infantry Combat Vehicle technology

demonstrator (Figure 18) which

replaced most of the aluminum

structure of the M2 Bradley Fighting

Vehicle with a S-2 glass reinforced

composite. This allowed the

demonstration of bonding of titanium

diboride tiles to the hull sides for

14.5-mm protection. Tile spacing

and cutouts are visible.

Photo Courtesy UDLP

Figure 18. 1st Generation CIFV with

TiB2 tiles

The 2nd

generation UDLP

ceramic composite armors can be

seen in the well-designed ballistic

protection of the M8 Armored Gun

System (Figure 19). The M8 is fitted

with bolt-on appliques and boxes

that can provide different levels of

protection from KE penetrators to

hand-held shaped charge warheads.

The ceramic composite shows

multiple impacts of 7.62-mm AP

projectiles on the test panel.

Photo Courtesy UDLP

Figure 19. M8 passive/shaped charge

armors being tested


The 3rd

generation UDLP armor is seen in the Composite Armored Vehicle

(CAV) technology demonstrator (Figure 20). The CAV incorporates full spectrum

protection into the vehicle design, including 7.62-mm hull protection as well as

enhanced 30-mm protection to the crew station. The multi-hit performance of the

hex tiles used in the hull design as well as the excellent multi-hit protection of the

silicon carbide/titanium crew station armor against 30-mm threats is shown.

Photos Courtesy UDLP

Figure 20. CAV Technology demonstrator and hull and crew station

protection tests

CONCLUSIONS

This paper has presented an abbreviated overview and discussion of the

ballistic requirements, ceramic design factors and chronology of significant U.S.

developments over the last 30 years. The applications of ceramics for armor are

growing rapidly as the need for lighter and more agile combat vehicles increases.

Ceramic armor technology offers the best potential for meeting future protection

requirements, particularly for the U.S. Army’s Future Combat System.

REFERENCES1.

C. Donaldson, “The Development of a Theory for the Design of Lightweight

Armor”, Aeronautical Research Associates of Princeton, Inc., Technical Report

AFFDL-TR-77-114.2.

A. L. Florence, “Interaction of Projectiles and Composite Armor”, Stanford

Research Institute, AMMRC-CR-69-15, August 1969.3.

A. M. Prior, “The Penetration of Composite Armor by Small Arms

Ammunition”, Proceedings of the International Ballistic Symposium, 1986. 4.

A. K. Wong and I. Berman, “Lightweight Ceramic Armor - A Review”, Army

Materials and Mechanics Research Center, Report No AMMRC-MS-71-1, 1971. 5.

Final Report, “Demonstration of Cast, Composite Ceramic Armor (C3A), BRL

Contract DAAA-15-86-C-0014, 1990. 6.

P. Woosley, “Ceramic Materials Screening by Residual Penetration Ballistic

Testing”, 13th

International Symposium on Ballistics, June 1992.


7. B. Morris and C. Anderson, “The Ballistic Performance of Confined Ceramic

Tiles”, 1991 Ground Vehicle Survivability Symposium, April 15, 1991. 8. G. Hauver and J. Dehn, “Interface Defeat Mechanisms in Delayed

Penetration”, 14th DEA-G-1060 Armor/Anti-armor Workshop. 9. W. Gooch, J. Prifti, P. Woolsey, J. Mackiewicz and W. Perciballi, “Tandem

Ceramic Armor for Defeat of Kinetic Energy Penetrators, ARL-TR-1946, May

1999.10.

E. Horwath and W. Bruchey, “The Ballistic Behavior of HIP Encapsulated

Ceramic Tiles”, 8th Annual TARDEC Ground Vehicle Survivability Symposium,

Monterey, CA, March 1997. 11.

W. Gooch and M. Burkins, “Ballistic Development Of U.S. High Density

Tungsten Carbide Ceramics”, Dymat 2000, Krakow, Poland, 23-29 September

2000.12.

M. Wilkins, C. Honodel and D. Sawle, "An Approach to the Study of Light

Armor", UCRL-50284, June 1967. 13.

M. Wilkins, C. Honodel and D. Sawle, "Second Progress Report of Light

Armor", UCRL-50349, Nov 1967. 14.

M. Wilkins, C. Honodel and D. Sawle, "Third Progress Report of Light Armor

Program", UCRL-50460, July 1968. 15.

G. Hauver, P. Netherwood, R. Benck, W. Gooch, W. Perciballi and M.

Burkins, "Variations of Target Resistance During Long-rod Penetration into

Ceramics", 13th Int. Ballistics Symposium, Stockholm, Sweden, 1992. 16.

G. Hauver, P. Netherwood, R. Benck and L. Kecskes,"Ballistic Performance

of Ceramics", U.S. Army Symposium on Mechanics, Plymouth, MA, 17-19

August 1993. 17.

G. Hauver, P. Netherwood, R. Benck and L. Kecskes, "Enhanced Ballistic

Performance of Ceramics", 19th Army Science Conference, Orlando, FL, 20-24

June 1994. 18.

J. Furlong, C. Westbury and E. Phillips, “A Method for Predicting the

Reflection and Refraction of Spherical Waves across Planar Interfaces”, J. of

Applied Physics, Vol. 76, July 1994. 19.

M. Alme, Alme Associates, private communication. 20.

K. Leighton, R. Franz, and A. Gerk, “Laminated Ceramic-Titanium

Macrocomposite Armor”, 8th Annual Ground Vehicle Survivability Symposium,

Monterey, CA, 24-27 March 1997. 21.

W. Gooch, M. Burkins and R. Palicka, “Development And Ballistic Testing Of

A Functionally Gradient Ceramic/Metal Applique”, NATO Applied Vehicle

Technology Panel, Loen, Norway, 7-11 May 2001. 22.

M. Burkins and W. Gooch, “U.S. Ceramic Ballistic Test Methodology and

Data”, TTCPWTP1 Meeting, Maribyrnong, Australia, 10 May 1995.


23. W. Bruchey and E. Horwath, “System Considerations Concerning the

Development of High Efficiency Ceramic Armors”, 17th Int. Sym. on Ballistics,

Midrand, South Africa, March 1998. 24.

DoD Contract DAAE07-00-D-M051, 16 November 2000, Brigade Combat

Team web site.


ARMOR CERAMICS UNDER HIGH-VELOCITY IMPACT OF A

MEDIUM-CALIBER LONG-ROD PENETRATOR

Hans-Jürgen Ernst, Volker Wiesner and Thomas Wolf

French-German Research Institute of Saint-Louis (ISL)

P.O. Box 34, F-68301 SAINT-LOUIS CEDEX (France)

ABSTRACT

In the first part of this paper, continuous measurements of the cratering

process in unconfined targets (Al2O3) of different lateral dimensions as well as in

completely confined constant-volume targets (Al2O3, B4C, SiC, and TiB2) are

given; they are achieved with a penetration gauge developed at ISL. The

confinement and material-related influences on the penetration resistance are

shown.

Secondly, the protective power of these ceramics is quantified and compared

to that of other inert materials. By means of an exponential fitting function, which

is based on the assumption that the very beginning of the penetration process is

not influenced by the geometrical armor configuration, appropriate ballistic

material parameters, called ductile limit of the space equivalence factor, are given.

Based on this parameter, a ballistic screening of materials is presented, which

enables a configuration-independent comparison of the protective power of the

investigated ceramics with that of other inert materials. In conclusion, it shows

that brittle materials are still interesting for light-weight armor design.

INTRODUCTION

The ballistic performance of a single ceramic material depends on both the

thickness of the ceramic block and its constructive environment, often called

confinement1. It is known that the ballistic performance of thick targets decreases

for most of the ceramic materials with increasing thickness 2, 3, 4

. The more a

ceramic material is confined, the more it tends to behave like a ductile one,

achieving a higher ballistic performance 5.

Time-dependent measurements of penetration into steel-confined ceramic

targets are not easily realized, as material thickness and relatively small density

differences exceed the capabilities of common measuring techniques (i.e. X-rays).



That is why, a special terminal ballistic gauge for continuous penetration

measurements has been developed at ISL6, 7

.

This paper has two aims: firstly, a closer insight into the time-dependent

penetration behavior of some ceramic materials is given, based on new results

with the penetration gauge. Secondly, a simplified choice of brittle materials for

light-weight armor design is proposed by introducing a ballistic material

parameter that allows a

configuration independent

evaluation of the protective power

of inert armor materials.

EXPERIMENTAL BASIS

The upper left-hand part of

figure 1 shows a photograph of the

kinetic energy projectile BMU G

154 developed at ISL. Below a

sectional drawing of the rod is

presented. Some penetrator

material data are given in the upper

right-hand table. In the lower left-

hand diagram the reference

penetration Pref is plotted versus

the impact velocity vz

kept constant in this

paper8: at 1800 m/s an

RHA penetration of

about 165 mm is

achieved. The Pref(vz)-

function and its fitting

parameters a and b can

be found in the lower

right-hand corner.

Figure 1: KE projectile BMU G 154

As seen in figure 2,

ceramic blocks of

different thicknesses

(tile thickness: 20 mm,

lateral dimensions:

100x100 mm2;

materials: Al2O3, B4C,

SiC and TiB2) have

been investigated inFigure 2: target set-up and ceramic data


three configurations: "unconfined", “laterally” and "totally confined". Some

ceramic material data are found in the table. In every experiment the ceramic

block thickness tz, and the residual penetration Pres are measured; the reference

penetration Pref is evaluated with the Pref(vz)-function.

Continuous penetration measurements can be made advantageously by means

of a special gauge developed at ISL7. Figure 3 shows in its upper left-hand part a

schematic drawing and below a sketch explaining the working mechanism of this

gauge. It consists of a metallic tube with an electrical resistance wire inside placed

in a hole in the target material (Ø approx. 1 mm) along the expected penetration

axis. The projectile dynamically closes the contact between tube and wire. The

ongoing penetration process causes an electrical resistance decrease that is

measured as a time dependent tension variation.

The upper right-hand part of figure 3 shows a photograph of the gauge; the

evaluation of the gauge signals is explained in the lower right-hand picture. Apart

from the transient beginning of the penetration process and its final phase (from

ceramics into RHA) the experimental penetration curve is approached by a

polynomial P(t) fit function, which yields the cratering velocity u(t) after

differentiation.

Figure 3: ISL penetration gauge


PROTECTIVE POWER DEFINITIONS

Based on experimental results (see figure 2), the total penetration Ptot , the

average density tot of the test set-up, as well as an RHA layer thickness tref ,

which is equivalent to the ceramic block, can be derived 9:

Ptot = tz + Pres , tot = (tz · z = Pres · ref ) / Ptot and tref = Pref – Pres . (1)

Normalized formulations for ceramic thickness tz,n , residual penetration Pres,n and

target density z,n

tz,n = tz / Pref , Pres,n = Pres / Pref and z,n = z / ref , (2)

as well as for total penetration Ptot,n , reference layer thickness tref,n and total

density tot,n

Ptot,n = tz,n + Pres,n , tref,n = 1 – Pres,n and tot,n = tot / ref , (3)

will mostly eliminate the experimentally caused scattering.

Though the protective power of a target set-up against a defined threat can

directly be quantified by the total penetration, normalized ballistic factors are

advantageously used for this purpose. Here, equivalence factors F describe the

volume gain (subscript s) and the mass gain (subscript m) of the ceramic block

under consideration, as compared to the equivalent reference material layer:

Fs = (1 – Pres,n ) / tz,n and Fm = Fs / z,n . (4)

In order to complete the formulations, efficiency factors E are added, which

describe the volume gain (subscript s) and the mass gain (subscript m) of the total

penetration in the test set-up Ptot , as compared to the reference penetration Pref :

Es = 1 / (tz,n + Pres,n ) and Em = Es / tot,n . (5)

Based on the assumption that the very beginning of the penetration process is

not influenced by the geometrical armor configuration, an appropriate ballistic

material parameter, called ductile limit of the space equivalence factor, has been

introduced 10

. The exponential fitting function for the space equivalence factor Fs

and the appropriate formulation for the normalized residual penetration Pres,n are:

Fs = Fs (0) · exp( ·tz,n ) and Pres,n = 1 – Fs (0) ·tz,n · exp( · tz,n) (6)


Fs(0) is the

configuration-

independent ballistic

material parameter. It

quantifies the space

equivalence that is

reached, if the ceramic

material behaves during

penetration like a ductile

one. The approximation

coefficient depends on

both the material and the target configuration.

Figure 4: thickness-dependent residual penetrationand space equivalence of inert materials

Figure 4 shows qualitative diagrams of the proposed approximation function

Fs (left-hand side) and of the resulting Pres,n (right-hand side) dealing with a

hypothetical brittle, ductile and composite material, in order to describe different

types of penetration behavior schematically.

RESULTS

Figure 5: confinement influence on the ballistic

performance of brittle materials

Confinement Influence

When a projectile hits an

unconfined ceramic target, the

ceramic material in front of the

projectile-target interaction zone

is increasingly fractured by the

shock wave and its reflections11

. Depending on the lateral tile

dimensions, the predisturbed

ceramic material can expand

radially, thus causing a density

decrease in the material.

Sufficient lateral dimensions

and/or a well-designed

confinement are able to reduce

or even to stop this expansion

completely5.

The upper diagram of figure

5 firstly shows the decrease in

residual penetration behind 120-

mm thick unconfined blocks of

glass and Al2O3 as a function of

the lateral tile dimension a.


Secondly, a decrease in residual penetration can be seen for laterally confined

blocks (a=100 mm). By following the horizontal dotted lines, significantly higher

lateral tile dimensions for the unconfined configuration are found.

The lower left-hand diagram of figure 5 shows a comparison of gauge

measurements for totally confined and unconfined Al2O3 blocks of equal

thickness. Appropriate second-order polynomial fits of the experimental curves

are given in the table. In the lower right-hand diagram the corresponding cratering

velocity comparison can be seen. The confined configuration shows the lowest

slope of the penetration curve and a significantly smaller cratering velocity, thus

indicating a higher penetration resistance as compared to the unconfined

configuration.

Influence of the Ceramic Material

Further time-dependent measurements were made in order to compare the

penetration resistance of 120-mm-thick totally confined blocks of some often used

armor ceramics (Al2O3,

B4C, SiC and TiB2). In

the left-hand diagram

of figure 6 penetration-

versus time plots are

given for these

ceramics and,

additionally, one for

mild steel. The

equations of the

second-order

polynomial fits

corresponding to the

measured curves are

presented in the table

below. The right-hand

diagram shows the

corresponding cratering

velocities as a function of time. Though the transient region of the penetration

beginning is not well defined, it is obvious that the penetration curves of all

ceramics have a significantly lower slope than that of mild steel, signifying a

higher penetration resistance. SiC and TiB2 have the highest curvature.

Figure 6: penetration gauge measurements in differentceramic materials

Differentiation of the P(t)-polynomials yields straight lines that indicate the

average cratering velocities. The u(t)-curves of all ceramic materials start at lower

values, and with the exception of Al2O3, they have higher slopes than that of mild


steel. Two - eventually accumulating - types of ceramic penetration behavior may

explain these effects:

1. the lower the cratering velocity at the beginning of the penetration process,

the more the ceramic reacts like a rigid target (examples: TiB2 and SiC);

2. the higher the slope of the u(t)-curve the more the projectile is decelerated

during the ceramic penetration process (example: B4C).

Protective Power of Some Ceramics

Detailed results of the DOP experiments with TiB2 and B4C, Al2O3 and SiC are

presented in the four upper diagrams of figure 7. Each of them shows

experimentally determined Fs-data and Fs(tz,n)-curves calculated by using equ. (6)

for the three

investigated

configurations. In

the lower part of

the figure the

protective power

hierarchy of inert

materials is

presented in the

form of a

diagram, in

which the ductile

limit of the space

equivalence is

plotted against

the normalized

density.

Figure 7: protective power of some brittle materials

It can be seen

that TiB2 has the

highest Fs(0)-

value of the

investigated

ceramics; a better

confinement may

still increase its

penetration

resistance. The

space equivalence

factors of Al2O3


are similar to those of B4C. The latter is more interesting for armor use because of

its significantly lower density. In the case of

SiC, the Fs(0)-values range between those of TiB2 and B4C.

In the lower diagram it can be observed that the Fs(0)-value of TiB2 is higher

than those of high-hardness steels, the latter being followed by that of SiC. No

significant differences exist between the ductile limits of the aluminas, B4C,

Si3N4, and titanium; that of glass, aluminium and GFRP materials share a lower

position. It can be observed that the (generally used) Fm(0)-values of high-

hardness steels and GFRP materials are comparable, but their Fs(0)-values

(signifying the ballistic result) differ significantly. Obviously, the good Fm(0)-

values of B4C and of the GFRPs are mainly due to their low densities. Though

GFRPs additionally profit from their increasing thickness-dependent space

equivalence 12

, it can be concluded that ceramics are still useful as light-weight

armor materials.

CONCLUSIONS

Continuous gauge measurements offer a closer insight into the penetration

process of ceramics impacted by a LRP at 1800 m/s. A heavy confinement cuts

the initial cratering velocity in Al203 down like a rigid target and it increases the

deceleration of the projectile. These two effects may explain the protective power

of different armor ceramics too: TiB2 and SiC have relatively low initial cratering

velocities; B4C decelerates the projectile due to a continuously decreasing u(t)-

curve. By optimally using the material intrinsic and/or the structural confinement,

the quasi-ductile penetration behavior of some ceramics might be approached.

A diagram, in which the ductile limit is plotted against the normalized density,

is perhaps more useful to quantify the protective power of ceramics (and other

inert materials) than the solely used mass factor. This graph shows on the one

hand that well-confined ceramics have comparable (SiC) or even higher ductile

limits (TiB2) than high-hardness armor steels and on the other hand it also shows

that the good protective power of B4C is mostly due to its relatively small density.

REFERENCES 1

2

3

Westerling L., Lundberg T., "The Influence of Confinement on the

Protective Capability of Ceramic Armour at Two Different Velocities", 15th ISB,

Jerusalem, Israel, 1995

Andersen Jr. C.E., Walker J.D., Lankford J., "Investigations of the Ballistic

Response of Brittle Materials", SWRI-Technical Report, 1995

Yaziv D., Partom Y., "The Ballistic Efficiency of Thick Alumina Targets

against Long-Rod Penetrators", 14th ISB, Quebec, Canada, 1993, Vol. 2, pp. 331-

340


4

5

6

7

8

9

10

11

12

Hauver G.E., Netherwood P.H., Benck A.F., Gooch W.A., Perciballi W.J.,

Burkins M.S., "Variation of Target Resistance During Long-Rod Penetration into

Ceramics", 13th ISB, Stockholm, Sweden, 1992

Ernst H.-J., Hoog K., Wiesner V., "Ballistic Impact Behavior of Some

Ceramics in Different Environments", EURODYMAT 94, Oxford, UK, 1994

Wiesner V., “Erfassung der Projektilbewegung im Ziel mit

Widerstandssonden“, MEBAL 85, ISL R 116/85, 1985

Ernst H.-J., Hoog K., Wiesner V., Wolf T., “DOP and Continuous Cratering

Measurements in Differently Confined Ceramics”, EAFV Symp., Shrivenham,

UK, 1996

Rapacki E.J., Hauver G.E., Netherwood P.H., Benck R.F., “Ceramics for

Armors – a Material System Perspective”, 7th TARDEC Ground Vehicle Symp.,

USA, 1996

Ernst, H.-J., Merkel Th., “Zur Vereinheitlichung der Anwendung

ballistischer Faktoren“, ISL RT 519/2000, 2000

Hoog K., Ernst H.-J., Wolf T., “A New Parameter Characterizing the

Ballistic Performance of Ceramics”, EURODYMAT 97, Toledo, Spain, 1997

Bless S.J, Subramanian R., Partom Y., Lynch N., “Effects of Radial

Confinement on the Penetration Resistance of Thick Ceramic Tiles”, 15th ISB,

Jerusalem, Israel, 1995

Ernst H.-J., Wolf T., Unckenbold W., “Protective Power of Thick

Composite Layers Against Medium-Caliber Long-Rod Penetrators”, 19th ISB,

Interlaken, Switzerland, 2001


PRACTICAL ISSUES IN CERAMIC ARMOUR DESIGN

Dr Bryn James

Defence Science and Technology Laboratories

Chobham Lane

Chertsey, Surrey, KT16 0EE

United Kingdom

ABSTRACT

The performance of ceramic armour is heavily dependent upon the

configuration of the system. Generally, compromises must be made in factors

such as single-shot performance in order to obtain the best overall system

performance and to accommodate practical requirements such as multi-hit

capability.

This paper will discuss some of the factors involved in designing practical

armour systems and will use experimental results to illustrate some design

techniques and improvements that may be made. Factors to be addressed include

the physics of stress propagation across an interface, acoustic impedance

matching, optimisation of ceramic tile edge profile, general rules on ceramic

armour design and choice of ceramic material.

INTRODUCTION

Ceramic materials are capable of displaying significantly better protective

performance than an equivalent weight of metal armour. The ability to perform so

well depends partly upon the very strong dependency, inherent in all ceramics, of

the yield stress on ambient pressure (1). It is therefore apparent that, for an armour

system to display the performance intrinsically available from the ceramic

elements, the armour configuration must be such that the ambient pressure is

maintained at the highest levels possible. In addition, the ambient compressive

stress should be highly homogeneous, in order to avoid stress gradients giving rise

to shear and tensile stresses which can lead to early catastrophic failure of the

system.

It is possible to devise an experimental system for mechanically constraining

ceramic materials so that the required conditions are met. Several examples of

these confinement systems are reported in the literature [2, 3, 4, 5, 6]. Such



confinement systems may be capable of maintaining a uniform compressive state

within the ceramic so that the highest levels of strength may be obtained resulting

in complete erosion of the penetrator on the surface of the ceramic tile. This

performance would be of great benefit in an armour system. However, the

confinement system required to maintain such a stress state generally has

significant mass and relies upon accurate impact of the projectile on a

predetermined site. These two factors indicate that the commonly used,

experimental confinement systems are inappropriate for practical ceramic armour.

Instead, to maximise the performance of a practical ceramic armour system, it is

necessary to attempt to mimic the effects of massive confinement by suitable

manipulation of the stresses generated during impact, using the lightest possible

configuration.

The requirements of practical ceramic armour systems generally go beyond

the maximisation of single impact performance. Often, an additional requirement

is that the system should be capable of defeating several impacts within a given

area, i.e. it must have a multi-hit capability.

IMPEDANCE MATCHING

The compressive wave launched from the site of a projectile impact

propagates as a crude approximation to a spherical wave (in the far field). As this

compressive wave impinges upon any interface defined by a change in acoustic

impedance, a number of transmitted and reflected waves are produced. In general,

any interface between two dissimilar materials will give rise to a set of both shear

and longitudinal reflected and transmitted waves, even if there is a very close

match in acoustic impedance.

Figure 1. Stress transformation at an interface


Figure 2. Shear failure of adhesive bond between alumina and aluminium

Metal backed ceramic armour is often constructed using a polymer adhesive.

In such cases, any advantage of close impedance matching of the metal and

ceramic layers is lost due to the inclusion of the low impedance glue layer. The

change of impedance at this layer gives rise to a strong tensile reflection into the

ceramic which acts to shatter the ceramic layer and to provide energy for ejection

of the shattered material. In addition to this, a strong shear wave is set up at the

interface which serves to ‘unzip’ the adhesive interface. An example of such shear

‘unzipping’ can be seen in Figure 2, where concentric bands of shear failure of the

toughened epoxy adhesive layer can be seen.

Very few adhesive materials exist with an acoustic impedance close to that of

metals or ceramics. The class of adhesives with the closest acoustic match are the

high temperature use ceramic adhesives. Typically these may have an impedance

of 6 MRayls compared to 37 MRayls for alumina and 46 MRayls for RHA steel,

reducing the stress reflected back into the ceramic by 33%. However, even this

relative mismatch is considerably better than that for polymeric materials for

which the stress reflected back into the ceramic may be 98%. Unfortunately, such

ceramic adhesives are not as strong as polymer glues and they must therefore be

used as a matching layer in a mechanically bonded system. When such a matching

layer is used, the improvement in performance can be considerable. An example

of the difference in impact performance when a suitable matching layer is

incorporated can be seen in Figure 3.


Figure 3. The effects of acoustic matching

Figure 3A shows a typical armour system coverplate impacted by a 30mm

calibre APFSDS projectile when no matching compound is used. The illustration

at Figure 3B shows the same coverplate when a few grams of acoustic matching

cement are used. Similarly Figure 3C shows the resultant ceramic debris when no

matching compound is used and Figure 3D shows the intact tiles (side view) after

impact when the matching compound is incorporated.

All our experience shows that an adhesive with an enhanced level of acoustic

impedance pays dividends in improving the multi-hit performance, the structural

integrity and, to a lesser extent, the ballistic mass effectiveness of composite

ceramic armour. It remains a technological challenge to produce a high strength

adhesive with such an acoustic impedance.

TILE EDGE GEOMETRY

Typical ceramic armour packs are built from arrays of tiles bonded in some

way to a backing. Due to lack of constraint, and stress wave reflections at the tile

edges, the protective capability for projectile impact at the tile edges is often

reduced. In order to provide a specified minimum level of protection, steps must

be taken to manage this performance reduction by such means as thickening the


tile edges or using thicker tiles. Both of these approaches are costly in terms of

weight or price.

Figure 4. Edge cracking of ceramic tile due to tensile reflections

In principle, if the impact induced stress wave can be propagated efficiently

across the tile boundary, the edge integrity may be maintained for an extended

time during the impact, enhancing the edge performance.

Figure 4 indicates failure induced at the tile edge due to stress waves

propagating away from the impact site being reflected at the edge. Impact closer

to the edge exacerbates this behaviour.

We used a 7.62mm, precisely made, experimental steel projectile

(performance closely matched to 7.62mm APM2) to test the performance of a

number of edge profiles of alumina tiles glued to 8mm aluminium alloy (7017)

plates. The residual energy of the emerging projectile was measured using high

speed photography and a measurement of the residual mass.

A number of different edge profiles were investigated, including profiling the

tile thickness and the tile width. The most successful profile was shown to be a

45 chamfer on the tile edges. Results can be seen in Figure 5, where the triangles

labelled LAT 45 indicate the performance of the 45 edge chamfer system on a

6mm tile. For comparison, the performance of an 8.5mm thick tile with standard

edges can also be seen, showing that, close to the tile edge, the 8.5mm tile

protection is only as good as that in the centre of a 6mm tile. By using chamfered

edges, a weight saving of 9.2 kg m-2

or 30% of the ceramic mass was


demonstrated, for a specified protection level, (7) without resorting to thickened

edge tiles.

Figure 5. Relative protective performance of different edge profiles


Figure 6. Configurations used in edge optimisation evaluation

OPTIMUM CERAMIC/BACKING RATIO

For many years, armour designers have used a general rule that 2/3 of the

mass of a composite ceramic armour system should be invested in the ceramic

front layer, whilst the backing system should consist of 1/3 of the total mass. For

an alumina/aluminium system this corresponds to a ceramic/metal thickness ratio

of 1.5. Whilst this relationship gives a good first estimate for the optimum

configuration for low velocity impact at normal incidence, we have found that it

does not give an adequate estimation of the optimal ratio over a range of velocity

or for oblique impact.

There is little information available in the literature that bears direct

comparison, but analysis of the available data shows a significant correlation for a

wide variety of projectile types impacting an alumina/aluminium system.

Hetherington (8) refers to work by Ali (9) showing an optimum ceramic

thickness, in the ballistic limit configuration, for the defeat of 7.62mm AP rounds

impacting alumina/aluminium systems at ~850 ms-1

. The latter showed

experimentally that maximum ballistic limit velocity, VBL, was obtained for a

ceramic/metal plate thickness ratio, Tcer/Tmet, of 1.5 for normal impact (VBL =

850 ms-1

), reducing to 1.0 for 30 obliquity.

Hohler, Stilp and Weber (10) used a somewhat more complex target structure,

incorporating a thin RHA and rubber front layer. However, their results using an

8.2mm diameter tungsten sinter alloy rod, with an enlarged central section,

impacting at 1500 ms-1

, give optimum thickness ratios, Tcer/Tmet, of

approximately 2.0 at 0 obliquity, 1.25 at 45 obliquity, and 0.82 at 60 obliquity.

The results of Hohler et al. (11) in another study show an optimum

performance for a ceramic/metal plate thickness ratio of 0.71 for alumina on

aluminium at 60 obliquity and at 1450 ms-1

using a 71mm long, L/D=20, tungsten

alloy rod. This value is consistent with that obtained by Hetherington and by

Hohler, Stilp and Weber. The latter similarity is not surprising as the impact

conditions were similar. The similarity in the optimum Tcer/Tmet ratio found by

Hohler et al. with that of Hetherington is, however, quite surprising given the

difference in impact conditions. It can be seen in Table 1, that the optimum

Tcer/Tmet ratio is highly dependent upon the impact conditions. It can be seen

that this ratio changes to 1.7 for Al2O3/RHA at 1450ms-1

and 60 obliquity, to 4.9

for SiC/Al at 1450ms-1

and 60 obliquity and to a ratio of 15.0 for SiC/Al at 2200

ms-1

and 60 obliquity. This ratio is an indicator of the relative performance of the

ceramic and metal fractions. It can be seen that the relative performance of the

ceramic increases with impact velocity and decreases with obliquity.

We can now revise the general rule that the hard ceramic front layer should

contain 2/3 of the system mass whilst the supporting back layer contains 1/3 of


the mass to include the effects of impact and velocity. A more useful

approximation to the optimal thickness ratio has been devised by fitting the,

admittedly sparse, available data to the following simple equation:

angle)Impact90(x000,60

Velocity)optimum(

TT

met

cer (3)

where Velocity is in ms-1

and Impact angle is in degrees.

The applicability of this fit can be seen by reference to the following table:

Table 1. Optimum thickness ratio for alumina/aluminium armour systems

Velocity

ms-1

Impact Angle

degrees

Experimental

Optimum Ratio

Tcer/Tmet

Optimum Ratio

From

Equation 3

Source

850 0 1.5 1.28 Ref. (8)

850 30 1.0 0.85 Ref. (8)

1450 60 0.71 0.73 Ref. (11)

1500 0 2.0 2.26 Ref. (10)

1500 45 1.25 1.13 Ref. (10)

1500 60 0.82 0.75 Ref. (10)

CHOICE OF CERAMIC MATERIAL

The ballistic mass effectiveness of ceramic materials is dependent upon the

armour configuration and the threat projectile. However, if the experiment is well

designed, a reliable general ranking of mass effectiveness may be measured

across a range of threats. We have performed such measurements using 14.5mm

heavy machine gun rounds and 30mm and 40mm APFSDS projectiles. Some of

the materials investigated are detailed in Table 2. An average ballistic mass

effectiveness has been calculated for impact from these projectiles in a number of

configurations, for these materials. Results can be seen in Figure 7. From this

figure it would appear that titanium diboride should be the ceramic of choice.

However, it is no secret that titanium diboride is expensive.

A cost analysis was performed using the best, informed estimate of the price

of ceramic materials in ‘production’ quantities (sufficient supply for 100 generic

light armoured vehicles). It can be seen that the hot pressed non-oxide materials

are significantly more expensive than sintered materials, but where either mass

(Figure 9) or thickness of armour (Figure 10) are critical, use of a non-oxide


ceramic is indicated. It should be noted that this cost estimate depends upon

several factors, the most significant being the use of a given ceramic for non-

armour applications. Armour is generally a relatively small sales area for a

ceramic production company. The cost of production of a material can only be

reduced if large quantities are required for another, non-armour application.

Identification of a bulk-use application for a specific material could result in a

dramatic reduction in the price of a ceramic material for armour.

Table 2. Ceramic materials used in cost/mass/thickness analysis

Name Material type

RHA United Kingdom RHA HV30 = 3.39 Gpa 7840 kg m-3

Alumina 1 Sintered 95% alumina 3680 kg m-3

Alumina 2 Sintered 98% alumina 3780 kg m-3

Novel alumina DSTL developed novel sintered alumina 3690 kg m-3

RB-SiC Reaction bonded silicon carbide 3210 kg m-3

TiB2 Hot pressed titanium diboride 4520 kg m-3

B4C Hot pressed boron carbide 2520 kg m-3

SiC Hot pressed silicon carbide 3230 kg m-3

AlN Hot pressed aluminium nitride 3270 kg m-3

When we calculate the mass of material required to defeat a given threat

(Figure 9), there is surprisingly little variation across a wide range of materials. It

can be seen that we pay a lot more for a small increase in performance. If we

calculate a figure of merit (1/(cost x mass2) (as mass is more important than cost

in our application) for all of these materials (Figure 11), it can be seen that

alumina becomes the most attractive material.

Figure 7. Ballistic Mass Effectiveness


Figure 8. Total cost of material required to defeat a given threat

Figure 9. Mass required to defeat a given threat

Figure 10. Thickness required to defeat a given threat


In Figure 11, we see that “Novel alumina” has the highest figure of merit of

all the materials studied. This material, a sintered alumina with modified

microstructure, was developed within the DSTL Armour Physics Group as part of

a programme to study the effects on performance of changing the microstructure

of alumina. It can be seen that it is possible to make significant improvements to

the performance of alumina to improve its attractiveness as an armour material. It

is believed that yet further improvements are possible.

Figure 11. Figure-of-merit considering cost and mass

ACKNOWLEDGEMENT

The work upon which this analysis is based was funded by the UK

Government Corporate Research Programme.

I would like to thank my colleagues Antony Barker, of DSTL, and Christian

LeGallic, of DGA, France, for their contribution to the experimental work.

REFERENCES

1T. J. Holmquist, G. R. Johnson, W.H. Cook, “A computational constitutive

model for concrete subjected to large strains, high strain rates and high

pressures”, 14th International Symposium on Ballistics. Sept. 1993 2

S.J.Bless, M.Benyami, L.S.Apgar and D.Eylon, “Impenetrable ceramic targets

struck by high velocity tungsten long rods”, Structures under shock and impact

II, Ed. P.S.Bulson, Computational Mechanics Publications, June 1992.3

G.E.Hauver, P.H.Netherwood, R.F.Benck and L.J. Kecskes, “Ballistic

performance of ceramic targets”, Proc. Army Symposium on Solid Mechanics,

Plymouth, Mass. USA, Aug 1993.


4 P. Lundberg, R Renstrom and L. Holmberg, “An experimental investigation of

interface defeat at extended interaction time”, Proc. 19th

International

Symposium on Ballistics, pp. 1463-1470, May 2001. 5 B.James, "The influence of the material properties of alumina on ballistic

performance", 15th International Symposium on Ballistics, May 1995. 6 N.S.Brar, H.D.Espinosa, G.Yuan and P.D.Zavattieri, “Experimental study of

interface defeat in confined ceramic targets”, Proc. APS Topical conference on

shock compression of condensed matter, July 1997. 7

GB Patent Application Number 0026710.4, “Ceramic Tile Armour”,

26th

October 2000, B. James 8 J. G. Hetherington, Two component composite armours, Proc. Light Weight

Armour Systems Symp., Shrivenham, UK, (1995) 9 M. S. B. Ali, “Optimisation of composite armour for normal and oblique

impact”, MSc Thesis, 21 Military Vehicle Technology Course, RMCS,

Shrivenham, UK, (1993) 10

V. Hohler, A. J. Stilp and K. Weber, “Ranking methods of ceramics and

experimental optimization of a laminated target with ceramics”. Proc. Light

Weight Armour Systems Symp., Shrivenham, UK, (1995) 11

V.Hohler, K. Weber, R. Tham, B. James, A.Barker, I. Pickup, “Comparative

analysis of oblique impact on ceramic composite systems”, Proc. Hyper

Velocity Impact Symposium, Nov. 2000


BALLISTIC DEVELOPMENT OF TUNGSTEN CARBIDE CERAMICS

FOR ARMOR APPLICATIONS

Dr Pierre-François Peron

Etablissement Technique de Bourges

Route de Guerry

18021 Bourges Cedex

France

ABSTRACT

In the frame of a cooperative research project agreement, France and the

United States of America are developing and optimising ballistically a new class

of ceramics which offers a very high space effectiveness. These ceramics have a

higher density than armor steel (about 15) and should solve protection weaknesses

on vehicles due to space restrictions. In this paper, the elaboration process and the

mechanical characteristics of these “high density” ceramics are first detailed.

Their ballistic performances against 44 APFSDS medium caliber projectile are

then presented.

INTRODUCTION

A cooperative research project agreement has been signed in December 1996

between the Minister of Defense of the French Republic and the Secretary of

Defense of the United States of America as regards the study of “high density”

ceramic technology for armor applications. The aim was to develop and to

optimise a new class of ceramic which offers a good mass effectiveness and

mainly a high space efficiency. These ceramics should solve vehicle protection

weaknesses related to space restrictions. They have a greater density than Rolled

Homogeneous Armor (RHA) steel and are designed for applications on medium

armor and high armor vehicles.

France and the United States have independently developed high density

ceramic belonging to tungsten carbide ceramics (WC) which density is about 15.

The French WC materials are WC/metal cermets with low metal binder content

while the U.S. WC materials are high purity WC with no binder addition.

During the cooperation, dynamic properties of these two kinds of ceramics are

investigated and their ballistic performances are evaluated against kinetic energy



projectiles and shaped charges to optimise target parameters and to compare the

results with other ceramics. Some material exchanges are also carried out to

enlarge the range of threats.

The French research is conducted by the Etablissement Technique de Bourges

(ETBS), Bourges, France, and the Centre Technique d’Arcueil (CTA), Paris,

France. The U.S. research is conducted at the Weapons and Material Research

Directorate of the U.S. Army Research Laboratory (ARL), Aberdeen Proving

Ground, MD. This paper documents the development of the French ceramics and

provides some ballistic test results against 44 APFSDS kinetic energy threat.

HIGH DENSITY CERAMICS

The ceramic class designated as “high density” includes all the ceramics

which density is greater than that of RHA steel (7.85). A review of potentially

interesting ceramics was undertaken by CTA1 and showed that a large number of

ceramic oxides, nitrides, carbides and borides met the density criteria. Mechanical

properties of some of them are listed in table I.

However, most of them were difficult to process industrially or had prohibited

costs. Tungsten carbide ceramics exhibited high mechanical properties and had a

great deal of applications on the civilian and the military markets. This kind of

ceramic was thus chosen for further investigations.

Table I. Physical and mechanical properties of various high density ceramics.

Ceramic Density

Melting

point

(OC)

Modulus

(GPa)

Hardness

(kg/mm2)

Poisson

ratio

Longitudinal

wave

velocity

(m/s)

MoB 8.77 2600 400 2350 -- 6754

TaC 14.5 4170 285 1800 0.24 4810

TaN 14.3 3093 -- -- -- --

UO2 11 3140 193 6800 0.3 4450

WB 16 2900 -- -- -- --

WC 15.8 3050 660 2700 0.22 6900

WC-Co 14.95 1450 645 2600 0.22 6900


SANDVIK TUNGSTEN CARBIDE CERMETS

The French material is a tungsten carbide ceramic with a low content of

cobalt. The cobalt is used as a binder to process WC cermet at lower pressure and

temperature as compared to pure WC ceramic. During the cermet elaboration, the

binder is added as a liquid phase to the WC powder. The agglomerated powder is

then densified by iso-static compression and sintered between 1350°C and

1500°C.

The French WC cermets are produced by Sandvik Hard Material society

located in Epinouze, France. They are available in 200 mm square or 250 mm

diameter cylinder tiles and in thickness up to 40 mm.

Simulations conducted with the analytical code BREFIL2 showed that the best

ballistic performances should be reached with WC-Co cermets which had less

than 10 % in weight of binder and a fine microstructure. Three cermets were thus

chosen for ballisitic evaluations : H3F, H5F and H10F. Their physical and quasi-

static properties are provided in table II. These cermets have a cobalt content

between 3 % and 10 % in weight and a very fine grain size (less than 1 m). The

addition of cobalt entails a decrease of the ceramic hardness and of its

compressive strength but increases its flexural strength.

Table II. Physical and mechanical properties of Sandvik and Cercom WC ceramic.

H3F H5F H10FWC

Cercom

WC content

(% weight) 97 92.5 89.5

96.8% WC

2.8% W2C

Co / other contents

(% weight) 3 / 0 4.5 / 3 10 / 0.5 0 / 0.4

Grain size

Average size Extra fine

Extra fine

0.5 µm Extra fine

Extra fine

0.9 µm

Theoretical density

Measured density

15.3 15.

14.95

14.5 15.7

15.6

Melting Point (°C) 1450 2800

Hardness (HV30) 1925 1810 1600 2200

Flexural Strength (MPa) 2570 2640 3960 1100

Tenacity (MPa m) 7 9 13 7

Young modulus (GPa) 670 645 580 690

Poisson ratio 0.21 0.21 0.22 0.20

Sound speed (m/s)

Longitudinal

Transversal

6970

4222

6858

4300


CERCOM TUNGSTEN CARBIDE CERAMIC

The U.S. material is a pure tungsten carbide ceramic. Its elaboration process

has been developed by the Cercom incorporated of Vista, CA in USA and enables

to produce high purity WC ceramic with no binder. The Cercom WC ceramic

contains 96.8% WC and 2.8% W2C in weight and has a density of 15.6. Its

physical and mechanical properties are listed in the table II and compared to the

Sandwik cermets ones.

TEST CONFIGURATION

French ballistic evaluations of the Sandvik and Cercom materials were carried

out with the 44 APFSDS kinetic energy projectile produced by GIAT Industries

(Figure I).

Figure I. 44 APFSDS kinetic energy projectile.

233

The 44 APFSDS is a L/D 25

tungsten alloy rod laboratory

penetrator (Table III) which

represents the 105 APFSDS

projectile at third scale. Its has a

9.3 mm equivalent diameter and

weights 0.257 kg. Its baseline

penetration into RHA steel is 160

mm for an impact velocity of

1500 m/s.

Table III. 44 APFSDS physical and

mechanical characteristics.

44 APFSDS

Composition

(weight %) 93W-4,6Ni-2,4Fe

Density 17,6

Hardness 423 Hv30

Yield stress 980 MPa

Ultimate stress 1150 MPa

Elongation 9 %

It is nominally fired at 1500 m/s

from a 44 mm bore diameter gun

based on a 40 mm L 70 Bofors tube

and fitted with a 105 mm HM2 breech.

For the tests, the gun was put at 61 m from the target. The impact velocity was

measured by two optical barriers. The total yaw angle was calculated from the

trace of the fins in paraffin cardboard put at regular intervals close to the target.

The tests with a yaw impact superior to 1° were disregarded.


The ballistic tests were

conducted by using the Depth of

Penetration (DOP) technique

described in figure II. This

technique compares the

performances of a RHA steel semi-

infinite target to the residual

penetration obtained in a RHA

steel block put behind a ceramic

tile. More details on the test

configuration are given in the next

section.

RHA st

f

Figure II. Target configuration

in the DOP test.

Steel lateral

confinementRHA

WC ceramic

Tcer Pres

eel

ront confinement

The ballistic performances of the ceramic are calculated from the residual

penetration of the projectile in the RHA steel back-up and are represented by an

equivalent thickness (Eeq), an equivalent mass (Meq) and a quality factor (q2) :

where PRHA is the projectile penetration in a RHA steel

semi infinite target,cer

resRHA

eqT

PPE

cer

RHAeqeq EM

eqeq

2 EMq

Pres is the residual penetration of the projectile in

the RHA steel block behind the ceramic tile,

Tcer is the ceramic thickness,

RHA and cer are respectively RHA steel and

ceramic densities.

In the case of a RHA steel plate confinement put in front of the ceramic, the

thickness of this plate is added to the residual penetration (Pres).

The reference penetration is that obtained with the 44 APFSDS at 1500 m/s in

a semi-infinite RHA steel target. It corresponds to an equivalent thickness and an

equivalent mass equal to 1. Higher values indicate that the tested material has

better ballistic performances than RHA steel.

WC SANDVIK BALLISTIC PERFORMANCES : FIRST EVALUATIONS.

A first evaluation was carried out with the three Sandvik cermets in order to

determine the influence of the cobalt content on the ballistic performances.

The tests were conducted with cylindrical cermets tiles of 240 mm diameter and

30 mm thickness. The target was composed of a ceramic module and a RHA

semi-infinite back-up (Figure III). The module was constituted of two ceramic

tiles put between two RHA steel plates of 10 mm thickness and confined laterally

by 10 mm of mild steel. All the material surfaces in contact with the ceramic were

grounded to obtain a perfect contact between the steel plates and the tiles.


Figure III. Test configuration for the ballistic evaluation of Sandvik cermets

against 44 APFSDS.

10 10 40

RHA

RHA

Mild steel

PVC

WC ceramic

44 APFSDS

1500 m/s

The results of the ballistic evaluations are provided in the table IV and are

compared to those obtained with Al2O3 and SiC ceramics.

Table IV. Ballistic performances of several ceramics against 44 APFSDS.

Ceramic

diameter

(mm)

Ceramic

thickness

(mm)

Ceramic

surfacic

density

(kg/m2)

Impact

velocity

(m/s)

Residual

penetration

(mm)

Meq Eeq q2

RHA steel 0 0 1500 160 1.00 1.00 1.00

H3F ( 240) 2 x 30.5 933.3 1478 80.2 0.67 1.31 0.88

H5F ( 240)2 x 29.8

2 x 29.8

891

891

1503

1503

70.1

61.0

0.79

0.87

1.51

1.66

1.19

1.45

H10F ( 240) 2 x 30.3 877.2 1488 69.2 0.79 1.50 1.22

Al2O3 (94%) 50 181 1469 115.5 1.80 0.89 1.6

SiC 49.9 157.2 1469 119.2 2.04 0.81 1.66

This first evaluation performed with H3F, H5F and H10F proved that a WC

cermet with cobalt content of 5 % in weight has the most interesting ballistic

performances. The equivalent thickness is always superior to 1, which shows the

interest of these cermets in term of space. Their equivalent mass is quite lower

than 1. But the thickness of the cermet was not optimised and better performances

should be obtained with finer tiles.

Besides, because of their high density, the WC cermets ballistic performances

are specific and opposed to SiC and Al2O3 ones which are interesting in term of

equivalent mass.


FURTHER EVALUATIONS OF SANDVIK AND CERCOM WC CERAMICS

PERFORMANCES AGAINST 44 APFSDS.

The influence of the elaboration process was also studied by comparing the

ballistic performances of H5F and WC Cercom ceramics. The test configuration

was the same as before. The results of the evaluation are listed in the table V.

Table V. Ballistic performances of H5F and WC Cercom ceramics

against 44 APFSDS.

Ceramic

diameter

(mm)

Ceramic

thickness

(mm)

Ceramic

surfacic

density

(kg/m2)

Impact

velocity

(m/s)

Residual

penetration

(mm)

Meq Eeq q2

RHA steel 0 0 1500 160 1.00 1.00 1.00

H5F 250

(cover plate)

2 x 30

2 x 30

897

897

1491

1502

72.9

76.6

0.76

0.73

1.45

1.39

1.1

1.01

Cercom 200

(cover plate)

2 x 30.1

30.2+30.1

939

941

1495

1502

38.9

57.2

1.01

0.86

2.02

1.71

2.04

1.46

These tests confirmed the ballistic performance level of H5F. The WC

Cercom ceramic performances were scattered but were higher than H5F ones.

Besides, the observation of the targets after shot (Figure IV) gave us some

explanations on the performance differences between Sandvik and Cercom

ceramics and on the WC ceramic dynamic behaviour during the penetration of the

projectile.

For firing tests on H5F cermets, the crater in the cover plate had a diameter

slightly superior to the projectile caliber. The cermet was damaged in the area of

interaction with the projectile and seemed to have been slightly affected

elsewhere. For shots on WC Cercom ceramics, the hole in the front plate was far

wider than the projectile diameter and the ceramic was damaged in a large area

around the projectile penetration zone.

According to these observations, the H5F cermet seems to exhibit limited

resistance to the projectile penetration. The binder agglomerates the WC grains

and enables to get good quasi-static mechanical properties. But at high strain

rates, it may be a weak area in comparison to WC grains and thus a privileged

way for the damage propagation in the cermet.

For Cercom ceramic, WC grains are intimately linked between each other and

exhibit thus a higher resistance to the projectile penetration. In these conditions, a

larger area of the ceramic takes part in the projectile erosion. This leads to a wide

damaged area but also an ejection of the affected ceramic through the cover plate.


Figure IV. Interaction of 44 APFSDS with H5F Sandvik and WC Cercom ceramics.

Crater in the cover plate and damage caused to the ceramic.

Cover plate

WC ceramic damage

No Image.

Pres 72.9 mm Pres 38.9 mm Pres 57.2 mm

H5F WC Cercom WC Cercom

CONCLUSION

In the frame of a cooperative research program, France and the United States

are developing and optimising a new class of ceramic which have a high space

effectiveness. These materials are tungsten carbide ceramics and have a density

superior to that of RHA steel. They are designed for applications on medium

armor and high armor vehicles in areas where space restrictions are present. The

French materials are WC/metal cermets with low metal binder content while the

U.S. are high purity WC with no binder addition.

In France, these ceramics were evaluated in 60 mm thickness against

44 APFSDS projectile. The shots showed that both ceramics exhibit high space

effectiveness. WC Cercom and H5F ceramic performed an equivalent thickness

superior to 1.5 and an equivalent mass slightly inferior to 1. Higher ballistic

effectiveness should be obtained by reducing the ceramic thickness.

REFERENCES1 C. Cottenot, “State of art and evaluation of high density ceramics as armor

materials”, ETCA 93 R 153 (1993). 2 S. Fouquet, “BREFIL : an analytical model for the interaction between a

kinetic energy projectile and a brittle material”, ETCA 88 R 042 (1988).


BALLISTIC DEVELOPMENT OF U.S. HIGH DENSITY TUNGSTEN CARBIDE CERAMICS

William A. Gooch and Matthew S. Burkins U.S. Army Research Laboratory Weapons and Materials Research Directorate Aberdeen Proving Ground, MD 21005-5066, USA

Richard PalickaCercom Incorporated, 1960 Watson Way Vista, CA, 92083, U.S.A.

ABSTRACT The United States and France, under a cooperative research agreement have

developed a new class of high density ceramics which inherently provide high space efficiency and reduced susceptibility to damage accumulation effects in thick sections. While many ceramics were considered, this research has focused on tungsten carbide based ceramics. The U.S. Army Research Laboratory, in cooperation with Cercom Inc. has developed a hot-pressed tungsten carbide ceramic for ballistic applications. This paper will present a survey of high density ceramics, document the mechanical and elastic properties of the U.S. WC ceramic and baseline the ballistic performance.

INTRODUCTIONIn December 1995, a cooperative Project Agreement under the Memorandum

of Understanding between the Secretary of Defense of the United States of America and the Minister of Defense of the French Republic concerning Technology Research and Development Projects was signed to jointly develop and optimize a new class of ballistic ceramic materials that offer very high space effectiveness for applications where inherent space restrictions are present. These ceramics are defined as any ceramic with a density greater than that of rolled homogeneous armor (RHA) steel (7.85 g/cm3). The U.S. research is being conducted at the Weapons and Materials Research Directorate of the U.S. Army Research Laboratory (ARL), Aberdeen Proving Ground, MD and the French research is being conducted by the Établissement Technique de Bourges, Bourges, France and the Centre Technique d.Arcueil, Paris, France. This paper documents



the development of the U.S. ceramic as well as providing limited ballistic testing of the ceramic.

HIGH DENSITY CERAMICS While a number of ceramic oxides, nitrides and carbides meet the criteria of a

high density ceramic, most of the ceramics are difficult to process or the costs are prohibitive. A review of possible ceramics of interest was undertaken by Cercom [1] and Table I lists important properties for nominally pure ceramics whose densities are greater than 7.85 g/cm3. Of the twelve ceramics listed, only two have been tested ballistically. Limited testing of hafnium carbide (HfC) was conducted by Hauver [2] in 1/5 scale tests to study dwell. Rupert et al [3,4] with Nuclear Metals Incorporated of Concord, MA examined uranium oxide (UO2)ballistically in both sintered and hot isostatic pressed conditions. The work was successful in producing near-theoretical density UO2 in 10-mm disks of thickness of 11.3-mm, but further work was discontinued because of the associated radiation limitations imposed by these uranium-based materials. Limited testing of high metal content WC cermets were also conducted at ARL with 14.5-mm armor-piercing projectiles, but the performance was generally equal to equivalent areal weights of RHA.

The data in Table I, while compiled from laboratory test data, established the direction for further development. The tungsten carbide (WC) family was selected as the prime ceramic of interest because of the high density, excellent mechanical properties and the potential unique applications in both the military and civilian markets. The WC family had both the highest density and modulus and exhibited a longitudinal sound speed about half that of lower density ceramics, resulting in an impedance about 2.5 times that for RHA.

Table I. Elastic Properties/Melting Temperature of Selected High Density Ceramics

CERAMIC DENSITY (g/cm3)

MELTING POINT

(°C)

MODULUS (GPa)

HARDNESS(kg/mm2)

LONGITUDINAL WAVE VELOCITY

(m/s) MoB 8.77 2600 400 2350 6754Mo2C 9.18 2522 533 1499 7620NbN 8.31 2300 483 1525 7626TaB 14.19 3090 400 3130 5309TaC 14.40 3985 285 1720 4449TaN 14.36 3087 576 -- 6333HfB2 11.19 3380 500 2900 6685HfC 12.67 3890 360 3830 5330HfN 13.39 3000 500 1600 6112UO2 10.97 2850 -- -- -- WC 15.7 2800 696 2200 6600W2C 17.20 2785 420 2150 4940


U.S. TUNGSTEN CARBIDE CERAMICSTungsten and carbon form two ceramics of interest, tungsten monocarbide

(WC) and ditungsten carbide (W2C) as seen in the two narrow phase stabilityregions at 50 and 30 atom % carbon, respectively, in the binary phase diagram ofFig.1[6]. Both ceramics have a melting temperature of about 2800°C and WC hasa very high elastic modulus. Currentlyproduced WC materials are, in fact,cermets, alloys of ceramics and metalbinders that are sintered to form a hard dense material. These cermets containeight to ten percent cobalt by weight,added as a liquid-phase sintering aid toallow the material to be fully densified atlower temperatures and pressures ascompared to binderless WC. The cobaltaddition reduces yield strength andhardness, but increases toughness. WCcermets of these high metal content exhibitreduced ballistic performance, as the resistance to penetration is governed bythe percentage of metal content andlocation of the sintering aid in themicrostructure. The specific materialproperties and structure that make WCcermets valuable industrial materialsinherently degrade their performance asballistic materials.

Figure 1. W-C Phase Diagram

The U.S. WC ceramic processing technology was developed by CercomIncorporated of Vista, CA and the physical, mechanical and elastic properties ofthese ceramics provide a ballistic response similar to high quality ballistic ceramics of lower density. Cercom developed the processing technology to densify large, high purity ceramics without metal sintering aids, and adapted the process to the densification of WC [5]. Cercom first densified tiles of 100-mm x 100-mm size in thicknesses up to 25.4-mm. The process was scaled-up to 152-mm x 152-mm tile sizes in thickness up to 50-mm and 203-mm diameter tileswere produced in thicknesses up to 30-mm. Ditungsten carbide (W2C) was also densified in 152-mm x 152-mm size tiles of 26-mm thickness, but will not bediscussed in this paper. The French ceramic is a low-metal binder WC that wasdiscussed in the preceding paper.


The hot-pressed Cercom WC had a density of 15.6 g/cm3 and the tiles were analyzed to be composed of WC and 2.8% W2C, the latter a byproduct of the densification process. The nominal purity was 99.6% WC/W2C. The tiles were densified without metal sintering aids allowing near-theoretical density tiles to be hot-pressed in large tile sizes. The grain size was between 0.3-1.4µm with an average grain size of 0.9µm. The crystal structure of WC and W2C is hexagonal and matches other higher performing hexagonal ballistic ceramics such as -SiC,

-TiB2, AlN, and -Al2O3. The measured quasi-static mechanical and elastic properties of the Cercom WC are provided in Table II and are compared to hot-pressed silicon carbide densified by the same process.

Table II. Measured Mechanical and Elastic Properties of Cercom WC and SiC

WC SiC-B

DENSITY (g/cm3)THEORETICAL

AS-PRESSED 15.715.6

3.223.20

Average Grain Size (µm) 0.9 4.0

HARDNESS (VICKERS-1-kg) (kg/mm2) 2200 ± 20 2700

FLEXURAL STRENGTH (MPa) WEIBULL MODULUS

1100 ± 130 10.2

65518

KIC TOUHGNESS (MPam1/2)SINGLE ETCH NOTCHED BEAM

INDENTATION VICKERS 7.56 ± 0.51 6.86 ± 0.19

5.2--

TENSILE STRENGTH (MPa) 589 ± 57 592

ELASTIC MODULUS (GPa) 690.1 455

SHEAR MODULUS (Gpa) 287 195

POISSON RATIO 0.20 0.14

SONIC VELOCITY (km/s) LONGITUDINAL

TRANSVERSE 6.8584300

12.257.65

When compared to traditional low-density ceramics, the compactness of the ceramic is a direct function of the inherent densities of materials. Relative to steel, WC is one-half the thickness for the same areal density; one-fifth that of silicon carbide ceramic. The remainder of this paper will document the ballistic performance of WC for a representative tungsten long-rod penetrator.


TEST PROJECTILEThe 162-g tungsten

projectile, shown in Figure 2, has a Length to Diameter(L/D) of 13 and has beenused for many years as a testsimulant for a mediumcaliber long-rod projectile.Table III lists thecomposition and typicalmechanical property data onthis penetrator. The baseline RHA penetration of this rodhas been documented byGooch et al [7] and is governed by the followingtwo parameter exponentialequation where P is in mm and V is in km/s:

Figure 2. L/D 13 Tungsten Rod

Table III. Mechanical Properties for L/D13 Rod

Designation L/D 13 W A Rods

Composition (wt %) 93W-3Ni-2Fe

Density (g/cm3) 17.7

Hardness 40-45 RC

Yield Strength 1200 MPa

Ultimate Tensile Strength 1280 MPa

Elongation 8 %

2)/447.1(8.308P Ve

This equation, based on work of Lanz and Odermat [8] and improved by Goochet al [7], is accurate between 500-1800 m/s. The L/D13 rod, when nominally firedat 1550 m/s, has a baseline penetration into RHA of 129.2-mm.

EXPERIMENTAL SETUP The test penetrator was fired from a laboratory gun consisting of a 40-mm

L70 Bofors breech assembly with a 38-mm smoothbore barrel. A custom-builtpolypropylene sabot system was used to launch the projectiles. The gun waspositioned 1.5 meters in front of the targets and an orthogonal flash radiographicsystem [9] was used to measure projectile velocity, pitch, and yaw. Propellantweight was adjusted to achieve the desired striking velocity and ballistic resultswith 2° total yaw were disregarded.

The ballistic test data presented in this paper were conducted using the depthof penetration (DOP) technique developed by Woolsey et al [10] and shown inFig. 3. This technique compares the performance of a RHA baseline (PO) to theresidual penetration (PR) of ballistic tests of different thicknesses of ceramic. All tests are shot at 0° obliquity and the target had no cover plate, but was confinedlaterally by a steel frame in which the ceramic tile was epoxied. The DOP into the rear RHA plate (PR) was measured for each impact. Burkins and Gooch [11],


when examining the sources of variancein DOP data, determined that bond thickness is a major source of variabilityin ballistic data. This observation hasled to a modification in the assemblyprocedures for DOP testing at ARL. Thebond thickness is maintained at 0.5-mm for both the side confinement and the rear interface. The DOP technique iscost-effective, but only provides arelative performance indication; time-dependent effects, reported by Hauver etal[12] predominate in this methodology where no cover plate is used, interfaceeffects are not minimized, and ceramic confinement is limited. Generally, the result of these factors are a rapid reduction in the relative performance as ceramicthickness increases, accompanied with an increase in the scatter of the ballisticdata. While these factors are present in DOP testing of the lower-densitytechnical ceramics, particularly for thicker tiles, high density ceramics appear to have less scatter. The optimum ceramic thickness for maximum ballisticerformance in simple laminate target designs, such as the DOP configuration, has been observed to be between 25 to 40-mm for the low densityceramics and thicker tiles begin to loose performance as time dependentfactors predominate. This loss primarily relates to geometric considerations ofthe ceramic tile to the confinement, the sound speed of the ceramics and time forreflection from the ceramic back and side interfaces. The material has failed andmass and space effectiveness rapidly decrease as the penetrator encounters,essentially, granulated ceramic. As the RHA penetration performance of a rodincreases, this ceramic design problem becomes greater for the armor designer,as the residual penetration becomes larger as the ceramic gets thicker. As withmany armor designs, the target is driven by the space factor, not the massfactor. While Hauver et al [13] and Prifti et al [14] were successful in overcomingsome time dependent effects in ceramics, the total mass and/or space effectivenessvalues are still low if the total parasitic mass or space is included in theeffectiveness calculations. The problem resides with the low density of theceramics for these high performance applications in simple armor designs.

Figure 3. Depth of PenetrationTest Configuration

BALLISTIC CHARACTERIZATIONBallistic performance of armors or elements of armors are characterized by

dimensionless factors which compare the areal density (mass/area) and thicknessof the material to baseline RHA. Many variations and terminologies exist, butFrank [15] developed and described a concise set of mass and space effectiveness


factors whose conventions are in use at ARL (Figure 4). Since the DOP technique determines the equivalent RHA performance of the ceramic relative to the semi-infinite penetration of the rod, the ballistic characterization of the ceramic can be defined by the mass effectiveness (em), the space effectiveness (es) and the armorquality factor (q2) as described by the equations below; the small e indicating thatthe performance indices are elemental rather than system effectiveness. The termPRHA represents how much baseline RHA penetration was removed by ceramicthickness, TCER, at the same impact velocity and is obtained by subtracting theresidual RHA penetration depth (PR) from the baseline RHA penetration of the rod (PO), i.e., (PO - PR). The ceramic mass effectiveness can then be related to eS

by the ceramic density ( CER) and the RHA density ( RHA). RHA has an em and es

of 1.0 and higher indices indicate better ballistic performance. The quality factorhas significance for armor designers as this factor relates both the mass and spacefactors; values over 1.0 indicate armors or materials which are thinner and/or lighter than the baseline RHA performance and indicate superior armors or materials.

sm

RHA

CERAMIC

m

CERAMIC

RHAs

CERAMICCERAMIC

RHARHAm exeqxe

T

Pe

xT

xPe 2

Figure 4. Mass and Space Efficiency Parameters

EXPERIMENTAL RESULTS/DISCUSSIONTable IV documents 12 DOP tests for three thicknesses of WC tiles that were

152-mm X 152-mm in lateral size. The effectiveness factors for each test havebeen determined from the impact velocity. These 12 tests represent a consistentset of data and demonstrate a number of interesting observations. First, theballistic performance for the 10-mm and 20-mm WC tiles was very similar withthe mass effectiveness near 1.55 and the space effectiveness near 3.0. The 20-mmtiles were slightly better performers. The 30-mm WC tiles demonstratedsignificantly better performance than would be predicted from the thinner tilesand an optimum thickness may not yet have been reached. However, the residualpenetration is approaching zero and a higher performing rod will have to beutilized to determine the optimum tile thickness for WC. For the 30-mm tiles, the mass effectiveness is over two and the space effectiveness is over four. The highspace factors in these tests mean that the rod is being stopped in a ceramicthickness that is about one-fourth the penetrator length.


Table IV. L/D 13 DOP Ballistic Test Data for WC

SIZE (mm)

CERAMIC THICKNESS

(mm)

CERAMIC AREAL

DENSITY (kg/m2)

IMPACT VELOCITY

(m/s)

AVERAGE PR (mm)

em es q2

RHA 0 0 1550 129.2 1.00 1.00 1.00

152

10.310.310.210.2

160.1160.1158.1159.3

1531154315521554

95.097.099.099.0

1.531.541.501.52

3.053.022.993.01

4.684.654.494.57

152

20.120.220.120.2

314.2314.6314.2315.0

1541154515541561

62.062.069.071.0

1.651.651.521.49

3.283.293.022.96

5.415.444.614.40

152

30.230.230.030.1

471.5470.7467.6470.3

1543154315431549

7.05.87.7

11.4

2.022.042.031.97

4.014.054.033.91

8.108.268.177.68

CONCLUSIONS This paper documents the U.S. development of a new class of ballistic

ceramics known as high density ceramics under a joint U.S/France cooperative research program. These ceramics have been defined to be any ceramic whose density is greater than that of rolled homogeneous armor steel (7.85 g/cm3) and are very interesting from an armor standpoint, as very compact targets are possible. The WC family of high density ceramics was selected as the prime ceramic of interest because of the high density, excellent mechanical properties, and the potential applications in both the military and civilian markets. The tungsten carbide ceramic densification technology was developed by Cercom Incorporated of Vista, CA, who succeeded in producing high purity WC ceramics in large tile sizes. Ballistic testing with a L/D13 tungsten rod demonstrated very high space effectiveness factors.

With increasing battlefield threats, current and future combat vehicles will require armor technologies which obtain maximum protection with compact structures and armors. The development of this ceramic provides armored system developers with a very space efficient material for use against higher-performing medium caliber and full-scale rods in applications such as add-on appliques, roof appliques, hatch appliques or hull and turret side armors.


REFERENCES 1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

S. Schneider, Engineered Materials Handbook, Vol.4, Ceramics and Glasses, American Society for Materials International, 1991

G. Hauver, Private Communication N. Rupert, R. Schoon, .Evaluation of High Density Ceramics for Ballistic

Applications., Conference on Dynamic Loading in Manufacturing and Service, Melbourne, Australia, 1993

N. Rupert, M. Burkins, W. Gooch, M. Walz, N. Levoy, E. Washchilla, Development of High Density Ceramic Composites for Ballistic Application., Inter. Conference on Advanced Composite Materials, Wollengong, Australia, 1993

Cercom Patents 5,302,561, 5,358,685, 5,354,536, Monolithic, Fully Dense SiC Material and End Uses.

E. Rudy, AFML-TR-65-2, Part IV, Compendium of Phase Diagram Data, pg. 192, Air Force Materials Laboratory, Wright-Patterson Air Force Base, June 1969

W. Gooch, M. Burkins, K. Frank, .Ballistic Performance of Titanium against Laboratory Penetrators., 1ST Australasian Congress on Applied Mechanics, Melbourne, Australia, 1996

W. Lanz and W. Odermat, .Penetration Limits of Conventional Large Caliber Antitank Guns/Kinetic Energy Projectiles., Proc. 13th Inter. Symposium on Ballistics, Stockholm, Sweden, 1992

C. Grabarek and L. Herr, .X-Ray Multi-Flash System for Measurement of Projectile Performance at the Target., U.S. Army Ballistic Research Laboratory Technical Note 1634, September 1966

P. Woolsey, S. Mariano, and D. Kokidko, .Alternate Test Methodology for Ballistic Performance Ranking of Armor Ceramics., 5th Annual U.S. Army Tank-Automorive Command Survivability Conference, Monterey, CA, 1989

M. Burkins and W. Gooch, .Ceramic Testing Methodology., U.S. Army Research Laboratory Workshop, June, 1995

G. Hauver, W. Gooch, P. Netherwood, R. Benck, W. Perciballi and M. Burkins, Variations of Target Resistance During Long-rod Penetration into Ceramics., Proc. 13th Inter. Symposium on Ballistics, Stockholm, Sweden, 1992

G. Hauver, P. Netherwood, R. Benck, .Ballistic Performance of Ceramic Targets., 13th Army Symposium on Solid Mechanics, Plymouth, MA, 1993

J. Prifti, P. Woolsey, W. Gooch and W. Perciballi, .Advanced Ceramic/Metallic Armor Systems for Defeat of Long Rod Penetrators., Second Ballistic Symposium on Classified Topics, Johns Hopkins University, 1993

K. Frank, .Armor-Penetrator Performance Measures., Armament Research and Development Command /Ballistic Research Laboratory Report MR-03097, 1981



INITIAL TESTS ON CERAMICS IN COMPOSITE ARMOR

W. Lanz

RUAG Land Systems (formerly Swiss Ordnance Enterprise)

Allmendstrasse 86

CH-3602 Thun, Switzerland

ABSTRACT

The intended development of the new Swiss Main Battle Tank at the

beginning of the Seventies instigated a major move in terminal ballistics

research. The existing homogeneous armor steel was at its limits; therefore,

new armor materials had to be selected to ensure tank crew protection.

Moreover, the influence of different geometries on terminal ballistics was

investigated. In this research, ceramics were included as candidate materials to

protect against both shaped charges and kinetic energy projectiles. Mainly,

model tests served to select a suitable material or suitable material

combinations. The first candidate ceramic armor material was alumina,

followed by porous silicon nitride (Si3N4) and silicon carbide (SiC). The SiC

composite armor was also tested in full scale; although it did not perform as

well as in the model tests, it still demonstrated a very high stopping power.

INTRODUCTION

The Near East wars of 1967 and 1973 clearly

showed how vulnerable even heavy tanks had

become. Especially, the massed deployment of

guided long range antitank missiles had a

devastating effect (Figure 1).

The frontal armor of heavy tanks of that period

was homogeneous steel of about 250-mm

thickness, measured in the horizontal plane. On

the other hand, the opposing shaped charges had a

penetration performance of 500-mm or more

(Figure 2). Even the KE (kinetic energy)

projectiles fired from tanks penetrated around 300-

mm RHA (rolled homogeneous armor).

Figure 1. Effect of an 80-mm Shaped

Charge Impact on a 40-tonne Tank


Any further thickness

increase of the conventional

armor would have led to

prohibitive vehicle weights and

would have been virtually

ineffective in view of the shaped

charge penetration power. In

order to protect the tank crews

from this threat a totally different

approach had to be chosen.Figure. 2. Before 1980: Antitank Projecti

Armor Protection of Heavy Tanks

EARLY ATTEMPTS

Of course, the problem described above was well known to all related

development institutions which carried out extensive research on projectile/target

interaction. The main objective was to design cost effective armor protection with

significantly higher stopping power than steel. The primary purpose was the

protection against the high performance shaped charge, but also against KE

rounds as well. These cause less penetration, but have a far higher energy content.

Basic analytic and experimental terminal ballistics investigations had been

described in 1, 2, 3 and other unclassified literature. Its principal statement was

the hydrodynamic approximation formula for a shaped charge jet penetration

behavior:

t

p

L

P

P = Penetration Depth p = Projectile Material Density

L = Projectile Length t = Target Material Density

The approximate formula contains no material strength values, since the

dynamic pressures at very high velocity impacts are an order of magnitude higher

than the material strengths. This hydrodynamic approach also forms the basis for

the well known "Odermatt Formula" for KE projectiles 4 this formula includes

the influence of material properties since these cannot be neglected after all.

Figure. 2. Before 1980: Antitank Projectiles vs.

Armor Protection of Heavy Tanks


FIRST TERMINAL BALLISTICS INVESTIGATIONS AT RUAG LAND

SYSTEMS

At the beginning of the Seventies, the Swiss Defense Procurement Group

initiated the development of an indigenous main battle tank. This also entailed a

terminal ballistics research program at RUAG Land Systems which concentrated

on the evaluation of new armor protection materials and on the influence of

geometry on terminal ballistics.

When this program was initiated, sufficiently accurate terminal ballistics

calculations were not believed to be feasible. Thus, an experimental approach was

chosen. Our main model ammunitions are shown in Fig. 3.

1972

Cal. 40 mm Shaped Charges, ca. 90 g of HE

Pentastite P = 180 mm RHA

Octastite P = 225 mm RHA

1975APDS

Ø20 90 mm Tungsten, mP = 300 g

v 1300 m/s

P 90 mm RHA

1978

APDSFS

v 1500 m/s

2 Types (Tungsten):

Ø9 130 mm, mP 150 g, P 130 mm RHA

Ø8 170 mm, mP 150 g, P 165 mm RHA

at 100 mm

stand off

Figure 3. RUAG Model Ammunitions

NEW ARMOR MATERIALS EVALUATION

A closer look at the hydrodynamic penetration formula readily reveals a

feasible way to enhance armor stopping power without increasing weight, i.e.,

using lower density material. However, a drawback is the correspondingly higher

thickness required. This is shown in Figure 4 as the penetration depth PCM of a

given projectile type versus the density ratio of candidate material and steel


CM/ St for different composite materials. The penetration PSt of the same

projectile in RHA steel of 825 MPa tensile strength serves as a reference value.

As defined in the Space Equivalence Factor,

St

CM

CM

StS

P

PF

describes the protection performance of a candidate material relative to RHA

steel. For practical reasons, application of the Mass Equivalence Factor is

preferred,

CM

StSM FF

which directly indicates the relative mass reduction when replacing steel with a

lighter material for the same protection against the same projectile. Equating the

above terms yields (for the hydrodynamic approximation):

S

MF

F1

Figure 4 shows FS values versus the density ratio CM/ St. The candidate materials

are:

CM g/cm3

titanium 4.50

alumina 3.92

aluminum 2.80

glass fiber reinforced plastic (GFRP) 1.92

magnesium 1.75

polyvinyl chloride (PVC) 1.10

The round dots in Figure 4 represent the theoretical FM values for the specific

materials. In accordance with the hydrodynamic theory, PVC, as the lightest

material, yields the best mass equivalence protection factor.


Figure 4. Space Equivalence Factor FS vs CM/ St

FIRST EXPERIENCES WITH AL2O3 CERAMICS

According to theory, the only ceramic material presented here is, obviously,

not very attractive. However, the experimental results obtained by firing at the

candidate materials with high precision shaped charges are more interesting, as

seen by the data notated by the triangles in Figure 4. Here the material strength

comes into play:

Due to the low strength of PVC, the measured value practically coincides with

the theoretical

Titanium reaches the same space equivalence, FS, as steel and a mass

equivalence FM = 1.744

GFRP has an astonishingly high FM = 3.3

Ceramics (alumina) need less space than steel with FS = 1.1 and at FM = 2.2, a

mass less than half the steel block


This positive result becomes even better when using a reduced thickness. In

the described tests alumina blocks of 150-mm thickness were used. With blocks

of 40-mm thickness the mass equivalence mounts to FM = 4 and the space

equivalence to FS = 2. This proves that alumina is an ideal protection material

against shaped charges, being lightweight and needing little space. The reason for

this thickness-dependence is already described in 1 . It should be noted however

that the alumina blocks were housed in confinements.

The results of these investigations allow the conception of composite armor to

oppose shaped charges with values of FS 1 and FM 3 (Figure 5). Indeed, the

performance of such a target composition was confirmed in live 40-mm shaped

charge firing, the charge having a reference penetration of 180-mm RHA. For

correctness, the confinement mass of the side walls was neglected for the FM

calculation.

Next, the negative aspect of ceramic materials, namely their brittleness, had to

be dealt with. The area to be protected was subdivided into "tiles" in order to

confine the destruction of the ceramic armor to a small area. As described in 7,

8 two types of tile arrangements were tested against both model shaped charges

and heavy metal rods:

Triangular alumina tiles without damping elements at the boundaries and

The same tile shape with a knopped 1-mm thick rubber layer in between.

Figure 5. Composite armor: Aluminum

Confinement with Alumina/GFRP Inserts


The damped version reduced the damage radius considerably (see Figs. 6, 7).

In the interim, this version was patented by RUAG Land Systems and model

armor specimens for trial firings were built 8 . In spite of these positive results,

no specific alumina applications were realized. On the one hand, this material is

rather expensive; on the other hand, there are other less delicate promising

materials which may protect against both shaped charges and KE projectiles.

Nevertheless, research studies continued on the protection effectiveness of

ceramics.

Figure 6. Damage to Ceramic Tiles

after Shaped Charge Impact with

Direct Contact between the Tiles

Figure 7. Damage to Ceramic Tiles

after Shaped Charge Impact with

Intermediate Layer between Tiles.

TEST RESULTS WITH SILICON NITRIDE (SI3N4)

The SI3N4 ceramic was provided by HTM of Biel/Bienne, Switzerland, under

the trademark "Hatemit". This firm offers this material as armor protection against

small arms projectiles 9 . Porous, hot isostatically pressed Si3N4 was tested

against large caliber shaped charges and KE projectiles 10 .

Against 105 mm KE rods, FM values between 2.5 and 3.2 were obtained,

nearly double the alumina values. However the extremely high price combined

with the reduced multihit capability have prevented practical applications so far.

Moreover porous Si3N4 with a density of only 2.3 g/cm3, requires a relatively high

armor thickness.


TESTS WITH SILICON CARBIDE

In the Eighties, extreme protection requirements arose. One outstanding

example was the gunshield of a tank; the restricted depth required a maximum

space equivalence factor, at the same time the gun unbalance moment needed to

be kept to a minimum, resulting in a maximum mass equivalence factor

requirement. The only advantage of this optimization problem was the small

volume of the gunshield, which reduced teh probability of the multihit problem.

Cercom Inc. of Vista, California, USA offered their range of SiC materials

already in series production. This fact promised low prices, apart from the high FS

and FM values to be expected. The Cercom materials properties have been

described in 11 . SiC blocks of 150-mm x 150-mm x 30-mm were tested against

model shaped charges and KE rods.

Target 1 (Figure 8) consisted of two blocks in a 10-mm RHA confinement

with a 20-mm backing. Target 2 (Figure 9) was a combination of two SiC blocks

and 150-mm of GFRP in an identical confinement. RHA witness plates were

placed behind each target.


Figure 8. Target 1

Figure 9. Target 2

CONCLUSIONS

The firing tests showed very high protection factors (see Table 1), the pure

SiC target naturally displaying higher values than the combined one.

Table 1: Test results (mean values)

L/D 20 Tungsten rod

Pref = 165-mm RHA

50- mm/50° Shaped charge

Pref = 330 mm RHA

Target 1: FM (SiC) 4.8 6.6

Target 2: FM (SiC + GFRP) 3.0 4.2

These good results encouraged us to try the 1:1 scale. Cercom provided SiC

blocks of 100-mm x 450-mm x 450-mm without problems. Targets of similar

layouts as the model targets were tested against large caliber rounds.

Concurrently, model tests were conducted at the Institute Saint Louis, France.

Partial results are published 12 . The protection capability of the full scale armor

was reduced (FM 3.2 instead of 4.8).

Table 1: Test results (mean values)

L/D 20 Tungsten rod

Pref = 165-mm RHA

50- mm/50° Shaped charge

Pref = 330 mm RHA

Target 1: FM (SiC) 4.8 6.6

Target 2: FM (SiC + GFRP) 3.0 4.2


REFERENCES

1 H. G. Hopkins and H. Kolsky, "Mechanics of Hypervelocity Impact of

Solids", 4. Symposium on Hypervelocity Impact, 1960

2 G. Weihrauch, "Behaviour of copper rods impacting various materials with

velocities between 50 and 1650 m/s", ISL, Rapport - Bericht 7/71, 1971

3 C. L. Grabarek, "Penetration of Armor by Steel and High Density

Penetrators (U)", Ballistic Research laboratories, Aberdeen Proving Ground, Maryland, Memorandum Report No. 2134, October 1971

4 W. Odermatt, "Penetration Formula for Long Rod Penetrators", Defence

Procurement Agency, Report No. 1546, 14.04.2000

5 H.-J. Ernst and V. Wiesner and T. Wolf, "Armor Ceramics under High-

Velocity Impact of a Medium-Caliber Long-Rod Penetrator", Presented at Pac

RiM 4 Ceramic Conference in Maui/Hawaii, November 2001

6 R. Ochsenbein "Behaviour of Alumina targets impacted by shaped charge

jets", RUAG Munition (formerly Eidg. Munitionsfabrik Thun), Report No. X 010

042/1-67, 04.03.1980

7 R. Jeanquartier, "Behaviour of Alumina targets impacted by tungsten

rods", RUAG Munition (formerly Eidg. Munitionsfabrik Thun), Report No. FA X

010 027, 18.10.1979

8 R. Jeanquartier and B. Lehmann, "Firing tests with 35 mm APDS vs.

Composite targets", RUAG Munition (formerly Eidg. Munitionsfabrik Thun),

Report No. FA X 010 095, 08.04.1983

9 Dr. Hr. Thieme, "Silicon-Nitride as armor material against small calibre

munitions", RUAG Land Systems (formerly Eidg. Konstruktionswerkstätte Thun),

Report No. FB 00014, 14.02.1990

10 N. Schwizgebel, "Physical/Chemical Analysis of Silicon-Nitride",

Gruppe für Rüstungsdienst, Report No. FA-26-SIG Schw/ah-200/2270,

02.07.1984

11 Dr. H. Leber, "Material Properties of Silicon Carbide", RUAG Land Systems (formerly Schweiz. Unternehmung für Waffensysteme AG), Report No.

WTB 100009917, 11.09.2000

12 H.-J. Ernst and T. Wolf and W. Lanz, "SiC-Targets Against Differently

Scaled KE-Threats", RUAG Land, Report No. WTB 100009917, 11.09.2000


STRUCTURE AND PROPERTIES OF SHOCK-RESISTANT CERAMICS

DEVELOPED AT THE INSTITUTE FOR PROBLEMS IN MATERIALS

SCIENCE, NAS OF UKRAINE

B.A. Galanov, O.N. Grigoriev, S.M. Ivanov and V.V. Kartuzov

Frantsevich Institute for Problems in Materials Science,

National Academy of Sciences of Ukraine

3 Krzhyzhanovsky St.

Kyiv, Ukraine 03142

ABSTRACT

The results of investigation of mechanical properties of a number of new

composite materials developed in IPMS, NAS of Ukraine are presented. A

prognosis of their ballistic properties was fulfilled on the base of the work [1].

INTRODUCTION

A wide range of ceramic materials, composites with ceramic matrix and

products made out of those has been developed practically for all fields of

economy (engineering, metallurgy, electrotechnics, chemical production,

environmental protection, etc.) and introduced by the Institute for Problems of

Materials Science, NAS of Ukraine within the period of 60-th of XX century and

up to present days. Actually, the Institute is one of those who developed bullet-

proof vests of all known protection classes. Bullet-proof vests production was

established by the Institute at a number of Ukrainian and Russian plants in the 80-

th.

R&D results on new ceramic materials, prospects for an employment in armor

and also for neighboring applications (wear-resistance and radiation protection)

are presented in this paper.

The materials under investigation are:

1. Ceramics and ceramic matrix composites (CMC) on the base of boron carbide;

2. Ceramics and CMC on the base of borides;

3. Ceramics and CMC on the base of silicon carbide;

4. Ceramic materials on the base of nitrides.



INVESTIGATIONS AND TECHNOLOGY

R&D was performed by:

Calculations and experimental determination of fields of internal stresses in

ceramics and CMC;

Optimization of structural and stress-strain states of CMC by an employment

of appropriate thermo-mechanical CMC model.

RAW MATERIALS

A wide range of refractory powder compounds supplied by different

manufacturers were used (Donetsk Plant of Chemical Reagents, Zaporozhye

Abrasive Combinat, Institute for Problems of Materials Science and H.C. Stark

(Germany)).

SINTERING

Within a framework of present R&D efforts a high-speed hot pressing

technology for products made out of different ceramics has been developed. The

hot pressing process was carried out on pilot installations with induction heating

in graphite molds without protective atmosphere.

CERAMICS PROPERTIES

Strength at Low Temperatures and Optimization of Structure and Composition of

Ceramics

Practically in all actual development programs, ceramics are not a single-

phase but represent some form of ceramic matrix composites. Elastic interaction

of phases at temperature and pressure changes during production and under

external thermo-mechanical effects results in a complex stress-strain state of a

material, which determines the features of its mechanical behavior. Therefore the

optimization of ceramics structure goes directly towards optimization of its fields

of internal stresses.

A mathematical formulation of fracture toughness criteria for the composite

with ceramic matrix and optimization methods for composite composition and

structure were proposed by B. Galanov and O. Grigoriev [2]. The introduction of

a high- component into the composite is accompanied by an increase of fracture

toughness. The maximum value of fracture toughness is shifted to the lower

second phase contents ( 10-30%) with an increase of CTE (coefficient of thermal

expansion) mismatch, the elastic characteristics, and grain sizes of the composite

phases. At high concentrations of the second phase, the minimum K1c with the

value of K1c 0 is found. It is caused by spontaneous failure under the effect of

thermal stresses with flaw size approximately equal to the grain size ( 10 m):


For example, in SiC-TiB2 system, an improvement of composite properties

can be expected in the range of 20-30 vol. % of TiB2, and grain size of 5-10 m.

The analysis considers thermal stresses, partly generated by the difference in CTE

of the constituting phases of the composite.

The stress calculations present only an estimation of stress levels in phases,

without accounting for viscous-elastic relaxation. Therefore there is an important

role for experimental methods to determine internal stresses.

The experimental determination of thermal stresses gives us not only quantitative

information on internal stresses in phases, but also data on viscous-elastic

relaxation and the state of grain boundaries. A comparison of the experimental

and theoretical values of internal stresses shows a good agreement between theory

and experiment only if take a rather small T value ( T 1300 C) in the

calculation. It means that the temperature of the viscous-elastic transition is fairly

low (Tve 1300 C) and the composite possesses has a significant relaxation

ability at high temperatures (T 1300 C).

Boron Carbide Based Ceramics

The ceramic materials based on boron carbide are being employed thanks to

their high hardness, low volume mass, high effective capture cross-section of

neutrons and the like. Hot pressing is principal technique to obtain dense ceramic

out of boron carbide at temperatures 2000 – 2150 . That is why to produce the

materials on the base of boron carbide of special attention are methods of

activation of sintering. In present paper the activation of sintering process was

provided by borides additives (W2B5, TiB2 and ZrB2)

Composites of the B4C-TiB2 and B4C-ZrB2 Systems

Mechanical properties of the composites are considerably lifted up if the

second phase is of 15% (See Figure 1). The further lifting of borides

concentration brings to the reduction of fracture toughness and other

characteristics since the inner stress exceed an optimum level. Temperature

dependencies of hardness display that hardness of the composites on the base of

boron carbide at temperatures higher than 1100 exceeds the diamond and

boron nitride hardness (See Figure 2).

Composites of the B4C-W2B5 System

The ceramic was investigated in the volume content interval of tungsten

boride from 10 up to 90%, with a step of 10%. Table I displays the properties of

the materials presented.


Table I. Properties of B4C-W2B5 system composites

Composition of

ceramics, vol. %

Density,

g/cm3

Bending

strength, MPa

HV, P=10N

GPa

90W2B5-10B4C 7.00 660 34

80W2B5-20B4C 6.75 500 35

50W2B5-50B4C 5.65 690 40

40W2B5-60B4C 5.10 590 38

10W2B5-90B4C 3.24 565 52

Figure 1. Variations of Young's modulus E, bending strength f, hardness HV,

fracture toughness 1 of ceramics on the B4C base versus second phase content:

– TiB2, – ZrB2

Figure 2. Temperature dependences of hardness for some superhard materials: 1-0

– B4C, 1-1 – B4C + 30% ZrB2, 1-2 – B4C + 40% ZrB2, 2 – diamond (Berkovich

indenter, = 70 ), 3 – diamond ( = 65 ), 4 – BN (hexanit), 5 – BN (elbor), 6 – BN

(PTNB)

The materials are obtained at comparatively low temperatures of hot pressing

(1800 C). Yet the mechanical properties, especially hardness, are turned to be

considerably higher than hot pressed boron carbide has (bending strength < 450

MPa, HV < 35 GPa).

Composites of the B4C-TiB2-W2B5 System

The materials of the above system apart from they have a high level of

mechanical properties (strength, hardness, impact- and wear-resistance) are

characterized by high linear coefficients of neutrons and -rays absorption.

Production conditions and properties of both one-phase materials – TiB2 and

W2B5, and composites TiB2-W2B5 and B4C-TiB2-W2B5 as well were studied.

Table II shows the properties of the materials obtained.


Of particular interest is high strength of two- and three-phase materials (850-

1100 MPa) even at big grain size, when the synthesized powder was not grinded

before hot pressing.

Preliminary investigations have shown that the proposed material with its

ballistic and radiation protection characteristics exceeds the previously employed.

Table II.

Composition of

ceramics, vol. %

Density,

g/cm3

Bending

strength, MPa

HV, P=10N

GPa

TiB2 4,51 365 36

W2B5(W+B+Ni) 8.62 525 12

W2B5 8.37 590 24

50W2B5-50 TiB2 5.88 1110 26

70W2B5-25TiB2-5B4C,

Grain size-20 m

7.06 850 19.6

70W2B5-25 TiB2-5B4C,

Grain size-7 .m

6.76 900 20

Coarse Heterogeneous Composites Ceramic Metal

For operation under shock loading we developed materials with coarse

heterogeneous structure – granules from borides as wear resistant component,

binded with tough matrix.

The structure of materials is shown on the Figure 3, its strength is more than

500 MPa, hardness 15 GPa, density is in the range 4.8-4.9 g/cm3. Composites

advantages are related to ceramic carcass, providing resistance at high-speed

shock, while metallic component provides increased toughness of material and

products of them.

a) b)

Figure 3. TiB2 granules and coarse-heterogeneous composite after

abrasive-wear test


DISCUSSION AND CONCLUSION

An analysis of penetration resistance was performed on the base of the

modified Alekseevskii-Tate model for nonstationary penetration of long rods into

targets [1].

“Static” component of resistance to penetration into target Rt was evaluated.

Total resistance to penetration is defined by three factors: static Rt, kinematic Pk

(0.5 U2) and dynamic Pd, relative contributions of those are varying during the

penetrator – target interaction.

In accordance with the accepted model the value Rt is defined by the system of

elastic and strength characteristics of target material: , E, , Y, f, etc. Table III

shows some of those.

0 1 2 3 4 5 6

-20

0

20

40

0.0 0.1

-20

0

20

Rt

Pk

Pd

Pc= Rt+Pk+Pd

Penetration, mm

Pressure, GPa

Figure 4. Structure of penetration resistance (contact pressure Pc) in Al2O3.

Solid line – data from [1], dashed – from table III. Impact velocity 1600 m/s (steel

penetrator).


In case target material possesses a tangible plasticity value Y that defines

material behavior on the boundary plastic — elastic

material is the yield stress of this material. In the

case of brittle material, the comminuted (pulverized)

zone is formed in the contact area as a result of

multiple fragmentation and value Y is defined by the

strength characteristics of material and is obviously

close to strength limit under compression. It was

established that at least for porous ceramic H

compression. Analogously, during quasi-static

indentation into brittle materials, semi-spherical

fragmentation zone is formed in the area of

deformation core (See Figure 5) with a pressure on

its boundary = Y, that defines the impression size

and, consequently, contact pressure and material's

hardness.

Figure 5. Core structure in

TiN/AlN ceramics


Table III. The properties of developed ceramic materials Materials Additions,

second

phases,

vol.%

Density,

g/cm3

Young's

modulus,

E, GPa

Hardness,

HV

(P=5N),

GPa

Yield

stress,

Y, GPa

E/Y Strength, fbend,

RT,MPA

Penetration

hardness

HP, GPa

HP B4C - 2.5 450 30 20 22.5 500 2.65

HP

B4C/MeB2

ZrB2 or

TiB2(up to

30%)

up to 2.7 460 35 23 20 500-800 2.66

HP

B4C/CaB6

CaB6 up to

100%

2.5 450 25-30 17-20 25 400-500 2.65

HP CaB6 - 2.4 460 25 17 26 400 2.63

TiB2 - 4.5 550 27 18 30 500-700 2.68

TiB2/CaB6 CaB6 up to

30%

4.4 530 45 30 17 700 2.68

W2B5

W2B5

(1%Ni)

- 10-13

10-13

775

775

30

12

18

8

43

97

500

500

3.05

3.73

TiB2/

W2B5/

B4C

- 4-10 600 30 20 30 800-1000 2.68

3.0

TiB2/

W2B5

W2B5 up to

50%

4.5-10 600 35 23 26 700-1000 2.68

2.86

HP SiC B4C up to

5%

3.2 460 20 13 35 300-500 2.72

HP SiC/

MeB2

ZrB2 or TiB2

(up to 30%)

3.3 450 25 17 26 600-700 2.66

RS

SiC/MeB2

The same 3.2 440 20 13 34 500 2.7

HP

TiN/AlN

AlN up to

100%

3.2 400 20 13 31 400-500 2.7

HP AlN - 3.2 280 12 8 35 300-400 3.07

HPSi3N4/

ZrO2/Y2O3

ZrO2 (up to

30%), Y2O3

(up to 10%)

3.3 320 16 12 26 700 2.78

S -

SIALON

z = 2-5 3.15 220 14 9 24 500 2.93

HP -

SIALON

AlN/Y2O3

(up to10%)

3.2 350 21 14 25 600 2.73

Al2O3 - 4 400 15 10 40 500 3.04

Al2O3 [1] - 3.5 373 2.62 142 262 7.45

Al2O3/

TiB2/

ZrO2

TiB2 up to

30%, ZrO2

up to 30%

4.5 430 16 12 35 800 2.96


In correspondence with indentation models of Tanaka, when radius of semi-

spherical core approximately equals to the radius of contact area, the ratio of

hardness to Y is ~ 1, which allows to evaluate the upper boundary of Y with

hardness value. The lower boundary of Y is defined evidently by the relationship

HV~3Y for plastic material.

The authors are also of the opinion that if under impact radial cracks are

formed outside the comminuted zone then their formation must be defined not by

material's tensile strength but by «contact strength» of material tested precisely

under the contact loading and which can be evaluated using the value of length of

radial cracks around the hardness indents. The technique of contact strength

measurement is presented in [3].

Resistance to penetration was characterized by the penetration work per the

unit volume of extruded material: "penetration hardness" HP= 1/P PcdP (P –

depth of penetration).

The analysis of the results has shown that under investigated conditions of

impact the following conclusions can be made:

1. During penetration the pressure on the contact surface increases.

2. High-strength ceramics in the wide range of variation of its characteristics

demonstrates small change in the "penetration hardness" – 2.5–3 GPa.

3. "Penetration hardness" increases with the growth of parameter E/Y (up to 7

GPa at E/Y=140), however, the depth of penetration also sharply increases.

ACKNOWLEDGMENT.

The authors would like to acknowledge the support from ARL under the

contract 68171-01-M-5848 and scientific coordinator Dr. W. Gooch.

REFERENCES

1. B.A. Galanov, S.M. Ivanov, and V.V. Kartuzov. "On one new

modification of Alekseevskii-Tate model for nonstationary penetration of long

rods into targets". Proc. of HVIS'2000, Journal of Impact Engineering, 26 201-10

(2001) (to be published).

2. B.A. Galanov, O.N. Grigoriev, and V.I. Trefilov, "Ceramic Matrix

Composites: theoretical fundamentals", in Ceramic- and Carbon-matrix

Composites, Edited by V.I. Trefilov, Chapman & Hall, 3-29, 1995.

3. B.A. Galanov, O.N. Grigoriev, and E.G. Trunova, "Contact strength and

statistikal fracture mechanics of ceramics"; p.19 in Proc. of Int. Conf. “Current

Problems of Strength”, 3-5 July, 2001, Kiev.


CERAMIC ARMOR WITH SUBMICRON ALUMINA AGAINST

ARMOR PIERCING PROJECTILES

E. Strassburger, B. Lexow

Fraunhofer-Institut für

Kurzzeitdynamik Ernst-Mach-Institut

(EMI)

Am Klingelberg 1

D-79588 Efringen-Kirchen, Germany

A. Krell

Fraunhofer-Institut für Keramische

Technologien und Sinterwerkstoffe

IKTS, Winterbergstr. 28

D-01277 Dresden, Germany

ABSTRACT

In a joint project of the Fraunhofer Institute for Ceramic Technologies and

Sintered Materials (IKTS) and the Ernst-Mach-Institute (EMI), aluminum oxide

ceramics with submicron grain size were developed and tested ballistically. In

DOP-tests with tungsten alloy projectiles, the new ceramics revealed a ballistic

efficiency superior to commercial alumina grades.

Additionally, the ballistic performance of the new submicron and commercial

alumina against armor piercing (AP) steel core projectiles was investigated. The

ceramic/aluminum targets were also tested in a Depth of Penetration (DOP)

configuration. The influence of ceramic layer thickness and sequence was

determined with laminated targets.

INTRODUCTION

High strength ceramics are employed as ballistic protection material when a

high protective strength is required at a low weight. In order to improve the bal-

listic performance of a ceramic, it is necessary to know the correlations between

the microstructure and the ballistic resistance. However, for studying the correla-

tions between microstructure and ballistic resistance of one type of ceramic, it is

essential to have well defined materials, where individual parameters like grain

size, purity, porosity and density can be adjusted with high accuracy. The results

of previous studies1 have indicated, that only high purity ceramics with relative

densities > 98.5 % should be used in investigations on the influence of grain size

and hardness on ballistic performance. On one hand, earlier tests have demon-

strated that the ballistic resistance of ceramics increases with increasing

hardness2. On the other hand, it is known that the hardness of polycrystalline



ceramics increases with decreasing grain size3. Therefore, high purity aluminum

oxide ceramics with sub-µm grain size were developed and ballistic tests were

conducted in a collaboration between IKTS and EMI. The first part of the

investigations comprised materials with 1 µm, 0.5 µm and 0.3 µm grain size.

The material with 1 µm grain size was supplied by Dornier GmbH, Germany. The

sub-µm grain size materials were developed by IKTS. The commercially

available AD995 (trade name CAP-3) of Coors, Golden, Colorado, was used as

reference material because it had exhibited the highest ballistic mass efficiency

among previously tested commercial alumina grades. The ballistic resistance of

these materials was tested in a Depth of Penetration (DOP)-configuration with an

armor steel backing by means of tungsten alloy projectiles.

A second part of the study focuses, on one hand, on the investigation of sub-

µm alumina with improved strength, due to improved manufacturing processes

which lead to a significant reduction of the number of microscopic flaws. With

respect to light armor applications of the materials, the objective of the

investigations was to determine the potential of the materials for the defeat of

steel core projectiles and possible armor efficiency improvements by laminated

target configurations.

MATERIALS

The relative density, grain size, hardness and bending strength of the tested

materials are specified in Table I. The materials designated “S-“ were manufac-

tured at IKTS by means of spray drying, cold isostatic pressing and unpressurized

sintering in air. The “D-0.9” material was supplied by Dornier GmbH, Frie-

drichshafen, Germany. A more detailed description of the materials is provided in

an additional paper by A. Krell4.

Table I. Material specifications

Relative density Grain size Hardness

HV10

4-point

bending strength

(%) (µm) (GPa) (MPa)

S-0.3 92.5 0.32 15.0 Not determined

S-0.5 99.3 0.53 19.3 203 16

S-0.7 99.5 0.71 19.1 526 55

D-0.9 98.7 0.92 15.7 244 41

AD995 (CAP-3) 98.8 10-20 12.3 350 25

BALLISTIC TESTING

Tungsten projectiles versus ceramic/steel targets

In the first part of the investigations, different types of alumina were tested in

a DOP-configuration with a RHA (Rolled Homogeneous Armor steel) backing of


hardness 300 HV30. A tungsten alloy cylinder with a hemispherical nose was

used as projectile. The diameter of the projectile was 10 mm, the length 32 mm

and the mass was 44 gram. The impact velocity was 1250 m/s nominally. As a

figure-of-merit for ballistic performance, the ballistic mass efficiency, Em,, was

chosen, which is determined from the residual penetration PR, the penetration into

the reference steel target, Pref, the thickness of the ceramic, TCer and the densities

St, Cer of the steel and the ceramic. Figure I shows a schematic of the test

configuration. The definition of the mass efficiency is given in equation (1).

Figure I. Schematic of DOP-Test configuration

RStCerCer

refStm

PT

PE (1)

The DOP data of all tested materials are presented in Figures II and III. On the

left hand side, the residual penetration is shown as a function of ceramic

thickness, whereas on the right hand side, the mass efficiency is plotted versus the

ceramic weight fraction Fcer/ Ftot = cerTCer/( cerTCer + StPR). The diagram for the

reference material AD995 (Fig. II) exemplifies the behavior observed with all

commercial aluminas. A linear decrease of residual penetration as the ceramic

thickness increases is associated with a linear increase of mass efficiency with

increasing ceramic weight fraction. A linear extrapolation to the point, where the

projectile is stopped just at the ceramic-steel interface (PR = 0), yields the

maximum mass efficiency Em,max for that material. For AD995 Em,max was 2.1

with the projectile/ target combination considered here. The sub-µm alumina

exhibited a different behavior (see Fig. III). Compared to AD995, a slightly

higher residual penetration PR was observed at a ceramic thickness up to 10 mm.

However, with a ceramic thickness of more than 15 mm, the residual penetration

decreased rapidly, which implies a higher ballistic efficiency. For this sub-µm


material, both dependencies, PR-TCer and Em versus ceramic weight fraction can

be approximated by a second order polynomial fit.

Figure II. DOP data with AD995 (CAP-3)

0

10

20

30

0 10 20

TCer [mm]

PR [

mm

]

30

AD995, monolithic

AD995, 6.7 mm + 15 mm

AD995, 15 mm + 6.7 mm

1,0

1,4

1,8

2,2

2,6

3,0

0,0 0,2 0,4 0,6 0,8 1,0

FCer/ Ftot

Em

AD995, monolithic

AD995, 6.7 mm + 15 mm

AD995, 15 mm + 6.7 mm

Figure III. DOP data with sub-µm aluminas

0

10

20

30

0 10 20

TCer [mm]

PR [

mm

]

30

D-1.0

S-0.7

S-0.5

S-0.3

Fit S-0.5

black symbols: monolithic and

10 mm + 10 mm

grey symbols: 5 mm + 15 mm

open symbols: 15 mm + 5 mm

1,0

1,4

1,8

2,2

2,6

3,0

0,0 0,2 0,4 0,6 0,8 1,0

FCer/ Ftot

Em

D-1.0

S-0.7

S-0.5

S-0.3

black symbols: monolithic and

10 mm + 10 mm

grey symbols: 5 mm + 15 mm

open symbols: 15 mm + 5 mm

The extrapolation of the Em-curves for D-0.9, S-0.7 and S-0.5 resulted in

significantly higher Em, max values compared to AD995. Extrapolation, based on

the results with 10 mm + 10 mm targets, yielded a maximum mass efficiency of

2.6, and even 2.9 could be achieved with the 5 mm + 15 mm configuration for

each of the three materials.

Two important conclusions can be drawn from the presented data. The results

with D-0.9, S-0.7 and S-0.5 show that hardness is much more important for the


ballistic efficiency than bending strength. The fact that Em,max of the very fine

grained S-0.3 material is higher than with AD995, but significantly lower

compared to the coarser, but dense, ceramics S-0.5 and S-0.7 demonstrates, that

there is no separate influence of grain size on Em,max beyond the hardness effect.

An additional effect was observed with laminated ceramic targets of 20 mm

total thickness. The lowest residual penetration was observed with the 5 mm + 15

mm targets, the highest PR occurred with the 15 mm + 5 mm plate sequence. The

lamination effects can be explained qualitatively as follows. When the projectile

first hits a thin front layer, this layer will be fragmented very rapidly and will

exhibit only a low ballistic resistance. However, pre-damage to the second plate

will be reduced so that the projectile has to penetrate a material with higher

ballistic resistance compared to the case of a monolithic target. When the thick

plate is at the front, the penetration of this plate is like that in a monolithic target.

However, the thin plate at the back will be shattered by several wave reflections

that results in a strongly reduced ballistic resistance of this plate. With two layers

of equal thickness both effects, stronger pre-damage of the first plate and reduced

pre-damage of the second plate, appear to compensate one another.

Steel core projectiles versus ceramic/aluminum targets

Three types of alumina were tested in a DOP-configuration with armor

piercing (AP) steel core projectiles of 14.5 mm caliber at an impact velocity of

1045 15 m/s. The projectiles had a total mass of 64.1 g, whereas the mass of the

steel core was 40.5 g. Aluminum (AlCuMg1) of tensile strength 400 MPa was

used as backing material. The ceramic tiles were glued to the backing by means of

the polyurethane glue Sikaflex and the joint between the lateral steel confinement

and the edge of the ceramic was filled with epoxy.

Figure IV shows the residual penetration versus ceramic thickness, not only

for monolithic targets of AD995, D-0.9 and S-0.7, but also for laminated targets

consisting of two or three plates of equal thickness or two plates with a thickness

ratio of ½. The dashed line indicates the overall behavior with monolithic ceramic

targets. Two sections can be distinguished where the decrease of PR with

increasing ceramic thickness can be approximated linearly. However, for ceramic

thickness 10 mm < TCer 15 mm, the slope is only half of that in the range from 5

mm to 10 mm thickness. When TCer is 5 mm or less, the ballistic resistance is very

small as the results with D-0.9 and S-0.7 indicate. In the range from 5 mm < TCer

15 mm, no significant difference was observed with the three types of alumina.

That means, with respect to the ballistic resistance, no benefit of the increased

hardness and strength compared to AD995 was achieved for this type of ballistic

test with D-0.9 and the high strength sub-µm grain size S-0.7. This result

indicates, that there is a saturation with respect to the influence of hardness on the

ballistic resistance against steel core projectiles. When the hardness is sufficiently


high erosion and/or break up of the steel core is initiated. A higher hardness does

not lead to a more efficient erosion of the steel core. The fact, that there was no

benefit of the improved strength could be attributed to the design of the DOP test

configuration. The quasi semi-infinite backing prevents or reduces bending of the

ceramic specimen during the first phase of the projectile/target interaction. The

role of the strength of the undamaged material decreases during penetration as the

pre-damage of the material by stress waves progresses. However, that does not

exclude benefits of the high strength in “thin” targets, where bending of the

backing occurs.

Figure IV. Residual penetration versus ceramic thickness

0 5 10 15 20

TCer [mm]

AD995, monolithic

AD995, 2 x 6.7 mm, plates loosely stacked

AD995, 2 x 6.7 mm, plates glued with Sikaflex

AD995, 2 x 6.7 mm, plates ground on both sides, stacked

D-0.9, monolithic

D-0.9, 3 x 5 mm, glued with Sikaflex

D-0.9, 5 + 10 mm, glued with Sikaflex

S-0.7, monolithic

arb

itra

ry u

nit

s

With AD995, three different target types consisting of two plates of 6.7 mm

thickness were assembled and tested. The targets differed in the flatness of the

ceramic plates and in the way the components were joined. All two layer targets

exhibited a significantly lower ballistic resistance compared to the resistance

expected for monolithic targets of the same total thickness. The highest residual

penetration was observed when the ceramic plates were only loosely stacked.

Joining the two ceramic plates with polyurethane glue resulted in a reduction of

PR. However, the best performance of the two layer targets was observed when the

two plates were ground on both sides and stacked without glue. This observation

implies that the ballistic resistance against AP projectiles is the better the closer

the target is to the monolithic case.

Targets consisting of three layers of equal thickness (3 x 5 mm) and two layer

targets (5 mm + 10 mm) of D-0.9 were tested. In both configurations, the ceramic


plates were glued together with Sikaflex and the bottom ceramic plate was glued

to the aluminum backing. The 5 mm + 10 mm targets performed significantly

worse than the monolithic targets, whereas the ballistic resistance of the three

layer targets was heavily degraded compared to the monolithic ones. The relative

performance of the laminated and monolithic targets is summarized in Figure V,

where it is presented as the ratio of the residual penetrations with monolithic and

laminated targets (PR, monolith/PR, laminate). Since the ballistic resistance of single 5

mm plates of D-0.9 was very low, the poor performance of the 3 x 5 mm targets

could be expected.

Figure V. Relative performance of laminated and monolithic targets

0

0,2

0,4

0,6

0,8

1

stacked glued ground, no glue monolithic

AD995

2 x 6.7 mm

0

0,2

0,4

0,6

0,8

1

3 x 5 mm 5mm+10mm monolithic

D-0.9

The efficiency of the ceramic is connected to its ability to erode and break up

the penetrating steel core of the projectile. Thus, the residual mass mR of the steel

core is also a measure of the efficiency of the armor. In Figure VI, the residual

mass of the steel core is plotted versus ceramic thickness. With the monolithic

targets, a strong decrease of mR was observed at a ceramic thickness of 10 mm.

At lower ceramic thickness, typically, an eroded steel core of mR > 30 g was left,

whereas with TCer > 10 mm, only 5-10 fragments of the steel core with a total

mass of less than 15 g could be found (see Fig. VI).

SUMMARY

The influence of grain size, hardness and strength on the ballistic performance

of Al2O3-ceramics was determined by means of DOP-tests.

Tungsten alloy projectiles were employed in order to determine the ballistic

resistance of ceramic/steel targets.

In this projectile/target combination, alumina with sub-µm grain size exhibited

significantly higher maximum mass efficiencies than commercially available

alumina tested under the same conditions.

The results clearly indicate that there is no separate influence of grain size or

of flaws beyond their impact on hardness.

The ballistic resistance against tungsten projectiles can be increased by

laminated targets consisting of a thin front layer and a thick rear layer.


14.5 mm AP steel core projectiles were used in order to assess the ballistic

resistance of ceramic/aluminum targets. No significant difference was observed in

the penetration behavior of the three alumina types tested.

That means, with respect to the ballistic resistance no benefit of the increased

hardness and strength compared to AD995 was achieved with D-0.9 and the high

strength sub-µm grain size S-0.7.

Lamination of the ceramic targets resulted in significant losses of ballistic

performance.

Figure VI. Residual steel core mass and state of the residual projectiles

0

10

20

30

40

50

0 5 10 15 20

TCer [m m ]

mR [g

AD995, monolithic

AD995, 2 x 6.7 mm, stacked

AD995, 2 x 6.7 mm, glued

AD995, 2 x 6.7 mm, ground,no glue

D-0.9, monolithic

D-0.9, 3 x 5 mm

D-0.9, 5 mm + 10 mm

S-0.7, monolithic

TCer, monolithic

< 10 mm

TCer, monolithic

> 10 mm

REFERENCES1B. James, “The influence of the material properties of alumina on ballistic

performance,” pp. 3-9 in Proceedings of the 15th

International Symposium on

Ballistics (Jerusalem/Israel, 1995 published by the Organizing Committee).2I. Faber, K. Seifert and L.W. Meyer, “Correlation between the mechanical

data of ceramics and their protective power against impact loading” (in German),

Final Report EB 6/95 (part 3), Technical University Chemnitz-Zwickau,

Department of Engineering Materials, 1995. 3A. Krell and P. Blank, “Grain Size Dependence of Hardness in Dense

Submicrometer Alumina,“ J. Am. Ceram. Soc. 78 4 1118-20 (1995).4A. Krell and E. Strassburger, “High-Purity Submicron -Al2O3 Armor

Ceramics – Design, Manufacture, and Ballistic Performance”, Proceedings of

PAC RIM IV, Ceramic Armor Materials by Design (Wailea, Maui, Hawaii,

2001).

Figure VI. Residual steel core mass and state of the residual projectiles

0

10

20

30

40

50

0 5 10 15 20

TCer [m m ]

mR [g

AD995, monolithic

AD995, 2 x 6.7 mm, stacked

AD995, 2 x 6.7 mm, glued

AD995, 2 x 6.7 mm, ground,no glue

D-0.9, monolithic

D-0.9, 3 x 5 mm

D-0.9, 5 mm + 10 mm

S-0.7, monolithic

TCer, monolithic

< 10 mm

TCer, monolithic

> 10 mm


ARMOR ALUMINA CERAMICS

Eugene Medvedovski

Ceramic Protection Corporation

3905 – 32nd

Street N.E.

Calgary, Alberta, T1Y 7C1, Canada

ABSTRACT

Dense alumina ceramics are still one of the most cost-effective armor

materials among different structural ceramics used for ballistic protection. They

have high mechanical properties and excellent manufacturability. High-alumina

ceramics with an Al2O3 content ranging from 97 to 99.6-wt.% and alumina-

zirconia ceramics produced by Ceramic Protection Corporation have been

successfully used as an armor material for personnel and vehicular ballistic

protection. They are manufactured by slip casting and pressing technologies

depending on the required shape and quantity. The main properties of the

ceramics, which affect ballistic performance, and ballistic test results are

examined and analyzed. Only the combination of all physical properties and

microstructure, as well as the optimization of the manufacturing process, should

be considered for selection and evaluation of armor ceramics.

INTRODUCTION

Ceramic armor was originally developed for “bulletproof vests” and seat-

armor in helicopters. At the present time ceramic armor is mainly used for

personnel and vehicular ballistic protection in military forces and by tactical

police teams, for protection of some critical parts of aircraft and helicopters and

for blast protection against landmines.

The mechanisms of ballistic protection for ceramic and metal armor are

significantly different. Metals absorb the energy of projectile by a plastic

deformation mechanism. In the case of ceramics, the kinetic energy of the

projectile is absorbed by a fracture energy mechanism. Usually the ceramic armor

system consists of the monolithic ceramic or composite ceramic-metal body

covered by ballistic nylon and bonded with a high tensile strength fiber lining

such as KevlarTM

, SpectraTM

or fiberglass. Also some soft metals (e.g. aluminum

thin sheets) may be used as a backing material. Upon impact of the bullet



(velocity greater than 700-800 m/sec), the hard-facing ceramic body is cracked

and broken, and the residual energy is absorbed by the soft reinforced backing

material. This backing material also must support post-impact fracturing of the

ceramic body caused by the bullet and the bullet itself.

Consideration of ballistic protection systems must take into account several

factors: the type of ballistic threat, the ability to manufacture the armor system

and the properties of the armor system components. These include such factors as

threat level, multi-hit performance, environmental conditions, space limitations,

manufacturing challenges, cost and weight limitations, physical properties of both

facing and backing material and the overall ballistic performance of the system.

Different armor ceramics including monolithic ceramics and ceramic-matrix

composites are described in [1-7]. Among them alumina ceramics are of low cost

and may be manufactured using a variety of methods, i.e. slip casting, pressing,

injection molding and some others, without the use of expensive equipment, e.g. a

kiln with special controlled atmospheres. Despite elevated density (up to 3.95

g/cm3), alumina ceramics are used for ballistic protection. In this paper, the high-

alumina and alumina-zirconia ceramics commercially produced by Ceramic

Protection Corporation (CPC) are reviewed and studied. Different armor products

(tiles and monolithic curved plates) are manufactured from these ceramics with a

high quantity (e.g. several hundreds plates and several thousand tiles per day).

They are designed and manufactured in accordance with the specific customer

demands depending on the application, required performance and level of

protection; they may be obtained as bare ceramic products, or they may be laid-up

with backing materials. They have been successfully used for personnel and

vehicular ballistic protection.

EXPERIMENTAL

Materials

The studied armor alumina ceramics are based on the systems Al2O3-SiO2-

CaO-MgO and Al2O3-MgO with an Al2O3 content approximately 97, 98, 98.5 and

99.6-wt.%. The alumina-zirconia ceramics is based on a specially selected ratio

between Al2O3 and ZrO2 (Y2O3 is used as a stabilizing agent). The starting

alumina powders producing by Pechiney - Altech (France) and Alcoa World

Chemicals (USA) have a high purity (minimum 99.8-wt.% of Al2O3); the -form

content is 95-wt.% or greater depending on the alumina grade. The average

median particle size and crystal size of the aluminas range from 0.35-0.45 to 1.1-

1.4 µm, and their specific surface BET ranges from 8-11 to 2.8-3.3 m2/g for the

used alumina powders, respectively. In the case of the alumina-zirconia ceramics,

the zirconia powder producing by Tosoh Corp. (Japan) is also used as a raw

material. The partially Y2O3-stabilized zirconia powder has the median particle

and crystal size of 0.3-0.4 µm and the specific surface BET of 8-9 m2/g.


Manufacturing

Manufacturing methods used for production include slip casting and dry

pressing processes depending on the shapes and quantity of ceramic products to

be made. Experimental, pilot-scale and production studies allowed for

optimization of the following manufacturing steps:

Ceramic water-based slip preparation depending on the batch composition

(consisting up to 77-80-wt.% of solid), including the development of the

dispersant and binder systems;

Slip casting process providing manufacturing of single, double, and triple

curve plates with the custom designed shape and dimension;

Spray drying process providing a powder yield of up to 96%, press-powder

preparation providing “donut”-free spherical particles with adjustable sizes;

Uniaxial pressing process;

Drying and firing processes (firing temperature less than 1550oC), including

kiln loading, depending on the shape and size of the products and the optimal

firing curve;

Bonding process of ceramics with a backing material including adhesive

preparation, Kevlar, fiberglass and nylon preparation, thermal treatment of the

glued ceramic product with a backing material in an autoclave where

temperature, pressure, and vacuum are applied;

Quality control system which provides overall quality control and possible

adjustments at each manufacturing step.

The manufacturing of armour products at CPC is ISO 9002 certified. Each

step of the manufacturing process must be accompanied by the corresponding

quality control procedures. The quality control starts with the raw materials

verification. The following ceramic manufacturing parameters are controlled;

some are adjusted individually in order to achieve required parameters:

Sequence and duration of the starting materials mixing and milling, specific

gravity, viscosity, pH of the initial slips;

Binder and plasticizer component contents, sequence of addition to the initial

slip, specific gravity, viscosity, pH of the resultant slips (particularly if the slip

is used for slip casting or for spray drying), and casting rate of the slip if a

new lot of raw materials is started using for manufacturing;

Spray drying parameters (air pressure, inlet and outlet temperatures, flow rate,

etc.);

Granulation process parameters;

Particle size distribution, bulk density, powder flow rate, moisture content for

spray dryed powders and for powders ready-to-press (RTP); compression

coefficient for the RTP-powders;


Slip casting and pressing processes and parameters;

Dimensions and weight for green products;

Firing parameters, including firing curve and final firing temperature, oxygen

level, air pressure, and kiln loading.

The following parameters are tested for the fired ceramics:

Fired and total shrinkage, dimensions and shape parameters (curvature for the

body armor and special vehicular armour plates, flatness for tiles);

Density and open porosity for the products and the witness samples made

from the same material as products from each kiln fired at different levels of

the kiln;

Physical properties (Vickers hardness, fracture toughness, sonic velocity,

Young’s Modulus, flexural strength) for the witness samples from the

selected firings using the specially developed testing protocol;

Ballistic performance for the products in accordance with testing protocols.

Testing

Microstructure was studied using transmission and scanning electron

microscopes. Density, porosity, and water absorption were tested using the water

immersion method based on Archimedes law. Four-point flexural strength was

tested in accordance with ASTM C1161. Young’s Modulus and sonic velocity

were tested by the ultrasonic technique measuring the longitudinal ultrasonic

velocity in accordance with ASTM C769 and by the resonant frequency method

in accordance with ASTM C885. Vickers hardness was tested in accordance with

ASTM C1327 at loads from 0.3 to 50 kg; the load 10 kg was used as a “standard”.

Fracture toughness KIc was also determined using the indentation technique under

the load of 10 kg. The test samples with required dimensions were cut from actual

products or from the test tiles produced by the mentioned technologies.

Ballistic performance of ceramics bonded with appropriate backing materials

was tested in accordance with the NIJ 0101.03 and NIJ 0101.04 standards using

the weapons M16, KAR 98K, AK47 and some others (caliber 0.30). Depending

on the application and the required level of protection, the ammunition 7.62x51-

mm NATO Ball Full Metal Jacket (FMJ) with a lead core, 7.62x63-mm Armor

Piercing M2 FMJ with a tungsten carbide core, 7.62x39-mm Russian Ball FMJ

with a steel core and some others were used. Depending on the ammunition, the

bullet weight, velocity and energy are varied. The bullet velocity was controlled

using a chronograph. The trauma after shooting was measured using a Roma

Plastilina modeling clay supported armor system on the back; the trauma in clay

duplicated the trauma in armor. The damage zone of the ceramics, including

ceramic fragmentation and the bullet were observed. Considering ceramic armor

systems for ballistic testing, the flat tiles (100x100 mm or greater) with a

thickness of 7-15 mm were used for the single shot testing. Also, the tiles (50x50


or 100x100 mm) assembled as a flat panel, and flat tiles (155x200x8-9 mm), as

well as the actual plates with different configuration, were used for multi-hit

ballistic testing (with approximately 50 mm spacing between hits).

RESULTS AND DISCUSSION

Microstructure and Physical Properties

All studied ceramics are fully dense (water absorption is not greater than

0.02%) after firing at temperature less than 1550oC. Phase composition and

microstructure of the AL97ML, AL98 and AL98.5 alumina ceramics (a number in

the ceramic composition denotes an approximate Al2O3 content) are similar, and

they consist of corundum grains (the major phase) bonded by a small amount of

anorthite crystals and a silicate-based glassy phase. A small amount of mullite

crystals is also present in the AL97ML ceramics. The AL99.6 alumina ceramics

consists of corundum grains bonded by spinel crystals and a very small amount of

a glassy phase that formed due to the presence of oxides-impurities.

The ultimate grain size of the alumina ceramics depends on the initial batch

composition, initial particle size and particle size distribution of the starting

alumina powders. As expected, as an alumina powder with a smaller particle and

median crystal size was used, a fine-crystalline structure with a smaller grain size

was achieved. The average corundum grains are ranged from 1-3 µm for the

AL99.6 (mostly isometric) to 3-6 µm (isometric) and (2-3)x(5-8) µm (short

prismatic) for the AL97ML ceramics. A glassy phase is distributed between

grains uniformly and, as expected, the amount of a glassy phase increases as the

alumina content decreases.

The alumina-zirconia AZ ceramics based on the special ratio between alumina

and partially stabilized zirconia (PSZ) does not have a glassy phase; zirconia

grains with a size less than 1 µm are uniformly distributed between corundum

grains with a size of 1-2 µm. The zirconia phase, probably, inhibits the corundum

grain growth during sintering.

All these microstructure features affect physical properties and ballistic

performance of the ceramics. Physical properties depend on the Al2O3 content, the

size and shape of corundum grains, the amount, composition and distribution of a

glassy phase cemented the crystalline phase, the presence and composition of the

“secondary” crystalline phases, and closed porosity. They also depend on the

“stressed conditions” at the boundary of the corundum grains and a glassy phase.

These factors are governed by the wetting of alumina particles by a liquid phase

and by the interaction between them during sintering, firing and cooling, as well

as by the difference in thermal expansion between crystalline and glassy phases.

The major properties of the studied ceramics are presented in Table 1.

Young’s Modulus, sonic velocity, and flexural strength of the studied alumina

ceramics tend to increase as the Al2O3 content increases and with a smaller grain


size. For example, the notable difference between Young’s Modulus and sonic

velocity data for the AL98 and AL98.5 ceramics, despite their closeness in Al2O3

content, can be explained by a smaller grain size, a higher densification, a higher

uniformity of microstructure and lower closed porosity for the AL98.5 ceramics.

The AZ ceramics demonstrates the highest value of flexural strength (>500 MPa)

due to the presence of the PSZ phase and fine-crystalline structure.

Hardness depends on the composition and microstructural features and also on

the load used for measuring. As the load is higher, as HV number is lower.

Hardness values tend to increase as the Al2O3 content and the corundum grains

content increase for the studied alumina ceramics. In the case of a higher glassy

phase content, more beneficial conditions for the corundum grain growth may

occur and that may result in a decrease of hardness. As expected, the ceramics

manufactured with a higher content of the starting alumina powder with a lower

particle size and a higher specific surface have more uniform microstructure and,

as a sequence, higher hardness and other physical properties. Also, the ceramics

with a smaller grain size have a narrower standard deviation in hardness. The

maximal hardness values (HV10 greater than 1550 kg/mm2) are observed for the

AL99.6 ceramics and for the AZ ceramics, which have a very uniform

microcrystalline microstructure with practically no glassy phase.

Indentation fracture toughness of the studied alumina ceramics tends to

increase with the Al2O3 content like Young’s Modulus and hardness; however,

this rise is not significant. As expected, the presence of the zirconia phase in the

alumina matrix results in an increase of fracture toughness.

Slip cast alumina ceramics demonstrate higher values of mechanical

properties such as flexural strength, hardness, Young’s Modulus and sonic

velocity, than pressed ceramics. It is explained by a higher level of densification

and uniformity and less stress and fewer defects formed during slip casting and

binder-burn out stages.

Ballistic Performance

The fracturing process of ceramics during impact and penetration at bullet

velocities ranging from 700 to 5000 m/s has several stages, and it includes [3]:

1) initial impact with hydrodynamic flow of penetrator and armor ceramics;

2) breakup and continued flow of penetrator and high speed jetting of debris;

3) ceramic fracture, formation of Hertzian cone cracks, and tensile cracks on the

back face with continued penetrator breakup and flow;

4) erosion of penetrator and widespread fracture of ceramics.

With increasing bullet velocities, the energy transmission through ceramic

armor and across boundaries via shock waves becomes more valuable, i.e. the

ability of ceramics to dissipate the bullet kinetic energy and to prevent the crack

propagation is very important. Energy dissipation during bullet impact and


fracturing of ceramics depends on many factors dealt with ballistic situation

(including initial bullet kinetic energy, bullet material properties, etc.) and

properties and microstructure of ceramics. Regarding ceramics, it should have

some level of properties, which include density and porosity, hardness, fracture

toughness, Young’s Modulus, sonic velocity, mechanical strength, and some

others. Any single property does not have a direct correlation with ballistic

performance because the fracture mechanism during the bullet impact is very

complicated, the crack formation is caused by different stress factors and it occurs

in an extremely short time. In short, the microstructural features affecting physical

and ballistic properties strongly influence crack propagation and energy

dissipation mechanisms and ultimately ballistic performance. Hence, all relevant

properties, as well as ceramic microstructural features, must be carefully

considered in assessment of ballistic performance of protective systems.

For dense homogeneous armor alumina ceramics in order to achieve

acceptable consistent level of ballistic performance Vickers hardness HV10

should exceed 1220-1250 kg/mm2 (i.e. significantly exceed the projectile

hardness). Sonic velocity indicating the ability of hard ceramics to dissipate

energy from the impact area should be greater than 10,000 m/s (preferably,

10,500-11,500 m/s). Young’s Modulus should be greater than 325 GPa (usually

350-450 GPa dependent on the Al2O3 content). The impedance I= c=( E)1/2

[3]

(where is density, c is sonic velocity, E is Young’s Modulus) indicating a wave

propagation in a material should have a level similar to steel (400 MPa.s/m).

Flexural strength should be greater than 220 MPa. Although many authors [1-3,7]

indicate that armor ceramics should have low fracture toughness, it seems that KIc

should not be lower than 3 MPa.m0.5

. Some “balance” between levels of hardness

and fracture toughness needs to be maintained.

There were a number of attempts to describe ballistic performance using

mathematical modeling (e.g. [1-4,7]). All of these had different approaches but

they did not fully describe ballistic performance. However, these models help to

understand better the fracturing mechanism, indicate the important properties

relevant to ballistic performance and allow for a preliminary evaluation of

ballistic performance. For example, Neshpor, et al., 1995 [7] proposed the semi-

phenomenological criterion of evaluation of ability of ceramics to dissipate

ballistic energy using the formula: D = 0.36 (HVcE)/KIc2. The approximate values

of the ballistic energy dissipation criterion have been calculated (Table 1). This

formula and the calculated D-criterion values show that the highest hardness, or

the lowest fracture toughness, is not the dominant factors affecting ballistic

performance. The optimal combination of relevant factors should be considered

for the promising armor ceramics. As an example, the AZ ceramics with high

hardness and elevated fracture toughness (in comparison with alumina ceramics)

demonstrate high ballistic performance.


All the studied ceramics demonstrate a high level of ballistic performance.

The armor systems based on these alumina ceramics bonded with appropriate

aramid-based and fiberglass backing materials are capable of defeating 7.62x51-

mm and 7.62x63-mm AP FMJ ammunition and 7.62x39-mm and 7.62x51-mm

Ball FMJ ammunition, and they provide ballistic protection to Level III or Level

IV dependant on the ceramics and backing material thickness (Level IV in

conjunction with a ballistic vest). The armor systems for personnel protection

have satisfactory multi-hit ballistic performance (up to 6 hits to one body-armor

plate). Trauma for the armor plates for personnel protection made from these

materials occurred at acceptable levels (i.e. not greater than 44-mm deformation

in accordance with NIJ Standards). The alumina ceramics with higher hardness

demonstrated less trauma and bullet intrusion. However, in this case, a greater

degree of a crack growth is observed, probably, due to a higher “ratio” between

hardness and fracture toughness. As mentioned above, a bullet is distorted and

eroded during the initial contact with the ceramics; the erosion of the projectile is

greater as hardness of ceramics increases. As expected, the highest level of bullet

erosion was observed for the hardest ceramics such as alumina-zirconia AZ and

alumina AL99.6 and AL98.5 ceramics that correlates well with the data [6].

Different kinds of cracks are formed during the ballistic impact, which depend

on type of a bullet and properties of ceramics. A locus of conoid coaxial cracks

starts at the impact point; radial tensile cracks are initiated at the back surface

close to the axis of impact. Star cracks are formed at the side of conoids.

Tangential spall cracks occurred due to shear stress waves reflected from the

edges of a tile and due to the formation of the cone cracks; lateral spall cracks

may also form due to longitudinal stress waves reflected from the backing

support. Comminution and erosion of ceramics occur at the cone area. The

thickness of ceramics may also affect crack formation and development. Usually

more conoids occur with greater thickness. The nature and thickness of backing

materials (high-strength aramid-based fabric such as Kevlar, aluminum sheet,

polymer block, or others) may have a significant influence on crack propagation

due to their different abilities to reduce the stress. Fragments of damaged

ceramics with different sizes ranging from big chunks to a fine powder were

observed after fracturing. The chunks with bigger sizes were observed for the AZ

ceramics and for the ceramics with a relatively lower content of crystalline

phases, such as AL97ML and AL98. By contrast, dense boron carbide and silicon

carbide armor ceramics commonly demonstrate explosive shattering at the

shooting. This transforms to a powder at the damaged area with a minimum

amount of large ceramic chunks, which do not remain with the backing material

despite a high level of bonding.

The compaction of the comminuted ceramics under the compression resulting

from impact with a projectile affects penetration resistance. The comminuted and


compacted area of the studied alumina ceramics had relatively small chunks and

agglomerated powder. Microscopic observation showed that the ceramics had

multi-grain fragments ranging from 10 to 100 µm, with micro-cracks and elevated

porosity. In contrast, the uncomminuted area had only macroscopic cracks. The

alumina ceramics with a higher content of a glassy phase such as AL97ML and

AL98 had more chunks; ceramic fragments had more micro-cracks, which

develop mostly through a glassy phase. The fine-crystalline ceramics with an

insignificant amount of a glassy phase had fewer chunks; micro-cracks grew

through the grain boundary and even through grains. Some grains were pulled out

which resulted in elevated porosity. The AZ ceramics had more chunks, but

micro-cracks with a relatively short length grew through the grain boundaries.

Considering armor systems with satisfactory ballistic performance, ceramic

damage at impact should be more conical than cylindrical. At the same time, the

hole caused by a bullet should have a small size. This indicates that the bullet

velocity decreases significantly after the contact with hard ceramics and, hence,

trauma should be minimized. Cracks in the ballistically tested ceramics are

desired to be shorter with small cones. In this case, the residual part of a

ballistically tested ceramic plate will have less damage, and, therefore, a ceramic

plate used for personnel protection has a better probability to resist subsequent

shots. This is true when a ceramic and a backing material are still bonded after

shooting; i.e. the adhesive and the bonding technique are optimized. Alumina

ceramics with different compositions and structure may be used for particular

ballistic applications. Ceramics with a higher alumina and less glassy phase

content and higher hardness are more beneficial for armor tile manufacturing and

for single-hit applications. Ceramics with a higher glassy phase content and lower

hardness values (or, probably more correct, with a lower hardness/fracture

toughness “ratio”), are more suitable for multi-hit ballistic applications despite

possibly demonstrating greater trauma. However, these recommendations are

broad generalizations, and again, all relevant properties, including the ability to

dissipate the bullet energy, must be considered for complete analysis of required

ballistic properties. The curvature of monolithic armor plates may affect the

fracturing, macro-crack propagation, and multi-hit performance.

SUMMARY

The developed and studied alumina ceramics with an Al2O3 content of 97-

99.6-wt.%, as well as the alumina-zirconia ceramics, demonstrate a high level of

physical properties and high ballistic performance. High performance of these

ceramics is achieved by maintaining the proper composition, including the use of

raw materials with optimal parameters, and microstructure, as well as through

optimization of the manufacturing process and quality control points. Properties

affecting ballistic protection and ballistic test results are discussed. A combination


of relevant properties for ballistic protection, including microstructure features

should be considered in the evaluation and selection of ceramics used in armor

applications.

REFERENCES

[1]. C.F. Cline and M.L. Wilkins, “The Importance of Material Properties in

Ceramic Armor”; p.p. 13-18 in DCIC Report 69-1; Part I: Ceramic Armor, 1969.

[2]. Soon-Kil Chung, “Fracture Characterization of Armor Ceramics”, Amer.

Ceram. Soc. Bul., 69 [3] 358-366 (1990).

[3]. D.J. Viechnicki, M.J. Slavin, and M.I. Kliman, “Development and Current

Status of Armor Ceramics”, Amer. Ceram. Soc. Bul., 70 [6] 1035-1039 (1991).

[4]. I.Yu. Kelina and Yu.I. Dobrinskii, “Efficiency of the Use of Silicon

Nitride Ceramics as an Armor Material”, Refractories and Technical Ceramics (in

Russian), [6] 9-12 (1997).

[5]. B. Matchen, “Application of Ceramics in Armor Products”; p.p. 333-342

in Advanced Ceramic Materials; Ed. by Hamid Mostaghasi; Key Engineering

Materials, Vol. 122-124, 1996. Trans. Tech. Publications, Switzerland.

[6]. R.G. O’Donnell, “An Investigation of the Fragmentation Behaviour of

Impacted Ceramics”, J. of Materials Science Letters, [10] 685-688 (1991).

[7]. V.C. Neshpor, G.P. Zaitsev, E.J. Dovgal, et al., “Armour Ceramics

Ballistic Efficiency Evaluation”; p.p. 2395-2401 in Ceramics: Charting the

Future. Proc. 8th

CIMTEC Florence, 28 June-4 July 1994; Ed. by P. Vincenzini,

Techna Srl., 1995.


Table 1. Some physical properties of the studied alumina and alumina-zirconia armor ceramics

Property AL97ML AL98 AL98.5 AL99.6 AZ

Density,* g/cm3

3.74- 3.76 3.78-3.82 3.81-3.84 3.90-3.91 4.35-4.39

Young’s Modulus, GPa 280-300 325-360 370-420 400-450 310-340

Sonic Velocity, km/s 9.5-9.9 10.0-10.5 10.6-11.3 10.7-11.6 9.8-10.0

Vickers Hardness HV10, kg/mm2

1230-1260 1250-1330 1320-1420 1520-1560 1520-1580

Fracture Toughness KIc, MPa.m1/2

3.0-3.3 3.2-3.3 3.3-3.4 3.1-3.4 3.9-4.0

Flexural Strength, MPa - 250-350 270-360 320-380 500-560

Ballistic Energy Dissipation Criterion

Dx10-12

, 1/s (calculated)

1.70-1.95 1.50-1.60 1.80-1.95 2.20-2.40 1.15-1.20

* Water absorption is not greater than 0.02%

These data are performed for the materials manufactured by slip casting and pressing


BALLISTIC PERFORMANCE OF ALUMINA CERAMIC ARMORS

Murat Vural and Zeki Erim B. A. Konduk

Istanbul Technical University Bogazici University

Dept. of Aeronautics Institute of Biomedical Eng.

Maslak-Istanbul 80626 Dept. of Materials

Turkey Bebek-Istanbul 80815

Turkey

A.H. Ucisik

Bogazici University

Institute of Biomedical Eng.

Dept. of Materials

Bebek-Istanbul 80815

Turkey

ABSTRACT

High quality alumina ceramic tiles, backed with semi-infinite aluminum

blocks were ballistically tested with armor piercing 7.62 mm projectiles. The

failure mechanism, ballistic efficiences and fragmentation behavior of ceramics

were investigated under impact loading conditions. The thickness and projectile

velocity were essential. Ballistic efficiency was affected by the thickness and

projectile velocity. Upon impact, radial, cone and lateral cracks form and

disintegrate the ceramic tile. A ceramic conoid zone within the innermost cone

crack interacts with the projectile. In the present study, ballistic efficiencies that

quantify the normalized performance of ceramic against the impacting projectile,

have been found to be a function of the projectile velocity and ceramic tile

thickness. High values of ballistic efficiencies were achieved for thinner ceramic

tiles and for higher impact velocities. These effects of projectile velocity and

ceramic tile thickness on the ballistic efficiency are thought to be extremely

important when making merit ratings between armor ceramics tested at various

velocities or thickness.



INTRODUCTION

Impact, penetration, perforation and trauma effects imposed on materials,

including living materials, are important to many fields of engineering, including

biomedical engineering, orthopedic and traumatology and to the military for

armor application. This interest comes from the desire of both increasing the

penetration capability of projectiles and making protective armor systems

resistant to certain types of threats. Penetration may be defined as the entrance of

a projectile into a target without completing its passage through the body.

Perforation, on the other hand, implies the complete piercing of a target by the

projectile. Upon striking of a projectile, a target can fail by a variety of

mechanisms, depending on a long series of parameters such as impact velocity,

geometry of interacting bodies, and material properties of both the projectile and

the target.

Armors are a means of protection against penetrators. The item to be protected

may be a human body, a vehicle or a fixed building. Body armor, the most

important one, is intended to protect individuals primarily against fragments from

high-explosive artillery shells, grenades, fragmenting mines, as well as projectiles

from small arms and rifles.

The evaluation of ballistic performance is one of the most important issues in

the selection of an armor material. However, the measurement of ballistic

performance has always been a very difficult task because of the destructive

nature of ballistic testing and the many variables involved, such as the type and

velocity of threat, and the types of target and target support.

A method to determine the ballistic performance that found early acceptance,

especially for small arms threats, is the V50 test. In this test, the efficiency of the

tested ceramic tile is determined by the magnitude of the ballistic limit velocity

(VBL), defined as the impact velocity at which 50 percent of the projectiles do not

penetrate the target. The experimental target configuration for V50 testing consists

of bonding a ceramic tile to a backup plate of comparable thickness and shooting

projectiles at these targets. However, as Rosenberg and Yeshurun [l] pointed out,

the V50 test with a thin-backing configuration is not a good test for screening the

ballistic performance of brittle materials.

Another method commonly used to evaluate the ceramics for armor

applications is to fire a reference shot into a thick reference backup target and a

second shot through the candidate ceramic tile which is bonded to the same

backing material, and to compare the residual depths of penetration. The

application of this ballistic testing technique [1-4], i.e. so-called thick backing

technique, has seen the use of various projectile types, backing types and

conditions of lateral confinement. The thick backing technique, originally

introduced by Rosenberg et al. [see Ref. l] and shown schematically in Figure la,


also enables a convenient measure for the ballistic efficiency ( ) of ceramic tiles,

which can be expressed as

= CC

RBB

t

P-P (1)

where B and C are the densities of the backing (aluminum in the present study)

and ceramic respectively, tC is the ceramic thickness, and (PB -PR) is the reduction

in thickness of backing penetrated due to the ceramic tile being in place, i.e. the

difference between the reference depth and the residual depth.

In the present work, ceramic tile thickness and projectile velocity are altered

in order to determine their influence on the ballistic efficiency parameter. A prime

objective is to understand the reliability of the ballistic merit ratings based on

thick backing technique.

EXPERIMENTAL STUDY

The ceramics used included two grades of alumina (prefix AD) and are listed

with their physical and mechanical properties in Table 1. The ceramic tiles were

50 mm square and of six different thicknesses, ranging between 4.1 and 14.7 mm.

These tiles were provided from Kaleporselen A.S., Istanbul, and their listed

properties were taken from the previous study of Birbilen et al. [5].

Table I. Properties of ceramic plates [5].SinteringCeramic

Plate Temp (°C) Time (h)

Density

(gr/cm3)

Hardness

(GPa)3-Point Bending

Streng. (MPa)Compressive

Streng.(MPa)

AD-96 1650 3 3.80 14.5 360 1460

AD-99.8 1680 3 3.90 15.0 400 1600

The ceramic tiles were bonded to clean 6061-TO aluminum alloy backing

blocks using a neoprene-based adhesive. Two configurations of target setup,

referred to as ''thick backing'' method, are schematically shown in Figure 1a and

1b. In all cases, the targets were impacted by a 0.30 calibre (7.62 mm) conical-

nosed armor piercing projectile at velocities ranging from 576 to 803 m/s. A

schematic of the projectile cross section is shown in Figure 2. The projectile has a

total mass of 9.56 0.08 g. This mass contains a 3.5 g hardened steel penetrator

with a conical nose.


Figure 1. Schematics of measured penetration depths; (a) in bare aluminum

blocks, (b) in target panels with "ceramic tiles (thick backing

configuration). The depths were measured by x-ray radiography and/or

depth gauging measurements.

Figure 2. Schematic of the projectile cross section

showing the steel core and jacket.

EXPERIMENTAL RESULTS AND DISCUSSION

Experimental data is presented in Table 2, in the form of reference depth into

the backing plate, residual depth into the backing following perforation of the

ceramic tile, and ballistic efficiency parameter, , as defined by Eqn.1. The

results presented in Table 2 are for successful shots on ceramic tiles smaller than

8.3 mm in thickness; ceramics of greater thickness could not be perforated by the

projectiles in the velocity range investigated. Therefore, ballistic efficiency for

those thicker tiles is not calculated since it is not appropriate. In some cases, a

greater sign (>) is shown just prior to value. These tiles were too thick for the

7.62 mm AP projectile and there was no penetration into the aluminum backing.

Thus, a lower bound on the ballistic limit can be calculated as if the ceramic tile

were just perforated.

Table 2. Penetration data of AD-99.8 and AD-96 ceramic tiles.


Specimen

Code

Projectile

Velocity

(m/s)

Residual

Penetration

(mm)

Reference

Penetration

(mm)

Ballistic

Efficiency

( )

X9914.15/4 795 46 95.3 8.4

X9914.1/2 803 42 96.6 9.5

X9914.1/3 797 33 95.6 10.9

X9914.1/5 788 34 94.2 10.4

X9914.1/6 672 20 75.5 9.6

X9914.1/8 601 21 64.1 7.5

X9916.5/1 802 8 96.4 9.7

X9916.4/2 802 6 96.4 10.0

X9916.4/3 785 5 93.7 9.8

X9916.5/5 694 3 79.1 8.3

X9916.5/6 680 3 76.8 8.1

X9916.45/8 585 1 61.5 6.7

Y9918.0/1 800 2 96.1 8.4

Y9918.0/3 800 4 96.1 8.2

Y9918.2/4 793 4 95.0 7.9

Y9918.1/5 692 1 78.7 6.8

Y9918.0/2 803 5 96.6 8.1

Y9918.1/7 615 0 66.4 >5.8

Y9918.3/8 595 0 63.1 >5.4

Y9918.1/6 693 1 78.9 6.8

X9614.2/7 595 16 63.1 8.0

X9614.2/2 792 33 94.8 10.5

X9614.2/3 803 37 96.6 10.1

X9614.2/6 659 23 73.4 8.5

X9616.0/7 592 8 62.7 6.5

X9616.0/3 804 11 96.7 10.2

X9616.0/1 804 9 96.7 10.4

X9616.0/2 795 6.5 95.3 10.5

X9616.0/4 789 7.5 94.3 10.3

X9616.0/8 577 4 60.2 6.7

X9616.0/6 677 6 76.3 8.3

Y9618.3/3 785 1 93.7 7.9

Y9618.3/4 778 5 92.6 7.5

Y9618.3/6 658 4 73.3 5.9

Y9618.3/2 800 0 96.1 >8.2

Y9618.3/5 678 0 76.5 >6.5

Y9618.3/7 592 0 62.7 >5.4

Y9618.3/8 576 0 60.1 >5.1


Table 2. Penetration data of AD-99.8 and AD-96 ceramic tiles.

Figure 3 shows the variation of ballistic efficiency data in Table 2 against the

projectile velocity for the ceramics of various thicknesses. These data suggest that

ballistic efficiency of a ceramic tile depends on both the impact velocity and the

thickness of ceramic tile used. It is well established by previous workers that the

ballistic efficiency of ceramics is affected by the kind of ceramic used [1] and the

projectile geometry [4]. The results of the present study (see Figure 3) reveal that

the thickness of the ceramic and the velocity of projectile are two new factors that

significantly effect the ballistic efficiency ( ) of ceramics. Increased projectile

velocity or decreased tile thickness results in increased ballistic efficiency. This

result is of considerable importance because it suggests that one must be very

careful when making merit ratings based on thick backing technique for ceramics

of different thickness (which is the case in Ref. [1]) and for ceramics impacted at

different velocities (which is the case in Ref. [4]).

Rosenberg and Yeshurun [1] compare the ballistic efficiencies of different

types of ceramics such as SiC, B4C, TiB2 and Al2O3 by using the thick backing

technique. However, the thickness of the ceramics in their work ranges between 6

and 10 mm. In the light of the results of present study, this type of a comparison is

open to speculation. Findings of the present study suggest that a ceramic tile of 6

mm will show greater ballistic efficiency than the 10 mm tile. Therefore, a

comparison between the types of ceramics without regarding the effect of

thickness will result in unfair merit ratings, which are based on thick backing

technique and Eqn.1.

In a similar manner, Woodward and Baxter [4] investigated the effect of

projectile geometry on the ballistic efficiency of Al2O3 ceramics. They also used

the thick backing technique in their experimental setup. Even though they held the

ceramic tile thickness constant, the projectile velocity ranged between 899 and

1243 m/s in their study. The results of present study show that increased projectile

velocity produces increased ballistic efficiency for Al2O3 ceramics in the range

between 576 and 803 m/s. If this trend holds for the range between 899 and 1243

m/s, some of the comparisons made in the work of Woodward and Baxter [4]

seem to be open to speculation as in the work of Rosenberg and Yeshurun [1].

CONCLUSIONS

Terminal ballistic tests were performed on high quality alumina ceramic tiles

backed with thick aluminum plates, i.e., the so-called thick-backing method. The

two main parameters were the projectile velocity and the thickness of ceramic

tiles. Results clearly show that the ballistic efficiency parameter of ceramics is not

constant and, contrary to common assumption in scientific literature, it

significantly varies as a function of both the tile thickness and the projectile

velocity, at least for the range of velocities between 576 and 803 m/s and

thickness between 4.1 and 8.3 mm. These effects of velocity and thickness on the


ballistic efficiency are thought to be extremely important when making merit

ratings among the armor ceramics tested at various velocities or thickness.

Figure 3. Ballistic efficiencies (defined by Eqn.3.1) of ceramic tiles for varying

projectile velocities: (a) AD-99.8 ceramics, (b) AD-96 ceramics.

REFERENCES1

Rosenberg, Z. and Yeshurun, Y., “The Relation Between Ballistic

Efficiency and Compressive Strength of Ceramic Tiles”, Int. J. Impact Engng.,

Vol. 7, No.3, 357-362, (1988).


2 Rosenberg, Z., Bless, S.J. and Brar, N.S., “On the Influence of the Loss of

Shear Strength on the Ballistic Performance of Brittle Solids”, Int. J. Impact

Engng., Vol.9, No.l, 45-49, (1990). 3 Rosenberg, Z. and Tsaliah, J., “Applying Tate's Model for the Interaction

of Long Rod Projectiles with Ceramic Targets”, Int. J. Impact Engng., Vol.9,

No.2, 247-251, (1990). 4 WOODWARD, R.L. and BAXTER. B.J., “Ballistic Evaluation of

Ceramics: Influence of Test Conditions”, Int. J. Impact Engng., Vol.15, No.2,

119-124, (1994). 5 Birbilen, M., Yildirim, I. and Valenta, L., Introduction of High-Tech

Ceramics into Turkish Industry (in Turkish), in M.L. Ovecoglu & H. Yaparlar

(eds), Proc. 2nd

Int. Ceramic Congress, Istanbul, 24-28 Oct. 1994, Turkish

Ceramic Society Press, Istanbul, Vol.2, 469-474, (1994).


Penetration and Ballistic Testing

AN OVERVIEW OF BALLISTIC TESTING METHODS OF CERAMIC

MATERIALS

Dr. Michael J. Normandia

Armor Mechanics Branch,

Terminal Effects Division,

Weapons & Materials Directorate,

Army Research Laboratory

AMSRL-WM-TA

Aberdeen Proving Ground

APG, MD 21005

Mr. William A. Gooch

Armor Mechanics Branch,

Terminal Effects Division,

Weapons & Materials Directorate,


AMSRL-WM-TA

Aberdeen Proving Ground

APG, MD 21005

ABSTRACT

An overview of impact testing techniques used to characterize or evaluate

engineering structural ceramics for armor applications is presented. The required

brevity of this paper restricts the detail to a listing of the commonly used testing

methods, a categorization of ballistic techniques, and an extensive, but far from

complete, listing of key references appears in alphabetical order, and we

apologize for any omissions. Individual speakers have been invited to this

conference, and they will provide greater detail of the testing techniques, the

evaluation procedures, and of the armor system design methodologies. In

addition, the presentation provides typical testing configurations, typical results,

and test objectives. A similar presentation and companion paper on ceramic

armors by Gooch also provides information on how this data is typically used to

construct armor systems.

A categorization of the testing techniques is provided to classify testing

methods into those that attempt to characterize a particular ceramic material’s

properties, those that attempt to evaluate and rank a ceramic material’s

performance for armor applications, and those techniques that actually evaluate

the ceramic in configurations more representative of armor systems. Finally, we

discuss some of the difficulties in utilizing these testing techniques for ranking

ceramic materials, particularly due to the fairly recent discovery of dwell, which

has had profound effects on data evaluation. Dwell describes the behavior of an

eroding penetrator prior to penetration and in certain circumstances a penetrator

can completely erode on the ceramic surface without penetration.



EDITORIAL NOTEA subset of the presented vuegraphs appears in a separate appendix in this manuscript.

These are presented in a Powerpoint report format beginning on page 131.

INTRODUCTION

During the past quarter-century, ceramics have seen limited use in armor

applications, mostly for lightweight armor applications such as body armor,

helicopter seats, and appliqués on land vehicles for additional threat protection.

Renewed interest and an increase in applications have occurred recently due to the

emphasis on weight reduction. Numerous ballistic testing techniques have been

used to ascertain the effectiveness of ceramics in armor applications, and, in

particular, to rank the performance of the various candidate materials. Typical

impact threats are fragments, which are representative of exploding warheads,

soft-core bullets, hard-core bullets, medium to long-rod, kinetic-energy

penetrators, and high-velocity, shaped-charge jets. Different defeat mechanisms

dominate each of these different threats, and for each threat the dominant ceramic

failure mechanism in a typical armor system is also likely to be different.

The ranking of ceramic (and other brittle) materials is significantly

affected by the test configuration’s geometry. Since armor designers utilize

different armor configurations to exploit different threat defeat mechanisms, the

ranking of ceramic materials is also clearly threat dependent. These are

significant complicating factors, which affect the results of even fundamental

testing methods, as the particular choice of threat and geometric configuration

may affect the ranking. Thus, evaluation of a material for a particular application

may require a test that best represents the mechanisms associated with that

application.

The influence of target, or test configuration geometry, is primarily due to

the material’s very strong pressure dependent strength (in both intact and

damaged states), and the very weak tensile failure strength. Geometry affects the

response to the initial shock, confining pressure, and greatly affects the onset of

tensile failure. Tensile failure may occur for a variety of different reasons, such

as local shear deformation, global ceramic bending, reflected tensile waves from

boundaries or free surfaces, etc. Once failed, the ceramic strength, depends very

strongly on the confinement (pressure) and the fragments can easily be displaced

from the penetrator path, sometimes providing very little resistance to penetration,

if the geometry permits. These effects introduce different time scales into the

testing, such as the time to fracture, the propagation of damage, and the time

response of the containment system, relative to the penetration time. The

outcome of a particular test depends upon when a particular failure mechanism

occurs, in both an absolute and a relative sense, which depends upon the

particular materials used, making comparisons relative rather than absolute. A

change of impact conditions, or material thicknesses may alter the ranking.

These are the most likely reasons why the numerous attempts to correlate

actual armor ballistic performance to fundamental material properties have been


unsuccessful. However, partial success in ranking materials ‘potential’ has been

made using a variety of techniques where a particular class of threats and a

common defeat mechanism were used in the testing technique, and where the

geometry was carefully controlled.

TEST CATEGORIZATION

For simplification (and not for completeness), the types of testing can be

categorized, subjectively, as either phenomenological, armor-material

characterization, or armor-design oriented.

Phenomenological experimental methods attempt to determine or obtain

specific material properties and include shock physics impact data, such as wave

profiles in normal or oblique plate impact. These properties are then utilized to

evaluate a material’s potential performance in an armor system, often utilizing

numerical simulations. They are non-ballistic and in non-armor configurations,

but are important and essential tests, particularly for constructing constitutive

models used in numerical simulations for armor design.

Armor-material characterization experimental methods, attempt to

determine a particular ceramic material’s resistance to penetration (often in an

integrated sense), and include the traditional depth-of-penetration experimental

technique. These tests are dynamic and are sub-divided into non-traditional tests,

with border on phenomenological, and traditional ballistic testing methods, which

are often used to validate numerical simulations. It is unfortunate, that at this time,

these test results are necessary to develop constitutive models, mainly to

determine the strength of the damaged ceramic.

Armor-design oriented experimental methods include the traditional, MIL-

SPEC ballistic limit velocity experiments, to determine a v50, for example. This

category also includes armor design testing methodologies to help isolate an

optimum armor configuration to defeat a specific threat.

A listing of the techniques in each of the three categories appears below,

followed by a brief description of the objectives. The experimental methods can

be (subjectively) categorized as either phenomenological (Table I), armor-

material characterization (Table II), or armor-design oriented (Table III). Ballistic

experimental techniques are bolded and will be discussed in slightly more detail.

Phenomenological Experiments

All of the experimental techniques listed in this category are not ballistic

testing methods, but they are necessary to determine the fundamental material

properties or behavior under shock loading. These properties and behavior are

utilized to develop constitutive models or to validate numerical simulation tools.

The often-attempted goal of ballistic ceramic performance testing methods is to

relate the performance back to these more fundamental characteristics. This has


been partially successful, but not generally recognized or utilized. These tests can

also be considered to be material characterization tests in that they characterize a

particular material property.

Table IA Categorization of Testing Techniques – Phenomenological Experiments

CATEGORIES OF CERAMIC TESTING UTILIZED TO EVALUATE

ARMOR PERFORMANCE OF CERAMICS

Phenomenological Experiments

(1) Pressure-volume

(2) Plate impact (normal, oblique, or multiple impacts-reshock)

(3) Split Pressure (Hopkinson or Kolsky) Bar (Compression)

(4) Bar impact (typically bar impacting bar)

(5) Tensile or Torsion

(6) Quasi-static three- or four-point bending tests

(7) Quasi-static indentation

A combination of these fundamental test results, provides significant

information, which, when utilized with numerical simulations, provides a

realizable-hope that ranking can be obtained with minimal testing. These

methods typically provide information about the best achievable performance.

One of the most promising is the indentation techniques (static or dynamic) to

measure stress-strain relations, dynamic yield strength, and apparent plasticity.

There are a variety of other phenomenological testing methods used that

do not readily fit into any of these categories. However, while this categorization

is likely to be incomplete, most traditional tests are included. Perhaps more

importantly, the variety of categories and testing methods demonstrates the

variety of experimental techniques and data that are typically generated for a

single processed ceramic material. Further complications are due to variations

attributed to the starting powder source and impurities, the batch processing

technique and the particular manufacturer. Additional differences may be due to

non-uniformities within a large processed sample cut up into smaller samples,

surface preparation, ceramic geometric configuration, and geometric testing scale.

Improved testing techniques have been developed that take advantage of

numerical simulation tools to design a test to prevent failure before the

measurement and to identify a particular, desired stress-state. Two examples

were demonstrated by Sandia National Labs to examine impact of confined

cylindrical ceramics at high strain rates using graded impactors (Chhabildas) and

to provide a ramp-loading time history in Split-Hopkinson Pressure Bars at

intermediate strain rates (Forrestal, Frew).


Underlying the numerical simulations are constitutive models, which

describe the material behavior in both a fully intact state, and in a damaged state,

both of which have pressure dependent material strength. The behavior of

ceramics in an intact, damaged or failed state, particularly under high strain-rate

and high-pressure loading, typical of ballistic impact events, has led to

constitutive models that are empirically determined. Typical formulations require

the use of ballistic data to calibrate the model coefficients, in particular, the

criterion for the transition from an intact to a damaged or failed state, and the

strength of the partially or fully-damaged material. In addition, many models

accumulate damage and the partial damage states often degrade particular

material properties (e.g., moduli), for which no data exists. Additional testing

techniques attempt to characterize the failed material in non-ballistic experiments,

such as collapsing cylinders or spherically expanding cavities. The connections

between cavity expansion and penetration have made this class of experiments

very relevant to material performance as well.

Armor-Material Characterization Experiments

This category of testing methods utilizes dynamic impact. These tests are

also phenomenological in nature, but these tests are utilized to directly measure,

or determine from behavioral models, properties characteristic of target resistance

or resistance to penetration. These experiments typically control the geometry of

the test, although variations exist between testing agencies. These tests were

specifically developed to attempt to directly or indirectly evaluate, rank and or

compare ceramic performance for ballistic armor applications, due to the inability

to utilize the phenomenological experimental data for this purpose. The recent

addition of the dwell/penetration transition experiments were added to this

category, even though below some impact load (velocity) there is no penetration.

Table II Categorization of Testing Techniques – Armor-Material

Characterization Experiments

CATEGORIES OF CERAMIC TESTING UTILIZED TO

EVALUATE ARMOR PERFORMANCE OF CERAMICS

Armor-Material Characterization Experiments Ref(CEX) Cavity expansion or cylindrical collapse 11, 52-4

(DAM) Damage Propagation (edge on impact) 31

(IND) Indentation: dynamic or loading and unloading 37, 38, 56

(NDP) Non-deforming penetration (referred to as rigid penetration) 10, 12, 49

(PEN)Semi-infinite penetration vs. velocity time histories 21, 42-4, 57

(DOP)Modified depth-of-penetration experiments (quantifying dwell) See Table IV

(DWE) Complete dwell (for damage onset and for structural response) See Table IV

(DPT) Dwell/penetration transition (concerns about shock mitigation) See Table IV


Non-Traditional Experiments-The following four test techniques are non-standard

techniques used as an extension of the phenomenological models to provide

material properties directly applicable to the prediction of the performance in

armor systems.

CEX: Cavity Expansion or Cylinder Collapse: Cavity expansion models

have been successfully utilized in penetration modeling for brittle materials,

typically geologic materials, such as soils and limestone, and concrete. Extensive

utilization of these models for ceramics has also achieved significant progress.

The basic premise is that the pressure at the penetrator/target interface, which

provides the resistance to penetration, can be computed from integration of the

equations of motion over the entire target. Hence the quantification of the various

damage regions, brittle comminuted, brittle fractured, plastic, and elastic are

necessary to obtain this information. In addition, the theories have been extended

to account for the dynamic evolution of the cavity and the various regions and

account for the presence of finite boundaries as well.

DAM: Damage Evolution Experiments: The consequences of damage can

be described in several aspects. First, for thin tiles that may fail in tension due to

bending, for example, less resistance to penetration is provided, even though the

material is essentially still capable of providing a compressive stress (particularly

under confining pressure). In thicker tiles, or those subject to more lethal threats,

the extensive fracturing and comminution occurs early in the process, hence most

of the penetration occurs in damaged material, which is also a function of

confining pressure. Lastly, the evolution of a damage front has been identified,

and if a threat were to penetrate faster than this damage front, greater resistance to

penetration will be provided. Extensive data on shaped-charge impacts have

quantified an order of magnitude increase, and is presented in the references.

Edge on impacts have attempted to quantify the propagation velocity of the

damage front. Discussions in the literature refer to a damage wave, but that

concept is debatable and the subject of current research. The reference describes a

test technique to monitor the damage front from an edge-on impact.

IND: Indentation: Quasi-static and dynamic indentation experiments

typically are used to measure hardnesses of materials and are representative of the

materials compressive strength, a logical first property to examine when

attempting to rank materials. However, brittle materials resistance to indentation

change when they are cracked or fractured, and typical indenters create localized

shear that can make the measurements difficult to interpret. Hertzian contact has

also been used extensively (Lawn). The additional information provided by the

use of multiple indents is the generation of stress-strain curves, which yield

information about the transition from an elastic behavior to an inelastic one. The

behavior after this transition is equally important in the ceramics ability to resist

penetration. The comparison of the apparent plasticity to the dynamic yield


strength has been used to rank ceramics. These techniques are promising in

providing valuable, fundamental material properties, that potentially can be used

to quantify ceramics (particularly when combined with so-called first-principle

numerical simulation tools.

NDP: Non-Deforming Penetration: Non-deforming or rigid penetration

has been used by numerous researchers to measure the resistance to penetration of

brittle materials. In soil, geologic materials, and concrete, hardened steel

penetrators typically are used to penetrate in a non-deforming or rigid mode. The

benefit of this is that the penetration depth (typically normalized by length,

diameter, or cubed root of mass) is proportional to the ratio of the penetrator

strength to the target strength. Above the impact velocity where the penetrator

deforms (a measure of its dynamic yield strength), this penetration depth is

proportional to the square root of this ratio and approaches this ratio times the

penetrator length at high impact velocities, the often termed, hydrodynamic limit.

Another benefit to using these techniques is that the deformation and flow

characteristics of the penetrator do not affect the results. When the penetrator

deforms, this behavior must be understood to interpret the data obtained. Strain-

rate effects, adiabatic-shear failure, etc., all become necessary to quantify.

Traditional ballistic experimental methods: Typcially, when one thinks of

ballistic experimental methods, one thinks of four test techniques, three of which

are described below, the other is a ballistic limit test. Their description and

history could be, and perhaps should be, the entire focus of this paper. However,

we chose to provide a more complete listing and utilize extensive vuegraphs (a

subset appear in the Appendix) and references to steer the reader to the more

traditional experimental methods. The references provided will detail these

techniques in great detail, and most researchers and engineers within the armor

ballistics community are familiar with these techniques. The more recent

newcomer to this list includes penetrator dwell, which is a pre-penetration phase

that has very significant implications. We discuss these somewhat more

extensively later in this paper.

PEN: Semi-infinite Penetration: The generic ballistic material behavior of

a material is expressed in terms of a penetration vs. impact velocity curve. These

curves can be fit with empirical curves useful for systems-level modeling. Direct

and reverse ballistics techniques have been used extensively. Confinement is

necessary and sometimes affects the results. Nevertheless, a measurement of the

resistance to penetration, the consumption rate of the eroding penetrator, and of

the areal density per unit mass of the ceramic penetrated, are very useful to the

armor designer. An invited speaker details this technique, e.g., see Orphal, this

symposium.


DOP: Depth-of-Penetration: In the mid-1980’s an attempt to standardize

an experimental method to rank ceramics for armor applications led to the

development of this technique. Utilized by numerous researchers, it is typically

the most-often used data, but the relevance of the data generated is highly

debated, basically because the configuration is not representative of most armor

applications. We present detail in our presentation, and an invited speaker will

discuss the issues associated with this common, but not standardized ballistic

testing technique. An international conference was held to discuss this technique

and attempted to standardize it, but generally each country uses its own

techniques, penetrators, target configuration, attention to interfaces, and whether a

cover plate is utilized or not. Reader is referred to James, this symposium.

DWE: Dwell: The discovery of dwell on the ceramic surface prior to

penetration is one of the most significant. Simply stated, the ceramic will resist

penetration, until it fails in some manner due to a number of possible reasons,

after which penetration will commence. George Hauver has gone through great

expense to carefully prevent the ceramic from failing, and was able to completely

erode any penetrator on the ceramic surface with no penetration whatsoever. This

is termed interface defeat, and would be the envy of all armor designers. A great

amount of research is being expended to understand this behavior and its

implications, which we believe are very significant. In fact, the techniques used

to interpret data obtained using traditional ballistic test techniques may need to be

re-examined, due to the fact that penetrator dwell is often present.

DPT: Dwell/Penetration Transition: As the impact velocity is increased, a

level is reached where the compressive strength of the ceramic is exceeded, and

dwell is no longer possible, or any significant duration of dwell. The

experimental technique used to obtain this information has been described in the

last three International Symposium on Ballistics and in the open literature. An

invited speaker will discuss this in more depth; see Lundberg, this symposium.

Armor Design-Oriented Ballistic Experiments

This category of experimental techniques represents armor system applications,

and hence it can be referred to direct evaluation of a material in the particular

application. Therefore, these are the most important results. However, these are

the most difficult test methods to utilize to provide information on how to

maximize the performance of the material. The probabilistic nature of these tests

also requires a large number of tests to be conducted. It is for this reason, that

simpler screening experiments described earlier have been developed. These

experiments are often used to validate numerical and analytic models, but this is


cautioned, as discussed in Normandia at the 45th

US Army Sagamore Conference

on Armor Materials by Design.

Table III Categorization of Testing Techniques – Armor-Design Oriented

(Ballistic) Experiments and Methodologies

CATEGORIES OF CERAMIC TESTING UTILIZED TO EVALUATE

ARMOR PERFORMANCE OF CERAMICS

Armor Design-Oriented (Ballistic) Experiments and Methodologies

(8) (FTG) Fixed geometry (e.g., 1-4-3 thicknesses at 60-deg obliquity)

(9) (TCA) Tandem configurations (MTL/BRL patent)

(10) (VBL) Ballistic Limit Velocity tests (V50 or perforation test data)

(11) (BAD) Behind Armor Debris

(12) (TAD) Minimum Target Areal Density (different for each material combination)

a. (PAD) Protection Areal Density Testing

b. (TTE) Threshold Thickness Experiments

FTG: Fixed Target Geometry: Experimental techniques that attempt to

include the effects of finite thickness or impact obliquity are more representative

of the actual armor geometry and have been used to better compare and rank

ceramic materials for their intended use. Fixed geometry targets are utilized as a

method to rank materials in more-realistic armor environments, while

standardizing the test methods to avoid the probabilistic interpretations required

of the more traditional ballistic limit test methods. A common example is a target

denoted as 1-4-3, where the target consisted of a unit thickness metallic cover

plate, a ceramic of 4 times the unit thickness and a metallic backing plate of 3

times the unit thickness. Targets with this or other finite thicknesses (or weights)

were typically tested at normal impact, but some standardized testing has been

conducted at 60-degrees obliquity.

TCA: Tandem Composite Armor: The difficulty in utilizing ceramics in some

armor applications is the result of accumulative damage effects, and the pressure-

dependent behavior of comminuted and fragmented ceramic. Thus, thicker

ceramic tiles often do not perform as multiples of lesser thicknesses, making it

expensive and less mass efficient to use ceramics to defeat more lethal threats. A

technique to utilize a thinner ceramic tile backed by a metal backplate in a

repeated environment, was developed as a joint effort between the Ballistics

Research Laboratory and the Army Materials Research Laboratory, both now part

of the US Army Research Laboratory. The performance of the armor system

approached the multiplicative performance of each independent system, due to

isolation of each system from each other.


VBL: Ballistic Limit Velocity: v50 or vL tests determine the effectiveness of the

armor system, often in the configuration actually used. Typically, a minimum of 5

experiments are required within a tight velocity range with at least 2 partial or 2

complete penetrations to determine a v50, and a much larger number of

experiments for a v95, typically used as acceptance criteria for actual deliverables.

BAD: Behind Armor Debris: Tests that determine the vulnerability of the

contents behind the armor evaluate the performance in an overmatched situation,

likely to occur in typical ballistic environments. Less debris is desirable for the

armor, more for the penetrator. Tests that quantify this effectiveness have been

developed and used for this purpose. The use of behind armor residual penetrator

and target debris is an often-used evaluation of lethality or vulnerability and is

critical in systems level evaluations.

TAD: Target Areal Density: Protection Areal Density (PAD) and Threshold

Thickness Experiments (TTE) have been developed to understand the general

behavior in ceramics to defeat a particular class of threat. These tests fix impact

conditions and adjust the target material allocation until a penetration threshold

target is obtained. This threshold target delineates the minimum weight for the

given thickness of ceramic that will defeat the threat for the particular material

allocation and impact conditions. Repeating this process for various ceramic

thicknesses generates a threshold curve. Understanding the general behavior of

this curve permits the designer to predict the performance against any threat, with

the caveat that the defeat mechanism has not changed as one changes the threat.

The threshold curve, which separates armor failures and successes (within some

defined probability), shows that there exists an optimum target configuration (a

particular material allocation) that minimizes the target areal density (weight) to

defeat the threat. The general behavior of targets near this optimum are examined

and used predict minimum target areal densities against various threats and

impact conditions. An invited keynote presentation that describes one of these

techniques will illustrate this procedure. Normandia has also presented an

alternative theory based on the relative time to failure of the confinement system

to that of the penetration time. The reader is referred to the keynote speaker,

Adams, this symposium, for discussion on the PAD technique.

While all of the test methods provide useful information, when taken together,

the results better define a material’s capability in a practical armor configuration.

The tests that are the more traditional ballistic experimental techniques appeared

in bold type in Tables II and III, and are re-tabulated in Table IV along with the

common test objectives and information provided. Most of these will be presented

in separate papers in this conference, as will some of the testing methods utilized

in the first two categories, as well as the traditional analytic and numerical tools

and methods used to evaluate the ballistic response.


Table IV Ceramic Material Evaluation Summary of Ballistic Test Methods

TEST TEST TYPE INFORMATION OBTAINED REFNDP Non-Deforming

Penetration

Typically used for soft metals and hard targets, this

applies for concrete, limestone and other geological

materials. Various researches attempt to isolate

target resistance in this penetration mode.

10, 12,

49

PEN Penetration Depth

Direct or Reverse

Impact

Penetration-velocity curves, penetration resistance,

penetration rate, penetrator consumption rate.

21, 42-4,

57

DOP Modified Depth-of-

Penetration

Relevant for determination of performance goals as

a function of ceramic thickness – similar to TAD,

but in a semi-infinite configuration.

3, 4, 13,

18, 25,

39, 48,

50-1,

67-71

DWE Dwell Tests Total interface defeat conditions. 5-7, 9,

16-7, 19,

20,35-6,

41, 46-7,

60

DPT Dwell/Penetration

Transition

Velocity defines a load that is characteristic of a

failure shear strength of the ceramic, or of a

transition strain.

28-30,

32-34

FTG Fixed Target

Geometry

Generic material comparison experiment in armor-

like configurations, particularly at obliquity.

22-3, 72

TCA Tandem Composite

Armor

Configuration to minimize the use of damaged

material.

15,

VBL Ballistic Limit

Velocity (V50) or

Residual Data

Typical requirement for acceptable armor,

individual tests measure residual penetrator

characteristics.

14, 45,

55

BAD Behind Armor

Debris

Used to measure the lethality of the penetrator or

the vulnerability of the target to an overmatched

threat. Data quantification utilized in lethality

assessment tools.

TAD Target Areal

Density

Performance Maps

Helps determine near-optimal armor configurations.

Theories permit extrapolation to different threats.

Includes PAD, TTE, and TSM methodologies.

1, 8, 40

DISCUSSION

When dwell was first discovered, it was obtained under small-scale testing

in a well-controlled laboratory environment. Since that time, the dwell

phenomenon has both fascinated and captivated researcher’s interest over the last

decade, particularly when total interface defeat of a penetrator is achieved

[Hauver, et. al]. Tests that provide information about the dwell phenomenon have


recently been added to the traditional suite of ballistic tests, and in our opinion,

are necessary to fully characterize the performance of ceramics for use in armor

applications, even if the dwell mechanism is not being utilized for threat defeat.

The successful up-scaling of interface defeat for bullet and medium-caliber

applications, albeit not in armor configurations, has led to a proliferation of

worldwide research activities. Subsequently, attempts to model this behavior

using the numerical tools and recent constitutive model formulations has also

accelerated, with limited success. The generation of data demonstrating dwell

and/or total interface defeat of bullets impacting ceramics and experiments to

determine the transition impact velocity between dwell and penetration for long-

rod, kinetic-energy penetrators has demonstrated that dwell is a natural occurring

phenomenon, which must be accounted for in material constitutive behavioral

models. The extensive data generated by Hauver, et al. over the past decade has

not yet been widely disseminated in the open literature and is being compiled into

an ARL report [Rapacki, et al].

Briefly, the possible occurrence of dwell is likely to have been present for

some duration, however slight, during most traditional ballistic testing techniques.

The lack of accurate accounting for the presence or duration of dwell during the

ballistic experiment has clouded the use of this data for constitutive, numerical or

analytic model validation. If dwell were present during the test, which we think

likely in most traditional tests, subtle geometric differences can result in variable

performance. For example, if dwell duration were to vary between tests due to

subtle geometric differences in target construction by several micro-seconds, one

would measure variable penetration resistances, which are likely to show up as

time-dependent or velocity-dependent target resistance. Generally terms that

account for the dwell phenomenon and its duration are lacking in the models used

to deduce target resistance. Without accounting for dwell, an exaggeration of the

variability of the measured target resistance is highly likely, and the performance

effects due to geometric differences are highly pronounced. An example of this is

attempting to compare ceramics using constant finite thickness (or constant areal

density) tiles. If ceramic bending were the cause of ceramic tensile failure,

different loads would produce this failure, and if this failure affected dwell

duration, ceramic comparative performance would be affected.

The reliance on numerical tools to guide armor development requires that

the material constitutive behavior be accurately described. The dependence on

the use of ballistic test results to determine model constants is disturbing.

Potentially worse, in our opinion, is the fact that dwell was likely present in some

the experimental data used to calibrate these model constants. Since the

numerical modeling of ceramics utilizes constitutive model formulations that may

not accommodate dwell, or the numerical methods do not allow dwell to occur for

any length of time, these constants may be determined incorrectly. Since the


strength of the failed material was deduced from the experimental data that may

have had dwell, that value is also likely to be incorrect.

REFERENCES

1 Adams, M. A. and Zwissler, J. G., “An Overview of the Penetration

Mechanics Modeling Program,” Annual Report to U. S. Army Tank-Automotive

Command, September 1993. 2 Anderson Jr., C. E., Morris B. L., and Littlefield, D. L., “A Penetration

Mechanics Database,” SWRI Report 3593/001, January 1992. 3 Anderson, Jr., C. E. and Morris, B. L., “The Ballistic Performance of

Confined Al2O3 Ceramic Tiles,” Int. J. Impact Engng. vol. 12, No.2, pp. 167-187,

1992.4 Bless, S. and Rosenberg, Z., “Hypervelocity Penetration of Ceramics,” Int.

J. Impact Engng., vol. 5, pp. 165-171, 1997. 5 Bless, S. J., Benayami, M., Apgar, L. S., and Eylon, D., “Impenetrable

Ceramic Targets Struck by High Velocity Tungsten Long Rods” Structure Under

Shock and Impact, edited by P. S. Bulson (Computational Mechanics),

Southampton, U. K., 19926 Burkins, M. S., Dehn, J., “Modeling Armor That Uses Interface Defeat,”

Proceedings of the 6th

Annual TARDEC Combat Vehicle Survivability

Symposium, Vol. 1, pp. 443-448, Monterey, CA, March 28-30 (1995a) 7 Chocron Benloulo, I. S. and Sanchez-Galvez, V., “A New Analytical Model

to Simulate Impact onto Ceramic/Composite Armors,” Int. J. Impact Engng, vol.

21, No.6, pp. 461-471, 1998. 8 Contiliano, R. M., McDonough, T. B., Swanson, C. V., A.R.A.P. Report No.

368, Aeronautical Research Associates of Princeton, Inc., 8 1978. 9 Dehn, J., “Modeling Armor that Uses Interface Defeat,” Proc. 6th Annual

TARDEC Combat Vehicle Survivability Symp., 1995. 10

Donaldson, C. D., Contiliano, R. M., Swanson, C. V., “The Qualification of

Target Materials Using The Integral Theory of Impact,” A.R.A.P. Report No.

295, Aeronautical Research Associates of Princeton, Inc., Dec. 1976. 11

Forrestal, M. J., Askari, E. and Longcope, D. B., “A Cavity-Expansion

Penetration Model for Concrete Targets,” Int. J. Solids and Structures, Vol.22

(12), 1485-1500, 1986.


12 Forrestal, M. J., “Penetration into dry porous rock,” Int. J. Impact Engng.,

Vol.16 (5/6), 4127-4146, 1997. 13

Franzen, R. R., Orphal, D. L. and Anderson Jr., C. E., “The Influence of

Experimental Design on Depth-of-Penetration (DOP) Test Results and Derived

Ballistic Efficiencies,” Int. J. Impact Engng, vol. 19, No.8, pp. 727-737, 1997. 14

Gooch, W. A., Hauver, G. E., Netherwood, P. H., and Benck, R., “Dynamic

X-ray imaging of the penetration of boron carbide,” J. Phys. IV France, 10, Pr9,

583-588, 2000. 15

Gooch, W. A., Prifti, J. J., Woolsey, P., Mackiewicz, J. F., and Perciballi,

W. J., “Tandem Ceramic Armor for Defeat of Kinetic Energy Penetrators,” ARL-

TR-1946, 1999 (C) 16

Grace, F., “Elastic Response & Potential Damage Zones in Ceramic Target

Materials for Rod Impact Under Interface Defeat Conditions,” Proc. 13th

Ceramic Modeling Working Group Meeting, Aberdeen Proving Ground, MD.

199717

Grace, F. and Rupert, N. L., “Analysis of Long Rods Impacting Ceramic

Targets at High Velocity,” ARL-TR-1493, Aberdeen Proving Ground, Md., 1997 18

Hauver, G. E., Netherwood, P. H., Benck, R. F., Gooch, W. A., Perciballi,

W. J., and Burkins., M. S., “Variation of Target Resistance During Long Rod

Penetration into Ceramics”, Proceedings of 13th

Int’l Symposium on Ballistics,

Stockholm, Sweden, 1992. 19

Hauver, G. E., Netherwood, P. H., Benck, R. F., and Kecskes, L. J.,

“Ballistic Performance of Ceramic Targets”, Proceedings of Army Symposium

on Solid Mechanics, Plymouth, MA, USA, 1993. 20

Hauver, G. E., Netherwood, P. H., Benck, R. F., and Kecskes, L. J.,

“Enhanced Ballistic Performance of Ceramics,” Proceedings of the 19th

Army

Science Conference, Orlando, FL, 20-24 June 1994. 21

Hohler, V., Stilp, A. J. and Weber, K., “Hypervelocity Penetration of

Tungsten Sintered Alloy Rods into Alumina,” Int. J. Impact Engng, vol. 17, No.4,

pp. 409-418, 1995. 22

Hohler, V., Stilp, A. J. and Weber, K.. “ Ranking Methods of Ceramics and

Experimental Optimization of a Laminated Target With Ceramics,” Proceedings

of Light Weight Armor Systems Symposium, Shrivenham, U.K., 1995 23

Hohler, V., Weber, K., Tham, R., James, B., Barker, A., Pickup, I.,

“Comparative Analysis of Oblique Impact on Ceramic Composite Systems,” To

appear, Int’l. Journal of Impact Symposium, presented at 2000 Hypervelocity

Impact Symposium, Galveston, TX, 2000. 24

Holmquist, T. J, Rajendran, A. M., Templeton, D. W., and Bishnoi, K. D.,

“A Ceramic Armor Materials Database,” TARDEC technical report No. 13754,

Warren, MI Jan. 1999.


25 James, B., “The Influence of the Material Properties on Ballistic

Performance,” Proceedings of 15th

Int’l Symposium on Ballistics, Jerusalem,

Israel, 1995. 26

Johnson, G. R. and Holmquist, T. J., in Shock Waves and High Strain-Rate

Phenomena in Materials, edited by M. A. Meyers, L. E. Murr, and K. P.

Staudhammer (Marcel Dekker, New York), 1992. 27

Kozhushko, A. A., Orphal, D. L., Sinani, A. B., and Franzen, R. R.,

“Possible Detection of Failure Wave Velocity Using Hypervelocity Penetration

Experiments,” Int. J. Impact Engng, vol. 23, pp. 467-475, 1999. 28

LaSalvia, J. C., Howarth, E. J., Rapacki, E. J., Shih, C. J., and Meyers, M.

A., “Microstructural and micromechanical aspects of ceramic/long-rod projectile

interactions: dwell/penetration transitions”, Fundamental Issues and Application

of Shock-Wave and High-Strain-Rate Phenomena, Elsevier Science, 2001, 437, 4. 29

LaSalvia, J. C., and Normandia, M. J., “The Influence of Mechanical

Properties on the Dwell/Penetration Transition,” Proceedings of the 19th

Intl.

Symposium on Ballistics, Interlaken, Switzerland, 2001. 30

LaSalvia, J. C., and Normandia, M. J., and Meyers, M. A., “Theory for the

Influence of Microstructure/Mechanical Properties on the Dwell/Penetration

Transition of a Ductile Long-Rod Penetrator Impacting on a Brittle Solid: Semi-

infinite case”, submitted to J. Appl.Physics, 2001. 31

LeGallic, C., Cauret, M., Tranchet, J. Y., Chartagnac, P., Gil, F., James, B.

J., Pickup, I. M., Milton, A. L., and Carson, W. A. J., “A Consideration of

Damage in the Interaction Between Tungsten Rod Penetrators and Ceramic

Materials,” Proceedings of the 16th

Intl. Symposium on Ballistics, San Francisco,

Ca, 1996. 32

Lundberg, P., Holmberg, L., and Janzon, B. “An Experimental Study of

Long Rod Penetration into Boron Carbide at Ordnance and Hypervelocities”,

Proceedings of the 17th

Intl. Symposium on Ballistics, South Africa, Vol. 3, 251-

265, 1998. 33

Lundberg, P., Renstrom, R., and Holmberg, L., “An Experimental

Investigation of Interface Defeat at Extended Interaction Time”, Proceedings of

the 19th

Intl. Symposium on Ballistics, Interlaken, Switzerland, 2001. 34

Lundberg, P., Renstrom, R., and Lundberg, B., “Impact of metallic

projectiles on ceramic targets: Transition between Interface Defeat and

Penetration”, Int. J. Impact Engineering, 19, 1-13, 1997. 35

Malaise, F., Ph. D. thesis, University of Bordeaux, France, 1999 36

Mellgard, I., Holmberg, L., and Olsson, G. L., “ An Experimental Method to

Compare the Ballistic Efficiencies of Different Ceramics Against Long Rod

Projectiles,” Proceedings of the 11th

Symposium on Ballistics, Brussels, Belgium,

1989.


37 Milman, Yu. V. and Chugunova, “Mechanical Properties, Indentation and

Dynamic Yield Stress of Ceramic Targets,” Int. J. Impact Engng 23, pp.629-638,

1999.38

Milman, Yu. V., Chugunova, and Timofeeva, I. I., “The Resistance of

Silicon Carbide to Static and Impact Local Loading,” presented at HVIS 2000,

Galveston, TX, 2000 to appear in Int. J. Impact Engng. 39

Morris, B. L. and Anderson Jr., C. E., “The Ballistic Performance of

Confined Ceramic Tiles,” Proceedings of the Combat Vehicle Survivability

Conference, Gaithersburg, MD, Vol II, pp. 235-244, 1991. 40

Normandia, M. J., “Temporal Scaling Model”, Proceedings of the Combat

Vehicle Survivability Conference, Gaithersburg, MD, Vol. II, 1991. 41

Normandia, M. J., “Why Study Dwell?,” presented at the 45th

US Army

Materials Conference on Armor Materials by Design, St. Michaels, Md., (to

appear in proceedings), July, 2001. 42

Orphal, D. L., Franzen, R. R., Piekutowski, A. J. and Forrestal, M. J.,

“Penetration of Confined Aluminum Nitride Targets by Tungsten Rods at Impact

Velocities from 1.5 to 4.5 km/s,” Int. J. Impact Engng, Vol. 18, No.4, pp. 355-368,

1996.43

Orphal, D. L. and Franzen, “Penetration of Confined Silicon Carbide

Targets by Tungsten Rods at Impact Velocities from 1.5 to 4.6 km/s,” Int. J.

Impact Engng, Vol. 19, No.1, pp. 1-13, 1997. 44

Orphal, D. L., Franzen, R. R., Charters, A. C., Menna, T. L., and

Piekutowski, A. J., “Penetration of Confined Boron Carbide Targets by Tungsten

Rods at Impact Velocities from 1.5 to 5.0 km/s,” Int. J. Impact Engng, Vol. 19,

No.1, pp. 15-29, 1997. 45

Partom, Y., Anderson Jr., C. E., and Yaziv, “Penetration of AP Projectiles

Into Spaced Ceramic Targets,” Proceedings of the 19th

Intl. Symposium on

Ballistics, Interlaken, Switzerland, 2001. 46

Rapacki, E., et.al. ARL REPORT under preparation (subject is dwell

history) 47

Rapacki, E., Hauver, G. E., Netherwood, P H., and Benck, R. F., “Ceramic

for Armors – A Material System Perspective”, Proceeding of 7th

Annual

TARDEC Ground Vehicle Survivability Symposium, 1996. 48

Reaugh, J. E., Holt, A. C., Wilkins, M. L, Cunningham, B. J., Hord, B. L,

and Kusubov, A. S., “Impact Studies of Five Ceramic Materials and Pyrex,” Int.

J. Impact Engng, 1999. 49

Rosenberg, Z. and Yeshurun, Y., “The Relation Between Ballistic Efficiency

and Compressive Strength of Ceramic Tiles,” Int. J. Impact Engng, vol. 7, pp.

357-362, 1988. 50

Rosenberg, Z., Bless, S., Yeshurun Y., and Okajima, K. “A New Definition

of Ballistic Efficiency of Brittle Materials Based On The Use of Thick-Backing


Plates,” Proceedings of the International Conference on Impact Loading and

Dynamic Behaviour of Materials, Bremen, F.R.G., pp 491-498, 1987. 51

Rosenberg, Z., Yeshurun, Y. and Tsaliah, J. “ More on the Thick-Backing

Technique for Ceramic Tiles Against AP Projectiles,” Proceedings of the 12th

International Symposium on Ballistics, ADPA, San Antonio, Texas, pp. 197-201,

1990.52

Satapathy, S. S. and Bless, S. J. “Calculation of Penetration Resistance of

Brittle Materials using Spherical Cavity Expansion Analysis”, Mechs. of

Materials, Vol. 23, pp.323-330, 1996. 53

Satapathy, S. S. and Bless, S. J. “Cavity Expansion Resistance of Brittle

Materials Obeying a Two-Curve Pressure-Shear Behavior”, J. Appl. Physics,

Vol. 88, 7, p.4004, Oct 2000. 54

Satapathy, S. S., “Dynamic Spherical Cavity Expansion in Brittle

Ceramics”, Int’l J. Solids and Structures, Vol. 38, pp. 5833-5845, 2001. 55

Sherman, D., “Impact Failure Mechanisms in Alumina Tiles on Finite

Thickness Support and the Effect of Confinement,” Int. J. Impact Engng, vol. 24,

No.4, pp. 313-328, 2000. 56

Sternberg, J., “Material Properties Determining the Resistance of Ceramics

to High Velocity Penetration,” J. App. Phys. 65 (9), pp. 3417-3424, May 1989. 57

Subramanian, R. and Bless, S. J., “Penetration of Semi-Infinite AD995

Alumina Targets by Tungsten Long Rod Penetrators from 1.5 to 3.5 km/sec,” Int.

J. Impact Engng, vol. 17, No.4, pp. 807-816, 1995. 58

Tranchet, J. Y., Malaise, F., and Collombet, F., “Modeling of Aluminas

Behavior Through Pyrotechnical Spherical Divergent Tests,” Proc. 13th Ceramic

Modeling Working Group Meeting, Aberdeen Proving Ground, MD. 1997 59

Tranchet, J. Y. and Malaise, F., “Dynamic Confinement of Ceramics:

Ballistic Experiments and Numerical Simulations,” Proc. 14th Ceramic Modeling

Working Group Meeting, IAT, Austin, TX 1999 60

Walker, J.D. Grosch, D., and C.E. Anderson Southwest Research Institute,

2000, unpublished. (subject, photos of dwell during ceramic impact) 61

Wilkins, M. L., Cline, C. F., and Honodel, C. A. “Second Progress Report

of Light Armor Program,” Lawrence Radiation Laboratory (now LLNL), UCRL-

50349 (Revision 1) (1976). 62

Wilkins, M. L., Cline, C. F., and Honodel, C. A. “Third Progress Report of

Light Armor Program,” Lawrence Radiation Laboratory (now LLNL), UCRL-

50460 (1968). 63

Wilkins, M. L., Cline, C. F., and Honodel, C. A., “Fourth Progress Report

of Light Armor Program,” Lawrence Radiation Laboratory (now LLNL), UCRL-

50694 (1969).


64 Wilkins, M. L., Honodel, C. A., and Sawle, D. R., “An Approach to the

Study of Light Armor,” Lawrence Radiation Laboratory (now LLNL), UCRL-

50284 (1967) 65

Wilkins, M. L., Gibbons, J., Hohler, V., Stilp, A. J., and Cozzi, M.,

“Ballistic Performance of AlN, SiC, and Al2O3 Ceramic Tiles Impacted by

Tungsten Alloy Long Rod Projectiles,” Lawrence Radiation Laboratory (now

LLNL), UCRL-50980 (1971) 66

Wilkins, M. L., Landingham R. L., and Honodel, C. A., “Fifth Progress

Report of Light Armor Program,” Proceedings of the Combat Vehicle

Survivability Conference, Gaithersburg, MD, Vol II, 1991. 67

Woodward, R. L. and Baxter, B. J., “Ballistic Evaluation of Ceramics:

Influence of Test Conditions,” Int. J. Impact Engng, vol. 15, pp. 119-124, 1993. 68

Woolsey, P., Kokidko, D., and Mariano, S., “An alternate Test Methodology

for Ballistic Performance Ranking of Armor Ceramics,” MTL TR-89-43. U.S.

Army Materials Technology Laboratory, Watertown, MA, 1989. 69

Woolsey, P., Mariano, S., and Kokidko, D., “Alternate Test Methodology

for Ballistic Performance Ranking of Armor Ceramics,” Annual TACOM Armor

Coordinating Conference, Monterey, Ca. 1989. 70

Woolsey, P., “Ceramic Materials Screening by Residual Penetration

Ballistic Testing,” Proceedings of 13th

Int’l Symposium on Ballistics, Stockholm,

Sweden, 1992. 71

Yaziv, D. and Partom, Y., “The Ballistic Efficiency of Thick Alumina

Targets against Long Rod Penetrators,” Proceedings of Fourteenth International

Symposium on Ballistics, Quebec, Canada, pp. 331-340, Sept. 1993. 72

Yaziv, D., Chocron, S., Anderson Jr., C. E., and Grosch, D. J., “Oblique Penetration in Ceramic Targets,” Proceedings of the 19

th Intl. Symposium on

Ballistics, Interlaken, Switzerland, 2001.


Ballistic Test Techniques(IND) Indentation tests

• Micro-hardness indentations used by Milman to develop stress-strain relations, plasticity and dynamic hardness

• Hertzian indentation techniques used by NIST/Dr. Lawn andARL/Dr. Weresczak generates load/unload curve

• Used to screen and rank ceramics for armor

PushRod

DiamondHertzian

(spherical)Indenter

Specimen

CapacitanceDisplacementGages (3) &

Holder

Loads tohundreds of

pounds

Conducting Tape(i.e., 1/2 of parallel plate capacitor)

Ballistic Test Techniques(NDP) Non-Deforming Penetration tests

• Areal density penetrated DATA as a function of velocity usedsuccessfully to screen and rank armor ceramics

• Resistance to penetration is independent of penetrator strengthIF and WHEN penetration is rigid

• Theoretical expression estimated dynamic hardness• Rigid/deforming penetration models were used to compute

penetration resistance, which were about 1/3 of the theoreticalvalues computed (Sternberg, JAP 1989)

Metal backing

Ceramic

WCCeramicsphere

Heavily ConfinedMetal frame

Pen

Velocity

Vary velocityVary pressure

Analytic SolutionFor Rigid Penetration

Determines Target Resistance

DOP data



Ballistic Test Techniques(CEX) Cavity Expansion tests -SRI

• Expanding SRI cavity experiments funded by ARO in mid 90’s

• Collapsing cylinder experiments conducted by Nesterenko

• Cavity Expansion Theory computes the pressure at the penetrator/targetinterface from integration of static or dynamic stress fields in the target

• Linkage implies CEX tests can be used to directly measure penetrationresistance

HE

ceramictube

taperedbrass pin

Empty

space

foamattenuator

Collapsing tube experiment(variation of experiment by Nesterenko et al)

detonator

copperconfinementtubes

HE insphericalcavity

ceramic cylinder

brassconfinementcylinder

Expanding cavity experiment

Ballistic Test Techniques(PEN) Penetration tests: Direct Ballistics

• Direct penetration of thick, confined ceramics in either direct orreverse ballistic configuration

• Penetration time histories obtained from flash x-rays• Typical penetration vs. impact velocity curves obtained• Los Alamos National Laboratory Phermex Experiments


Penetration TestsReverse Ballistics Configuration

D. L. Orphal and R. R. Franzen, “Penetration of Confined SiliconCarbide Targets by Tungsten Long Rods at Impact VelocitiesFrom 1.5 to 4.6 km/s,” Int. J. Impact Eng., v19 (1997) 1-13.

Tungsten (99.95%)ρ = 19,300kg/m3

L = 15.24mm or11.43mmD = 0.762mm

(PEN) “Long Rod Penetration of Ceramics,”

Dr. Dennis Orphal

Int’l Research Associates, CA

Ballistic Test Techniques(DOP) Depth-of-Penetration tests

• DOP technique developed to evaluate ceramic materials(ceramic effectiveness factors on an areal density basis)

• Data corrected for to nominal impact velocity of 1500 m/s• Initial tests: no cover plate, semi-infinite RHA backing,

bonded interface, W or DU, 65 g L/D 10 penetrators

L/D=1065 g

W or DU

“Depth of Penetration Testing,” Dr. Bryn James, DSTL, United Kingdom“Armor Ceramics Under High Velocity Impact of a Medium Caliber LRP,”Ernst, H., Wiesner, W., and Wolf, T., ISL, France


Figure 2.1.6.1 Target andPenetrator Descriptions forSilicon Carbide DOPExperiments, Franzen et al.

Silicon Carbide

RHAbase target

Tungsten penetratorL = variedD = varied

Pr = residual penetration into RHA base target

tc

Figure 2.1.6.2 Target and PenetratorDescription for DOP Experiment,Rosenberg and Tsaliah [39].

Silicon Carbide(Mat. #118)

RHAbase target

Tungsten Alloypenetratorρ = 17800kg/m3

L = 80mmD = 8mm

12.7mm


Ceramic is bonded to theRHA base target

DOP Test Variations

SiliconCarbide(Mat. #113)

tc

HH RHAUTS = 1.45GPaRc~45

30mm

100mm

20m

m

10mmMild Steel

Tungsten Sinter-Alloy PenetratorV = 1700m/sρ = 17600kg/m3

UTS = 1.2 GPaelongation = 10%L = 72.5 mmD = 5.8 mmL/D = 12.5

1.5mm rubber

Pr= residualpenetrationinto HH RHA

Ceramic is gluedto RHA base

Target is square

Figure 2.1.6.4 Target and Penetrator Descriptionfor DOP Experiment, Rosenberg et al. .

Figure 2.1.6.3 Target and Penetrator Description for DOP Experiment, Reaugh et al.

Silicon Carbide (Mat. #110)

4340 Steel Base TargetRc = 33-37

Tungsten Sintered Alloy W2 Penetratorρ = 18360kg/m3 Rc = 28-31L = 25.4mm UTS = 0.88GPaD = 6.35mm Yield = 0.695GPaL/D = 4 Elongation to fracture = 5.5%

tc


Ceramic is bonded to theRHA base target using Stycast 1266

102 mm

152mm

64mm

V

T. Holmquist, R. Rajendran, D. Templeton, &, K. Bishnoi,“A Ceramic Armor Materials Database,” TARDEC report13754, Warren, MI, Jan. 1999

• George Hauver discovery began by determining the change intarget strength as a function of time

• Achieved interface defeat under strong confinement• Extensive recent activity has significant implications

Ballistic Test Techniques(DWE) Complete Dwell tests

TiC


Dynamic Confinement TestsSabot Preloads Ceramic

Tranchet and Malaise(Centre d’Etudes de Gramat and ENSAM/LAMEF, Talence, France)CMWG mtg. 8-10, 1999 (courtesy of Dr. Bless)

Ballistic Test Techniques(DPT) Dwell/Penetration Transition tests

• Similar to other reverse ballistic experiments in design,however, controlled heat-shrunk confinement and frontplate shock-mitigation was used

• Penetration-time histories established dwell occurrence andtransition to penetration

Rear Plug

Front Plug

Locking Rings

Tube

12φ

20 8

Ceramic

2028

Rear Plug

Front Plug

Locking Rings

Tube

12φ

20 8

Ceramic

2028Dimensions in millimetersTempered steel front and rear plugs

(750 MPa flow stress).Maraging steel tube and locking rings

(Mar 350, 2600 MPa flow stress).

P. Lundberg, R. Renstrom, & B. Lundberg, “Impact of Metallic Projectiles onCeramic Targets: Transition Between Interface Defeat and Penetration,” Int. J.Impact Eng., v24 (2000) 259-275.

P. Lundberg, R. Renstrom, & L. Holmberg, “An Experimental Investigation onInterface Defeat at Extended Interaction Time,” Proceedings of the 19th

International Symposium on Ballistics, edited by I. Rose Crewther, v3, 1463-1469.

(DWE) “An Analysis of the Transition between Interface Defeat and Penetration forA Given Combination of Projectile and Ceramic Material,” P. Lundberg, R.Renstrom, L. Westerling, Swedish Defense Research Agency, Sweden


Ballistic Test Techniques(FTG) Fixed Target Geometry tests

• ARL 1-4-3 tests in 1980’s– Determined standard limit velocity VL

– Equation estimates zero residual penetration velocity– 3 or 4 tests for each ceramic at normal obliquity

• EMI/DERA tests at obliquity– Determined optimum material allocation and weights

Ceramic25.4 mm(nominal)

Metal backing19-mm

High Hard Steel(or ESR or RHA)

Metal cover6.25 mm RHA

19-mmMild Steel

Confinement

Metal backing

Ceramic Tilesat obliquity

Penetrator

V. Hohler, K. Weber, R. Tham, B. James, A.Barker, &, I. Pickup, “Comparative Analysis ofOblique Impact on Composite Systems,” Int. J.Impact Eng., to appear, HVIS 2000proceedings.

ARL Standardized Ceramic Target: ATM-C


AD995 Ceramic Disk101.6mm diameter

12.7mm thick

Ti-6Al-4V Cover Plate127mm diameter

6.35mm thick

Ti-6Al-4V Confinement Ring

Ti-6Al-4V Cover Plate127mm diameter

6.35mm thick

Standardized Research TargetInitial Encapsulated Ceramic Configuration

Electron-beam weldedin vacuum

Representative,IdealizedConfiguration

ARL/NIST/SNL/LANL effort

Ballistic Test Techniques(TCA) Tandem Composite Armor tests

• BRL/MTL Damage Mitigation Configuration• Design approach isolates ceramic material to repeatedly achieve

material performance potential– ad hoc, intelligent solution that capitalized on empirical

testing observations– Trades off space for performance– Good performance levels achieved

• Component nature of design provides design flexibility

Metal backing

Ceramic honeycomb

Test bed for materials, models, and design optimization


Ballistic Test Techniques(VBL) Ballistic limit velocity tests

• Probably the most extensively used data technique• Statistical data analysis required multiple tests• Higher confidence levels requires significantly greater number

of experiments• Used for acceptance testing and is standardized• AMTL Mascianica Handbook available in digitized form

Silicon Carbide(Mat. #105)

6061-T6 Aluminum Back Plate

6.35mm

6.35mm

Allegheny Steel 609 Sharp PenetratorRc = 54-56M = 8.32gL ~ 29 mmD = 7.62 mmcone angle = 55o

Target Configuration

Figure 2.1.7.1 Target and Penetrator Description forPerforation Experiment, Wilkins et al. [26].

Bonded using polyurethane(Scotchcast 221)

Ballistic Test Techniques(TAD) Target Areal Density tests

• Systematic ballistictesting was used todevelop armor designmethodologies during the1980’s– JPL (figure shown)– DARPA contractors

• Honeywell (Alliant)• DuPont and GD teams• A.R.A.P./Abex/Norton

– LANL/LLNL

(TAD) “Theory and Experimental Test Methodsfor Evaluating Ceramic Armor Components,”Dr. Marc Adams, Jet Propulsion Laboratory, CA

THEORY AND EXPERIMENTAL TEST METHODS FOR EVALUATING

CERAMIC ARMOR COMPONENTS

Marc A. Adams

Jet Propulsion Laboratory

4800 Oak Grove Drive, Bldg 122-B3

Pasadena, CA 91109

ABSTRACT

The Ballistic Performance Map (BPM) and its derivatives are useful

constructs for understanding the function and performance of ceramic and other

components in hardfaced armor systems. By combining the BPM model with a

constant velocity ballistic testing approach and varying the areal densities of the

target components, the Protection Areal Density (PAD) testing methodology, the

relative and absolute performance of different ceramic materials are readily

evaluated. In addition, such methods enable adequate statistical analysis of the

unbiased test results to understand the basic uncertainty in the measured

performance and the stochastic behavior of any hardfaced armor material or

armor system. Examples of the ballistic performance of several classes of

hardface materials are given.

INTRODUCTION

This paper discusses the results of studies, sponsored by the U.S. Army

TACOM, that were conducted in the late 1980's and early 1990's. These studies

developed new methods for the ballistic evaluation of candidate armor materials

and evaluated the ballistic performance of a variety of ceramic materials. Before

initiating the experimental evaluation program, the various ballistic test methods

being used at the time for evaluation of armor materials and components were

critically examined. None were found to be adequate for the statistically

meaningful characterization of ceramic material performance. Often, large

uncertainties in the measured performance values were ignored and bias in the

performance measures, introduced by the testing methods, was not being

adequately addressed.



A new method for the ballistic evaluation of materials, components and armor

systems was developed. In addition to developing the experimental procedures

and statistical experiment design philosophies, analysis methods were developed

that enable the statistical interpretation of experimental test results and

identification of uncertainties associated with the measured values. This testing

method is called PAD, the determination of the protection areal density of a

material, component or armor system. The method fixes the test projectile

characteristics, e.g. projectile type, impact velocity, obliquity, yaw, and varies the

areal density of the targets used in the test series. The fundamental determination

made from the ballistic test data is the probability of partial or complete

penetration of a target of given design and areal density. The data is analyzed

with binomial statistical procedures or by maximum likelihood estimation

techniques.

The ceramic material evaluation studies characterized the performance of

ceramic and ceramic composite materials for use as components in armor systems

to protect against "small arms, kinetic energy threats". Armor piercing (AP)

projectiles with hard penetrator cores were used as the test projectiles. The .50cal

AP M2 projectile was used for the majority of the studies, although some

evaluations were also performed with .30cal and 14.5 mm AP projectiles. Over a

period of several years, ceramics companies, ceramic developers at universities

and government laboratories submitted specimens of various types of ceramic and

ceramic composite materials that were evaluated using the methods described in

this paper. The results of some of these evaluations are presented below.

USE OF CERAMICS IN ARMOR SYSTEMS

Ceramic containing armor systems typically have configurations similar to

that shown in Figure 1. The ceramic or cermet (hardface) component is usually

one of the first armor system components impacted by the projectile. Situated

behind the hardface component are one or more backing components that provide

support to the brittle hardface plate and affect the final defeat of the damaged

projectile and the ceramic debris. The shape and dimensions of the hardface

material vary from one armor system to another but the basic function of the

hardface remains that of damaging (cracking, shattering, eroding) the incident

projectile and turning or yawing the projectile from its incident trajectory.

Other features are often incorporated into ceramic armor systems. Typically,

a spall shield component is placed in front of the ceramic to suppress the ceramic

debris thrown off the front face during projectile impact. The ceramic is attached

to the backing component with adhesive or by other means. Many armor systems

have a requirement to defeat multiple hits of the threat projectile, some hits in

close proximity. Often, the individual plates of ceramic are structurally isolated


with material in the area between the plates to prevent adjacent ceramic plates

from being damaged by a hit. Additional dynamic isolation may be required

Ceramic Plates

Metallic or Polymer Composite Backing Plate(s)

Bond of Ceramic to Backing Plate

Material isolates ceramic plates

Thin Cover Plate

Figure 1. Typical Configuration of a Ceramic Armor System

between the ceramic plate and the backing components. Alternately, "tough"

ceramics and cermets have been investigated for their ability to defeat impacts of

projectiles and limit the lateral damage created in the ceramic such that

subsequent, proximate hits of the threat can be defeated by a single continuous

ceramic plate instead of the "tiled" array of plates shown in Figure 1.

UNDERSTANDING THE PERFORMANCE OF CERAMICS IN ARMOR

SYSTEMS

In general, the performance of a hardface material in an armor system cannot

be predicted from the intrinsic properties of the hardface material at the present

time. Few static properties correlate with the ballistic performance of ceramic

materials. Comprehensive physical models of the penetration event, which use

intrinsic material properties and closed form physical descriptions, are not capable

of accurately predicting the marginal conditions under which a given projectile

will completely penetrate the armor. Measurement of relevant dynamic material

properties sheds some light on the suitability of a ceramic material but the

relationship of these properties to actual performance in an engineered armor

system cannot be relied upon, at present, for design purposes. Some success has

been obtained in efforts to use discretized, finite element/finite difference

modeling or "hydrocode" modeling but these analytical tools are not adequate to

design armor systems or to adequately predict material performance. They simply

don't describe, with sufficient accuracy, all of the important phenomena that affect

performance. The basic dynamic material behavior, under the conditions of

projectile impact and penetration, is still imperfectly understood.

In order to accurately characterize the relative or absolute performance of

ceramic materials for use in armor systems, it remains necessary to evaluate the

materials by ballistic testing of the armor system or testing a target that is a good


surrogate of the system. All important phenomena that occur in the armor system

must occur during the projectile penetration of the target used for the evaluation;

the target design must insure this. Such testing is more akin to component

evaluation than to the determination of intrinsic material properties. The design of

the target influences the absolute performance level of the ceramic component

and affects any performance comparisons with targets using other materials.

EVALUATION OF THE BALLISTIC PERFORMANCE OF CERAMICS

All of the design features described above affect the ballistic performance of

the hardface armor component. Depending on the particular design of the armor

system, the hardface component may be more or less effective in damaging the

projectile and contributing to its defeat. The development of any efficient armor

system requires the complex, co-optimization of several design parameters and

materials selections. This fact complicates attempts to evaluate the relative

performance of ceramic materials tested in different target designs and

complicates evaluation for different armor system designs. A fixed target design

and test projectile should be used for comparative ballistic evaluation of hardface

materials. This target design must faithfully create the same penetration

phenomena as the armor system for which the ceramics are being evaluated.

Given that ceramic materials are best evaluated as "components" in a system,

the most unambiguous performance measure is whether the target used for the

evaluation, is partially penetrated or completely penetrated by the projectile.

Binomial data is taken in such tests. The target of a given configuration is either

partially penetrated or completely penetrated by a given projectile impacting at a

given velocity and obliquity. Other diagnostics can be used in these tests such as

capture of the damaged projectile to determine the level of breakup and

measurement of the deformation produced in the backing plate of partially

penetrated targets.

Ballistic testing approaches typically vary one of two principal variables in the

course of the experimental determination. The target design can be fixed and the

impacting velocity of the projectile varied. This testing is called ballistic limit

determination or the determination of "V50", by definition the velocity at which

there is a 0.5 probability that the target will be completely penetrated. The other

approach is to fix the impact velocity of the chosen test projectile and vary the

areal density (thickness) of the target components. The probability of partial

penetration as a function of target areal density is determined. This is the PAD

method of ballistic performance characterization.

The requirements for most new armor applications define the threat

projectile(s) and impact velocity(s). The program goals are usually to find the

lightest weight armor system; minimizing armor system areal density is of

greatest interest. Curiously, much of the ballistic testing conducted in these


programs employ the variable velocity approach and determine V50 quantities.

The relationships between target design, target areal density, impact velocity and

defeat of the projectile are complex. It is best to measure, directly, the principal

variable of interest and hold all other variables as constant as possible.

Figure 2 illustrates the nature of the two different test methods and the

features of the Ballistic Performance Map. On the base of the three axes plot are

the areal densities of ceramic and backing in the target. The vertical axis

represents the impact velocity of the projectile. The ballistic limit surface, shown

with constant velocity contours, represents the locus of Target Design Points

(ceramic and backing areal density) that will defeat the projectile (prevent

complete penetration of the target) at the given impact velocity 50% of the time.

Target tests used for the PAD test method and for the variable velocity test

method are shown in the cutaway area. The stochastic behavior of the complete

penetration event can be visualized as a "thickness" of the ballistic limit surface.

As one moves upward through the surface from below, the probability of

complete target penetration increases from near zero toward unity. The Ballistic

Performance Map is the base of this figure. The PAD line for an impact velocity

is the projection of that constant velocity contour of the ballistic limit surface onto

the base of the figure.

Figure 3 illustrates the design of PAD tests and the methods used to analyze

the ballistic test data. Appropriate Test Line(s) are chosen and the test lines are

populated with targets at selected Target Design Points. The density of Target

Design Points on the Test Line is a balance between the number of targets that

can be used and the degree of accuracy and statistical confidence required for the

determination of the partial penetration probability along the Test Line. Binomial

statistical analysis can be used to establish the uncertainty interval for the

measured value of partial penetration probability at each target design point.

Alternately, maximum likelihood estimation techniques can be used to analyze

the data on the Test Line to obtain the partial penetration probability (with its

uncertainty) as a function of target areal density on the Test Line. These test and

analysis methods were used to obtain the results presented in this paper.


Figure 2. Ballistic Limit Surface and Ballistic Performance Map


THE BALLISTIC PERFORMANCE OF VARIOUS CERAMIC MATERIALS

The target chosen for the ceramic evaluation program was a 4x4 inch plate of

ceramic bonded with solid film adhesive to an octagonal shaped backing plate

made from 5083- H131 aluminum alloy. If the ceramic test plate was not flat to

within 0.003 (in), the minimum required thickness of polyester resin was cast onto

the surface to make it flat and eliminate the possibility of voids in the bond line.

Material suppliers were requested to provide each of their ceramic plates in a

thickness that gave each plate an areal density of 11 (lb/ft2). The testing was

performed along a constant ceramic areal density Test Line. The backing areal

density was varied over a range that produced partial and complete penetrations in

the targets. The tests of the highest areal density targets with complete

penetration and the tests of the lowest areal density targets with partial penetration

were replicated as many times as possible to increase the statistical resolution and

decrease the uncertainty in the PAD determination. Measurements of the

permanent deformation of the aluminum backing plates were made for all

partially penetrated targets. Also, the degree of projectile core breakup was

measure for all tests in which the front and rear target containment boxes

adequately captured the pieces of the damaged projectile.

Figure 3. Design and Analysis of PAD Tests


Maximum likelihood estimation (MLE) techniques were used to analyze the

data set of partial and complete penetrations obtained for each material. MLE

was used to determine: (i) the expected value of target areal density for a

probability of partial penetration (Pp) equal to 0.5 and (ii) the 90% confidence

interval of areal density for Pp = 0.5. Some materials have a large uncertainty

(confidence interval) associated with their expected PAD0.5 values. This arose

from two sources. In some cases, insufficient material was supplied to perform an

adequate number of tests to reduce the statistical uncertainty. In other cases,

sufficient material was supplied to test an adequate number of specimens but the

material's behavior was not consistent. These two cases cannot be distinguished

in the data presented. In general, poorer performing materials (ones that required

higher target areal density to defeat the projectile) also demonstrated more

inconsistency in their performance and had larger uncertainty bands.

In all, thirty-six materials were evaluated including: aluminum oxide, boron

carbide, boron carbide/aluminum cermet, silicon carbide and aluminum nitride.

The materials were provided by nine different organizations. As indicated on the

figures presenting the analyzed data, both hot pressed (hp) and sintered (s)

materials were provided.

Figures 4 through 7 summarize the results of the experimental investigations.

These results are plotted on Ceramic Performance Maps that display target areal

density as a function of ceramic areal density. Ceramic materials that have their

PAD points at lower target areal densities are higher performing ceramics.

Figure 4 presents the results for the five sintered aluminum oxide ceramics

evaluated at approximately 11 (lb/ft2). The "AD995 Al2O3 (s)" material is Coors

sintered, CAP3 alumina that was used extensively in the program to study the

effects of many variables on the performance of hardface components in armor

systems. It is one of the two baseline materials for the study and its performance

has been characterized in every area of the Ballistic Performance Map for four

different velocities. In each plot summarizing the performance data for a

particular class of material, there is a heavy orange line represented the expected

PAD0.5 value for the AD995 Al2O3 (s) ceramic and two thinner orange lines

representing the upper and lower bounds of the 90% confidence interval for the

PAD0.5 value. Similar lines are presented in blue on each plot for the other

baseline material, a hot pressed boron carbide made by the Dow Chemical

Company.

The aluminas [H] and [V] have performance indistinguishable from the

baseline alumina. Material [B] most likely has a PAD0.5 slightly less than the

baseline alumina. Material [A], a lower grade alumina, demonstrated inferior

performance to the other aluminas. Even this compositionally inferior alumina

requires less than 10% higher areal density to equal the performance of the better

aluminas.


Figure 4. Expected Value and 90% Confidence Interval for

the PAD (Pp = 0.5) Values for Aluminum Oxide Ceramics

[B]

Al 2

O3(s

) [V]

Al 2

O3(s

)

[H]

Al 2

O3(s

)

[A] Al2O3(s)

AD995 Al2O3(s)

Baseline

80% confidence

AD 995

Al2O3

Dow B4C

15

16

17

18

19

20

21

10.7 10.8 10.9 11 11.1 11.2

Ceramic Areal Density (lb/ft^2)

Tar

get

Are

al D

ensi

ty (

lb/f

t^2)

Figure 5 summarizes the results of the evaluation of boron carbide containing

materials, two pure hot pressed boron carbides and one boron carbide/aluminum

cermet material. The two hot pressed materials have indistinguishable

performance that is about two lb/ft2 less than the baseline alumina. This

represents the performance difference between the best composition ballistic

ceramic for AP threats, B4C, and the lowest performing composition of ballistic

ceramic material, Al2O3. Several varieties of the B4C/Al cermet material were

evaluated. The one shown in Figure 5 was the best performer and, importantly,

was the most "ceramic like" in composition and microstructure of all the varieties

evaluated. It was found that, as the ductile metallic phases present in the cermet

body were increased, the toughness and lateral damage resistance increased;

however, the areal density required to defeat the threat, PAD0.5, increased

dramatically.

Figure 6 summarizes the results of the evaluation of six sintered and two hot

pressed aluminum nitride ceramics. No performance difference was observed


Figure 5. Expected Value and 90% Confidence Interval

of PAD (Pp = 0.5) Values for Boron Carbide Ceramics

and Cermet

Dow

B4C

(hp)

[M] B4C/Alcermet

[S]

B4C

(h

p)

80% confidence

AD 995

Al2O3

Dow B4C (hp)

13

14

15

16

17

18

19

10.2 10.4 10.6 10.8 11 11.2


Tar

get

Are

al D

ensi

ty (

lb/f

t^2)

between the hot press and sintered materials. All materials have PAD0.5 values 1

to 1.5 lb/ft2 lower than the baseline alumina. One hot pressed material, [Q],

fabricated with "improved" techniques, had such inconsistent performance that

statistical analysis could say little about its performance other than it was

extremely inconsistent. Several of the sintered materials represent compositions,

microstructures and processing that were painstakingly developed over a

considerable period of time to be "superior ballistic materials". All this

development was based on static property measurements. None of these materials

are better than [C], the cheapest material with the simplest processing and a larger

grain size. It is not easy to change the composition, microstructure or processing

of a ceramic body and improve its ballistic performance. Ballistic testing during

material development is absolutely required to guide the development.

Figure 7 summarizes the results of the evaluation of six sintered and three hot

pressed silicon carbide ceramics. The performance differences between all but

one of the materials are small. Inspection of the expected values of PAD0.5, the


Figure 6. Expected Value and 90% Confidence Interval for the PAD

(Pp = 0.5) Values of Various Aluminum Nitride Ceramics

[F]

AlN

(s)

[X]

AlN

(s)

[P]

AlN

(s)

[N]

AlN

(s)

[Y] AlN(s) 70%

confidence

80% confidence

[L] AlN(hp)

AD 995 Al2O3

Dow B4C

[C]

AlN

(s)

80% confidence

[Q]

AlN

(hp)

14

15

16

17

18

10.6 11 11.4 11.8 12.2


Tar

get

Are

al D

ensi

ty (

lb/f

t^2)

green data points, shows a general drift to higher target areal densities with

increasing areal density of the ceramic. This is an effect of target design and

points out the problem of comparing ceramic performance between ceramics


Figure 7. Expected Value and 90% Confidence Interval for the

PAD (Pp = 0.5) Values for Various Silicon Carbide Ceramics

[O]

SiC

(s) [W

] S

iC(s

)

[G]

SiC

(hp)

80% confidence

[D]

SiC

(s)

[K]

SiC

(hp)

70% confidence

[R] SiC(s)

AD 995

Al2O3

Dow B4C

[I]

SiC

(s) [U

] S

iC(h

p)

[Z]

SiC

(s)

70% Confidence

14

15

16

17

18

19

20

21

10 10.5 11 11.5 12 12.5 13 13.5


Tar

get

Are

al D

ensi

ty (

lb/f

t^2)

evaluated in different designs. As the areal density of the ceramic is increased in

the target, the areal density of the backing decreases to maintain a constant target

areal density. The weight fraction of ceramic in the target increases, reducing the

weight efficiency of the target, irrespective of the ceramic performance.

Translation of the blue baseline performance line up or the orange baseline down

shows that the increase in target areal density is due to the target design factor and

not the fact that the higher areal density ceramics have lower performance.


LONG ROD PENETRATION OF CERAMICS

D. L. Orphal

International Research Associates

4450 Black Ave.

Pleasanton, CA 94566

ABSTRACT

Reverse ballistic experiments were performed to measure penetration of long

tungsten rods into confined boron carbide, silicon carbide, and aluminum nitride

targets. Penetration depth and the length of the remaining rod were measured as

functions of time using flash X-rays. These data were used to determine the

velocity of penetration and the rate of rod erosion. Impact velocities ranged from

about 1.5 to 5 km/s. The experiments were performed using a two-stage light-gas

gun.

INTRODUCTION

Hohler and Stilp [1,2], Sorensen, et al. [3], and others have published data for

the penetration of long rods into steel and aluminum as a function of impact

velocity. The principle objective of this paper is to present similar data for

penetration of long tungsten rods into three confined ceramics; boron carbide

(B4C), silicon carbide (SiC) and aluminum nitride (AlN). The experiments were

performed in the reverse ballistic mode with multiple flash X-rays of the

penetration process. This approach results in data for both penetration depth and

length of eroded rod as a function of time. These data are used to determine the

velocity of penetration and rate of rod erosion, in addition to final penetration

depth. Impact velocities ranged from about 1.5 to 5 km/s.

A secondary objective is to briefly discuss the advantages and disadvantages of

reverse ballistic testing. Reverse ballistic testing is not a new technique but is

probably not used as often as would be beneficial.

EXPERIMENT DESIGN

These ceramic penetration experiments were designed to achieve several

objectives. Impact velocities were to range from about 1.5 to 5 km/s. Penetration

depth, p, and remaining rod length, Lr, were to be measured as a function of time,



allowing calculation of the velocity of penetration, u = dp/dt and the

“consumption velocity, vc = dLc/dt, where Lc = “consumed rod length” = L-Lr and

L = original rod length.

Given these objectives, it was decided to conduct the experiments in the

reverse ballistic mode. In direct ballistic experiments, the projectile, here a long

tungsten rod, is launched and impacts a stationary target. In reverse ballistics the

“target”, here a confined ceramic, is launched and impacts a stationary projectile.

In reverse ballistic experiments the size of the target is restricted by the size of the

gun and is thus necessarily small. The size of the projectile must be

correspondingly small. Thus reverse ballistics experiments are nearly always

small scale and the issue of scaling is very important.

A big advantage of small-scale reverse ballistics experiments is that, properly

designed, flash X-rays can be used to view the penetrator inside the target during

the penetration process. Multiple, independently timed flash X-rays provide

measurements of penetration depth and remaining penetrator length at known

times. In addition, the X-rays provide data relevant to the overall phenomenology

of the penetrator-target interaction, target hole size and growth, and the spatial

distribution of the eroded penetrator material. All these data are very difficult to

obtain in large-scale direct ballistics tests where X-rays cannot penetrate the target

and data is usually limited to final penetration depth and target hole geometry. An

excellent recent example of the use of reverse ballistic testing to study details of

the penetration process, specifically interface defeat by ceramics, is the work by

Lundberg, et al. [4] and Westerling, et al. [5].

Reverse ballistic testing has several other advantages. Penetration by very

complex penetrators, which would be very difficult to launch directly, can be

studied. Also, in reverse ballistic testing parameters such as angle of attack can be

precisely controlled. In the experiments reported here angle of attack was zero.

Target Geometry

A disadvantage of reverse ballistic testing is that because of the small scale and

the requirement to launch the target, target complexity is limited. In these tests

this limitation was not an issue. The ceramic targets were simple cylinders. The

diameter of the ceramic was selected to be as large as possible within the

limitations of the diameter and launch mass capability of the two-stage light-gas

gun. The targets for each of the three ceramics were basically the same, but

dimensions varied somewhat because of the different ceramic densities. Details of

the targets for each of the ceramics are reported in [6-8]. Here the B4C targets

used for v < 4.2 km/s are shown in Figure 1 to illustrate the essential features of

the target design. The longer target was used for lower velocity tests (1.5 v

2.77 km/s). The targets are composed of a ceramic cylinder radially confined by a

thin titanium sleeve and with 6061-T6 aluminum front and rear plates. Each


titanium sleeve was machined to achieve a tight press fit with its matching

ceramic cylinder. The targets were surrounded by a 38 mm. outer diameter Lexan

sabot. In all tests the tungsten rod was completely eroded and the hole in the

target fully formed well within the ceramic.

Fig. 1. Typical target geometry (dimensions in inches).

Ceramics

The B4C ( t = 2.51 g/cm3) was hot-pressed by the Norton Company with a

typical grain size of 9 m. The SiC ( t = 3.22 g/cm3) was “pressure assisted

densified” (PAD) by Cercom; typical grain size was about 2 m. The AlN ( t =

3.25 g/cm3) was hot-pressed by the DOW Company and had a typical grain size

of 1.5 m. Additional information on the ceramics tested is given in [6-8].

Penetrators

Penetrators were long rods (right circular cylinders) of pure tungsten ( p = 19.3

g/cm3). Penetrator diameter, D, was selected to be visible in the flash X-rays and

small enough to insure a sufficiently large ratio of ceramic to penetrator diameter.

Littlefield, et al. [9] and Anderson, et al. [10] performed numerical simulations for

L/D = 20 tungsten alloy penetrators symmetrically impacting armor steel targets

that show no significant effects of the radial boundaries if the target diameter is

greater than about 15 to 20 penetrator diameters.


For nearly all the tests reported here the rod diameter was 0.762 mm (0.030

inch). For some of the B4C tests in the velocity range 1.493 < v < 2.767 km/s a

penetrator diameter of 1.02 mm (0.040 inch) was also used. Thus the ratio of

ceramic to penetrator diameter was about 30 for nearly all the tests and was never

less that 20. In addition, examination of the flash X-rays from the tests reveals no

measurable radial expansion of the targets. Therefore it is believed that the targets

behaved as “well confined” targets.

Instrumentation

Primary instrumentation was four independently timed 450kV flash X-rays that

viewed the rod-target interaction. In addition, two continuous X-rays positioned

up-range of the impact and separated by 0.30 m were used to determine impact

velocity. The four flash X-rays contained fixed spatial fiducials and this plus the

known times the x-rays fired provided an independent measure of impact velocity

as well as the absolute zero time of impact. Impact velocities determined by the

independent measurements were always in excellent agreement.

ANALYSIS OF FLASH X-RAYS

Fig. 2 is a typical set of flash X-rays (Test 117, B4C at 3.134 km/s). The depth

of penetration, p, and length of remaining rod, Lr, is measured in each X-ray. The

length of rod “consumed, Lc, is Lc = L - Lr . Measured values for p and Lc for

Test 117 are given in Fig. 3. These data include the (0,0) point since time of

impact is independently measured.

Fig. 2. Typical flash X-rays Fig. 3. Data from flash X-rays


The data shown in Fig. 3 are used to determine the following parameters:

Penetration Velocity, u

As can be seen from Fig. 3 the first four points in the penetration depth-time

plot, including (0,0), are well represented by a straight line of slope dp/dt = 2.039

km/s = u, with a correlation coefficient, r = 0.999. This slope is defined as the

penetration velocity.

Consumption Velocity, vc

The first four points on the Lc versus time plot, including (0,0), are also well fit

by a straight line of slope dLc/dt = 1.111 km/s = vc, with r = 0.999. This slope is

defined as the “consumption velocity” for the rod.

Primary Penetration Depth, pprimary

To a good first approximation it may be assumed for these high L/D rods that

the rear of the rod does not significantly decelerate until it reaches the target

interface. With this approximation, and assuming u and vc are constant as shown

by the data, the rear of the rod reaches the target interface at time tc = L/vc. The

depth of penetration at time tc is defined as the primary penetration, pprimary = utc,

and is shown as the open circle in Fig. 3.

Total Penetration Depth, ptotal

For each test one of the flash X-rays was timed to fire long after the penetrator

was fully consumed. The depth of penetration in this X-ray is defined as the total

penetration depth, ptotal.

EXPERIMENTAL DATA AND ANALYSIS

Due to space limitations the general phenomenology observed in the

experiments is not discussed but can be found in [6-8]. In discussing the

experimental data comparisons will be made to “ideal hydrodynamic theory” [11]

for which: uhyro = v / [1+ ( t/ p)1/2

] and phydro = L ( p/ t)1/2

.

Penetration Velocity, u

Figures 4, 5, and 6 show u versus impact velocity for B4C, SiC and AlN,

respectively. For these ceramics it is observed that penetration is steady-state (i.e.

u = constant) to a high degree of approximation. For the lowest impact velocities

u sometimes appears to be slightly non-constant and the lowest correlation

coefficients for a linear fit to the penetration-time data typically occurs for the

lower impact velocities. For the data shown here r 0.990 for a linear fit to the

penetration-time data in all cases.


The dashed line is uhydro. As shown, for these ceramics u < uhyro over the entire

range of impact velocity. For each of the ceramics u is well fit by a linear

equation in v:

B4C: u = 0.757v - 0.406 (km/s) (1)

SiC: u = 0.781v - 0.510 (km/s) (2)

AlN: u = 0.792v - 0.524 (km/s) (3)

0.0

1.0

2.0

3.0

4.0

1.0 2.0 3.0 4.0 5.0

Impact Velocity, km/s

Pen

etr

ati

on

Velo

cit

y, u

,

km

/s

Fig. 4. Penetration velocity versus impact velocity for B4C.

0.0

1.0

2.0

3.0

4.0

1.0 2.0 3.0 4.0 5.0


Pen

etr

ati

on

Velo

cit

y, u

,

km

/s

Fig. 5. Penetration velocity versus impact velocity for SiC.


0.0

1.0

2.0

3.0

4.0

1.0 2.0 3.0 4.0 5.0


Pen

etr

ati

on

Velo

cit

y, u

,

km

/s

Fig. 6. Penetration velocity versus impact velocity for AlN.

Consumption Velocity, vc

Plots of vc versus v are given in [6-8] but are not included here because of

space limitations. For each of the ceramics vc > vchydro = v - uhydro over the entire

range of impact velocity. There is significant scatter in the vc data but for each

ceramic the data are well fit with a linear equation in v:

B4C: vc = 0.219v + 0.333 (km/s) (4)

SiC: vc = 0.240v + 0.383 (km/s) (5)

AlN: vc = 0.216v + 0.434 (km/s) (6)

Primary Penetration, pprimary

Figures 7, 8, and 9 show pprimary normalized by original rod length versus

impact velocity for the three ceramics. The horizontal dashed line on each plot is

the ideal hydrodynamic penetration, ( p/ t)1/2

. For each of the ceramics pprimary/L <

phydro/L over the entire range of impact velocity. This reflects the strength of the

ceramics as discussed in [12].

The measured pprimary/L data can be well fit by a cubic equation in v:

B4C: pprimary/L = -1.213 + 2.178 v - 0.512 v2 + 0.044 v

3 (km/s) (7)

SiC: pprimary/L = 0.747 – 0.049 v + 0.185 v2 - 0.024 v

3 (km/s) (8)

AlN: pprimary/L = -1.258 + 1.842 v - 0.342 v2 + 0.022 v

3 (km/s) (9)

These cubic equations fit the data well [6-8] and are convenient but, of course,

they are purely empirical, do not asymptote to the hydrodynamic limit as v ,

and should not be used outside the range of the data.


0.0

1.0

2.0

3.0

1.0 2.0 3.0 4.0 5.0


Pp

rim

ary

/L

Fig. 7. Primary penetration versus impact velocity for B4C

0.0

0.5

1.0

1.5

2.0

2.5

3.0

1.0 2.0 3.0 4.0 5.0


Pp

rim

ary

/L

Fig. 8. Primary penetration versus impact velocity for SiC.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

1.0 2.0 3.0 4.0 5.0


Pp

rim

ary

/L

Fig. 9. Primary penetration versus impact velocity for AlN.


Total Penetration, ptotal

Figures 10, 11, and 12 show the normalized total penetration, ptotal/L versus

impact velocity for the three ceramics. Again the horizontal dashed line is the

ideal hydrodynamic penetration. For each of the ceramics ptotal/L is less than the

hydrodynamic value up to impact velocities of about 4 km/s or higher.

Strictly speaking hydrodynamic penetration only applies to what is called here

primary penetration. As noted above pprimary/L is less than the hydrodynamic

penetration even at impact velocities as high as 4.6 km/s. A comparison of Fig. 7-

9 with 10-12 shows that except for a few of the lowest velocity tests ptotal/L >

pprimary/L. Total penetration is the sum of the primary penetration plus what has

been called “secondary penetration”, “residual penetration”, “after-flow”, or as

preferred here “Phase 3 penetration”, after Eichelberger and Gehring [13]. As

discussed by Orphal [14] Phase 3 penetration potentially involves two distinct

phenomena. The first phenomena is often called “after-flow” after Pack and

Evans [15] and later Tate [16] and is the extension of the target hole due to

momentum in the target material at the time the rod is fully eroded. The second

phenomena was called “secondary penetration” by Christman and Gehring [17]

and Allen and Rogers [18] and is the further penetration of the target by the

eroded rod debris in the case when p > t. Phase 3 penetration for these three

ceramics is discussed in some detail by Orphal [14 ] .

The measured ptotal/L data were also fit by cubic equations in v:

B4C: ptotal/L = -2.338 + 3.256 v - 0.821 v2 + 0.077 v

3 (km/s) (10)

SiC: ptotal/L = -1.438 + 1.904 v - 0.367 v2 + 0.030 v

3 (km/s) (11)

AlN: ptotal/L = -1.393 + 1.954 v - 0.365 v2 + 0.029 v

3 (km/s) (12)

The admonition above about the application of these purely empirical fits applies

to ptotal/L as well.

0.0

1.0

2.0

3.0

4.0

1.0 2.0 3.0 4.0 5.0


Pto

tal/L

Fig. 10. Total Penetration versus impact velocity for B4C.


0.0

0.5

1.0

1.5

2.0

2.5

3.0

1.0 2.0 3.0 4.0 5.0


Pto

tal/L

Fig. 11. Total Penetration versus impact velocity for SiC.

0.0

0.5

1.0

1.5

2.0

2.5

3.0

1.0 2.0 3.0 4.0 5.0


Pto

tal/L

Fig. 12. Total Penetration versus impact velocity for AlN.

SIZE SCALING

These reverse ballistic experiments are small scale. The issue of scaling is best

addressed by performing full-scale tests, if possible, or at least larger scale tests.

Ten larger scale direct ballistic tests were performed against B4C targets. These

tests are described in detail in [6, 19]. The target design for these larger scale tests

was based on the reverse ballistic targets. The ceramic diameter was 87.5 mm. In

nine of these tests L/D = 13 tungsten alloy rods were impacted against the

confined B4C targets at 1.38 v 3.76 km/s. These rods had D = 4.22 mm in five

tests (1.38 v 2.96 km/s) and 2.92 mm in four tests (3.21 v 3.76 km/s). In

the tenth test (2.79 km/s) an L/D = 20, D = 2.79 mm rod was used. Thus in these

tests the ratio of ceramic diameter to rod diameter was 20-30. These tests were

essentially 5.5 times larger in scale than the reverse ballistic tests. Figure 13

compares the penetration velocity, u, measured in both the reverse ballistic tests

(Fig. 4) and these larger scale direct ballistic tests (labeled GRC). Figure 14


compares ptotal/L for the two sets of experiments. The agreement between the data

from the small-scale reverse ballistic tests and the larger scale direct ballistic tests

is considered very good.

Seven similar larger scale direct ballistic tests were performed for AlN targets

by Piekutowski and Forrestal [20]. While not included here because of space

limitations the agreement between the reverse and larger-scale direct ballistic

experimental data for AlN is similarly very good [8].

0.0

1.0

2.0

3.0

4.0

1.0 2.0 3.0 4.0 5.0


Pen

etr

ati

on

Velo

cit

y, u

,

km

/s

GRC

Fig. 13. Comparison of penetration velocity with larger scale direct ballistic tests.

0.0

1.0

2.0

3.0

4.0

1.0 2.0 3.0 4.0 5.0


Pto

tal/L

GRC

Fig. 14. Comparison of total penetration with larger scale direct ballistic tests.

NORMALIZATION OF TOTAL PENETRATION FOR B4C, SiC AND AlN

For hydrodynamic penetration P/[L( p/ t)1/2

] = 1. To compare P/L for different

ceramics it is reasonable to attempt to normalize by the factor ( p/ t)1/2

. Figure 15

shows Ptotal/[L( p/ t)1/2

] versus velocity for B4C, SiC and AlN. As can be seen, for

these three ceramics this normalization approximately collapses the data to a


single curve. Although not shown here, similar results are obtained for

Pprimary/[L( p/ t)1/2

].

Fig. 15. Ptotal / [L( p/ t)1/2

] versus impact velocity for B4C, SiC and AlN

0

0.2

0.4

0.6

0.8

1

1.2

1 2 3 4 5


Pto

tal/

[L*(

rho

p/r

ho

t)**

0.5

]

B4C

SiC

AlN

SUMMARY

Penetration depth and remaining rod length as functions of time were measured

in reverse ballistic tests for long tungsten rods impacting confined B4C, SiC and

AlN at velocities from 1.5 to 5 km/s. Penetration is steady state (or very nearly

steady state at the lowest impact velocities), i.e. penetration velocity = constant.

Penetration velocity and rate of rod erosion are both well described by linear

functions of impact velocity. Primary penetration is less than the ideal

hydrodynamic value over the entire range of impact velocity. Total penetration is

less than ideal hydrodynamic for impact velocities less than about 4 km/s.

Dividing penetration depth by the factor L( p/ t)1/2

nearly collapses the

penetration versus impact velocity data for the three ceramics to a single curve.

REFERENCES1V. Hohler and A. J. Stilp, “Hypervelocity impact of rod projectiles with L/D

from 1 to 32,” International Journal of Impact Engineering, 5, 323-331 (1987).2V. Hohler and A. J. Stilp, “Long rod penetration mechanics,” Chapter 5 in

High Velocity Impact Dynamics, Edited by Jonas A. Zukas. John Wiley, 1990.3B. R. Sorensen, K. D. Kimsey, G. F. Silsby, D. R. Scheffler, T. M. Sherrick

and W. D. deRosset, “High velocity penetration of steel targets,” International

Journal of Impact Engineering, 11, 107-119 (1991).4P. Lundberg, R. Renstrom and B. Lundberg, “Impact of metallic projectiles on

ceramic targets: transition between interface defeat and penetration,”

International Journal of Impact Engineering, 24, 259-275 (2000).


5L. Westerling, P. Lundberg and B. Lundberg, “Tungsten long rod penetration

into confined cylinders of boron carbide at and above ordnance velocities,”

International Journal of Impact Engineering, 25, 703-714 (2001). 6D. L. Orphal, R. R. Franzen, A. C. Charters, T. L. Menna, and A. J.

Piekutowski, “Penetration of confined boron carbide at targets long rods at impact

velocities from 1.5 to 5.0 km/s,” International Journal of Impact Engineering, 19,

15-29 (1997). 7

D. L. Orphal and R. R. Franzen, “Penetration of confined silicon carbide

targets by tungsten long rods at impact velocities from 1.5 to 4.6 km/s,”

International Journal of Impact Engineering, 19, 1-13(1997). 8

D. L. Orphal, R. R. Franzen, A. J. Piekutowski, and M. J. Forrestal,

“Penetration of confined aluminum nitride targets by tungsten long rods at 1.5-4.5

km/s,” International Journal of Impact Engineering, 18, 355-368 (1996). 9

D. L. Littlefield, C. E. Anderson, jr., Y. Partom, and S. J. Bless, “The

penetration of steel targets finite in radial extent”, International Journal of Impact

Engineering, 19, 49-62 (1997). 10

C. E. Anderson, Jr., J. D. Walker, and T. R. Sharron, “The influence of edge

effects on penetration”, Proceedings: 17th

International Symposium on Ballistics

(Midrand, South Africa), pages 3-33 to 3-40, March 23-27, 1998 11

G. Birkoff, D. P. MacDougall, E. M. Pugh, and Sir. G. Taylor, “Explosives

with lined cavities,” Journal of Applied Physics, 19, 563-582 (1948). 12

C. E. Anderson, Jr., D. L. Orphal, R. R. Franzen, and J. D. Walker, “On the

hydrodynamic approximation for long rod penetration’” International Journal of

Impact Engineering, 22, 23-43 (1999). 13

R. J. Eichelberger and J. W. Gehring, Journal of the American Rocket

Society, 32, 1583-1591 (1962). 14

D. L. Orphal, “Phase three penetration,” International Journal of Impact

Engineering, 20, 601-616 (1997). 15

D. C. Pack and W. M. Evans, Proceedings Physical Society of London, B64,

298-302 (1951) 16

A. Tate, “A theory for the deceleration of long rods after impact,” Journal

Mechanics and Physics of Solids, 15, 387-399 (1967). 17

D. R. Christman and J. W. Gehring, “Analysis of high-velocity projectile

penetration mechanic,” Journal of Applied Physics, 37, 1579-1587 (1966). 18

W. A. Aleen and J. W. Rogers, “Penetration of a rod into a semi-infinite

target,” Journal Franklin Institute, 272, 275-284 (1961). 19

T. L. Menna, A. C. Charters, and A. J. Piekutowski, “Penetration

performance of confined boron carbide by continuous and segmented rods,”

General Research Corporation Report SB-90-0105 (1990). 20

A. J. Piekutowski and M. J. Forrestal, “Penetration into aluminum nitride

targets with L/D = 10 tungsten rods at impact velocities of 1.7, 2.2 and 2.7 km/s,”


Report SAND91-0088.RS123/90/00007, Sandia National Laboratories,

Albuquerque, NM.


DEPTH OF PENETRATION TESTING

Dr Bryn James


Chobham Lane

Chertsey, Surrey, KT16 0EE

United Kingdom

ABSTRACT

The Depth of Penetration (DOP) test has been widely used for many years for

ranking the protective value of materials, most notably ceramics. This is

essentially a simple and straightforward test with definitive results. In practice

however, a significant number of factors must be taken into account to achieve

reliable and comparable results. Often, published results cannot be utilised to

augment other data sets as insufficient detail is given to allow the necessary

correction or normalisation to be made.

The aim of this paper is to provide details of the DOP measurement system

devised at the Defence Science and Technology Laboratories, Chertsey, UK and

to present the methodology for correction and normalisation of the data.

Guidelines will be given for choice of configuration of the target assembly.

INTRODUCTION

There are two principal methods by which a material may be tested

ballistically. The ballistic limit configuration reported in the literature (1) consists

of a relatively thin layer, or composite, system which is defeated by the

penetrator. Performance of this configuration is measured by the residual length

and velocity of the projectile, or by penetration of a witness pack. This

configuration is often used to investigate the effectiveness of specific armour

systems. Later configurations published in the literature are variations on the

semi-infinite backstop method first suggested by Bless and Rosenberg et al. (2,

3). Performance in this configuration is measured by the residual Depth of

Penetration (DOP) of the projectile into a backblock of reference material for

which the penetration depth of the projectile for direct impact is known. The

backblock dimensions are large ('semi-infinite'), such that the penetration is not

influenced by the proximity of edges or interfaces.



In order to assess the intrinsic performance of bulk material, the latter

configuration is preferred. Target dimensions are determined such that the

majority of the steady state penetration phase is accommodated within the

material under test. A small amount of this phase and the deceleration phase are

contained within the semi-infinite back-block to give a depth of penetration

measurement. Massive containment is often used for ceramic tiles to mimic the

effects of a laterally infinite target. This configuration is designed to avoid the

effects of elastic reflections from the lateral extents of the target assembly

affecting the penetrator, and to maintain impact induced pressure.

A problematical situation exists concerning medium scale, long rod type

penetration testing for which the backblock material is often rolled homogeneous

armour steel (RHA). Unfortunately the specification for RHA is different in

virtually every country, leading to many different values for ballistic efficiency

being quoted for the same material. We have investigated the use of alternative,

more universally available and better specified materials, but unfortunately, RHA

still seems to be the ideal material for this application. This situation also occurs

for other backblock materials for which an internationally agreed standard has not

been resolved.

Other factors that affect the derived value for ballistic efficiency include the

variation of depth-of-penetration with projectile yaw at impact, with projectile

velocity and of course with the overall target configuration.

BALLISTIC TESTING

In order to mimic the performance of a semi-infinite array of ceramic tiles, a

massive containment system was developed at Dstl, Chertsey (Figure 1).

Lateral containment

Section of rig showing

lateral and axial containment

Ceramic

RHA backblock

Figure 1. Ceramic containment system


It should be noted that the containment ring is included to minimise the reflection

of impact induced stress waves from the periphery of the ceramic tile and to

maintain impact pressure. Any precompression of the ceramic has a negligible

effect upon the intrinsic ceramic performance as the small amount of pressure that

can be applied is negligible compared to the 5-20 GPa required to increase

significantly the failure strength of the material.

The corner of the containment ring, in our system, is relieved to allow the

ceramic tile to fit freely and to allow space for the introduction of a 1.0mm thick,

fully annealed, brass shim between the steel containment and the ceramic. This

shim has a similar acoustic impedance (18.2 MRayls) to that of steel

(25.4 MRayls), alumina (21.4 MRayls), boron carbide (22 MRayls) and silicon

carbide (25.2 MRayls) but being soft, will conform to any small irregularities in

the mating surface between the steel and the ceramic, providing an excellent

acoustic interface (Figure 2).

Figure 2 Brass shimming of ceramic within containment

The containment rig is assembled with the surface machined RHA (or

aluminium alloy) backblock between the rear of the ceramic tile (or tiles) and the

backplate of the rig. All bolts are tightened, in sequence, to a given torque.

IMPACT YAW CORRECTION

The penetration of a projectile subject to yaw at impact will be less than that

for an exactly normal impact. In order to allow for the experimental variation in

projectile yaw, depth-of-penetration measurements are corrected for yaw

according to the analysis of Bjerke et al (4). Note that in the following, 'yaw'

refers to the total yaw of the projectile, i.e. the angle between the longitudinal axis

of the rod and the velocity vector of the centre of mass.


Bjerke’s correction factors are based upon the analysis of a very large number

of low yaw normal impacts of long rods into semi-infinite RHA over a range of

velocities from 1.2 to 4.7 kms-1

and a wide range of sizes and aspect ratios.

Upon impact, a penetration channel of diameter H is formed. If the projectile

yaw is such that the tail of the projectile of length L gouges the side of the

penetration channel, energy and mass of the projectile will be expended enlarging

the channel thus diminishing the total depth-of-penetration. As the penetration

channel is larger than the projectile diameter D, this condition will not occur until

a critical yaw angle has been reached. The critical yaw value cr is given by the

following relationship:

]2L / D)-(H[sin= -1

cr(1)

It is obvious from the above that the critical yaw value depends upon the

penetration channel diameter. It is not necessary to determine this value for each

impact as Bjerke et al. have performed an empirical fit to their large database of

impact geometry’s and velocities V. The penetration channel diameter may be

approximated by:

(2) 2V0.1286+V0.3388+1.1524=D / H

Where V is in kms-1

.

Bjerke calculated relative yaw (i.e. / cr) for a large number of impact

experiments so that an empirical fit could be made and the effective penetration as

a function of yaw could be calculated, eliminating the influence of velocity and

relative penetrator dimensions. Effective penetration factor, PENeff , is then given

by:

(3) ) / 11.46(cos=PEN creff

The yaw corrected penetration DOPyaw-corr may then be calculated from the

measured penetration DOPmeas using the following relationship:

(4) PEN / DOP=DOP effmeascorr-yaw

The original Bjerke analysis was performed for the impact of tungsten alloy

rods into steel. However, high intensity X-radiography has shown that the

dynamic crater diameter in ceramic materials is very similar to that in steel. We


have also shown that use of the yaw correction significantly reduces the standard

deviation in the data set. It is therefore considered that this correction is valid.

VELOCITY NORMALISATION

The depth-of-penetration is a function of impact velocity. This function is

generally not linear over a wide range, and so a reference DOP assessment must

be made into the backblock material over the velocity range at which the material

will be tested. In order to obtain uniformity we must normalise both reference

DOP and residual DOP to a nominated test velocity.

If it is deemed necessary (i.e. in the case of large non-linearity within the

velocity range used), the full form of the DOP vs. velocity curve must be

determined, and a polynomial should be fitted. In our case, the relationship has

been found to be highly linear over a relatively large velocity range and so a

straight line fit may be used. The velocity normalised depth-of-penetration

DOPvel-norm at the reference velocity Vref is calculated from the measured depth-of-

penetration at the measurement velocity Vmeas by:

(5) )V-Vm.(+DOP=DOP measrefcorr-yawnorm-vel

Where m is the empirically derived slope of the DOP vs. velocity curve for

the specific penetrator.

CALCULATION OF CORRECTED AND NORMALISED DOP

The following details indicate the method for calculating the yaw corrected

depth of penetration and normalisation to a given reference velocity:

i/ Generate curve of DOP into the backing material vs. Velocity for the

velocity range required, using yaw corrected DOP's. Calculate straight line

regression or more complex function if required.

ii/ For each experiment, calculate yaw-corrected DOP.

iii/ Normalise yaw corrected DOP to Vref.

iv/ Determine DOPref at Vref using curve generated in i/.

Calculation of yaw-corrected DOP

i/ Calculate H/D at impact velocity using equation 2, hence find H.

ii/ Calculate cr at impact velocity using equation 1.

iii/ Calculate PENeff using equation 3 and the measured total yaw at

impact, hence find DOPyaw-corr using equation 4.


FACTORS AFFECTING RESULTS

Rolled Homogeneous Armour Steel. The single most important factor affecting

the measured DOP results rests in the mechanical properties of the backing block.

The properties of RHA differ widely from nation to nation and in general the

specification for RHA is very wide. Often, the only reason for DOP results to be

valid is that, for economical reasons, the manufacturer supplies material with

properties as close as possible to the lowest limit of the specification. We have

found that, at the very least, a hardness measurement should be taken for every

batch of DOP material used, all backblocks must come from the same batch and

they must all be prepared in the same fashion.

During the course of experimentation at Dstl, Chertsey, (approximately 1000

DOP experiments), some interesting anomalies have been encountered in the

properties of RHA. Through thickness hardness transects generally show uniform

hardness in the bulk of the material with a slightly harder layer at the surface.

Occasionally, RHA plates have been seen with soft surface layers or with a

hardness gradient through the material. These plates appear visually to be the

same as standard plates, but may show a DOP up to 20% greater than standard

material. The importance of acceptance testing for each plate is apparent.

Ceramic Tile Size. Significant differences in the impact performance of a

given type of ceramic will be seen dependent upon the size of the sintered or

pressed tile. The furnace conditions necessarily must be different for different

sizes of tile, resulting in different residual stress states for different sized tiles. In

general, the larger the tile, the worse the ballistic performance, due to residual

stress in the material. We have encountered very large ceramic tiles with so much

residual stress that inadvertent damage during handling could result in

catastrophic fracture. To implement scaling experiments the ceramic material

should be cut from one large piece or large tiles should be machined to produce

smaller tiles. However, the surface state should be the same in each case.

Surface Preparation. The firing process for ceramic tiles often has an

advantageous effect on the surface, leaving a relatively flaw free layer with a

residual compressive stress. The presence of this layer tends to increase the

ballistic performance of the tile (up to ~5%). Often, this layer is machined away

in order to create a more precisely defined surface for interfacing with another

layer. It should be noted that any increases in performance gained by better

geometrical tolerancing may be offset by the removal of the beneficial surface

layer to be replaced by a surface microscopically damaged by the machining

process.

Lateral dimensions. In order for the lateral dimensions to have no influence

upon the measured DOP result, ideally the ceramic tile smallest lateral dimension

should be greater than 30 times the projectile diameter for relatively low velocity

impact. As the impact velocity increases, the lateral dimensions may become


smaller. Above an impact velocity of 1600 ms-1

the tile size can be reduced until

at an impact velocity of 1800 ms-1

the lateral dimension of the tile may be only 15

times the projectile diameter. This should be considered a minimum. If no tiles of

adequate size are available, the tile may be clad in a supportive frame to mimic a

larger tile. Should a frame be fitted, the ideal minimum width of this frame should

be 0.5 x (30 x projectile diameter – the ceramic width). Of course, if the

calculated frame thickness is very thin, it may be omitted.

Ideally, any frame should be securely clamped to the ceramic with an acoustic

impedance matched soft shim between the ceramic and frame. It should be noted

that the lateral dimension guidelines are applicable in the case of a centre strike on

the target assembly. If there is appreciable shot dispersion from the launch

system, the target array should be larger to accommodate this dispersion.

RECORDING DATA

In order for published or recorded data to be valuable to other workers, the

following data should be reported:

Backing material Minimum: Manufacturer, type, hardness, density, axial and

lateral dimensions.

Desirable: Material model data.

Ceramic material Minimum: Manufacturer, type, density, axial and lateral

dimensions.

Desirable: Porosity, Material model data.

Projectile material Minimum: Manufacturer, type, density, dimensions.

Desirable: Hardness, strength, material model data.

Target configuration Minimum: All lateral and thickness dimensions and details

of any interlayers. Method of support.

Desirable: Tolerances and finish of mating surfaces,

tightening torques.

Impact Minimum: Impact velocity, yaw at impact, strike position.

Desirable: Yawing rate at impact.

CONCLUSIONS

The DOP test is a useful ranking test for ceramic materials. If attention is paid

to the detail and reproducibility of the target configuration, and if simple

corrections and a normalisation are carried out on the results, a significant amount


of the inherent scatter in the data can be removed. The ranking measured in one

configuration is applicable to other configurations, even though the absolute

values of protective capability may not be the same for all configurations.

If a certain minimum data set is published for each experimental series, the

results may be easily utilised by other workers to supplement their own data sets.

ACKNOWLEDGEMENT

The work upon which this analysis is based was funded by the UK

Government Corporate Research Programme.

REFERENCES

1 J. A. Zukas, T. Nicholas, H. F. Swift, L. B. Greszczuk and D. R. Curran,

Impact Dynamics, J. Wiley and Sons, New York 1982

2 Z.Rosenberg, S.Bless, Y.Yeshurun, K.Okajima, “ A new definition of

ballistic efficiency of brittle materials based on the use of thick backing

plates”, in Impact loading and dynamic behaviour of materials Vol. 1 (ed.

C. Y. Chiem, H. D. Kunze and L. W. Meyer) pp. 491-498. DGM

Informationsgesellscaht mbH, Oberursel 1988

3 S.Bless, Z.Rosenberg, B.Yoon, “Hypervelocity penetration of ceramics”

Int.J.Impact Engng., 5 pp.165-171, 1987

4 T. Bjerke, G. Silsby, D. Scheffler, R. Mudd, “High yaw penetration

performance of long rod penetrators”, pp 191-198, Vol 3 13th

Int. Symp.

Ballistics, June 1992


TRANSITION BETWEEN INTERFACE DEFEAT AND PENETRATION FOR

A GIVEN COMBINATION OF PROJECTILE-AND CERAMIC MATERIAL

Patrik Lundberg, René Renström and Lars Westerling

Swedish Defence Research Agency, FOI

Weapons and Protection

SE-147 25 Tumba, Sweden

ABSTRACT

At a certain impact velocity, the surface load generated by a projectile be-

comes critical and transition from interface defeat to penetration occurs. This

transition impact velocity is estimated by combining two models, one for the con-

tact load during interface defeat and one for the yield condition in the ceramic

target. The effects of the model parameters are studied with the aid of numerical

simulations, and the transition impact velocity is determined as a function of the

ceramic yield strength.

INTRODUCTION

By using devises for chock attenuation and load distribution in combination

with supporting confinement, it is possible to design ceramic targets capable of

defeating high velocity projectiles on the ceramic surface, so-called interface de-

feat or dwell [1-5]. The possibility to maintain interface defeat against long rod

projectiles for long interaction times (hundreds of s), which results in a nearly

static loading of the ceramic target material, has been shown experimentally [5].

The transition impact velocity, i.e., the impact velocity at which interface de-

feat ceases and penetration starts, has been determined experimentally for differ-

ent ceramic materials [4]. Also, analytical models for estimation of this velocity

have been derived [4,6]. The results of the experiments show that the transition

from interface defeat to penetration is distinct and related to the surface load.

In order to investigate the state of stress in the target material under conditions

of interface defeat, it is necessary to determine the contact pressure (negative of

normal stress at the contact interface) generated by the flowing projectile material.



This contact pressure and the corresponding state of stress is studied here by

means of continuum-dynamic simulations. First, a relation is derived for the

maximum contact pressure during interface defeat as a function of impact velocity

and projectile density, bulk modulus and yield strength. Then, using a simple con-

stitutive model for the ceramic material and the contact pressure distributions

from the simulations, a relation is established between the maximum contact pres-

sure and the state of stress in the target corresponding to incipient plastic yield

and large-scale plastic yield. With the aid of these relations, the transition impact

velocity is estimated as a function of the ceramic yield strength.

ANALYTICAL MODEL

In [4], analytical models were presented which make it possible to estimate

the maximum contact pressure generated by a long-rod projectile during inter-

face defeat and the corresponding state of stress in the ceramic target.0p

The projectile material was treated as linear elastic perfectly plastic, obeying

von Mises yield criterion. It was characterised by its density bulk modulus

and yield strength . The target surface was considered to be flat, rigid and

friction-free, and the axis of symmetry of the projectile was oriented perpendicu-

larly to the target surface.

,p

pK yp

It was shown in [4] that the maximum contact pressure at r = 0, z = 0 can

be expressed as0p

, (1) 0 1pp q

where2 2p p pq v , (2)

and is the impact velocity. The functionspv and , which represent

elastic and yield strength effects, respectively, were determined to be 1 2

and 3.27 , where

,p p yp pK q q . (3)

The radial distribution of the projectile load was approximated by one deter-

mined experimentally for a low-velocity water jet [7]. This load distribution, in

combination with Boussinesq’s elastic stress field solution [8] and a plastic slip-

line solution [9] for the indentation of a flat rigid die, led to the transition interval


(4) 0low yc high ycp

for the transition contact pressure, where is the yield strength of the ceramic

in uniaxial compression. The coefficients and were determined in [4] to

be 1.47 (for Poisson’s ratio ) and 2.85, respectively. The ends of the in-

terval (4) correspond to incipient and large-scale plastic yield, respectively, of the

ceramic material. Because of relations (1) and (2) there is a corresponding interval

yc

low high

0.16

(5)low p highv v v

for the transition impact velocity. In Figure 1, the domain where yield occurs in

the ceramic target is illustrated correspondingly.

p lowv vincipient

plastic yield

low p highv v v

(a) (b)

p highv vlarge-scale

plastic yield

(c)

Fig. 1 Impact velocity (a) equal to the lower transition velocity , (b) in-

between the lower and higher transition velocity and (c)

equal to the upper transition velocity . The shape and location of the

domain where yield occurs is schematic.

lowv

highvlow pv v

highv

NUMERICAL SIMULATIONS

The AUTODYN code [10] was used for determining the distribution of con-

tact pressure and the resulting state of stress in a ceramic target during interface


defeat. The simulations were two-dimensional with cylindrical symmetry. Two

types of simulations were performed. First, the contact pressure on a flat, rigid

and friction-free target surface was calculated using Eulerian simulations. A line-

arly elastic perfectly plastic constitutive model and von Mises yield criterion with

associated flow rule was used for the projectile material. A frictionless boundary

condition was used for the impact surface, and an inflow boundary condition was

used to simulate an infinitely long projectile. The simulations were performed

until a stationary contact pressure was reached. Different combinations of impact

velocities vp, bulk moduli Kp and yield strengths yp were used together with a

projectile density 17600 kg/mp3 in order to separate the influence of the com-

pressibility and the yield strength. By this technique, the functions and

were evaluated.

One of the contact pressure distributions was used in a second set of Lagran-

gian simulations in which the coefficients and were determined. It was

used as a boundary condition and the contact pressure was increased linearly in

order to follow the formation and growth of a plastic region, which finally

reached the contact surface. The target material was modelled as a linear elastic

perfectly plastic material with density kg/m

low high

3215c3, bulk modulus

GPa, Poisson's ratio and yield strength . A con-

tact pressure distribution corresponding to a Hertz indent [11] was also used for

comparison.

221cK 0.16 10 GPayc


The maximum contact pressure p0 is shown in Table I for different combina-

tions of projectile bulk modulus Kp, yield strength yp and impact velocity vp.

Table I. Maximum contact pressure p0 for different combination of bulk

modulus Kp, yield stress yp and impact velocity vp.

p0 (GPa)Kp

(GPa)yp

(GPa) vp =1000 m/s vp = 1500 m/s vp = 2000 m/s vp = 2500 m/s vp = 3000 m/s

285 1 11.59 23.70 40.80 63.85 93.99

285 0.001 8.99 20.57 37.33 60.43 90.78

28500 1 11.63 23.04 38.67 58.49 82.87

28500 0.001 8.77 19.74 35.14 54.93 79.21


The distribution of the contact pressure p and the normalised contact pressure

p/p0 for different impact velocities vp is shown in Figure 2. The curve for 1500

m/s in Figure 2(a) is the one used for the determination of the coefficients

and .low

high

The normalised contact pressure p/p0 for two different combinations of projec-

tile bulk moduli and yield strengths are shown in Figure 2(b). If the influence of

compressibility and yield strength is suppressed, corresponding to high bulk

modulus and low yield strength ( ), the contact pressure

distribution corresponds well to the one measured for a low velocity water jet [7].

On the other hand, low bulk modulus and high yield strength will give a narrower

pressure distribution as shown in the figure ( ). As a compari-

son, the pressure distribution for a Hertz indent [11] is also shown.

3240, 0.000114

32.4, 0.114

0 0.5 1 1.5 2 2.5 3

r/a

0

20

40

60

80

100

p (G

Pa)

1000 m/s

1500 m/s

2500 m/s

2000 m/s

3000 m/sp = 17600 kg/m 3

Kp = 285 GPa

yp = 1 GPa

0 0.5 1 1.5 2 2.5 3

r/a

0

0.2

0.4

0.6

0.8

1p/p

0Low velocity water jet

= 3240, = 0.000114

= 32.4, = 0.114

Hertz

(a) (b)

Fig. 2 (a) Contact pressure p and (b) normalised contact pressure p/p0 versus

normalised radius r/a. Filled circles in (b) represent the contact pressure

distribution from a low-velocity water jet [7], and dashed curve corre-

sponds to a Hertz indent [11].

The functions and obtained from the simulations are plotted in

Figure 3.


0 0.05 0.1 0.15 0.2 0.25 0.3 0.35

-1

0

0.05

0.1

0.15

0.2

0 0.025 0.05 0.075 0.1 0.125 0.150

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

1 1.93 23.62 7.04

(a) (b)

Fig. 3 The contribution of (a) compressibility versus for small values of

(< 0.000114) and (b) material strength versus for large values of

(> 360). Filled circles are simulations and the solid curves are least square

fits to the data.

1

The coefficients and for incipient and large-scale plastic yield are

shown in Table II.low high

Table II. The coefficients and .low high

Hertz indent Projectile load

Simulation Data in [4] Simulation

low 1.46 1.47 1.48

high - 2.85 2.73

From least square fits to the data in Figure 3 together with the data from Table

II, the estimates

1 1.93 , 23.62 7.04 , , (6) 1.48low 2.73high

based on the simulations are obtained.

Relations (1) to (6) give an interval for the transition impact velocity v as a

function of the ceramic yield strength. This interval is shown in Figure 4 together

with experimental data for silicon carbide [4]. The transition impact velocity for

silicon carbide determined experimentally is between 1645 m/s and 1705 m/s [4]

and the maximum yield strength, determined from plate impact experiments, is in

p


the range of 12.5-14.5 GPa [12,13]. The density, bulk modulus and yield strength

used for the projectile material was =17600 kg/mp3, 285 GPa and

GPa, respectively.pK

1.2yp

10

yc

yc

lowv

1.49low

high

0 51000

1250

1500

1750

2000

v p

(m/s

)

SiC

Tungsten projectile

vlow

vhigh

15 20 25

(GPa)

Fig. 4 Transition velocity versus ceramic yield strength . The curves for

and correspond to incipient and large-scale plastic yield.pv

highv

The values in Table II for the coefficients and differ from the ones

obtained in [4].low high

The contact pressure distribution for a low-velocity water jet [7] used in [4],

gave a slightly lower value of than the projectile pressure distribution used

here, the curve for 1500 m/s in Figure 2(a). The factor can be solved analyti-

cally with Boussinesq’s elastic stress field solution [8] for arbitrary load distribu-

tions. This method gives for the Hertz pressure distribution and

for the projectile pressure distribution.

low

w

low

1.46lo

The coefficient corresponds to large-scale yield, viz., the instant when

the plastic region beneath the projectile reaches the surface and penetration takes

place. In the simulations, this coefficient was determined when the plastic region

reached the target surface on the axis of symmetry. This evaluation method gave a

slightly lower value of

high

high than that obtained in [4]. The differences can be re-

lated to the use of different material models and contact loads. In [4], the coeffi-

cient was based on the solution of a flat rigid cylindrical die indenting a

semi-infinite rigid-plastic medium [9], while an elastic perfectly plastic constitu-

tive model was used in the simulations together with a bell-shaped surface load.

Values of this coefficient between 2.5 and 3.0 have been reported from simula-

tions of the total loading history when a projectile starts to penetrate a target [14].


CONCLUSIONS

A relation between the yield strength of the ceramic material and the transi-

tion impact velocity with regard to interface defeat has been obtained from nu-

merical simulations. This relation provides two limits, one lower, corresponding

to incipient yield in the target material, and one higher, corresponding to large-

scale yielding. These limit values agree well with analytical results published ear-

lier [4]. The transition impact velocity obtained experimentally for silicon carbide

falls in-between these two limits.

REFERENCES

[1] G. E. Hauver, P. H. Netherwood, R. F. Benck and L. J. Kecskes. Ballistic

performance of ceramic targets. Army Symposium On Solid Mechanics.

USA 1993.

[2] G. E. Hauver, P. H. Netherwood, R. F. Benck and L. J. Kecskes. Enhanced

ballistic performance of ceramic targets. 19th

Army Science Conference.

USA 1994.

[3] E. J. Rapacki, G. E. Hauver, P. H. Netherwood and R. F. Benck, Ceramics

for armours- a material system perspective. 7th

Annual TARDEC Ground

Vehicle Survivability Symposium. USA 1996.

[4] P. Lundberg, R. Renström, B. Lundberg. Impact of metallic projectiles on

ceramic targets: transition between interface defeat and penetration. Int J

Impact Engng 2000;24:259-275.

[5] P. Lundberg, R. Renström, L. Holmberg. An experimental investigation of

interface defeat at extended interaction times. Proc 19th Int Symp on Ballis-

tics, Switzerland: 2001;3:1463-1469.

[6] LaSalvia JC, Horwath EJ, Rapacki EJ, Shih CJ, Meyers MA. Microstruc-

tural and micromechanical aspects of ceramic/long-rod projectile interac-

tions: dwell/penetration transitions. Fundamental Issues and Applications of

Shock-Wave and High-Strain-Rate Phenomena, Staudhammer KP, Murr

LE, Meyers MA, Elsevier Science, pp 437-446, 2001.

[7] Reich F. Omlenkung eines freien Flussigkeitsstrahles an einer zur

Strömungsrichtung senkrecht stehenden ebenen Platte. Diss Hannover:

1926, (oder VDI-Forsch. –Heft 290).

[8] Y. C. Fung. Foundations of solid mechanics. Prentice-Hall, 1965.

[9] R. T. Shield. On the plastic flow of metals under conditions of axial symme-

try. Proc R Soc, A, 233, 267, 1955.

[10] K. Birnbaum, M. S. Cowler, M. Itoh, M. Katayama and H. Obata,

AUTODYN - an interactive non-linear dynamic analysis program for micro-

computers through supercomputers. Ninth International Conference on Struc-

tural Mechanics in Reactor Technology. Lausanne, Switzerland, (1987).


[11] K. L. Johnson, Contact Mechanics, Cambridge University Press, 1985.

[12] N. Bourne, J. Millett, I. Pickup. Delayed failure in shocked silicon carbide.

J Appl Phys. 81(9), 1 May 1997.

[13] R. Feng, G. F. Raiser and Y. M. Gupta. material strength and inelastic de-

formation of silicon carbide under shock wave compression. J Appl Phys.

83(1), 1 January 1998.

[14] Z. Rosenberg, E. Dekel. Material similarities in long-rod penetration me-

chanics. Int J Impact Engng 2001;25:361-372.


Shock and High Strain Rate Dynamic

DYNAMIC FRACTURE OF CERAMICS AND CMC

Albert S. Kobayashi

University of Washington

Department of Mechanical Engineering

Seattle, Washington 98195-2600

ABSTRACT

This paper reviews the limited literature on dynamic fracture mechanicscharacterization of ceramics and ceramic matrix composites (CMC). Dynamic

fracture toughness, KId, at room and elevated temperature of reaction bonded and

hot-pressed Si3N4, Al2O3, partially stabilized zirconia (PSZ), TiB2-particulate

reinforced SiC (TiB2p/SiC), and SiC-whisker reinforced Al2O3 (SiCw/Al2O)) are

presented. Dynamic stress intensity factor, KID versus crack velocity relations at

room and elevated temperature of Al2O3, SiCw/Al2O3, PSZ and Si3N4 are also

discussed. Dynamic crack arrest stress intensity factor, KIa, was only detected in

PSZ.

INTRODUCTION

Dynamic fracture mechanics encompasses the three phenomena of dynamic

crack initiation, i.e. crack initiation under dynamic loading, rapid crack

propagation, and arrest of a rapidly propagating crack. While early papers on

dynamic fracture mechanics dates back to the 1950’s1-4

, studies on dynamic

fracture mechanics started in the 1970's with the need to predict the extent of rapid

crack propagation in a nuclear power pressure vessel under emergency core

cooling and the effectiveness of a crack arrester in a large marine structure. As a

result of such concerted efforts, much is known on the dynamic response of a

rapidly propagating crack in metals and polymers. Unfortunately, the same cannot

be said about ceramics and CMC due to their extremely low static initiation

fracture toughness, i.e. KIC. With no tough ceramics in sight, design of a safe-fail

ceramic components is based on avoiding fracture all together or to promote the

use of ceramic components as a one-time energy absorber through fragmentation.

Both applications circumvent research in dynamic fracture mechanics of ceramics

and CMC.



In the following sections, a cursory review of the state of science of dynamic

fracture mechanics will be given. Procedures for dynamic fracture mechanics

characterization and properties peculiar to ceramics and CMC will be discussed.

HISTORICAL REVIEW

The early papers in dynamic fracture mechanics were simple extensions of

Griffith's instability criterion for predicting the onset of crack propagation. Mott1,

Roberts and Well3

and Berry4 added varying forms of an estimated kinetic energy

rate term to Griffith's balance of energy rate equation to account for the global

kinetics associated with a moving crack. This approach did not account for the

dynamic crack-tip state of stress and the possible difference in the static and the

dynamic fracture processes.

The moving Griffith's crack, which was derived by Yoffe2

during this early

period, did provide a crack velocity independent stress intensity factor for a crack

velocity dependent crack tip stress field. Using her solution, Yoffe predicted a

crack kinking angle of about 63o at a crack velocity of about sixty percent of the

shear wave velocity thus leading to a crack branching criterion which depended on

a critical crack velocity. While Yoffe's solution was a historical first, the anomaly

of her modeling resulted in an infinite energy release rate as the crack velocity

approached the Rayleigh wave velocity. Subsequent solutions for a constant

velocity crack initiating from zero and finite crack lengths by Broberg5 and

Baker6, respectively showed that the energy release rate approached zero as the

crack velocity approached the Raleigh wave velocity. The corresponding crack tip

stress fields were also characterized by crack velocity dependent stress intensity

factors.

Early views on crack arrest considered the arrest to be an inverse of the onset

of crack propagation7, namely that a propagating crack would arrest when the

instantaneous static stress intensity factor KI < KIC. Many tests and research

programs were conducted to verify or discredit this postulate with raging

controversies at times on the physical significance of dynamic crack arrest stress

intensity factor, KIa. Ensuing experimental8 and numeical

9 analyses, however,

suggested that KIa is a material property and that static analysis is not sufficient for

predicting the arrest of a propagating crack.

FUNDAMENTAL EQUATIONS IN DYNAMIC FRACTURE

Ceramics exhibits cleavage fracture at room as well as at elevated

temperature. This is fortunate since most of the theoretical developments in

dynamic fracture are confined to linear elastic fracture mechanics (LEFM).

However, the additional complexities of fiber pullouts and fracture involved in

dynamic fracture of CMC are yet to be addressed. Available theoretical solutions

in dynamic fracture are few, some of which are discussed in the following


sections, and are limited to a self-similar crack extending at a constant velocity in

an infinite solid. Despite these limitations these solutions can be used to deduce

the crack tip state of stress as well as to extract the dynamic stress intensity factor.

Stationary Crack Impacted by a Tension Wave

The dynamic initiation stress intensity factor, KId, of a stationary semi-infinite

crack, which is impacted by square plane tension wave of duration t, in an infinite

solid was given by Freund10

. For a ramp tensile pulse loading, KId is a simple

superposition of the discrete KId values of the corresponding incremental plane

tension waves. Unlike its static counterpart, this KId does not involve a

characteristic length dimension. If, however, the crack starts to propagate rapidly

after an incubation time, then the resultant stress intensity factor, KID, is modified

by a scalar function of the crack velocity.

Crack Propagating at Constant Velocity

The state of stress at the tip of a crack propagating at a constant velocity in a

two dimensional, isotropic, homogeneous elastic material has been derived by

Nishioka and Atluri11

who provided the asymptotic crack tip stress and

displacement fields in infinite series. The singular, first order term in the infinite

series with a dynamic stress intensity factor, KID, is the most significant term in

the crack tip stress field. In addition, the second order term has been shown to

govern crack kinking and branching angle12,13

.

DYNAMIC INITIATION FRACTURE TOUGHNESS

Dynamic initiation fracture toughness, KId, is commonly determined by impact

loading a ceramic or CMC fracture specimen by a split Hopkins bar or by a drop-

weight.

Split Hopkinson Bar Tester

The split Hopkinson bar tester, which has been used extensively for impact

testing of metals and ceramics, was modified by Duffy et. al.14,15

, as shown in

Figure 1, to impart tension directly to the specimen without a prior compression

wave. The compressive wave developed by an explosive charge is shaped into a

tensile pulse through multiple reflections and then propagates down the steel bar.

Duffy et al14,15

determined the dynamic initiation fracture toughness, KId, of

precracked Al2O3 and SiCw/Al2O3 bar specimens. Table I shows their Al2O3

results at room and elevated temperature testing. By adding a pre-torque to the

bar specimen, the test setup was also used to measure the fracture toughness under

combined modes I and III fracture 16

.


Figure 1 Split Hopkinson bar test for dynamic fracture testing14,15

.

Table I. Dynamic and static initiation fracture toughness of Al2O314,15

.

Temp KId KIC KId/KID

(oC) (MPa m

1/2) (MPa m

1/2)

20 3.5 2.7 1.3

900 3.4 2.2 1.5

1100 3.1 2.2 1.4

1300 2.0 1.4 1.4

Drop Weight Test

An early study on KId of ceramics was based on a static evaluation of the

impact data obtained from drop-weight loaded, single edge-notched (SEN), three-

point bend (TPB) specimens17

. This static analysis was subsequently replaced by

a dynamic finite element (FE) analysis of a pre-cracked21

, SEN TPB specimen to

which the impact load and the crack extension histories were prescibed18-20

. As

the load was measured outside the furnace for elevated temperature testing, the FE

model also included the impact rod in its load train. The crack extension history

was monitor by a calibrated laser interferometric displacement gage system22

.

The KId of Al2O3, TiB2p/SiC and SiCw/Al2O3 CMC thus obtained are listed in

Table II.

Table II Dynamic and static initiation fracture toughness of Al2O320

and CMC19

.

Mat’l Temp KId KIC KId/KIC

(oC) (MPa m

1/2) (MPa m

1/2)

Al2O3 20 5.7 4.3 1.3

1000 4.3

TiB2p/SiC 20 5.7 5.2 1.1

1000 5.1

SiCw/Al2O3 20 6.2 6.2 1.0

1000 6.1


Similar drop weight loading system and SEN-TPB pre-cracked21

specimens

were used to determine KId of PSZ and Si3N4 through a temperature of

1200oC

23,24. Caustics method combined with an ultra-high speed camera was

used to determine KId as well as KID during rapid crack propagation. These results

will be discussed together in a later section on dynamic crack propagation.

A novel variation of the drop weight testing is the one-point bend (OPB), pre-

cracked21

specimen which is suspended by ceramic threads in an infrared image

furnace25,26

. Since the thin threads are broken at the instant of impact, the

specimen breaks without any constraint at the two end supports. A pair of

semiconductor strain gages on the impact rod was used to measure the impact

force. KId was obtained from the equation of motion of the impacted OPB

specimen. The KId of five ceramics at room temperature and 1200oC are shown in

Table III. The KId rate was about 1.2 x 105 MPa m

1/2/sec. KId of SiC, Si3N4 and

Al2O3 remained constant but KId of PSZ decreased substantially at 600oC.

Table III Dynamic and static initiation fracture toughness of ceramics24

.

Ceramics Temp KId KIC KId/KID

(oC) (MPa m

1/2) (MPa m

1/2)

SiC 20 6.3 5.5 1.2

1200 5.9

Si3N4 20 6.0 6.0 1.0

1200 6.0

PSZ 20 7.0 4.0 1.8

600 3.0

Al2O3 20 5.2 4.5 1.2

1200 5.2

Al2O3/ZrO2 20 10.2 6.5 1.6

Instrumented Charpy Impact Test

Quasi-dynamic analysis of an instrumented Charpy impact test has been used

by T. Kobayashi et. al.27,28

for KId determination at room temperature. Unlike the

other results, their KId of Al2O3 and Si3N4 remained essential constant with

increasing KId rate and then suddenly increased at a KId rate of 105 MPa m

1/2/sec

as shown in Figure 2.

DYNAMIC FRACTURE OF CERAMICS AND CMC

Impact failures of ceramics and CMCs are characterized by shattering which is

a complex phenomenon involving a multitude of simultaneous micro-crack

generation, growth and coalescence into macro-cracks which in turn grow, branch

and coalesce. Intact fibers in CMC do not necessarily arrest a propagating crack

in the brittle ceramic matrix, as the propagating crack is known to tunnel around


Figure 2 Effect of KId rate on KId28

.

the fibers with little crack opening. Theoretically, a CMC could have a larger KId

and a dynamic crack arrest stress intensity factor, KIa, in order to resist rapid crack

propagation. Once the laws governing a single crack is known, a statistical or a

fractal analysis of the many branched cracks and the laws governing fiber

fracture/pull-out can be used to predict the overall dynamic response of the

impacted structural component.

Dynamic fracture mechanics study of a rapidly propagating crack in ceramics

and CMC, however, are virtually non-existent except for the papers by Shimizu et.

al.23,24

and the author and his colleagues18-20

. The experimental procedures used

were discussed in the previous section and thus only the results are presented in

the following.

Figure 3 shows the resultant crack velocity versus the KID relation for Al2O3

where little differences are noted between the data of room temperature and

1000°C. If the cluster of data at the left end did not exist, then the well-known

gamma shape curve, which has been observed in metals and polymers, could have

been obtained. Figure 3, however, shows that the crack continues to propagate

slowly, i.e. at speeds ranging from 10 to 40 m/s under a KID less than the KIC and

is consistent with previous findings29

. Also shown in this figure is the KID versus


crack velocity relation for statically loaded specimens under fixed displacement

loading at room temperature.

Figure 3 Crack velocity versus KID of Al2O319

.

Figure 4 shows the resultant crack velocity versus KID relations of TiB2p/SiC

CMC impacted at room temperature and 1200°C. The crack velocity under

impact loading is relatively constant during the entire crack propagation history.

Also shown is the crack velocity versus KID relation for a statically loaded

specimen.

Figure 4 Crack velocity versus KID of TiB2p/SiC19

.


Figure 5 shows KID versus the resultant crack velocity of PSZ impacted at

room temperature. KIa, which was not observed by others, was obtained after the

crack had propagated about 100 m/sec and arrested at Kia = 4.0 MPa m1/2.

.

Figure 5 KID versus crack velocity relation of PSZ23

.

Figure 6 shows the KID versus crack velocity relation of Si3N4. The crack

velocity observed in Figures 5 and 6 are order of magnitude higher than those of

Figures 3 and 4, possibly due to the sharp but machined notch tip, which obliterate

the trailing fracture process zone of a real crack30

and yielded a larger KId and

hence higher stored energy to drive the crack. While the trend of a decreasing

crack velocity with decreasing KID is observed, the available data in Figures 2, 4

and 5 do not indicate the existence or lack of existence of a KIa in these ceramics

and CMC which lack the stress induced transformation of PSZ. Crack arrest,

however, has been observed in chevron-notched, three point bend specimens,

which were machined from the same SiCw/Al2O3 ceramic composites and which

were loaded under an extremely small displacement rate of 0.01 mm/min29

. The

run-arrest events in this test were characterized by small crack jumps of about 0.8

mm, which initiated at the sharp crack tip in the chevron notched specimens. Once

the excess driving force had been dissipated during rapid crack propagation under

static loading and the crack had entered a region of KID < KIC, crack arrest was to

be expected. Figures 3 and 4 show that such was not the case.


Figure 6 KID versus crack velocity relation of Si3N424

.

FRACTURE MORPHOLOGY

The lack of crack arrest was then attributed to the difference in the fracture

morphologies of extremely slow and rapid crack extensions. This postulate was

tested by extensive fractography analysis of the statically and impact loaded Al2O3

specimens31

. While intergranular fracture was the dominant failure mode in both

specimens, some transgranular fracture was observed in all regions of the fracture

surface. The percentage areas of transgranular fracture decreased from an average

of 16% during the initiation phase to an average of 10% at slower crack

propagation in the impacted specimen. For the statically loaded specimen, the

percentage of transgranular areas decreased from 5 to 2%. The higher percentage

areas of transgranular fracture during the initiation phase can be attributed to the

higher crack velocity and the higher KID due to the overdriving force generated by

the blunt crack tip. Fractography analysis also showed that rapid crack

propagation is always accompanied with transgranular fracture regardless of the

magnitude of the driving force, i.e., KID and the crack velocity. In contrast, the

fracture morphology for stable crack growth showed the dominance of

intergranular failure. The percentage area in excess of 10% at the lowest KID of

about 1.5 MPa m1/2

was obtained from the impacted specimen. This data suggests

that the continuous input of work during the fracture process generated a higher

percentage area of transgranular failure with little chance of crack arrest.

The failure energy of a single crystal ceramic, i.e., energy required for

transgranular fracture, is generally higher than that of a polycrystalline ceramic

thus suggesting that transgranular failure requires more energy than intergranular


failure. Transgranular failure thus provides a larger driving force but also a

competing higher resistance.

These results on Al2O3 showed that under rapid crack propagation,

transgranular fracture does occur both at high as well as at a low. KID was

associated with the corresponding high and low crack velocities, respectively.

However, the kinematic constraint of a rapidly extending flat crack front must

have enforced a locally moderate transgranular failure and drove the crack at a

lower KID, thus reducing the chance for crack arrest even at KID < KIC. A low

percentage area of transgranular failures, i.e. 2%, thus continued to drive the crack

at a subcritical KID.

CONCLUSIONS

The paucity of data is symptomatic of the developing state of this field where

few laboratories throughout the world have the resources to undertake dynamic

and impact characterization of ceramic composite at an operating temperature in

excess of 1200o C. Thus, much of the actual test data was limited to room

temperature testing of Al2O3 and Si3N4.

The limited data suggests that:

1. Dynamic initiation fracture toughness, KId, of most ceramics and CMC is

slightly larger than the static initiation fracture toughness, KIC.

2. Significant increase in KId of PSZ at room temperature was observed.

Likewise, the KId of PSZ at 600oC decreased significantly.

3. For a rapidly propagating crack in PSZ at room temperature, the dynamic

crack arrest stress intensity factor, KIa, was almost equal to its KIC.

ACKNOWLEDGMENT

The author gratefully acknowledges the financial support of the Office of

Naval Research through ONR Contract N00014-87-K-0326 through which many

of the results reported in this paper were generated.


REFERENCES1N.F. Mott, "Fracture of Metals: Theoretical Considerations," Engineering, 165,

16-18 (1948).2E.H. Yoffe, "The Moving Griffith Crack," Philosophical Magazine, 42, 739-

750 (1951).3D.K. Roberts and A.A. Wells, "The Velocity of Brittle Fracture," Engineering,

178, 820-821 (1954).4J.P. Berry, "Some Kinetic Considerations of the Griffith Criterion for

Fracture," Journal of Mechanics and Physics of Solids, 8, 194-216 (1960).5K.G. Broberg, "The Propagation of a Brittle Crack," Arkiv for Fysik, 18, 159-

192 (1960).

6B.R. Baker, "Dynamic Stresses Created by a Moving Crack." ASME Journalof Applied Mechanics, 29, 449-458 (1962).

7G.R. Irwin, "Basic Concepts for Dynamic Fracture Testing," ASME Journal ofBasic Engineering, 519-524 (1969).

8.J.G. Kalthoff, J. Beinert, S. Winkler and W. Klemm, "Experimental Analysisof Dynamic Effects in Different Crack Arrest Test Specimens," Crack ArrestMethodologies and Applications, eds. G.T. Hahn and M. E. Kanninen, ASTM STP711, 109-127 (1980).

9L. Dahlberg, F. Nilsson and B. Brickstad, "Influence of Specimen Geometryon Crack Propagation and Arrest Toughness," Crack Arrest Methodologies andApplications, eds. G.T. Hahn and M.E. Kanninen, ASTM STP 711, 89-108(1980).

10L. B. Freund, “Crack Propagation in an Elastic Solid Subjected to GeneralLoading IV-Stress Wave Loading,” Journal of Mechanics and Physics of Solids,21, 47-61 (1973).

11T. Nishioka and S.N. Atluri, "Numerical Analysis of Dynamic CrackPropagation: Generation and Predictions Studies," Engineering FractureMechanics, 16 [3], 303-332 (1982).

12M. Ramulu and A.S. Kobayashi, "Dynamic Crack Curving - A PhotoelasticEvaluation," Experimental Mechanics, 23, 1-9, (1983).

13M. Ramulu, A.S. Kobayashi and B.S.-J. Kang, "Dynamic Crack Branching -A Photoelastic Evaluation," Fracture Mechanics; Fifteen Symposium, ed. R.J.Sanford, ASTM STP 833, 130-148 1(984.)

14J. Duffy, S. Suresh, K. Cho, and E. Bopp, "A Method for Dynamic FractureInitiation Testing of Ceramics, " Transaction of ASME, 110 (4), 325-331 (1989).

15S. Suresh, T. Nakamura, Y. Yeshurun, K.-H. Yang and J. Duffy, "TensileFracture Toughness of Ceramic materials: Effects of Dynamic Loading andElevated Temperatures," Journal of American Ceramic Society, 73 (8), 2457-66(1990).

16S. Suresh and E. K. Tschegg, "Combined Mode I-Mode III Fracture ofFatigue Precracked Alumina," Journal of American Ceramic Society, 70 (10) 726-733 (1987).

17S.T. Gonczy and D.L. Johnson, "Impact Fracture of Ceramics at HighTemperature," Fracture Mechanics of Ceramics, 3, edited by. R.C. Bradt, D.P.H.Hasselman and F.F. Lange, Plenum Press, 495-506, New York (1978).

18A.S. Kobayashi, A.F. Emery, and B.M. Liaw "Dynamic Fracture Toughnessof Reaction Bonded Silicon Nitride," Journal of American Ceramic Society, 66 (2),151-155 (1983).

19K.-H. Yang, A.S. Kobayashi and A.F. Emery, "Effects of Loading Ratesand Temperature on Dynamic Fracture of Ceramics and Ceramic MatrixComposites," Ceramic Materials and Components for Engines, edited by V.J.Tennery and M.K. Ferber, American Ceramic Society, 766-775 (1989).

20Y. Takagi and A.S. Kobayashi, ”Further Studies of Dynamic FractureResponses of Alumina and SiCw/Al2O3 Composite,” Proceedings of the


Symposium on Elevated Temperature Crack Growth, edited by S. Mall and T.Nicholas, ASME MD-Vol. 18, 145-148 (1990).

21T. Nose and T.Fujii, “Evaluation of Fracture Toughness for CeramicMaterials by a Single-Edge-Precracked-Beam Method,” Journal of the AmericanCeramic Society, 71 (5), 328-333, (1988).

22M.G. Jenkins, A.S., Kobayashi, M. Sakai, K.W. White and R.C. Bradt,"Fracture Toughness Testing of Ceramics Using a Laser Interferometric StrainGage," Bulletin of American Ceramic Society, 66 (12), 1734-1738 (1987).

23K. Shimizu, S. Takahashi and M. Suetsugu, “On the dynamic fracturebehavior of the structural ceramics at various temperatures by caustics,” RecentAdvances in Experimental Mechanics, edited by S. Gomes et. al., A.A. Balkema,Rotterdam, 1329-1334 (1994).

24M. Suetsugu, K. Shimizu and S. Takahashi, “Dynamic Fracture Behavior ofCeramics at Elevated Temperature,” Experimental Mechanics, 38 (1), 1-7 (1998).

25M. Sakata, S. Aoki, K. Kishimoto, Y. Fujino and T. Kiba, “Measurement ofDynamic Fracture Toughness of Ceramic Material at Elevated Tempratures byImpact Test with Free End Bend Specimen,” Materials, 37 (419), 34-39 (1988), inJapanese,

26S. Takahashi, S. Aoki and K. Kishimoto, “Impact Fracture Toughness Test ofCeramics and Ceramic Reinforced Metals at Elevated Temperature,” Proceedingsof the VII International Congress on Experimental Mechanics, II, 1581-1586(1992).

27T. Kobayashi, K. Matsunuma, H. Ikawa and K. Motoyoshi, “Evaluation ofStatic and Dynamic Fracture Toughness in Ceramics,” Engineering FractureMechanics, 31 [5], 873-885 (1988).

28T. Kobayashi, “Toughness problems in advanced materials,” InternationalJournal of Materials and Product Technology, 14 (1/2/3), 127-146 (1999).

29M.G. Jenkins, A.S. Kobayashi, K.W. White and R.C. Bradt, “Crack Initiationand Arrest in SiC Whisker/Al2O3 Matrix Ceramic/Ceramic Composites,” Journalof American Ceramic Society, 70 (6), 393-395, (1987).

30Z.K. Guo, J.C. Hay, K.W. White and A.S. Kobayashi, “Fracture ProcessZone Modeling of Monolithic Al2O3,” Engineering Fracture Mechanics, 63, 115-129 (1999).

31W.J. Yang, C.T. Yu and A.S. Kobayashi, “SEM Quantification ofTransgranular vs Intergranular Fracture,” Journal of the American Ceramic Society,74 (2), 290-295 (1991).


COMPRESSIVE FRACTURE OF BRITTLE SOLIDS UNDER SHOCK-WAVE

LOADING

G. I. Kanel S. J. Bless

Institute for High Energy Densities The University of Texas at Austin

IVTAN, Izhorskaya 13/19, Institute for Advanced Technology

Moscow, 127412 Russia 3925 W. Braker Lane, Suite 400

Austin, Texas 78759

ABSTRACT

The behavior of different kinds of brittle materials, including single crystals,

glasses, and ceramics, under shock wave loading (uniaxial strain conditions) and

impact loading under uniaxial stress conditions, is reviewed and compared from

the viewpoints of mechanisms and criteria of plastic deformation and compressive

fracture.

GENERAL BEHAVIOR OF BRITTLE MATERIALS UNDER COMPRESSION

Mechanisms of inelastic deformation of brittle materials under compression

were initially investigated for rocks (see review papers1,2,3

). It was found that

fracture under one-dimensional stress conditions or at relatively low confining

pressure often occurs by axial splitting. For greater confining pressures, failure

occurs by shear faulting at an angle less than 45˚ to the loading axis. Extensive

compressive fracture is preceded by microcracking. Orientations of microcracks

are predominantly within 10˚ of the direction of compression.4 Crack density

increases as macroscopic deviatoric stress increases above a distinct threshold

level. Faults and other macroscopic fractures appear to form after attainment of

the ultimate compressive stress, which is called the failure stress. Beyond the

point of peak load, the failure becomes unstable. In the post-failure region of

compression, the load-carrying capacity drops rapidly to a low value.

Since cracks occupy volume, their formation is accompanied with a decrease

in the average matter density. This nonlinear inelastic volume change is

commonly referred to as dilatancy or bulking.5 The bulking effect grows with

increasing deviator stress and decreases under confining pressure. Typically the

onset of the dilatation region occurs between one-third and two-thirds of the

failure stress. Unloading from this stress region yields a permanent residual

volume increment.



The dilatancy is accompanied by a hysteresis in physical properties between

loading and unloading which is manifested mostly in the lateral strains, while the

axial strain is nearly elastic and almost completely recoverable. The lateral

dilatantional strains are attributed to opening of axial cracks. Formation of open

axial cracks at a fraction of the maximum stress is also suggested by variations of

sound velocity in axial and radial directions: velocity in the axial direction is

hardly changed by stress, whereas the sound velocity in the radial direction begins

to decrease at about half the failure stress and may drop 10 to 20%.5

A confining pressure strongly affects the strength and inelastic behavior of

brittle materials. The deviator stresses at which microcracking starts or failure

occurs increase as the confining pressure increases. At a sufficiently high pressure

a transition from brittle to ductile response usually occurs. For example, Heard

and Cline 6 observed failure followed immediately after essentially elastic

deformation of Al2O3, AlN, and BeO when the confining pressure was low, but

there was a transition to more ductile response at high pressures. The ultimate

compressive strength of ceramics increases rapidly with pressure below the brittle-

ductile transition; above this threshold, the ultimate strength is nearly constant.

These ceramics also exhibit increasing ductility when the confining pressure is

above the brittle-ductile transition. The pressure of transition from brittle to

ductile response is different for different materials. Alumina, for example,

remains brittle at confining pressures at least up to 1.25 GPa. However, extensive

evidence of ductility by both slip and twinning was observed in alumina and

sapphire at room-temperature indentation deformation.7,8

POSSIBLE MECHANISMS OF MICROCRACKING UNDER COMPRESSION

Open cracks, like other voids, may nucleate and grow only when at least one

principle stress is tensile. Even in the case of overall compressive loading, small

regions of tension may appear inside the body as a result of modification of the

stress field by different concentrators, such as grain boundary contacts,

microcracks, and cavities in the incident materials, etc.9,10

Hence, cracks may

grow in response to this local tensile stress.

Intuitively, it appears that if easy shear is allowed within a limited band with

fixed tips inside a body, rarefaction and compression regions will be created near

the tip, as illustrated schematically in Fig. 1. There should also be concentration of

shear stresses in the crack plane ahead of the tip. The rarefaction may initiate a

tensile crack which can grow in the direction perpendicular to maximum tension

out of the crack plane, whereas localized shear may propagate further in the crack

plane. For different materials and various loading rates, Kalthoff observed that

both of these modes of shear failure initiated at the crack tip.11


Rarefaction

Compression

Shear band

Tensile

crack

Figure 1. Schematic of the failure initiation at mode-II crack tip.

Griffith postulated that isotropic materials contain randomly oriented flaws or

cracks in all directions which significantly alter the stress field within the

material.9

The basic hypothesis of Griffith’s model is that fracture occurs when

the most vulnerably oriented crack begins to extend under applied stress. The

extension of the crack is assumed to occur when the maximum tensile stress

component at any point around the crack reaches the critical value needed to

overcome the interatomic cohesion of the material. The Griffith theory, or at least

its basic premise that fracture starts from flaws, is fundamental to all

investigations of brittle fracture.

Brace and Bombolakis observed the growth of cracks in glass and polymer

plates under compression.12

They found that the most severely stressed cracks

were inclined at about 30˚ to the axis of compression. The cracks, when either

isolated or placed in an array, grow along a curved path that becomes parallel with

the direction of compression. When this direction is attained, growth stops. The

resultant kinked crack consisted of a central crack with sliding surfaces, which is

inclined to the direction of compression, plus two cracks emanated from its ends,

which are called “wing cracks”. Modern theories of brittle fracture and dilatancy

under compression are mostly based on the development of the wing crack model.

More recently, a series of similar experiments was performed by Nemat-

Nasser and Horii.13,14

They also have shown that the relative sliding of the faces

of one or even an array of pre-existing cracks leads to the formation of tension

cracks which grow in the direction of maximum compression. A lateral

compression reduces the final crack length, whereas even small lateral tension

increases it. In the model experiments with specimens containing a number of


randomly oriented cracks or rows of inclined coplanar cracks, axial splitting,

rather than localized shear failure, was observed.

Thus, in both natural and man-made materials, crack growth is directed

preferably along the compression and does not immediately produce a mechanical

instability as it does in tension. The tensile axial cracks which may open and grow

at stress concentrators under overall compression cannot be considered

themselves as a mechanism of an inelastic shear strain, but they may facilitate

shear and rotation of blocks of matter relative to each other and in this way to

contribute to deformation. The stress required to cause additional crack extension

increases after some crack growth has been initiated. Macroscopic faults are

formed out of systems of cracks.

DYNAMIC STRENGTH PROPERTIES OF SINGLE CRYSTALS AND

GLASSES

First consider plate impact experiments on sapphire and ruby samples backed

by water, in which the rear surface velocity histories were measured.15,16

Results

are shown in Fig. 1. These data exhibit most of the peculiarities of the response of

single crystals of hard brittle materials to shock-wave loading. In the experiment

with ruby, the peak stress did not exceed the Hugoniot elastic limit (HEL). The

interface velocity history is smooth and mimics the shape of the stress pulse inside

the sample. The high velocity pullback indicates the dynamic tensile strength (the

spall strength), is as high as 10 GPa. In the other shot, the peak stress exceeded the

HEL. The spall strength drops practically to zero, and irregular oscillations appear

in the wave profile. Vanishing resistance to tension after shock compression

above the HEL was observed for quartz single crystals as well.17

Presumably, the

absence of fracture nucleation sites enables high spall strength at peak shock

stresses below the HEL. However, fracture nucleation sites obviously appeared

during shock compression above the elastic limit.

The high-frequency particle velocity jitter is evidence for heterogeneity of the

inelastic deformation process. Similar records have been obtainable for quartz17

and olivine18

. Another characteristic feature is the significant stress relaxation

behind the elastic precursor front caused by intense multiplication of the

deformation carriers. This also is typical for brittle crystals.18,19

Figure 3 presents the results of measurements of shock compressibility of

sapphire crystals. Above the HEL, the Hugoniot shows a collapse toward the

isotropic compression curve (hydrostat): the stress offset above the HEL is 3.8 to

4.3 GPa, while at the HEL it ranges from 5.5 to 11 GPa. The collapse of the

Hugoniot indicates a collapse of shear strength. Quartz22

, magnesium oxide19,23

,

zirconia24

, iron-silicate almandine-garnet25

, and olivine single crystals18

also

The HEL is limiting stress for linear elastic compression under uniaxial strain;

e.g., it is the compressive strength for full lateral confinement.


exhibit this phenomenon. Grady19

found that in magnesium oxide crystals,

strength persists at the Hugoniot state and the release paths deviate significantly

from the hydrostat; the initial elastic release velocities was some below expected

longitudinal (elastic) velocities but was substantially above expected bulk

(inelastic) sound velocities.

0.0 0.2 0.4 0.6

0.0

0.2

0.4

0.6

0.8

1.0

Expected

spall signal

Sapphire

Ruby

Ve

loc

ity

, k

m/s

Time, s

0,85 0,90 0,95 1,00

0

15

30

45

K0=226 GPa,

K'=4.0

Str

ess, G

Pa

V/V0

Figure 2. The wave profiles at interface between alumina single crystals and water

window at shock wave loading up to various peak stresses.

Figure 3. Stress-volume relations for sapphire under shock-wave compression.

Points present the data.20,21

The dot-dashed line shows the isotropic compression

curve.21

Wang and Mikkola examined recovered sapphire samples with transmission

electron microscopy after shock compression up to 23 GPa.26

They observed a

significant number of slip bands in different crystallographic directions and

suggested that a large amount of plastic deformation had occurred at shock

stresses of 12 GPa and more. Anan’in et al. have revealed glass-like interlayers

between quartz blocks in recovered single crystals after shock loading.27

This

lamellae structure indicates a heterogeneous nature of shock deformation of

quartz, accompanied by melting. Grady has developed a model of localized

dissipation of elastic strain energy in low thermal conductivity strong solids.28

The dissipation leads to local temperature growth, which reduces the local flow

stress, causing the shear strain and the energy release to be localized within

narrow bands where the temperature may reach melting. However, no signs of

local melting were observed in sapphire and other hard materials.

Thus, hard single crystals show a more or less substantial reduction in shear

strength at shock compression beyond their Hugoniot elastic limits. Within the

elastic range, they demonstrate very high dynamic tensile strength, which is


attributed to lack of flaws and heterogeneities. At shock compression above the

HEL, they show vanishing dynamic tensile strength.

Unlike crystals, silicate glasses maintain high tensile strength after shock

compression above the HEL. Figure 4 presents the free surface velocity profiles

for K8 crown glass.29

Spallations were not observed in these shots, which means

that the spall strength of the glass exceeds 6.8 GPa when the shock stress is below

the HEL, and it remains very high above the HEL. For comparison, the static

tensile strength of glasses is around 0.1 GPa. The reason for such a large

discrepancy is that the fracture nucleation sites in homogeneous glass are

concentrated on the surface. These incident microcracks are activated and

determine the strength magnitude in the static measurements, whereas spall

strength is an intrinsic property of matter.

0,0 0,5 1,0 1,5 2,0

0,0

0,6

1,2

1,8

Reflection from at a failed layer

1Simulation

2

Calculated rereflectionat the impact surface

Fre

e S

urf

ace

Ve

locity (

km

/s)

Time ( s)

Figure 4. Experimental results for K8 glass samples 6.1 mm thick at the impact

velocities of 670 30 m/s (profile 1), and 1900 50 m/s (profile 2). Impactors are

steel 0.9 mm thick (1) and aluminum 2 mm thick backed by paraffin (2). Dashed

line shows results of computer simulations assuming no failure.

At high pressures, brittle glasses become ductile. Ductility of glasses is

possible because there is a loose microstructure with a large amount of molecular-

size voids. It is known that glasses show gradual structural changes, resulting in

increased density.30

Irreversible densification of some glasses also occurs under

shock compression above the HEL.31,32

It is supposed that the irreversible

densification and compaction in the silicate structure are responsible for the

plastic flow properties of glasses under high pressure.33

Once the plastic flow

starts, stress relaxation reduces the stress concentration at crack tips and thus stops

the propagation. The high spall strength revealed in the stress range above the

HEL means that the ductility is preserved during the subsequent tensile loading.


Single crystals and glasses are initially homogeneous in bulk so the only way

fracture nucleation sites may be formed is in the course of plastic deformation. In

this sense the difference between single crystals and glasses is that hard crystals

have only a limited number of crystal planes and directions in which the usual

mechanisms of ductility may work, whereas the ductility of amorphous glasses is

completely isotropic. The impossibility of plastic shear along arbitrary directions

in crystals results in stress concentration at points of intersections of slip bands or

twins that, in turn, may result in cracking at compression or unloading.

Comparison of the measured free surface velocity history (curve 1 in Fig. 4)

with results of computer simulations shows that the elastic wave reverberation

inside the glass plate sample occurs earlier than expected. The early arrival of

second compression pulse is due to reflection from a failure wave at some

distance from the impact surface.

The failure wave is a network of cracks that are nucleated on the surface and

propagate into the stressed body. There are many observations of fracture front

propagation in glasses under tensile stresses. Schardin recorded expansion of

fractured areas with a sharp front formed by bifurcated cracks.34

Galin et al.

reported an explosion-like fracture under bending of high-strength glass with

removed surface defects.35

The explosion-like fracture was treated by Galin and

Cherepanov, who termed it a self-propagating failure wave.36

The similar fracture mode under compression was revealed in shock-wave

experiments with glass plates. Some results of observations of failure wave

phenomena were reviewed recently.29,37

The observations may be summarized as

follows: (i) failure waves are observed when the impact stress exceeds some

threshold but is still below the Hugoniot elastic limit of glass; (ii) failure waves

nucleate at a plate surface; (iii) decrease of deviator stresses and vanishing of

tensile strength occur behind the failure wave front; (iv) propagation of the failure

wave stops when the stress in front of it decreases; and (v) the failure wave

velocity is much less than the sound speed. Many measurements give the failure

wave speed equal to an ultimate speed of growth of cracks (~1.5 km/s for glass),

but higher and lower velocity values were reported as well. Both constant and

decreasing propagation velocities were reported.

Failure waves present a mode of catastrophic fracture in an elastically

compressed media that is not limited to impact events. One hopes that the

investigations of failure waves in shock-compressed glass will provide

information about the mechanisms and general rules of nucleation, growth, and

interactions of multiple cracks and lead to better understanding of experiments

with other hard brittle materials, such as ceramics and rocks.

DYNAMIC STRENGTH PROPERTIES OF POLYCRYSTALINE CERAMICS

Modern shock-wave tests of ceramics include measurements of the Hugoniot

over a wide stress range, shock front rise time, Hugoniot elastic limit (HEL),


stress state immediately after shock compression, high pressure stress strain path,

tensile (spall) strength after shock compression below and above the elastic limit,

and post-test examination of recovered samples. Fig. 5 shows typical particle

velocity histories for SiC and B4C ceramics.38

The wave profiles in Fig. 5 exhibit two extreme examples of behavior of

ceramics in plane shock waves. The response of silicon carbide is very similar to

that of ductile materials. There is an initial elastic arrival whose amplitude (the

HEL) is the limit stress for elastic behavior, there is a second shock corresponding

to bulk compression, followed by elastic and bulk unloading waves that originate

from the rear surface of the flyer plate. Post-yield strength of silicon carbide,

determined by comparison of uniaxial strain and calculated hydrodynamic

response, increases considerably beyond the initial dynamic yield. The release

trajectories for silicon carbide indicate reverse yielding (e.g. reversal of the sense

of shear) and continued elastic-plastic bulk behavior, probably with a Baushinger

effect at higher peak stresses. The shock response of B4C is quite different. The

bulk compression wave is much slower. According to Grady, the Hugoniot

collapses to the hydrostat at stresses approaching about twice the HEL.39

A

dispersed character of the unloading wave indicates inelastic strain starts almost

immediately behind the rarefaction wave front. The stress-strain trajectory for the

B4C ceramic 38,39

shows evidence of dilatancy when the compressive stress

approaches zero on unloading.

Correspondingly, post-yield characteristics of the materials are qualitatively

contrasted by the shape of the bulk compression waves. For silicon carbide,

positive slope of the wave demonstrates strain hardening. The stress drop after the

HEL in boron carbide, in contrast, indicates post-yield softening. Spall strength is

sustained for shocks above the HEL in SiC, but not in B4C.

0,0 0,5 1,0 1,5

0,0

0,5

1,0

1,5

SiC

B4C

Pa

rtic

le V

elo

city, km

/s

Time, s

Figure 5. Particle velocity profiles for SiC and B4C ceramics measured at the

interface with a LiF window.38


The dynamic compressive properties of many ceramics have been summarized

recently in our report.44

Here we’ll consider, as an example, the published data

for alumina. Table I and Figs. 6 and 7 present the shock data (longitudinal sound

speed, cl, the Hugoniot elastic limit, HEL, and the Von Mises yield stress,

Y = HEL (1-2 ) / (1- ) as functions of the initial density, 0) for different Al2O3

ceramics. Whereas the general trend is HEL reduction with increasing ceramic

porosity, the impurity content, the grain size, and material processing also

influence the HEL value. The compaction of more porous ceramics occurs within

the stress range from the yield point to about 30 GPa. At the higher stresses the

states of all alumina ceramics are practically described by one curve in stress-

volume coordinates. Beyond the compaction region, the yield strength, Y,

estimated from the stress offset between the Hugoniot and isotropic compression

curve is comparable to but somewhat smaller than the yield strength at the HEL.

The profiles of shock compression waves propagating through alumina

ceramics exhibit an elastic jump and a subsequent dispersed rise to the bulk wave

which compresses the matter to a final state. This gradual transition from the

elastic to inelastic portions of the compression wave is typical for strain hardening

materials. Cagnoux and Longy measured the free surface velocity profiles for

alumina at various rise times of the compression wave entering into the sample.45

The HEL was found to be independent of the wave propagation distance, the peak

shock stress, and the entering stress gradient; this means that there is no influence

of strain rate on the yield strength of alumina in a range of 5 104 to 6 10

5 s

-1. On

the other hand, Furnish and Chhabildas found evidence of rate-dependent

behavior of AD995 ceramic at step-like compression.46

According to many

measurements, the unloading wave front in shock-compressed alumina is elastic;

however, there is no sharp distinction between the elastic and inelastic parts of

unloading.

Longy and Cagnoux found that alumina ceramics with 2% porosity but with

grain size of 5 or 60 m exhibit a HEL of 8.5 GPa and 5 GPa, respectively.56

Microscopic examination of impure alumina showed microcracks in the inter-

granular glassy phase after shock stress 0.9 HEL, and there was no correlation

between the HEL and microcrack levels.

Cagnoux carried out microscopic examination of alumina samples of two

different grain sizes (4.7 m and 10-20 m) with 99.7% Al2O3 content and 3.91

g/cm3 density.

52The samples were recovered after compression above their HEL

by spherical shock waves. In the region of maximum peak stresses, the fine-grain

alumina remained uncracked, whereas the coarse-grain sample was

microfragmented. SEM photographs of fine-grain samples showed reduction in

porosity, with no slip-nucleated microcracks; in the coarse-grained sample,

numerous twins were observed. It was concluded that twinning is favored by large

grain size, while slipping by small grain size.


Table I. Hugoniot elastic limits of aluminas

Material (wt. fractionAl2O3), Grain Size

0, g/cm3,

(Porosity, %)

cl, km/s Poisson’s

ratio,

HEL,GPa

Y, GPa Ref.

Lucalox (99.8%) 3.98 (<0.2) 10.95 0.2363 11.2 1.3 7.8 43

Lucalox (99.9%),

25-40 m

3.969 10.92 9.1 0.4 6.0 44

MTU JS-I (99.99%), 1.5

m

3.974 10.9 0.237 11-11.9 7.6-8.2 45

D999 (99.9%), 4 m 3.99 10.82 0.232 13-14 9-9.8 46

Carborundum hot pressed 3.92 (0.8) 10.59 0.243 9.2-16 47

Wesgo Al-995 (99.5%) 3.81 (3.5-4.3) 10.2 0.218 8.3 0.5 6.0 43

D975 (97.5%), 4 m 3.8 10.3 0.234 7.5-9 5.2-6.2 46

Coors AD995,aluminosilicate glassbinder

3.88 (2) 10.56 6.7 0.1 48

Coors AD995 3.89 10.59 0.234 6.2 0.4 4.3 49

Coors AD-85 3.42 (6.6) 8.84 0.256 6.1-6.5 4.1 47

Coors AD-85 (84%) 3.42 (6.6) 4.7-6.1 50

H880 (88%), 2 m 3.55 9.1 0.226 5.5-6.5 3.9-4.6 46

Diamonite P-3142-1 3.72 (5.5) 9.98 0.234 7.2-8.1 47

Desmarquest alumina 3.62 (5.3) 9.45 4.5 45

ENSCI, 4.7 m 3.91 (2) 10.63 8.7 0.4 51

ENSCI, 1 m 3.54 (11) 9.34 5 51

ENSCI, 0.6 m 3.31 (17) 8.55 4-5 51

ENSCI T60 (99.7%), 5-

125 m

3.85 (3.5) 10.32 5 51

UL500 (93.8%), 11 m 3.62 (6.2) 9.77 6.5 51

3.2 3.4 3.6 3.8 4.0

8.5

9.0

9.5

10.0

10.5

11.0

cl, k

m/s

Density, g/cm3

Figure 6. Longitudinal sound speed, cl, in different Al2O3 ceramics as a function

of their density.


3.2 3.4 3.6 3.8 4.00

3

6

9

12

15

Lucalox

5-125 m

AD995AD-85

0.5-5 m

HE

L,

GP

aDensity, g/cm

3

Figure 7. Hugoniot elastic limits of Al2O3 ceramics of different density and grain

size.

Table II summarizes the results of rod impact experiments which provide the

dynamic failure threshold under uniaxial stress conditions.76

Comparison of data

of Tables I and II shows that the failure threshold for rod impacts is lower than the

dynamic stress Y at the HEL for the same material. For the ceramic rods mounted

within a close-fitting high-impedance sleeve, the maximum axial stress may

approach the HEL value.

Table II. The failure thresholds for alumina ceramics in rod impact experiments.

Material, impact

conditions

Confine-

ment

Failure Threshold, GPa Reference

AD94, direct impact - 2.7 54

AD99, direct impact - 4 54


AD995, direct impact Ta 5.8 6.3 55

AD995, dispersed impact - 3.5 4.2 53

AD995, dispersed impact Steel 4.6 53

AD995, direct impact - 3.6 3.7 56

AD995, direct impact Steel 4.2 56


Figure 8 presents results of spall strength measurements as a function of

normalized peak stress for alumina ceramics.45,48,58,59

It seems the spall strength

undergoes a transition, first decreasing near the HEL, then increasing with

increasing pressure above the HEL. The reduction in spall strength value near the

HEL is especially significant for aluminas with a large glassy phase content. This

observation correlates with observed 51

microcracks in the inter-granular glassy


phase at shock stress of 0.9 HEL and over. Within the elastic region, the spall

strength decreases with increasing porosity and grain size.

0,1 1

0,0

0,5

1,0

1,5

0,3 0,5 432

Pure hot pressed,

HEL=10-12 GPa

MTU JS-1, HEL=11.9 GPa

AD995, HEL=6.7 GPa

AD85, HEL=6GPa

Spa

ll S

tre

ngth

, G

Pa

(Peak Stress) / HEL

Figure 8. Spall strength of alumina ceramics as a function of peak stress. Data for

AD8558

, for and for hot pressed pure alumina.45,59

BRITTLE FAILURE CRITERIA AND MODELS

Reviews of models and criteria of brittle fracture in quasi-static compression

are available.3,60

Most models consider failure as a consequence of three

sequential events: crack initiation, crack propagation, and crack coalescence.

According to the Griffith’s criterion, crack initiation occurs when the highest local

tensile stress at the longest crack of the most dangerous orientation reaches a fixed

critical value. For a biaxial stress state, the corresponding relationship is

( 1- 2)2

8 f ( 1+ 2)=0, (1)

where 1, 2 are principal stresses, f is a material constant, which is assumed to

be the ordinary tensile stress for uniaxial stressing. Thus, Griffith’s criterion

predicts the value of the uniaxial compressive strength to be eight times the value

of the uniaxial tensile strength. This ratio is smaller than the ratio commonly

measured for rocks and other brittle materials.

Chen and Ravichandran have found that Mohr-Coulomb-like criterion,

generally used as a yielding criterion for sand-like materials, fits rather well the

experimental data on failure of ceramics under lateral confinement.61,62

The

criterion is expressed by a relationship


+ p = 0, (2)

where is the absolute value of shear stress, p is the pressure or the mean

stress, 0 is the shear strength of the material without any pressure, and is the

internal friction coefficient. It is expected that the value is between -1.5 and 0.

For AlN ceramic, = -1; for Macor glass ceramic, = -0.74. Actually, in the

Coulomb’s interpretation 63

, there should be normal stress component

perpendicular to the shear direction used in Eq. (2) instead of the mean stress. In

this case, the competition between the resolved shear force and the friction

forces p explains the failure angle <45 observed in rocks and granular materials

under compression.

In general, theoretical criteria, such as that of Griffith, often give an

inadequate fit to the data. Because of that, empirical criteria have been developed

to meet the practical requirements of accurate strength prediction and simplicity of

use. Since different mechanisms of inelastic deformation and failure may operate

depending on the region of stress space, different strength criteria should be used

at different levels of stress. For computations of impact problems, continuum

damage mechanics models are needed. Such models operate with defined damage

parameters and include an evolutionary equation for the damage and a constitutive

equation that relates the stress and strain to the damage. In this sense the model of

Ashby and Sammis is very illustrative (Figure 9).64

Yield

YieldFailure

Damage initiation

Failure

Damage initiation

-1

-2=-

3

Figure 9. Failure and yield surface according to Ashby and Sammis.64

COMPARISON OF 1-D-STRESS AND 1-D-STRAIN DATA FOR CERAMICS

AT VARIOUS STRAIN RATES

To what extent do the strength properties of ceramics depend on strain rate?

For many metals the strain rate sensitivity of the flow stress increases steeply

above ~103 – 10

4 s

-1. This is interpreted as a transition in the rate-controlling


mechanism of dislocation motion. For low rates the motion is aided by thermal

fluctuations; at very high strain rates, viscous phonon drag becomes dominant.

For brittle materials, cracking accompanies the inelastic deformation, so the

kinetics of cracks nucleation and growth may contribute into a total strain-rate

dependency of the resistance to inelastic deformation.

Conclusions about rate sensitivity of ceramics should be based on comparison

of compressive strength properties over a wide strain rate range. However, this

procedure is not easy. In Fig. 10 the yield strength, Y, determined from the HEL

by the Von Mises relationship, is compared with the failure stress values

measured in uniaxial stress conditions (quasistatic tests, Hopkinson bars, rod

impact) for alumina. The data indicate weak rate dependencies at strain rates less

< 103 s

-1 and > 10

5 s

-1 with a sharp transition between these ranges.

10-6

10-3

100

103

106

0

3

6

9

12

Rod impact

Y/(1-2 )

Y

Uniaxial Stress

Un

iax

ial s

re

ain

(S

ho

ck

)

A lumina

Fa

ilu

re

Str

es

s,

GP

a

Strain Rate, s-1

Figure 10. Dynamic failure properties of alumina ceramic under quasi-static and

dynamic compressive loading. Solid points reproduce data65

for failure strength

measured under uniaxial stress conditions and the yield strength at HEL (1D

strain) calculated with Von Mises yielding criterion. Hatched rectangle shows the

region of rod impact data. Open points are the failure strength values multiplied

by a factor of 1/(1-2 ) according to Rosenberg.66,67

Rosenberg, however, has suggested the use of Griffith’s failure criterion for

reconciliation 1D stress and 1D strain data instead of yield criteria.66,67

In this

case the HEL should be

HEL = (1- )Yc / (1-2 )2, (3)

where Yc is the compressive strength under 1D stress conditions. Thus, the

Griffith’s failure criterion predicts the HEL strength to be higher by a factor of

1/(1-2 ) than the yield criteria used for ductile materials, and this reduces the

discrepancy between the 1D stress and 1D strain data. Figure 10 shows that the


Griffith’s failure criterion provides very good agreement between 1D stress and

1D strain dynamic data.

While the rod impact and Hopkinson bar experiments permit a fairly confident

judgment that failure is brittle, interpretation of the shock-wave experiments

under 1-D strain conditions is more ambiguous. The Hugoniot data and the shock-

wave profiles are not by themselves sufficient to make definite conclusions about

ductile or brittle response. The continuity in spall strength values indicates that, in

many cases, the mode of deformation is mainly ductile.

DISCUSSION

It seems the compressive fracture is (or may be) a secondary effect of the onset

of plastic deformation in brittle materials. Experiments with single crystals

demonstrate that microcracks may nucleate at stress concentrators formed under

inelastic compression, even if initially the material does not contain any defects.

The comparison of the behaviors of single crystals and glasses shows that the

choice of ductile or brittle response is controlled by the capability of material to

accommodate local shears in different directions.

Most of the shock-wave tests of polycrystalline ceramics, with the exception

of boron carbide, do not show unambiguous evidences of fracture under uniaxial

compression. Only boron carbide manifests dilatancy effects on the stress-strain

trajectory when the compressive stress approaches zero on unloading. Even the

observed decrease in spall strength after shock compression above the HEL may

be the result of cracking not under compression but at the end phase of unloading

as an effect of transversal stresses when the axial stress becomes zero. Additional

efforts are needed to answer the question of whether or not microcracking of

ceramic materials occurs under shock compression.

Shock wave tests provide information about the resistance to inelastic

compressive deformation, which is very useful whether or not microcracking

occurs. Additional information over a wider range of stressed states and strains

may provide experiments with spherical 68

and cylindrical 69

divergent shock

waves and experiments with granular ceramics and ceramic powders 70

.

REFERENCES1

E.G. Bombolakis, “Study of the brittle fracture process under uniaxial

compression,” Tectonophysics, 18 231-248 (1973). 2

R.L. Kranz, “Microcracks in rocks: a review,” Tectonophysics, 100 449-480

(1983).3

E.Z. Wang and N.G. Shrive, “Brittle fracture in compression: mechanisms,

models and criteria,” Engng. Fracture Mechanics, 52 [6] 1107-1126 (1995). 4

W.R. Wawersik and W.F. Brace, “Post-failure behavior of a granite and

diabase,” Rock Mechanics, 3 61-85 (1971).


5W.F. Brace, B.V. Paulding, and C. Scholz, “Dilatancy in the fracture of

crystalline rocks,” J. Geophys. Res., 71 3939-3953 (1966). 6

H.C. Heard and C.F. Cline, “Mechanical behavior of polycrystalline BeO,

Al2O3, and AlN at high pressure,” J. Mat. Sci., 15 1889-1897 (1980). 7

B.J. Hockey, “Plastic deformation of aluminum oxide by indentation and

abrasion,” J. Amer. Ceram. Soc., 54 223-231 (1971). 8

H.M. Chan and B.R. Lawn, “Indentation deformation and fracture of

sapphire,” J. Ceram. Soc., 71 [1] 29-35 (1988). 9

A.A. Griffith, “The theory of rupture”; pp. 55-65 in Proc. 1st

Int. Congr. Appl. Mech. Delft, Holland, 1924.

10F.A. McClintock and A.S. Argon, Mechanical Behavior of Materials.

Addison-Wesley Publ., 1966. 11

J.K. Kalthoff, “Modes of dynamic shear failure in solids,” International

Journal of Fracture, 101 1-31 (2000). 12

W.F. Brace and E.G. Bombolakis, “A note on brittle crack growth in

compression,” J. Geophys. Res., 68 3709-3713 (1963). 13

S. Nemat-Nasser and H. Horii, “Compression-induced nonplanar crack

extension with application to splitting, exfolation, and rockburst,” J. Geophys.

Res., 87 [B8] 6805-6821 (1982). 14

H. Horii and S. Nemat-Nasser, “Compression-induced macrocrack growth

in brittle solids: axial splitting and shear failure,” J. Geophys. Res., 90 [B4] 3105-

3125 (1985). 15

G.I. Kanel, S.V. Razorenov, and T.N. Yalovets, “Dynamic strength of

ruby,” Chemical Physics, 12 [2] 175-177 (1993). 16

G.I. Kanel, S.V. Razorenov, A.V. Utkin, and V.E. Fortov, “Investigations of

mechanical properties of materials under a shock wave loading,” Mekhanika

Tverdogo Tela, 5 173-188 (1999). 17

G.I. Kanel, S.V. Razorenov, and V.E. Fortov, “The failure waves and

spallation in homogeneous brittle materials”; pp.451-454 in Shock Compression of Condensed Matter 1991. Edited by S.C. Schmidt et al. North Holland

Publishing Co., 1992. 18

M.D. Furnish, D.E. Grady, and J.M. Brown, “Analysis of shock wave

structure in single crystal olivine using VISAR”; pp. 595-600 in Shock Waves inCondensed Matter – 1985. Edited by Y.M. Gupta, Plenum Publishing Corp., New

York, 1986. 19

D.E. Grady, “Processes occurring in shock wave compression of rocks and

minerals”; pp. 389-437 in High Pressure Research Application in Geophysics.

Academic Press Inc., New York, 1977. 20

R.A. Graham and W.P. Brooks, “Shock-wave compression of sapphire from

15 to 420 kbar. The effects of large anisotropic compression,” J. Phys. Chem.

Solids, 32 2311-2330 (1971).


21T. Mashimo, Y. Hanaoka, and K. Nagayama, “Elastoplastic properties under

shock compression of Al2O3 single crystal and polycrystal,” J. Appl. Phys., 63 [2]

327-336 (1988). 22

R.A. Graham, “Shock-wave compression of X-cut quartz as determined by

electrical response measurements,” J. Phys. Chem. Solids, 35 [3] 355-372 (1974). 23

T.J. Ahrens, “High-pressure electrical behavior and equation of state of

magnesium oxide from shock wave measurements,” J. Appl. Phys., 37 [7] 2532-

2541 (1966). 24

T. Mashimo, A. Nakamura, M. Kodama, K. Kusaba, K. Fukuoka, and

Y. Syono, “Yielding and phase transition under shock compression of yttria-doped

cubic zirconia single crystal and polycrystal,” J. Appl. Phys., 77 [10] 5060-5068

(1995).25

R.A. Graham and T.J. Ahrens, “Shock wave compression of iron-silicate

garnet,” J. Geophys. Res., 78 [2] 375-392 (1973). 26

Y. Wang and D.E. Mikkola, “Response of alpha-aluminum oxide to shock

impact”; pp. 1031-1040 in Shock-Wave and High-Strain-Rate Phenomena in Materials. Edited by M.A. Meyers, L.E. Murr, and K.P. Staudhammer. Marcel

Dekker, Inc., New York, 1992. 27

A.V. Anan’in, O.N. Breusov, A.N. Dremin, S.V. Pershin, A.I. Rogacheva,

and V.F. Tatsii, “The effect of shock waves on silicon dioxide: I. Quartz,”

Combustion Explosive and Shock Waves, 10 372 (1974). 28

D.E. Grady, “Shock deformation of brittle solids,” J. Geophys. Res., 85 [B2]

913 (1980). 29

G.I. Kanel, S.V. Razorenov, A.V. Utkin, Hongliang He, Fuqian Jing, and

Xiaogang Jin, “Influence of the load conditions on the failure wave in glasses,”

High Pressure Research, 16 27-44 (1998). 30

J. Arndt and D. Stoffer, “Anomalous changes in some properties of silica

glass densified at very high pressures,” Phys. and Chem. of Glasses, 10 [3], 117-

134 (1969). 31

R.V. Gibbons and T.J. Ahrens, “Shock metamorphism of silicate glasses,” J.

Geophys. Res., 76 [23] 5489-5498 (1971). 32

J. Cagnoux, Deformation et Ruine d'un Verre Pyrex Soumis à Un Choc Intense. Thesis, University Poitiers, 1985.

33F.M. Ernsberger, “Role of densification in deformation of glasses under

point loading,” J. Amer. Ceram. Soc., 51 [10] 545-547 (1968). 34

H. Schardin, “Velocity effects in fracture”; pp. 297-330 in Fracture. Edited

by B.L. Averbach et al. MIT Press, Cambridge, Mass., and Wiley, New York,

1959.35

L.A. Galin, V.A. Ryabov, D.V. Fedoseev, and G.I. Cherepanov, “On failure

of a high-strength glass,” Dokl. AN SSSR, 169 [5] 1034-1036 (In Russian) (1966).


36L.A. Galin and G.I. Cherepanov, “On self-supporting failure of a stressed

brittle solid,” Dokl. AN SSSR, 167 [3] 543-546 (In Russian) (1966). 37

N.S. Brar, “Failure waves in glass and ceramics under shock compression”;

pp. 601-606 in Shock Compression of Condensed Matter – 1999, Edited by M.D.

Furnish et al., AIP CP 505, 2000. 38

M.E. Kipp and D.E. Grady, “Shock compression and release in high-

strength ceramics”; pp 377-380 in Shock Compression of Condensed Matter 1989.

Edited by S.C. Schmidt, J.N. Johnson and L.W. Davison. Elsevier Science

Publishers B.V., 1990. 39

D.E. Grady, “Shock-wave properties of high-strength ceramics”; pp. 455-

458 in Shock Compression of Condensed Matter 1991. Edited by S.C.Schmidt et

al. North Holland Publishing Co., 1992. 40

G.I. Kanel, S.J. Bless, and A.M. Rajendran, “Behavior of brittle materials

under dynamic loading,” Institute for Advanced Technology Report, IAT.R 0219,

2000.41

J. Cagnoux and F. Longy, “Is the dynamic strength of alumina rate-

dependent?”; pp. 293-296 in Shock Waves in Condensed Matter 1987. Edited by

S.C. Schmidt and N.C. Holmes. Elsevier Science Publishers B.V., 1988. 42

M.D. Furnish and L.C. Chhabildas, “Alumina strength degradation in the

elastic regime”; pp. 501-504 in Shock Compression of Condensed Matter – 1997.

Edited by S.C. Schmidt et al. AIP Conference Proceedings 429, 1998. 43

T.J. Ahrens, W.H. Gust, and E.B. Royce, “Material strength effect in the

shock compression of alumina,” J. Appl. Phys., 39 [10] 4610-4616 (1968). 44

D.E. Munson and R.J. Lawrence, “Dynamic deformation of polycrystalline

alumina,” J. Appl. Phys., 50 [10] 6272-6282 (1979). 45

J.M. Staehler, W.W. Predeborn and B.J. Pletka, “The response of a high-

purity alumina to plate-impact testing”; pp. 745-748 in High-Pressure Science and Technology – 1993. Edited by S.C. Schmidt et al. AIP Conference

Proceedings 309, 1994. 46

N.H. Murray, N.K. Bourne, and Z. Rosenberg, “Precursor decay in several

aluminas”; pp. 491-494 in Shock Compression of Condensed Matter – 1995.

Edited by S.C. Schmidt and W.C. Tao. AIP Conference Proceedings 370, 1996. 47

W.H. Gust and E.B. Royce, “Dynamic yield strength of B4C, BeO, and

Al2O3 ceramics,” J. Appl. Phys., 42 276-295 (1971). 48

D.P. Dandekar and P. Bartkowski, “Shock response of AD995 alumina”; pp.

733-736 in High-Pressure Science and Technology – 1993. Edited by S.C.

Schmidt et al. AIP Conference Proceedings 309, 1994. 49

D.E. Grady, “Impact strength and indentation hardness of high-strength

ceramics”; pp. 741-744 in High-Pressure Science and Technology - 1993, Edited

by S.C. Schmidt, et al. AIP Conference Proceedings 309, 1994.


50D. Yaziv, Y. Yeshurun, Y. Partom, and Z. Rosenberg, “Shock structure and

precursor decay in commercial alumina”; pp. 297-300 in Shock Compression of Condensed Matter 1987. Edited by S.C. Schmidt and N.C. Holmes. Els. Sc. Publ.

B.V., 1988. 51

F. Longy and J. Cagnoux, “Plasticity and microcracking in shock-loaded

alumina,” J. Amer. Ceram. Soc., 72 [6] 971-979 (1989). 52

J. Cagnoux, “Spherical waves in pure alumina. Effect of grain size on flow

and fracture”; pp. 445-448 in Shock Compression of Condensed Matter 1989.

Edited by S.C. Schmidt, J.N. Johnson and L.W. Davison. Elsevier Science Publ.

B.V., 1990. 53

L.C. Chhabildas, M.D. Furnish, W.D. Reinhart, and D.E. Grady, “Impact of

AD995 alumina rods”; pp. 505-508 in Shock Compression of Condensed Matter – 1997. Edited by S.C. Schmidt, D.P. Dandekar, and J.W. Forbes. AIP Conference


S.J. Bless, N.S. Brar, and Z. Rosenberg, “Failure of ceramic and glass rods

under dynamic compression”; pp. 939-942 in Shock Compression of Condensed Matter 1989. Edited by S.C. Schmidt, J.N. Johnson and L.W. Davison. Elsevier

Science Publ. B.V., 1990. 55

J.L. Wise and D.E. Grady, “Dynamic, multiaxial impact response of

confined and unconfined ceramic rods”; pp. 777-780 in High-Pressure Science and Technology – 1993. Edited by S.C. Schmidt et al. AIP Conference


C.H.M. Simha, S.J. Bless, and A. Bedford, “What is the peak stress in the

ceramic bar impact experiment?”; pp. 615-618 in Shock Compression of Condensed Matter – 1999. Edited by M.D.Furnish, L.C.Chhabildas, and

R.S.Hixson. AIP Conference Proceedings 505, 2000. 57

J.U. Cazamias, B. Reinhart, C. Konrad, L.C. Chhabildas, and S. Bless, “Bar

impact tests on alumina (AD995)”; in Shock Compression of Condensed Matter - 2001, in press.

58Z. Rosenberg, “The response of ceramic materials to shock loading”; pp.

439-444 in Shock Compression of Condensed Matter 1991. Edited by S.C.

Schmidt et al. North Holland Publishing Co., 1992. 59

H. Song, S.J. Bless, N.S. Brar, C.H. Simha, and S.D. Jang, “Shock

properties of AL2O3 and ZrO2 ceramics”; pp. 737-740 in High-Pressure Science and Technology - 1993. Edited by S.C. Schmidt et al. AIP Conference


V.N. Nikolaevskii and J.R. Rice, “Current topics in non-elastic deformation

of brittle materials”; pp. 455-464 in High-Pressure Science and Technology (Proceedings of Sixth AIRAPT Conference). Edited by K.D. Timmerhaus and

M.S. Barber. Plenum Press, New York, 1979.


61W. Chen and G. Ravichandran, “Dynamic compressive failure of a glass

ceramic under lateral confinement,” J. Mech. Phys. Solids, 45 [8] 1303-1328

(1997).62

W. Chen and G. Ravichandran, “Failure mode transition in ceramics under

dynamic multiaxial compression,” Internat. J. of Fracture, 101[1] 141-159

(2000).63

B. Paul, “Macroscopic criteria of plastic yielding and brittle fracture”; in

Fracture: an Advanced Treatise. Edited by H. Liebowitz. Vol. II “Mathematical

Fundamentals”, Academic Press, New York, 1968. 64

M.F. Ashby and C.G. Sammis, “The damage mechanics of brittle solids in

compression,” PAGEOPH, 133 [3] 490-521 (1990). 65

D.E. Grady, “Shock wave compression of brittle solids,” Mechanics of

Materials, 29 181-203 (1998). 66

Z. Rosenberg, “On the relation between the Hugoniot elastic limit and the

yield strength of brittle materials,” J. Appl. Phys., 74 [1] 752-753 (1993). 67

Z. Rosenberg, “On the shear strength of shock loaded brittle solids,” J. Appl.

Phys., 76 [3] 1543-1546 (1994). 68

J.Y. Tranchet and F. Collombet, “Behavior of pure alumina submitted to a

divergent spherical stress wave”; pp. 535-542 in Metallurgical Applications of Shock-Wave and High-Strain-Rate Phenomena. Edited by L.E. Murr, K.P.

Staudhammer, and M.A. Meyers. Elsevier Science B.V., 1995. 69

G.I. Kanel, S.V. Razorenov, A.V. Utkin, S.N. Dudin, V.B. Mintsev, S.

Bless, and C.H.M. Simha, “Investigation of mechanical properties of ceramics

using axi-symmetric shock waves”; pp. 489-492 in Shock Compression of Condensed Matter – 1997. Edited by S.C. Schmidt, D.D. Dandekar, and J.W.

Forbes. AIP Conference Proceedings 429, 1998. 70

L.W. Meyer and I. Faber, “Investigations on granular ceramics and ceramic

powder,” J. Phys. IV France 7 [C3] 565-570 (1997).

ACKNOWLEDGEMENT The research reported in this document was performed in connection with Contract number DAAD17-

01-D-0001 with the U.S. Army Research Laboratory. The views and conclusions contained in this document are those of the authors and should not be interpreted as presenting the official policies or position, either expressed or implied, of the U.S. Army Research Laboratory or the U.S. Government unless so designated by other authorized documents. Citation of manufacturer’s or trade names does not constitute an official endorsement or approval of the use thereof. The U.S. Government is authorized to reproduce and distribute reprints for Government purposes notwithstanding any copyright notation hereon.


RECENT DEVELOPMENTS IN SPLIT HOPKINSON PRESSURE BAR

TESTING

W. Chen, and B. Song D. J. Frew and M. J. Forrestal

The University of Arizona Sandia National Laboratories

Tucson, AZ 85721-0119 Albuquerque, NM 87185-1174

ABSTRACT

The split Hopkinson pressure bar (SHPB) technique has been widely used to

determine the dynamic properties of engineering materials. Recent applications

of this technique to materials with extreme properties (e.g., extremely soft or

extremely hard) have forced more careful examination of the SHPB apparatus and

mandated modifications to this well-established experimental technique in order

to obtain valid results. In this paper, we present a brief review of recent

developments that adapt this technique for testing a variety of engineering

materials under valid dynamic testing conditions. It is shown that, in order to

subject the specimen to uniform deformation at a nearly constant strain rate under

dynamic stress equilibrium, pulse shaping must be used in the SHPB experiments.

In addition, a more sensitive transmission bar must be employed to detect weak

transmitted signals when testing soft materials.

INTRODUCTION

The SHPB or Kolsky bar technique originally developed by Kolsky1, 2

has

been used by many investigators to obtain dynamic compressive properties of

solid materials. The evolution of this experimental method and recent advances

are discussed by Nicholas3, Ellwood et al.

4, Franz et al.

5, Follansbee

6, Nemat-

Nasser et al.7, Ramesh and Narasimhan

8, Gray

9, and Gray and Blumenthal.

10 This

technique has mostly been used to study the plastic flow behavior of metals that

undergo large strains at strain rates between 102 – 10

4/s. As discussed by Yadav

et al.11

, data for the compressive flow stress of metals are typically obtained for

strains larger than a few percent because the technique is not capable of

measuring the elastic and early yield behavior. Also, the method of measuring

strains with resistance strain gages mounted on the metal bar surfaces limits the

applicability of SHPB to test the dynamic properties of low-strength and low-

impedance materials. Recently, there has been a need to obtain reliable dynamic



material properties for brittle materials with failure strains are less than about 1.0

percent, such as ceramics12,13

and rocks14

, and for soft material that are too week

to generate a clear transmitted pulse, such as silicone rubbers15

and polymeric

foams16

. These brittle and soft materials require careful examination of the SHPB

technique and mandate necessary modifications to obtain valid data. In this

paper, we briefly review the SHPB technique, the challenges it faces, and the

remedies for obtaining valid experimental results.

As shown in Fig. 1, a conventional SHPB consists of a striker bar, an incident

bar, a transmission bar, and a sample placed between the incident and

transmission bars. Nicholas3, and Gray

9 present the working principles and the

equations that describe the sample response in terms of the measured strain

signals.

Striker

BarIncident Bar Transmission Bar

Specimen

( s, cs, As)

i r

( , c, A) ( , c, A)

2

t

1

lo

Fig. 1 A schematic illustration of a split Hopkinson pressure bar.

The equations that relate the strain gage signals to material responses

(Nicholas3, and Gray

9) are based on idealized 1-D wave propagation analysis. In

addition, it is assumed that the sample undergoes homogeneous deformation and

is in dynamic stress equilibrium. In a conventional SHPB experiment, the wave

propagation in the elastic bars is not as ideal as assumed. Figure 2 presents a set

of oscilloscope record of the incident, reflected and transmitted strain signal

during a typical SHPB experiment on a 1046 mild steel specimen. The signals in

Fig. 2 show that the incident pulse is not rectangular in shape as idealized in 1-D

wave theory. The reflected signal, which directly correlates to the strain rate in

the specimen, has significant fluctuations with its average amplitude decreasing

with time. This indicates that the specimen is not deforming at a nearly constant

strain rate. Furthermore, due to the significant fluctuations in the beginning

portion of the reflected pulse, the initial specimen strain from a SHPB test is not

considered to be reliable, leading to an unreliable Young's modulus of a material

measured with a SHPB. The dynamic stress state in the specimen can be

examined using a 1-wave, 2-wave method10, 17, 18

, which compares the axial force

history on the transmission side of the specimen (transmitted signal) to that on the


incident side (the difference between the incident and reflected signals). Figure 3

shows the results of such a 1-, 2-wave analysis for the experiment in Fig. 2.

Fig. 2 Oscilloscope record of a conventional SHPB experiment on a mild steel.

Fig. 3 Axial force histories on the specimen by 1-wave, 2-wave analysis.

The results shown in Fig. 3 illustrate that dynamic equilibrium in the

specimen is approximately reached during the later stages of the experiment (t >

~60 s). During the early stages of the experiment (t < ~30 s), the specimen is

not in stress equilibrium. This leads to uncertainties on the dynamic yield

strength as determined by SHPB tests. Gray and Blumenthal10

reached similar

conclusions regarding the equilibrium of a soft rubber specimen.


When the SHPB is used to test relatively brittle materials such as ceramics12, 13

and rocks14

, most of the material behavior of interest occurs at strains less than

about 1.0 percent, which is within the large error range for conventional SHPB

experiments. For other material such as polymers and shape-memory alloys,

loading history significantly affects the mechanical behavior. The initial portion

of the experimental duration cannot be ignored. Therefore, to use SHPB

technique for obtaining the dynamic properties of such materials, modifications

must be made to the testing technique to ensure that the specimen deforms

uniformly at a nearly constant strain rate under dynamically equilibrated stress.

Such modifications include pulse shaping, a sensitive transmission bar, and

dynamic equilibrium monitoring. Next, we briefly describe each of these new

developments in SHPB testing in which we have been involved.

PULSE SHAPING

The initial significant fluctuations in the reflected signal shown in Fig. 2 and

the initial non-equilibrium shown in Fig. 3 indicate that the incident loading pulse

profile needs to be controlled to facilitate stress equilibrium and uniform

deformation in the specimen. Some of the advantages and necessities for shaping

the incident pulse for SHPB experiments were discussed twenty years ago. Franz

et al.5 and Follansbee

6 wrote review papers that discussed pulse shaping for SHPB

experiments with metal samples. In these review papers, the authors emphasized

that a slowly rising incident pulse is preferred to a pulse that rises steeply in order

to minimize the effects of dispersion and allow the sample to achieve dynamic

stress equilibrium. Franz et al.5 and Follansbee

6 also discuss experimental

techniques for pulse shaping and a numerical procedure17

for correcting raw data

for wave dispersion in the bars. These authors5 and Ellwood et al.

4 show that a

properly chosen pulse shaper can also be used to generate a nearly constant strain

rate in the sample. Gray9 and Gray and Blumenthal

10 present additional

information in recent survey papers that include these subjects. However, Duffy

et al.19

were probably the first authors to use pulse shapers to smooth pulses

generated by explosive loading for the torsional Hopkinson pressure bar.

While pulse shaping techniques have been successfully used to achieve the

goals of many different experiments, pulse shapers are usually designed by

experimental trials that exclude a model to guide the design parameters. For

examples, Wu and Gorham18

used paper on the impact surface of the incident bar

to eliminate high frequency oscillations in the incident pulse for Kolsky

compression bar experiments. Togami et al.20

used a thin, Plexiglas disk to

produce nondispersive compression pulses in an incident bar, and Chen et al.21

used a polymer disk to spread the incident compressive pulses for experiments

with elastomers. Christensen et al.22

used striker bars with a truncated-cone on

the impact end in an attempt to produce ramp pulses. In contrast to other pulse


shaping studies, Nemat-Nasser et al.7 modeled the plastic deformation of an

OFHC copper pulse shaper, predict the incident strain pulse, and show good

agreement with some measured incident strain pulses. Frew et al.23

further

extended the model to describe the behavior of C11000 copper pulse shapers

driven to much larger strains. In the equations that govern wave propagation in

the striker bar, it was found that the added mass from the sabot must also be

considered24

.

Pulse-shaping techniques have been applied recently to obtain valid dynamic

material properties of a variety of materials, such as limestone14

, ceramics23

, a

shape-memory alloy25

, a polymeric foam16

, and rubbers15

. Pulse shaping has also

been used in dynamic tension experiments26

. As an example, Fig. 4 shows the

oscilloscope record of a pulse-shaped SHPB experiment on the same 1046 mild

steel. The incident pulse was created by placing a combination of hardened 1046

steel and C11000 copper disks on the striking end of the incident bar. The nearly

flat reflected signal shows minimum fluctuations, which indicates that a nearly

constant strain rate has been achieved in the specimen. Furthermore, a detailed

examination of the reflected signal reveals that, without the fluctuations

associated with the reflected signal (Fig. 2), the reflected signal in Fig. 4 actually

is composed of two plateaus: a small-amplitude plateau followed by a second,

larger one. Data reduction further reveals that the small plateau corresponds to

the elastic deformation of the specimen, whereas the larger one is associated with

the plastic flow. In addition to revealing the details in the reflected pulse, pulse

shaping also facilitates dynamic equilibrium. Figure 5 shows the 1-, 2-wave

analysis for the pulse shaped experiment, which indicates a nearly perfect

agreement between the front- and back-end force histories.

SENSITIVE TRANSMISSION BAR

When a soft material is tested with a SHPB, the transmitted signal can be too

weak to provide a stress history for the specimen21

. More sensitive transmission

bars are thus necessary. Low-impedance bars, such as polymer bars, will extend

the time for the sample to reach dynamic equilibrium14

. We have developed an

aluminum transmission tube21

and a quartz-crystal embedded aluminum bar27

to

provide high sensitivity of the transmission bar, while still maintaining the high

impedance mismatch between the specimen and the bar.

DYNAMIC EQILIBRIUM MONITORING

When the specimen is a very soft material, nearly all of incident signal is

reflected back into the incident bar. This introduces significant errors in the 2-

wave analysis, which takes the differences between the incident and reflected

signals. We have developed quartz-crystal methods to directly measure the front-


and back-end force histories in the specimen15, 16

, which directly monitors the

dynamic equilibrium process in the specimen.

Fig. 4 Oscilloscope record of a pulse-shaped SHPB experiment on a mild steel.

Fig. 5 Axial force histories after pulse shaping by 1-wave, 2-wave analysis.

SUMMARY

Pulse shaping must often be employed to obtain dynamic material properties

with a SHPB to ensure that the specimen is deforming uniformly at a nearly

constant strain rate under dynamic equilibrium. Proper modifications to a

conventional SHPB are necessary when testing hard or soft materials.


ACKNOWLEDGEMENTS

This work was sponsored by the U.S. Army Research Office through a grant

to The University of Arizona (G-DAAD19-00-1-0493) and the Sandia National

Laboratories Joint DoD/DOE Penetration Technology Program. Sandia is a

multi-program laboratory operated by Sandia Corporation, a Lockheed Martin

Company, for the U.S. Department of Energy under Contract DE-AC04-

94AL8500.

REFERENCES 1H. Kolsky, “An Investigation of the Mechanical Properties of Materials at

Very High Rates of Loading,” Proc. Royal Soc. Lond., B62 676-700, (1949).2H. Kolsky, Stress Waves in Solids. Dover, New York (1963).

3T. Nicholas, “ Material Behavior at High Strain Rates,” Impact Dynamics,

Chapter 8, John Wiley & Sons, New York, (1982). 4S. Ellwood, L. J. Griffiths, and D. J. Parry, “Materials Testing at High

Constant Strain Rates,” J. Phys. E: Sci. Instrum., 15 280-282 (1982).5C. E. Franz, P. S. Follansbee, and W. J. Wright, “New Experimental

Techniques with the Split Hopkinson Pressure Bar,” in the 8th

Int. Conf. on High

Energy Rate Fabrication, ASME, (ed. I. Berman and J. W. Schroeder), San

Antonio, TX, June 17-21 (1984). 6P. S. Follansbee, “The Hopkinson Bar,” Mechanical Testing, Metals

Handbook, 9th

ed., 8, Am. Soc. for Metals, Metals Park, Ohio, 198-217 (1985). 7S. Nemat-Nasser, J. B. Isaacs, and J. E Starrett, “Hopkinson Techniques for

Dynamic Recovery Experiments,” Proc. R. Soc. Lond., A435 371-391,(1991). 8K. T. Ramesh and S. Narasimhan, “Finite Deformations and the Dynamic

Measurement of Radial Strains in Compression Kolsky Bar Experiments,” Int. J.

Solids Structures, 33 3723-3738 (1996). 9G. T. Gray, “Classic Split-Hopkinson Pressure Bar Technique,” ASM

Handbook, 8, Mechanical Testing and Evaluation, ASM International, Materials

Park, OH, 44073-0002 (2000). 10

G. T. Gray and W. R. Blumenthal, “Split-Hopkinson Pressure Bar Testing of

Soft Materials,” ASM Handbook, 8, Mechanical Testing and Evaluation, ASM

International, Materials Park, OH, 44073-0002 (2000).

11S. Yadav, D. R. Chichili, and K. T. Ramesh, “The Mechanical Response of

a 6061-T6 Al/Al2O3 Metal Matrix Composite at High Rates of Deformation,”

Acta metall. Mater., 43 4453-4464, (1995). 12

W. P. Rogers and S. Nemat-Nasser, “Transformation Plasticity at High

Strain Rate in Magnesia-Partially-Stabilized Zirconia,” J. Am. Ceram. Soc., 73

136-139, (1990).


13W Chen. and G. Ravichandran, “Dynamic Compressive Failure of a Glass

Ceramic Under Lateral Confinement,” J. Mech. Phys. Solids. 45 1303-1328,

(1997).14

D. J. Frew, M. J. Forrestal, and W. Chen, “A Split Hopkinson Bar

Technique to Determine Compressive Stress-Strain Data for Rock Materials,”

Experimental Mechanics, 41 40-46, (2001).15

W. Chen, F. Lu, D. J. Frew, and M. J. Forrestal,“Dynamic Compression

Testing of Soft Materials,” ASME J. Appl. Mech., rivised (2001). 16

W. Chen, F. Lu, and N. Winfree, "High Strain-Rate Compressive Behavior

of a Rigid Polyurethane Foam with Various Densities," Experimental Mechanics,

accepted (2001).17

P. S. Follansbee and C. E. Franz, “Wave Propagation in the Split Hopkinson

Pressure Bar,” Trans. ASME, J. Eng. Mat. Technol., 105 61-66 (1983).18

X. J. Wu and D. A. Gorham, “Stress Equilibrium in the Split Hopkinson

Pressure Bar Test,” J. Phys. IV France, 7 (C3) 91-96 (1997).19

J. Duffy, J. D. Campbell, and R. H. Hawley, “On the Use of a Torsional

Split Hopkinson Bar to Study Rate Effects in 1100-0 Aluminum,” ASME J. Appl.

Mech., 37 83-91 (1971). 20

T. C. Togami, W. E. Baker, and M. J. Forrestal, “A Split Hopkinson Bar

Technique to Evaluate the Performance of Accelerometers,” J. Appl. Mech., 63

353-356 (1996).21

W. Chen, B. Zhang, and M. J. Forrestal, “A Split Hopkinson Bar Technique

for Low-Impedance Materials,” Exp. Mech., 39 81-85 (1999).22

R. J. Christensen, S. R. Swanson, and W. S. Brown,”Split-Hopkinson-Bar

Tests on Rock Under Confining Pressure,” Exp. Mech., 29 508-513 (1972).23

D. J. Frew, M. J. Forrestal, and W. Chen, “Pulse Shaping Techniques for

Testing Brittle Materials with a Split Hopkinson Pressure Bar,” Exp. Mech.,

accepted (2001). 24

M. J. Forrestal, D. J ,Frew, and W. Chen, “The Effect of Sabot Mass on the

Striker Bar for Split Hopkinson Pressure Bar Experiments,” Exp. Mech.,

submitted (2001). 25

W. Chen, Q. Wu, J. H. Kang, and N. A. Winfree, "Compressive Superelastic

Behavior of a NiTi Shape Memory Alloy at Strain Rates of 0.001 to 750 s-1

,"

International Journal of Solids and Structures, in printing (2001). 26

W. Chen, F. Lu, and M. Cheng, “Tension and Compression Tests of Two

Polymers Under Quasi-static and Dynamic Loading” Polymer Testing, in printing

(2001). 27

W. Chen, F. Lu and B. Zhou, “A quartz crystal imbedded split Hopkinson bar

for soft materials,” Experimental Mechanics, 40, (1) pp. 1-6 (2000).


USING BAR IMPACT TO DETERMINE DYNAMIC PROPERTIES OF

CERAMICS

Dr. Stephan J. Bless

Institute of Advanced Technology

The University of Texas at Austin

3925 West Braker Lane, Suite 400

Austin, TX 78759

ABSTRACT

Impact onto bars provides a useful means to study the properties of brittle

materials. Measurements may be stress (using piezoresistive gauges) or free

surface velocity. The amplitude of the stress that propagates in the bar is the

largest compressive stress that can be supported by the target material (in a one-

dimensional stress state); however, premature failure due to tension in the impact

zone may limit the peak stress. High-speed photography can be used to reveal the

morphology of the failure – which takes place by longitudinal cracks, transverse

cracks, or failure waves.

INTRODUCTION

It is difficult to develop empirical criteria for compression failure of ceramics

under impact loading. Strength depends on stress rate and stress state. Plate

impact experiments determine the compressive strength at very high rates and full

lateral confinement. Under these conditions, many brittle materials exhibit

considerable ductility.

In fact, high velocity impact induces compressive brittle failure, generally

accompanied by substantial lateral strain. For example, cavity expansion models

of ceramic penetration show that a critical stage of compressive failure takes place

under uniaxial stress [1].

In a plate-impact experiment, the failure stress is usually identified as the

Hugoniot elastic limit HEL. In a bar-impact experiment, the failure stress is

nominally equal to Y, the strength in one-dimensional stress. As discussed

already in this symposium [2], different micromechanical models predict different

relationships between HEL and Y. In some instances, strain-based failure criteria

have also been proposed.



BAR IMPACT TESTS

The bar impact geometry is shown in Fig. 1. A striker plate or bar impacts a

target bar. The impactor can be the same material as the target, or a hard steel;

more consistent results are usually obtained with metal strikers. If the bar does not

fail, then a stress wave is produced whose amplitude is determined by the

conventional impedance match solution. But the test is usually designed so that

the target bar fails in the vicinity of the impact plane. As in a plate impact test, an

elastic wave then propagates along the bar, and the amplitude of this wave is

equal to the largest stress in the bar before it failed. By definition this is the bar

impact strength, YB. In metals, YB=Y, but in brittle materials YB Y because

failure can initiate from transient impact-induced tensile stresses [3].

Most experiments have been done with round bars. However, tests have also

been performed with square, rectangular, or octagonal bars cut from plates. There

does not appear to be any systematic difference in strength associated with cross

section shape.

The distance from the impact plane to the measurement plane should be about

10 diameters. Shorter distances are probably possible in some materials, but in

[4] it appears that six diameters was too short. Using a layered striker to induce a

ramp wave loading of the target bar seems to enable use of shorter target bars [5].

11

00

.25

98

a LEXAN

GAUGE PLANETARGET ROD

STEEL

LEXAN LEXAN

CERAMIC

OR

Figure 1. Bar impact geometry. The ceramic target rod may be struck by a steel

plate or a ceramic rod.

The stress wave in the bar can be recorded by using an interferometer to

measure the motion at the free end, or by using an embedded stress gauge (see

record in Figure 2). The free surface technique provides the most faithful record

of the rise to peak stress. However, spall failure can occur near the free surface

soon after wave reflection, so if the goal of the test includes measuring the

crushing behavior of the failed column of material, this technique is not useful.


0

0

5

5

10

10

15

15

20

20

t [µs]

1 1

2 2

3 3

4 4

5 5

6 6

σ z[G

Pa]

4340

Steel

Gauge failed

AlON 4340 steel

10 mm sq. c.s.101 mm 42 mm

10 mm

51mm

Camera trigger

Figure 2. Gauge records from bar impacts on AlON.

Use of stress gauges in bars has been analyzed and validated by [6]. The best

records are obtained when the brittle bar is backed by a metallic witness bar. The

strength of the witness bar should be greater than the strength of the brittle

material; otherwise, the gauge measures the witness bar strength. For strong

ceramics, it is often necessary to place the gauge between two ceramic bars. This

emplacement, unfortunately, usually results in gauge failure just after the peak

stress signal. The reasons for premature gauge failure are not yet clear.

Bar impact tests have also been performed with lateral confinement. Steel and

tantalum have been used [5]. The measured bar impact strength increases when

the bars are confined. However, the interpretation of the strength is not

straightforward, since the confining stress is rather difficult to determine.

The split Hopkinson bar (SHB) also can be used to measure unconfined

compressive strength. While historically SHB test results have exhibited

considerable scatter, recent progress, reported in this symposium, on wave

shaping holds great promise for achieving reliable strength measurements [7].

Nevertheless, advantages of the bar impact over the Hopkinson bar include less

sample machining, avoiding shape changes, bigger samples, observation of failure

morphology, ability to test very hard materials, ability to study failure

propagation, and easier characterization of the post-failure behavior.

Another attribute of the bar impact test is that there is relatively little scatter.

In several test series, the shot to shot variation in YB has been < 10%. This much

better than usually observed in static unconfined compression tests. Thus, the bar

impact may be an economic means of quality control for ceramics.

Lastly, both the flyer plate and the witness bar can exhibit several different

post-impact appearances. They may be undamaged, indented, or cratered. These

behaviors are indicative of radically different failure modes among various brittle

materials.


RESULTS FROM BAR IMPACT TESTS

One of the most useful features of the bar impact test is visual access. High-

speed photos of the dynamic failure process are possible. Several categories of

failures have been observed. The most dramatic observations are of self-

propagating failure waves. Figure 3 shows an example in glass. In such a failure

wave, material apparently is transformed from an intact to a comminuted state.

Transparent material becomes opaque, and there is moderate radial expansion.

Figure 3. Failure wave in silica glass [8].

Failure waves can also be driven by the projectile, in which case standoff

between the impact face and the propagation front is almost contort. Figure 4

shows an example.

Figure 4. Example of a driven failure front, seen in granite [9].


The main use of the bar impact tests has been to measure compressive

strength. Table 1 provides a measurement of strength and observations of failure

modes in bar impacts. Less reliable strength data are shown in parentheses.

Table I. Measurements of bar impact strength of ceramics

Material References YB (GPa) Failure

Morphologies

(See Table 2)

Sintered alumina 3, 4, 5, 6, 10,

11, 12

3.6, 4.2 1, 2

Hot pressed alumina 4 4.1 1, 2

TiB2 4 4.9 3

Silicon carbide 4 4.8 3

Boron carbide 4 n/a 1

Soda lime glass 13 2.0 3, 5

Borosilicate glass 2, 4, 11 (1.5) 2.5 3, 4

Aluminum oxynitride 14 4.0 1, 4

Homalite 15 n/a 5

Table II. Failure modes

1. Axial splitting near impact face

2. Transverse faulting away from the impact zone

3. Self-propagating failure wave

4. Failure along central axis

5. Driven failure wave

In addition to these studies of compressive failure, impact-induced tensile

failure has been studied in ceramic bars by [16].

EMERGING NEW CAPABILITIES

At The University of Texas, we are expanding the repertory of bar impact test

techniques with experiments on a surrogate material – homalite. Homalite is a

brittle thermoset plastic, 1.23 g/cm3, Y = 0.15 GPa.

One promising new area is variations in bar cross section. By using tapered

bars, we achieve a condition in which the stress increases continuously along the

bar. This may avoid the problem of premature failure due to impact transients.

Figure 5 is a photograph of a test using a tapered homalite bar.


Figure 5. Photograph of impact induced failure in a tapered homalite bar.

Recovery of fractured material is also an area that has received little attention

except [11], but is potentially very useful. We have recently developed a

technique in which a 10-mm bar is sleeved in plastic. The particles are recovered

within the sleeve, and they can be correlated with fracture zones seen in

photographs of the bars. These include the comminuted region associated with

failure fronts, faulted material produced relatively late in the impact, and spall

fragments formed at the rear surface. So far in homalite, we have not seen

evidence of fractal behavior, nor of ductile flow. Rather, we observed that in the

comminuted region there is a well defined smallest particle size, about 20

microns. Particles from elsewhere along the bar are much larger, ranging in size

up to several mm.

It has also become clear that there are phenomena associated with bar impact

that are not yet understood. Figure 6, for example, shows precursor cracks that

will lead rapidly to an isolated zone of comminuted material. This type of failure,

which is not due to simple inversion of the loading pulse at the free end, has not

yet been modeled.

Figure 6. Spall failures developing in the interior of a homalite bar [13].


ACKNOWLEDGEMENTS

The research reported in this document was performed in connection with

Contract number DAAD17-01-D-0001 with the U.S. Army Research Laboratory.

The views and conclusions contained in this document are those of the authors

and should not be interpreted as presenting the official policies or position, either

expressed or implied, of the U.S. Army Research Laboratory or the U.S.

Government unless so designated by other authorized documents. Citation of

manufacturer’s or trade names does not constitute an official endorsement or

approval of the use thereof. The U.S. Government is authorized to reproduce and

distribute reprints for Government purposes notwithstanding any copyright

notation hereon. Additional data were provided by Rod Russell at The University

of Texas.

REFERENCES

[1] S. Satapathy and S.J. Bless, “Cavity Expansion Resistance of Brittle

Materials Obeying a Two Curve Pressure Shear Behavior,” Journal of Applied

Physics, 88 [1] 4004-4012 (1999).

[2] G.I. Kanel and S.J. Bless, “Compressive Fracture of Brittle Solids under

Shock-Wave Loading,” Int’l Conference on Advanced Ceramics and Glasses

(PacRim IV), Nov. 4-8, 2001, to be published by American Ceramics Society

(2002).

[3] C.H.M. Simha, S.J. Bless, and A. Bedford, “What is the Peak Stress in the

Ceramic Bar Impact Experiment?”; pp.615-618 in Shock Compression of

Condensed Matter – 1999. Edited by M.D. Furnish, L.C. Chhabildas, and R.S.

Hixson. American Institute of Physics, 2000.

[4] N.S. Brar and S.J. Bless, “Dynamic Fracture and Failure Mechanisms of

Ceramic Bars,” On Shock-Wave and High-Strain-Rate Phenomena in Materials

(EXPLOMET 90), Aug. 12-17, 1990, to be published.

[5] L.C. Chhabildas, M.D. Furnish, and D.E. Grady, “Impact of Alumina Rods

– A Computational and Experimental Study,” J. Phys. IV, 4 [C3] 137-143 (1997).

[6] Z. Rosenberg, P. Partom, and B. Keren, “Gauge Factor of Manganin under

Axial Stress Conditions,” J. Appl. Phys., 54 2824-2825 (1983).

[7] W. Chen, B. Song, D.J. Frew, and M.J. Forrestal, “Recent Developments

in Split Hopkinson Pressure Bar Testing,” Int’l Conf. On Advanced Ceramics and

Glasses (PacRim IV), Nov. 4-8, 2001, to be published by American Ceramics

Society (2002).

[8] S.J. Bless, N.S. Brar, G. Kanel, and Z. Rosenberg, “Failure Waves in

Glass,” J. Am. Ceram. Society, 75 [1] 1002-1004 (1992).

[9] L. Glenn and W. Janach, “Failure of Granite Cylinders under Impact

Loading,” Int’l. J. Fracture, 13 [1] 301-317 (1977).


[10] J.U. Cazamias, B. Reinhart, C. Konrad, L.C. Chhabildas, and S.J. Bless,

“Bar Impact Tests on Alumina (AD995),” Shock Compression of Condensed

Matter 2001, to be published by American Institute of Physics, 2002.

[11] H.D. Espinosa, Y. Xu, and N.S. Brar, “Micromechanics of Failure Waves

in Glass: I, Experiments,” J. Am. Ceram. Soc. 80 [1] 2061-73 (1997).

[12] J.L. Wise and D.E. Grady, “Dynamic, Multi-axial Impact Response of

Confined and Unconfined Ceramic Rods”; pp. 777-780 in High-Pressure Science

and Technology. Edited by S.C. Schmidt et al. 1993. AIP Conference Proceedings

309 (1994).

[13]. N.H. Murray, N.K. Bourne, J.E. Field, and Z. Rosenberg, “Symmetrical

Taylor Impact of Glass Bars,” Shock Compression of Condensed Matter – 1997,

American Institute of Physics (1998).

[14] J.U. Cazamias, P.S. Fiske, and S.J. Bless, “The Hugoniot Elastic Limit of

AlON,” Shock Compression of Condensed Matter-2001, to be published by

American Institute of Physics (2002).

[15] R. Russell, S. Bless, and T. Beno, “Impact Induced Failure

Phenomenology in Homalite Bars,” Shock Compression of Condensed Matter-

2001, to be published by American Institute of Physics (2002).

[16] J. Najar, “Dynamic Tensile Fracture Phenomena at Wave Propagation in

Ceramic Bars,” J. Physics IV 1 [C8] 647-652 (1994).


SHOCK COMPRESSION AND RELEASE PROPERTIES OF COORS AD995

ALUMINA

William D. Reinhart Lalit C. Chhabildas

Sandia National Laboratories Sandia National Laboratories

Weapons Science Applications Weapons Science Applications

PO Box 5800 PO box 5800

New Mexico, 87185-1181 New Mexico, 87185-1181

Dennis E. Grady Tsutomu Mashimo

Applied Research Associates, Inc. High Energy Rate Laboratory

4300 San Mateo Blvd. NE. Kumamoto University

Albuquerque, New Mexico, 87110 Kumamoto 860, Japan

ABSTRACT

An investigation of the shock compression, recompression and decompression

properties of Coors AD995 alumina (aluminum oxide) ceramic and single crystal

sapphire has been conducted. Well-controlled, planar impact experiments have

been performed in which stationary targets are impacted by ceramic plates to

pressures exceeding 100 GPa. In this study of Coors AD995 ceramic and single

crystal sapphire, dynamic material property data is obtained utilizing gun loading

techniques and high-resolution velocity interferometric tools. Techniques used to

determine the dynamic compression, recompression, and release behavior are

summarized herein.

INTRODUCTION

Ceramics in general have repeatedly demonstrated to be an effective armor

material due to its high dynamic yield strength compared to metals. However, like

many brittle materials ceramics are weak in tension as evidenced by the low spall

strength of the materials. There is also evidence that the transient strength of

many of these materials degrades - which has led to the concept of the existence

of failure waves in materials and is indicative of damaged material even when

under compression. Kanel et al [1] was the first to show conclusively from his

experiments on glass that the dynamic yield strength of glass decreases as the

shock dwell time increased. There is also evidence of the dynamic yield strength



degradation in boron carbide ceramic [2] when shocked above its Hugoniot elastic

limit.

Shock experiments are confined uniaxial strain experiments and are generally

referred to as the Hugoniot state of the material. To estimate the dynamic yield

strength of the material, the Hugoniot state is compared to a hydrostat - which is

determined by extrapolating the stress-strain behavior of the material determined

at lower hydrostatic pressures. Based on Von-Mises yield criteria the difference

between the Hugoniot stress and the hydrostatic pressure curve is defined as two-

thirds the dynamic yield strength. If this difference is (1) independent of the

shock-loading stress then the material exhibits elastic-perfectly plastic behavior,

(2) changing with increasing stress then the material exhibits a pressure-

dependent yield strength. An increase in yield strength may be attributed to many

factors such as rate-dependence and or a pressure dependent yield behavior, while

a decrease would be related to a softening behavior resulting from heterogeneous

deformation process and or from damage resulting from shock compression.

One of the objectives of the present study is to investigate the possibility of

determining, dynamically, the shock-hydrostat for ceramics. This technique has

been previously applied to investigate metals [3,4], and in particular has been

used extensively to characterize the strength properties of 6061-T6 aluminum and

tungsten in the shocked state. The method employs re-shock and release

experiments to be conducted from the same Hugoniot stress state to

experimentally evaluate the departure of the initial loading stress state from an

elastic plastic behavior. The asymmetry in the reloading and release path is then

used to determine the shock-hydrostat. In this study, well-controlled impact

experiments are performed on smooth-bore guns, and velocity interferometric

diagnostics [5] are used to acquire high-resolution shock compression, and

subsequent recompression or release data on alumina.

Aluminum oxide (Al2O3) is a widely used commercial ceramic because of its

useful electrical, and mechanical properties. It also has good optical properties

when used as a single crystal, commonly known as sapphire. In shock

experiments single crystal sapphire has been used as laser-interferometer

windows [6]. Extensive shock-Hugoniot equation-of-state studies have been

performed on aluminum-oxide primarily because of its wide applicability as an

armor ceramic. Sapphire was included in this study because it is the single crystal

form of Al2O3 and is the building block at a granular level. In this study, only

reshock and release experiments at 27 and 44 GPa on alumina are highlighted,

however, the Hugoniot experiments are reported to a peak stress level of over 1

Mbar (for alumina and sapphire).


MATERIAL

The aluminum oxide (Al2O3) used in this study is referred to as Coors AD995.

Its composition consists of 99.5% alumina and the remainder of the material is

aluminosilicate glass. The density of the material (Al2O3) was 3.89 g/cm3 and the

average longitudinal and shear wave speed was 10.56 km/s and 6.24 km/s

respectively. This yields an estimate of 7.71 km/s, 0.234 and 231.7 GPa for the

bulk wave velocity, Poisson’s ratio and the bulk modulus, respectively. Sapphire,

which is the single crystal form of Al2O3, has a rhombohedral-hexagonal crystal

structure with close-packed oxygen ions. Both c-axis and a-axis crystals were

used in this study. The elastic longitudinal wave speed for the c-axis and a-axis

crystal was determined to be 11.19 km/s [6] and 11.17 km/s [7]. The density of

the sapphire crystals used in this study was 3.98 gm/cm3.

EXPERIMENTAL METHOD

Compressive shock, reshock and release waves are produced in aluminum

oxide and sapphire with a single stage powder gun and a two-stage light gas gun.

The experimental configuration used for this study for both the powder gun and

the two-stage light gas gun is shown in Figure 1. The powder gun has an 89 mm

bore diameter and achieves impact velocities exceeding 2.3 km/s, while the two-

stage light gas gun utilizes a 28 mm bore diameter with projectile velocities

approaching 8 km/s. The powder gun projectile velocity is measured by three

electrical self-shorting pins, which are mounted on the target fixture, to accuracy

better than 0.5%. Additional electrical pins are incorporated to measure impact

planarity (typically about 1-2 milli-radians) and provide triggers to diagnostic

equipment. The two stage light gas gun incorporates a projectile velocity

measuring system call the Optical Beam Reflector [8] (OBR) that accurately

Ceramic

Low or High

Impedance Backing

Lithium

Fluoride

Window

Electrical Self

Shorting Pins

Projectile

VISAR

Ceramic

Low or High

Impedance Backing

Lithium

Fluoride

Window

Electrical Self

Shorting Pins

Projectile

VISAR

Figure 1. Experimental Configuration


measures projectile velocity to better than 0.2%. As with the target on the powder

gun, electrical pins are also used for impact planarity and diagnostic triggering.

SPT-2:

SPT-1:

SPT-4:

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Win

do

w V

elo

city

(k

m/s

)

0.25 0.30 0.35 0.40

Arbitrary time ( s)

CE58 ALRL3

ALRS2

0.5 1.0 1.5 2.0 2.5

Arbitrary time ( s)

0.0

0.5

1.0

1.5

2.0

Win

do

w V

elo

city

(k

m/s

)

SPT-2:

SPT-1:

SPT-4:

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Win

do

w V

elo

city

(k

m/s

)

0.25 0.30 0.35 0.40

Arbitrary time ( s)

CE58 ALRL3

ALRS2

0.5 1.0 1.5 2.0 2.5

Arbitrary time ( s)

0.0

0.5

1.0

1.5

2.0

Win

do

w V

elo

city

(k

m/s

)

ALRS1ALRS1a)

.

b).

SPT-3:CE60

SPT-3:CE60

Figure 2. (a) Window/interface particle velocity profiles for Al203, and (b) c-

axis and a-axis sapphire, respectively.

In Figure 1, the projectile is faced with the ceramic Coors AD995 or sapphire

and is backed with either a foam disk of low shock impedance, or a high shock

impedance material, tantalum, for reshock experiments. The target configuration

in Figure 1, will have a alumina (or sapphire) ceramic disk similar to that

mounted on the projectile and a single crystal lithium-fluoride is bonded with

epoxy to the back of the ceramic sample. The lithium-fluoride is an optical

quality disk, lapped and polished and is typically flat to within a few bands of

sodium light. One surface of the lithium-fluoride is diffused and approximately

100nm of aluminum is vapor deposited on the lapped surface before being glued

to the alumina disk. The particle velocity histories resulting from impact were

measured at the target/lithium-fluoride window [9] interface using a velocity

interferometer, VISAR [5]. The Doppler shifted interference fringes measured

with the VISAR are converted to a time-resolved velocity history and are shown

in Figures 2a and 2b for the experiments on alumina and sapphire, respectively.

The amplitude resolution is approximately

2% per fringe and typically two to three

fringes are achieved in the interface

acceleration resulting from the compressive

shock front.

EXPERIMENTAL RESULTS

Elastic Waves

The impact conditions for the

experiments in the current study are

summarized in Table I. The wave profiles

shown in Figures 2 are used to determine the

Elastic Shock

Elastic-Plastic

Deform ation Hugoniot State

M id-Point of shock

Elastic Shock

Elastic-Plastic

Deform ation Hugoniot State

M id-Point of shock

Figure 3. Shock wave profile

in alumina traversing from right

to left.


Hugoniot properties of the ceramic. Hugoniot refers to the peak stress states

achieved in the shock compression process. Figure 3 illustrates the particle

velocity (stress) wave profile traversing toward the left through the ceramic. In

this illustration, material to the left of the elastic shock front is undisturbed. The

leading edge of the precursor wave is used as a fiducial for the analysis in this

study. For alumina, the leading edge of the elastic wave traverses at the elastic

longitudinal wave speed of 10.74 km/s, a value that has been determined on

earlier studies in the alumina [10]. For sapphire the leading edge of the wave

traverses at 11.7 km/s, and is consistent with the linear elastic behavior reported

in the literature [6]. The Hugoniot elastic limit stress, ( hel), is determined using

the relation:

hel = ( o Cl ue ), (1)

where o is the initial density of the ceramic, Cl the elastic longitudinal wave

speed, and ue is the in-material particle velocity measurement prior to transition

to a plastic wave.

Table I. Summary of Impact Conditions

Exp.

No.

Impact

Velocity

(km/s)

Target

Thickness

(mm)

Impactor

Thickness

(mm)

ue

(km/s) e HEL

(GPa) e

Elastic

Strain

CE57 1.019 10.006 5.019 0.153 6.41 0.0143

CE58 1.572 10.008 5.008 0.154 6.44 0.0143

CE59 2.030 10.007 5.013 0.150 6.26 0.0140

CE60 2.329 9.998 5.005 0.178 7.44 0.0166

CE61 0.561 9.998 5.013 0.148 6.18 0.0136

CE62 2.211 9.987 5.005 0.175 7.32 0.0163

CE63 2.062 9.987 4.989 0.162 6.75 0.0151

ALRL3 2.158 7.988 3.070 0.169 7.06 0.0157

ALRS11

2.185 6.335 0.4992 3 3 3

ALRS2 2.208 6.337 4.211 0.162 6.78 0.0151

SAPT14

4.220 3.195 12.697 0.506 24.17 0.0432

SAPT24

4.431 3.193 12.708 0.527 24.48 0.0451

SAPT35

3.28 3.387 1.4665

0.5 24.00 0.0427

SAPT45

3.30 3.422 1.4875

0.532 25.50 0.04551 Reverse ballistic experiment 2 Aluminum Buffer for Window experiment

Table I. Summary of Impact Conditions

Exp.

No.

Impact

Velocity

(km/s)

Target

Thickness

(mm)

Impactor

Thickness

(mm)

ue

(km/s) e HEL

(GPa) e

Elastic

Strain

CE57 1.019 10.006 5.019 0.153 6.41 0.0143

CE58 1.572 10.008 5.008 0.154 6.44 0.0143

CE59 2.030 10.007 5.013 0.150 6.26 0.0140

CE60 2.329 9.998 5.005 0.178 7.44 0.0166

CE61 0.561 9.998 5.013 0.148 6.18 0.0136

CE62 2.211 9.987 5.005 0.175 7.32 0.0163

CE63 2.062 9.987 4.989 0.162 6.75 0.0151

ALRL3 2.158 7.988 3.070 0.169 7.06 0.0157

ALRS11

2.185 6.335 0.4992 3 3 3

ALRS2 2.208 6.337 4.211 0.162 6.78 0.0151

SAPT14

4.220 3.195 12.697 0.506 24.17 0.0432

SAPT24

4.431 3.193 12.708 0.527 24.48 0.0451

SAPT35

3.28 3.387 1.4665

0.5 24.00 0.0427

SAPT45

3.30 3.422 1.4875

0.532 25.50 0.04551 Reverse ballistic experiment 2 Aluminum Buffer for Window experiment 3 Elastic values not determined 4 C-cut single crystal sapphire 5 Tungsten ( = 19.2 g/cm3) used to Impact A-axis sapphire crystal.


Figure 4. Hugoniot results. a)

shock–velocity vs material

velocity b) stress vs. material

velocity c) stress vs strain

0

20

40

60

80

100

120

0.00 0.05 0.10 0.15 0.20 0.2Strain ( )

Str

ess (

GP

a)

C - cut Sapphire

Aluminum Oxide

A - cut Sapphire

0

20

40

60

80

100

120

0.0 0.5 1.0 1.5 2.0 2.5Particle Velocity (km/s)

Str

ess (

GP

a)

C- cut Sapphire

Aluminum Oxide

A - cut Sapphire

6

8

10

12

14

0.0 0.5 1.0 1.5 2.0 2.5

Particle Velocity (km/s)

Sh

ock V

elo

cit

y (

km

/s)

C - cut Sapphire

Aluminum Oxide

HEL-Sapphire

HEL-Al203

A - cut Sapphire

The in-material particle velocity is

determined through the impedance

matching relation:

ue = uw (Zw + Zm) / 2 Zm (2)

where uw, Zw, and Zm are the measured

velocity in the window material, and the

shock impedance of the lithium-fluoride

window and the ceramic material

respectively. The shock impedance of

the material is defined as the product of

its density and the shock velocity. The

equation of state of lithium-fluoride [10]

is used to calculate its shock impedance,

while the elastic shock impedance of the

alumina or sapphire is the product of o

and Cl, their respective density and the

elastic wave speed. The elastic strain, e,

is calculated using ue/Cl . The stress,

strain and the particle velocity results at

the elastic limit for the series of

experiments reported in this paper are

tabulated in Table I and shown in Figure

4.

Plastic Waves

The planar impact produces a

compressive wave of uniaxial strain,

which propagates across the target

specimen and into the lithium-fluoride

window. The measured velocity exhibits

a two-wave structure. The subsequent

structure following the elastic precursor

represents pressure hardening of the

material and this two-wave structure is

the result of a transition from elastic to

plastic deformation. As compression

within the shock increases during the

shock loading process, shear stresses

Figure 4. Hugoniot results. a)

shock–velocity vs material

velocity b) stress vs. material

velocity c) stress vs strain

0

20

40

60

80

100

120

0.00 0.05 0.10 0.15 0.20 0.25Strain ( )

Str

ess (

GP

a)

C - cut Sapphire

Aluminum Oxide

A - cut Sapphire

0

20

40

60

80

100

120

0.0 0.5 1.0 1.5 2.0 2.5Particle Velocity (km/s)

Str

ess (

GP

a)

C- cut Sapphire

Aluminum Oxide

A - cut Sapphire

6

8

10

12

14

0.0 0.5 1.0 1.5 2.0 2.5

Particle Velocity (km/s)

Sh

ock V

elo

cit

y (

km

/s)

C - cut Sapphire

Aluminum Oxide

HEL-Sapphire

HEL-Al203

A - cut Sapphire


will exceed the critical strength of the material (HEL) and plastic deformation

occurs in the observed second wave.

Because finite rise times are measured for the plastic wave, the plastic-

wave velocity, Usp, is taken at the center of the wave (Figure 3) and the

corresponding wave speed is given in Table II. Where symmetric impact

techniques are used, the particle (material) velocity, uph, behind the shock front is

exactly one-half the impact velocity. The Hugoniot stress, ph, and strain, ph,

behind the plastic-wave front are estimated using the following relations:

ph = e + [ o Usp (uph – ue)] (3)

ph = e + (uph – ue) / Usp (4)

The summary of all Hugoniot data in Tables I and II, are shown plotted in Figure

4, represents the study on Coors AD995 and agrees well with previous work

[10,12,13,14] of this material. It should be noted that the HEL of this material is

about 6.0 to 7.5 GPa while the HEL of the single crystal material is about 20 to

24 GPa. As indicated in Table II, the current studies span over the stress regime

of 18 to 100 GPa.

Table II. Hugoniot Summary

Exp. No. Usp

(km/s)

uph

(km/s) ph

(GPa) ph

Total Strain

CE57 8.38 0.510 18.02 0.0568

CE58 8.40 0.787 27.11 0.0896

CE59 8.49 1.015 34.83 0.1159

CE60 8.81 1.165 42.30 0.1286

CE61 9.18 0.281 11.01 0.0280

CE62 9.48 1.452 54.69 0.1502

CE63 9.72 1.457 55.83 0.1477

ALRL3 8.94 1.079 38.70 0.1176

ALRS1 8.86 0.762 26.27 0.0860

ALRS2 8.93 1.104 40.38 0.1178

SAPT1 10.54 2.110 90.62 0.1953

SAPT2 10.72 2.215 96.31 0.2025

SAPT3 10.92 2.285 101.3 0.2062

SAPT4 10.94 2.390 106.2 0.2153


Off-Hugoniot States

Most of the experiments conducted in this investigation, evaluate the off-

Hugoniot states of Coors AD995 alumina as it is allowed to release from

compression. Experiments ALRS1 and ALRS2 are the experiments on Coors

AD995 alumina as it is further recompressed from its original shocked Hugoniot

state. An incremental form of the conservation equations given by the relations:

= o c u (5)

= u/c (6)

is used to estimate the final released or reshocked stress, and strain, ,

respectively. The Lagrangian-wave velocity, c, corresponds to the material

particle velocity change, u. This is indicated in the x-t diagram in Figure 5a.

Although the figure emphasizes the reshock experimental configuration, the same

concept is true for a release diagram. The backing to the ceramic impactor is

replaced by carbon foam, a low shock impedance material, so that a reflected

release wave will propagate in the impactor material when the shock arrives at the

Ta

nta

lum

Ceramic

Win

dow

Shock

Elasti

c-Res

hock

Plasti

c-Res

hock

Time

X

Ta

nta

lum

Ceramic

Win

dow

Shock

Elasti

c-Res

hock

Plasti

c-Res

hock

Time

X

Tan

talu

m

Ceramic Ceramic

Win

dow

ElasticPlasti

c

Release Waves

Elastic-

Reshock

Plasti

c-Res

hock

te

tp

tf

tp

Different Wave Velocities

resulting from wave

interactions

High Stress:

SampleWindow Stress

Us effec

tive

Time

X

Tan

talu

m

Ceramic Ceramic

Win

dow

ElasticPlasti

c

Release Waves

Elastic-

Reshock

Plasti

c-Res

hock

te

tp

tf

tp


resulting from wave

interactions

High Stress:

SampleWindow Stress

Us effec

tive

Time

X

b)a)

Figure 5. X-t diagram typical experiments. a) depiction of symmetric impact

with wave interactions, b). alternate technique where by wave interactions are

eliminated.

Ta

nta

lum

Ceramic

Win

dow

Shock

Elasti

c-Res

hock

Plasti

c-Res

hock

Time

X

Ta

nta

lum

Ceramic

Win

dow

Shock

Elasti

c-Res

hock

Plasti

c-Res

hock

Time

X

Tan

talu

m

Ceramic Ceramic

Win

dow

ElasticPlasti

c

Release Waves

Elastic-

Reshock

Plasti

c-Res

hock

te

tp

tf

tp


resulting from wave

interactions

High Stress:

SampleWindow Stress

Us effec

tive

Time

X

Tan

talu

m

Ceramic Ceramic

Win

dow

ElasticPlasti

c

Release Waves

Elastic-

Reshock

Plasti

c-Res

hock

te

tp

tf

tp


resulting from wave

interactions

High Stress:

SampleWindow Stress

Us effec

tive

Time

X

b)a)


0

10

20

30

40

50

60

0.00 0.04 0.08 0.12 0.16

Strain ( )

Loading Path

Hugoniot Stress

Reshock

elastic

0

10

20

30

40

50

60

0.00 0.04 0.08 0.12 0.16

Strain ( )

Loading Path

Hugoniot Stress

Reshock

elastic

Str

es

s (

GP

a)

Str

es

s (

GP

a)

The results of companion

reshock and release experiments

conducted at approximately

27 GPa and 40 GPa are shown in

Figures 6 and 7, respectively.

The results of these experiments

will be highlighted in this paper.

Also shown in the figure is a

calculated hydrostat for the

alumina based on Murnaghan

equation of state where the bulk

modulus of Coors AD995

alumina is used. The bulk

modulus is based on the

ultrasonic sound speed

measurements on the samples. A value of four (4) is used for the pressure

backing/sample interface.

The analysis is

approximated by

representing the finite rise-

time of the shock in the

impactor as a single shock

wave traversing at an

effective shock velocity,

calculated using the

relation, Ueff = ph/( ouph),

where the stress ph and

particle velocity uph are the

first shocked states. The

Lagrangian-wave velocity

is estimated from the time difference between the arrival of the leading edge of

the release/reshock wave at the window/sample interface and the time at which

the effective shock arrives at the back surface of the impactor (Finite rise times

upon reshock are measured because an elastic-plastic wave is observed). As

indicated in figure 5, the release fan or the reshock is also perturbed by the

reflected release wave that emanates at the target window interface. Making the

ratio of the target sample dimension to the impactor dimension large, confines the

interaction zone towards the window interface, and minimizes this perturbation.

An alternate technique (Figure 5b) is to impact the window directly so that all

wave interactions are eliminated. This is specifically done in the reshock

experiment ALRS1.

0

10

20

30

40S

tress (

GP

a)

0.00 0.02 0.04 0.06 0.08 0.10 0.12

Strain ( )

Loading Path

Hugoniot Stress

Calculated Hydrosta

t

UnL

oadin

gPath

s

Reshock

0

10

20

30

40S

tress (

GP

a)

0.00 0.02 0.04 0.06 0.08 0.10 0.12

Strain ( )

Loading Path

Hugoniot Stress

Calculated Hydrosta

t

UnL

oadin

gPath

s

Reshock

elastic

Figure 6. Stress-strain plots for ALRS1 and

CE58 depicting loading and unloading paths.

UnL

oading P

ath

UnL

oading P

ath

Figure 7. Stress-Strain plots for

ALRS2 and CE60 depicting

compression, recompression and

unloading behavior.


derivative of the bulk modulus and is based on high-pressure x-ray diffraction

data on single-crystal sapphire [15].

DISCUSSIONS

In this study, we are reporting the shock Hugoniot of Coors AD995 alumina to

approximately 60 GPa, and that of sapphire to stresses over 100 GPa. The

alumina used in this study is the same batch of material used in an earlier

investigation [10,16]. Results from previous investigations [12,13,14] are not

reported herein mainly because the material studied may not be quite the same –

even though the results do suggest a good agreement with past studies on other

types of alumina.

Elastic- plastic waves

The Hugoniot elastic limit of Coors AD995 reported in Table I vary from

about 6.0 GPa to 7.5 GPa, while the corresponding values for single crystal

sapphire is approximately 24 GPa. This is consistent with previous studies on

single crystals in other materials [16,17] that report higher elastic limits than

polycrystalline materials. A good example is the previous study in crystalline

quartz [18] that has indicated different values of HEL for different crystal

orientation. This is because different slip systems are activated during the

dynamic yielding process. In a polycrystalline material, all these different slip

systems are randomly distributed and hence the results will be dominated by the

weaker slip systems (such as grain boundaries, presence of glass) that could yield

at a lower stress. The Hugoniot elastic limit in alumina will be overdriven when

the shock velocity measurement exceeds the elastic-wave velocity of 10.74 km/s.

Another interesting feature observed in this study is the transition from

elastic yielding to plastic deformation for the two different materials. The post

yielding process in polycrystalline alumina is considerably ramped suggesting a

work hardening type of process or heterogeneous yielding from different slip

systems followed by a plastic-wave that has a finite rise time even up to stresses

approaching 60 GPa. For single-crystal sapphire, however, the transition to the

plastic deformation state is very different. The post yielding process appears to be

either constant or suggests a decrease prior to the arrival of the plastic wave. This

decrease could either be due to elastic precursor decay or may be indicative of a

softening behavior in the single crystal. Also, the rise-time of the plastic-wave for

single crystal sapphire is extremely rapid. It should also be noted that even at

100 GPa, the elastic limit is not overdriven. In this instance, the elastic limit will

be overdriven when shock velocities exceed 11.6 km/s. The interface particle-

velocity measurement for single-crystal sapphire suggests fluctuations behind the

shock front and is presumably a result of heterogeneous deformation process in

the single crystal.


Shock Velocity vs Particle Velocity

The variation in shock velocity vs. particle velocity for both materials is

shown in Figure 4(a). As indicated in the figure the elastic limit for Coors AD995

and single crystal sapphire extend up to particle velocity of ~ 0.15 and 0.5,

respectively. The shock-velocity measurements below a particle-velocity

measurement of 0.5 km/s indicate a quasi-elastic behavior, for polycrystalline

alumina, indicates that the dynamic yielding process may not be totally complete.

The shock-velocity measurements above a particle velocity of 0.5 km/s suggest

that the yielding process may be very nearly complete. As indicated in Figure

4(a), the observed shock-velocity vs. particle-velocity variation for single-crystal

sapphire experiments at Mbar pressures appear to be consistent with the

measurements depicted for alumina at stresses up to 50 GPa. However, for a

conclusive interpretation there should be overlapping experiments in the same

shock-velocity/particle-velocity regime for the two different materials. A least

squares fit to the shock vs. particle velocity data beyond the elastic limit would

yield the relation:

Us (km/s) = 1.675up + 7.027 (7)

Stress vs. Strain

The values for stress vs. strain for both the materials tabulated in Table I

and II are shown plotted in Figure 4c. The HEL for the single crystal sapphire is

larger than the HEL of Coors AD995 alumina. As indicated in the figure, the

volume compression has being determined to stresses above 100 GPa and to about

22% strain. The experiments do suggest that the single-crystal sapphire

compression appears to be consistent with that of polycrystalline Coors AD995

alumina at high stresses. This implies that either the shear stress in the shocked

state are similar for both materials or that both materials have collapsed to the

hydrostat. It should also be noted that in the shocked state, melting cannot be

ruled out at about 100 GPa. It would, however, be desirable to have overlapping

experiments at the same stress. This also provides an incentive to conduct re-

shock and release experiments in the future for both materials at megabar stresses

to verify these assumptions.

Stress and Wave Speed vs. Particle Velocity

The stress vs. particle velocity behavior is shown in Figure 4b. As in

figures 4a and 4b, the figure shows a similar behavior as discussed above, namely

higher HEL’s for the single crystal sapphire when compared to the polycrystalline

material alumina. A least squares fit to the stress (GPa) vs. particle-velocity

(km/s) data beyond the elastic limit would yield the relation:


ph = uph uph2

(8)

The Lagrangian wave speed, Cl, in the material can be obtained by using the

relation:

Cl = (1/ o) (d ph / duph) (9)

which yields;

Cl (km/s) = 7.45 + 3.37 uph (10)

Equation (10) is plotted as a function of particle velocity in Figure 8 as a solid

line. Based on a constant Poisson’s ratio of 0.247, a Lagrangian elastic wave-

speed is calculated and is also shown (as a dashed line) in Figure 8. The

experimentally determined stress vs. particle-velocity relation can be used to

determine the Lagrangian bulk and elastic wave speed in the material (to obtain

the Eulerian wave speed one can use the relation Ce = ( o/ Cl.). As indicated in

Figure 5, we can also calculate the speed at which the leading edge of the release

or reshock wave traverses in the shocked material. The Lagrangian wave speeds

6

10

14

18

22

0.5 1.0 1.5 2.0 2.5Particle Velocity (km/s)

Lag

ran

gia

n W

ave S

peed

(km

/s)

Experiments

Experimentally Calculated

Elastic Wave Speed Assuming

Constant Poisson's Ration

Figure 8. Variation of Lagrangian elastic and bulk wave

speed as a function of particle velocity.


of the leading edge of the wave obtained from these experiments are shown as

triangles (fitted with a dashed line) in Figure 8. As indicated in the figure, the

wave-speed measurements obtained from the off-Hugoniot experiments indicate

that the material response remains elastic up to a particle velocity of 1.5 km/s. At

a particle velocity of 2.2 km/s it collapses to the hydrostat. This also explains why

the single-crystal sapphire stress-strain compression data at about a Megabar is

consistent with the polycrystalline stress-strain compression results (see figure

4c).

Reshock and Release States

The results of companion reshock and release experiments conducted at

approximately 27 GPa and 40 GPa are shown in Figures 6 and 7, respectively.

Also shown in Figure 6, is a calculated hydrostat for the alumina based on

Murnaghan equation of state where the bulk modulus of Coors AD995 alumina is

used. In both these experiments the leading edge of the reshock or release wave

traverses at an elastic wave speed (see Figure 8). The release path exhibits an

elastic release from the initial shocked state. The reloading path shows precursor

elastic recompression and the final reshocked state to about 37 GPa. This reshock

state lies above the calculated hydrostat and Hugoniot states. Similar results are

obtained for the reshock and release experiment at approximately 40 GPa. In this

experiment, within the experimental uncertainties, extrapolation of the static

hydrostat to very high pressures and the current experiments, the shocked state

appears to be on the hydrostat. During recompression from about 40 to 57 GPa,

the recompression wave exhibits an elastic recompression. The leading edge of

the release-wave traverses at an elastic wave velocity and as evidenced by the

elastic release. This is indicative of the loss of shear strength in the material; this

phenomena has been observed previously by Asay and Chhabildas in 6061-T6

aluminum, [2] and by Kanel [1] in glass. The damaged material in alumina

resulting from shock compression is presumed to cause the shear strength loss.

Therefore, the material is exhibiting strength recovery during the recompression

process. This phenomena has also been observed in Coors AD995 alumina even

at its Hugoniot elastic limit [17]. Although the shocked state of the material

exhibited in Figure 6 lies above the hydrostat – this technique can also be used to

experimentally determine a shock hydrostat [3] at very high dynamic stresses and

will be the subject of future discussions. This technique is anticipated to be more

accurate than extrapolating the hydrostatic data because the dynamic hydrostat

will be the mean value of the reshock and release end states from a common

Hugoniot state.


SUMMARY

Shock compression, recompression and decompression properties are

summarized. Well-controlled, planar impact experiments have been performed

on Coors AD995 ceramic and single-crystal sapphire ceramic plates to pressures

exceeding 100 GPa. In this study of Coors AD995 ceramic and single-crystal

sapphire, dynamic material property data are obtained utilizing gun loading

techniques combined with high-resolution velocity interferometric tools.

Substantial experimental data on the dynamic response of alumina and

sapphire exist. Results of these studies are unique in that they are probing the

strength properties of the material in the shocked state using reshock and release

test methodologies. This has allowed wave speed/sound speed measurements to

stresses above 1 Mbar. Strength loss in the Hugoniot state is evidenced by

precursor elastic compression in the recompression process. This should allow

the development of damage models needed for use as material models in

computational codes. This technique also can yield the dynamic shock hydrostat

to pressure in excess of 1 Mbar using experimental methods and will be the

subject for discussions in future investigations.

REFERENCES

1. G. I. Kanel, S.V. Rasorenov, V. E. Fortov, The Failure Waves and

Spallations in Homogeneous Brittle Materials, Shock Compression of Condensed

Matter, (1991), Schmidt, Dick, Forbes, Tasker, eds., Elsevier Science Publishing,

451-454, 1992.

2. D. E. Grady and R. L. Moody, Shock Compression Profiles in Ceramics,

Sandia National Laboratories Report, SAND96-0551, March 1996

3. J. R. Asay and L. C. Chhabildas, Determination of Shear Strength of

Shock-Compressed 6061-T6 Aluminum, Shock Waves and High-Strain-Rate

Phenomena in Metals, Myers and Murr, eds., Plenum Pub. Corp, New York, NY

(1981)

4. L.C. Chhabildas, J.R. Asay, L.M. Barker, Shear strength of Tungsten

Under Shock and Quasi-Isentropic Loading to 250 GPa, Sandia National

Laboratories Report, SAND88-0306, April, 1988.

5. L. M. Barker and R. E. Hollenbach, Laser Interferometer for Measuring

High Velocities of any Reflecting Surface, J. Appl. Phys. 43, (1972), pp. 4669-

4675.

6. L. M. Barker, R. E. Hollenbach, Shock-Wave Studies of PMMA, Fused

Silica, and Sapphire, J. of Applied Physics, Vol. 41, No. 10, 4208-4226,

September 1970.

7. Acoustic velocity measurements on a-axis single crystal sapphire

performed by J. H. Gieske, Sandia National Laboratories, Albuquerque, NM.


8. T. F. Thornhill, W. D. Reinhart, C. H. Konrad, L. C. Chhabildas, Accurate

Velocity Measurements of the Two-Stage Gun Projectile, 51st Aeroballistics

Range Association Meeting, September 17-21, 2000 Madrid, Spain.

9. J. L. Wise, L. C. Chhabildas, Laser Interferometer Measurements of

Refractive Index in Shock-Compressed Materials, Shock Waves in Condensed

Matter, Gupta, eds., (Plenum, New York), 441, 1986.

10. D. E. Grady, Dynamic Properties of Ceramic Materials, Sandia National

Laboratories Report, SAND88-3266, February 1995.

11. W. J. Carter, Hugoniot Equation of State of Some Alkali Halides, High

Temperature-High Pressure. 5:313 (1973)

12. T. Mashimo, Y. Hanaoka, K. Nagayama, Elastoplastic Properties under

Shock Compression of Al203 Single Crystal and Polycrystal, J. Appl. Physics. 63,

327 (1988)

13. R. A. Graham, W. P. Brooks, Shock-Wave Compression of Sapphire from

15 to 420 Kbar. The Effects of Large Anisotropic Compression., J. Phys. Chem.

Solids, Vol. 32, pps. 2311-2330 (1971), printed in Great Britain.

14. M. N. Pavlovskii, Shock Compression of Six Hard Substances, Soviet

Phys. Solid State, 12, 1736 (1971)

15. Y. Sato, S. Akimoto, Hydrostatic Compression of four corundum-type

compounds: Al203, V2O3, Cr2O3, and Fe2O3, J. Appl Phys., 50(8), August 1979.

16. M. D. Furnish, L.C. Chhabildas, Alumina Strength Degradation in the

Elastic Regime, Shock Compression of Condensed Matter, (1997), Schmidt,

Dandekar, Forbes, eds., pp. 501-504, 1997.

17. L. E. Pope, A. L. Stevens, Wave Propagation in Beryllium Single Crystals,

Metallurgical Effects at High Strain Rates, Rohde, Butcher, Holland, and Karnes,

eds., pp. 350-366 (1973).

18. G. I. Kanel, S. V. Razorenov, A. V. Utkin, V. E. Fortov, K. Baumung, H.

U. Karow, D. Rusch, and V. Licht, Spall Strength of Molybdenum Single Crystals,

J. Appl. Phys. 74 (12), December 1993.

19. G. R. Fowles, Dynamic Compression of Quartz, Journal of Geophysics

Res., 72, 5729, (1967).


COMPRESSIBILITY AND SHEAR STRENGTH OF TITANIUM DIBORIDE

UNDER PLANE SHOCK WAVE LOADING

D. P. Dandekar and E. J. Rapacki


AMSRL-WM-TD


ABSTRACT

Compressibility and shear strength of ceramics influence the potential

usefulness of these materials in protective systems against ballistic impact threats.

These properties of ceramics, due to their dominant brittle nature, can undergo

drastic changes under impact induced stress waves, and thus determine its

ultimate impact worthiness. This work brings together the results of shock wave

investigations on titanium diboride (TiB2) having a bearing on its compressibility

and shear strength. Further, the results of a limited number of shock–release–re-

shock (double-shock) experiments performed on TiB2 indicate that the observed

work–hardening response of TiB2 above the Hugoniot Elastic Limit (HEL) under

the initial shock compression augments its shear strength under the subsequent

shock wave loading. The results of this type of experiment are crucial for

developing a better understanding of the performance of ceramics under ballistic

impact, and for aiding the development of material models to predict the ballistic

performance of ceramics in armor configurations.

INTRODUCTION

Ceramic materials have received considerable attention from the ballistic

impact protection community for armor applications ranging from personnel

protection (body armor) to combat vehicle protection (integral composite armor).

Their very high strength to mass density ratio makes these materials attractive for

high performance, low weight armor system components. The asymmetry of the

compressive and tensile strengths of most ceramics, however, creates challenges

for their implementation. The compressibility and shear strength of a material

determine its dilatational and deviatoric deformation response to loading.

Because ballistic impact conditions produce propagating shock waves, which

subsequently are reflected from impedance mis-matched material interfaces, the



shock-release characteristics of armor materials must be known. The degradation

of strengths due to these load-unload conditions also provides insight for the

temporal longevity of the initial high strength characteristics. The shock-release-

reshock, or double-shock impact experiment1, 2, 3, 4

additionally provides a precise

methodology to probe a material’s retained strength, and hence its utility as an

armor system component, since ballistic loading times are typically of significant

duration.

This paper explains in some detail the rationale, implementation and analysis

of such double-shock experiments. A well-known high performance armor

ceramic, titanium diboride (TiB2) is the material that has been focused on in the

analysis of extant data, and the material investigated by the multi-shock impact

technique. The retained shear strength and work-hardening behavior observed in

this material’s response to plane shock wave loading helps to explain its excellent

performance as ballistic armor.

EXPERIMENTAL TECHNIQUE, ANALYSIS AND RESULTS

A schematic representation of double shock experiments is shown in Fig. 1.

Dandekar, Gaeta and Horie2 and Dandekar

3, 4 give detailed descriptions of the

double shock experiments. Briefly, an initially shocked material undergoes a

totally stress-free state for a pre-determined time duration before being shock

compressed a second time. There are two general configurations of this type of

experiment. In one (Fig. 1a), a thick impactor simultaneously impacts two

targets. One target consists of a single specimen, and the other consists of two

specimens separated by a measured, uniform gap between them. In the latter, the

firstly impacted target becomes the impactor of the second after stress release.

The free surface velocities are measured by multibeam interferometry, using

VISAR, (velocity interferometer system for any reflector). Alternatively, (Fig.

1b) two impactors, which are separated by a measured, uniform gap between

them, impact a target with an in-material gage to measure stress or particle

velocity in the target. Variables in both experimental configurations include: the

relative thicknesses of impactors and targets, the gap between the two targets or

two impactors, and the relative mechanical impedances of the impactors and

targets. The experimentally measured variables are: impact velocity, tilt of

impact, successive shock and release wave speeds, free surface velocity profiles

in the configuration of Fig. 1a, and stress wave profiles at various target locations

in the configuration of Fig. 1b. In either configuration shown in Fig. 1, a

transparent window could be used to monitor the particle velocity profile or stress

wave profile at the target-window interface. It should be mentioned that the

configuration shown in Fig. 1a does not permit observation of the change in wave

profile due to propagation of the second shock in a material, except in transparent

materials. The advantage of using an interferometer for inferring the shock


properties of a material under multiple shocks and release is that it does not

require any calibration. The analysis of in-material wave profiles, obtained from

the configuration shown in Fig. 1(b), is straightforward.

Figure 1. Schematic configurations of shock, total release, and re-shock

experiments; (a) single impactor/multiple targets, whose free surface velocities

are monitored by multiple VISAR beams, and (b) multiple, successive

impactors/single target, which uses an in-situ stress or particle velocity gage to

monitor the wave profile.

Shock wave impact experiments were performed with a ten (10) cm diameter

single-stage gas gun at the U. S. Army Research Laboratory. The circular disc

specimens of TiB2 were such as to satisfy the one dimensional strain condition for

the total duration of wave profile measurements.4 The 32-50 mm diameter

ceramic disks were lapped flat to 5 m, and their opposing faces were mutually

parallel to within two (2) parts in 104 over the lateral dimension of the disks. The

deviation of planarity of impact in any given experiment was around 0.5 mrad.

Impact velocity of a projectile was determined by measuring the time intervals

signaled by the electrical shorting of four charged pins of known separation

distance. The precision of impact velocity determinations was 0.5%. The wave


profiles were measured by means of a multibeam VISAR, or by means of a

Manganin stress gage. The precision of measurements of the particle velocity by

using VISAR, and of the stress measurements by means of Manganin gages, were

1% and 2.5%, respectively. The Manganin stress gages (Micro-Measurements,

Inc., Type LM-SS-125CH-048) are calibrated to nine (9) GPa.3

As mentioned above, the configuration of a repeated shock experiment shown

in Fig. 1(a) does not permit direct observation of the transmitted wave profile in a

material subjected to the repeated shock. However, the response can be

monitored through the material being subjected to the first shock and release.

Hence, the analysis is done in two stages. First, the response of the material

subjected to the first (initial) shock, generated by the impact of an impactor with

velocity v, and total release from this shock, are obtained as in a normal

transmission shock wave experiment. Incidentally, this can be conducted

simultaneously with the multiple shock experiments, as shown in Fig. 1(a), thus

insuring identical impact velocity for both experiments. In the second stage, the

analysis follows the procedure adopted for a front surface impact experiment, in

which the stress state attained during the second (subsequent) impact of the pre-

shocked material is obtained from the measured free surface wave profile, and the

measured response of the material during the first shock and its release.

Dandekar, in Ref. 4, presents the specific details of the analysis.

First Shock Response

Ambient properties and elastic compression: The ambient properties of the

TiB2 from various sources used in the various investigations to determine their

shock response are given in Table I. The two values of the Hugoniot Elastic

Limit (HEL) given in this table require some explanation. Plane shock wave

investigations on TiB2, irrespective of source and/or chemical impurities, show

the presence of two cusps prior to the onset of inelastic deformation. These two

cusps indicated two precursor waves, each propagating with velocity very nearly

the same as the elastic waves, but differing in magnitudes. The values of the first

and the second cusps were found to be between 4.2 - 5.9 GPa, and 8.0 - 17.0 GPa,

respectively. The first cusp was shown to be the limit of elastic deformation in

TiB2 in the sense that the material was damaged irreversibly above this first cusp.5

Compressive strength: Compressive strength or compressibility of a material

at different pressures is obtained from hydrodynamic compression of a material

under dynamic loading. Hydrodynamic compression of a material may be

represented by a functional dependence of the bulk modulus (K) on pressure. It is

well known that hydrodynamic compression of a hard material like a ceramic is

satisfactorily represented by its initial bulk modulus K0 and its initial pressure

derivative i.e., K0'. Thus, the values of K0 and K0' jointly convey the magnitude of

the compressive strength of a material. Dandekar and Benfanti6 analyzed the


existing data on shock compression of TiB2. The study of the pressure

dependence of the elastic constants of TiB2 by Frankel, Abbate and Dandekar7

yields a value of 2.02 0.18 for the pressure derivative of the bulk modulus, K0'.

This value of K0' is in agreement with the high pressure shock data reported by

Gust, Holt and Royce8 and Marsh

9. This value of K0' together with the value of

bulk modulus K0 given in Table 1 for TiB2 yield its compressive strength.

Table I. Properties of various source TiB2 ceramic at ambient pressure condition.

Manufacturer Ceradyne

Eagle-

Picher Cercom

Union

Carbide (unknown)

Data Reference [6,7] [10] [10] [8] [11]

Mass Density,

(Mg/m3) 4.49±0.01 4.45 4.51 4.515±0.002 4.36±0.03

Elastic Wave

Velocities (km/s):

longitudinal, CL 11.23±0.11 10.93 10.79 11.21±0.20 10.79±0.15

shear, CS 7.41±0.13 7.30 7.43 7.30±0.16 7.24±0.10

bulk, CB 7.27±0.24 6.96 6.54 7.39±0.37 6.82±0.25

Elastic Constants:

bulk modulus, K0

(GPa)

237±16 216 193 247±12 203±15

Shear modulus,

(GPa)

246 ±9 237 249 241 ±5 228 ±6

Poisson’s ratio, 0.114±0.011 0.097 0.049 0.131±0.012 0.090±.009

HEL stresses, HEL:

1st cusp (GPa) 5.9 4.7 - 5.2 - - 4.2 - 4.9

2nd cusp (GPa) 13.5 13.1-13.7 17.0 8.0 9.0

Shear strength: The stress offset between the hydrodynamic compression and

the shock hugoniot data permits calculations of the shear strength of a material.

Shear strength for Cercom TiB2, from the shock hugoniot measurements to 60

GPa reported by Grady10

was calculated in this manner and reported in Ref. 6.

The values of shear strength of the Cercom produced TiB2, obtained from

simultaneous measurements of longitudinal and lateral stress under plane shock

wave loading12, 13

are compared with the those obtained on the same material from

the offset between the shock Hugoniot and the hydrodynamic compression of

TiB2.6, 7, 10

The values of shear strength of the TiB2 used in the investigation by

Winkler and Stilp11

are included for completeness. The values of shear strength

retained, and lateral stresses developed, in the various TiB2 under the first shock

wave propagation are given in Table II. The values of shear strength, , and

lateral stress, Y, given in parentheses in Table II are calculated from linear


elasticity theory; i.e., Eqns. 1 and 2. In these equations X is the longitudinal, or

impact stress, and is Poisson’s ratio.

Table II. Shear strength and lateral stress imposed under plane shock wave

loading of TiB2; the materials are from the various sources indicated.

Compression

V/V0 (-)

Impact Stress

X (GPa)

Shear Strength

(GPa)

Lateral Stress

Y (GPa)

Eagle-Picher [Ref. 6 & 10]

0.9173 31.4 8.4 14.6

0.8929 46.7 14.6 17.5

Cercom [Ref. 6 & 10]

0.9825 8.6 3.8 (4.1)* 0.8 (0.4)*

0.9714 15.0 6.9 (7.1)* 1.2 (0.8)*

0.9428 24.5 9.3 5.9

0.9216 32.6 11.7 9.2

0.9207 32.2 11.2 9.8

0.8806 49.8 16.5 16.8

0.8635 61.0 21.2 18.6

Cercom [Ref. 12]

0.9876 6.8 3.2 0.4

0.9820 10.0 4.6 0.8

0.9615 19.5 8.4 2.7

0.9605 19.5 8.2 3.1

0.9474 24.0 9.2 5.6

Cercom [Ref. 13]

- 7.1 3.2 {2.5}** 0.7 {2.1}**

- 16.5 6.4 {4.5}** 3.7 {7.5}**

- 18.6 7.0 {6.0}** 4.6 {6.6}**

Unknown [Ref. 11]

0.9912 4.4 1.9 (2.0)* 0.6 (0.4)*

0.9800 9.1 3.7 (4.1)* 1.7 (0.9)*

0.9711 10.6 3.4 3.9

0.9678 14.3 5.6 3.1

Ceradyne [Ref. 3]

0.9846 8.7 3.7 (3.8)* 1.3 (1.1)*

0.9774 12.8 5.4 (5.6)* 2.0 (1.6)*

0.9545 19.1 5.6 7.9

0.9523 19.6 5.6 8.4


0.9532 19.5 5.7 8.1

*Parameter by elasticity calculation **parameter for failed material

= (1-2 ) X /2(1- ) (1)

Y = X /(1- ) (2)

These values are comparable to those obtained from the stress offset between the

shock hugoniot and hydrodynamic compression of TiB2. These data indicate that

the shear strength retained by the various source TiB2 increases with an increase

in the magnitude of the impact stress under plane shock loading. This is most

clearly evident in Cercom and Eagle-Picher material because shock experiments

in these materials were performed at impact stresses exceeding twice the

magnitude of their respective HEL’s; see Table I. Further, Bourne, Gray and

Millet13

measured lateral stresses at two (2) mm and six (6) mm from the impact

surface in TiB2 specimens, and their lateral stress profiles show a two-step

structure. They associate the first step with the shear strength of intact material,

and the second step with the shear strength of damaged material, respectively.

They assumed that the damaged material was generated by the propagation of a

failure front in the material. The values of shear strength and lateral stress

associated with failed material are given in curly brackets in Table II. Figure 2

shows a plot of shear strengths of various TiB2 as a function of impact stress.

Second Shock Response

Dandekar4 performed the double shock experiments on Ceradyne material

only, to support then ongoing work on that material at the Army Research

Laboratory. The primary purpose was to examine whether the work hardening

behavior exhibited by Cercom and Eagle-Picher TiB2 was also exhibited by

Ceradyne TiB2 material. An analysis of those single- and double-shock

experiments on Ceradyne material at ~19 GPa is given in Table III. The data

show that the first shock of magnitude ~19 GPa, and subsequent release there-

from, follows an elastic-plastic (work-hardening) deformation path. The

subsequent second shock of ~19 GPa in this material is attained through elastic

deformation, because the impedance magnitude for the second shock is 55

Gg/m2s, that is, equal to its elastic impedance. As a consequence, the shear

strength of this TiB2 increases from 5.6 GPa under the first shock, to 8.3 8.7

GPa under the second shock. Further, the estimates of lateral stress imposed on

this material during the first and second shock decrease from 8.0 GPa to 1.7 2.0

GPa. The values of shear strength and lateral stress given in parentheses in Table

III were obtained using the elastic relations, Eqns. 1 and 2, and assumed that the

Poisson’s ratio remained unaltered, i.e., equal to 0.114. The shear strength


value of 8.7 GPa and the lateral stress value of 1.7 GPa were obtained under the

assumption that the equation of state of an intact TiB2 is invariant. Using the new

value of shear strength , or lateral stress Y, in conjunction with the axial

compressive stress X, the new computed value of is 0.082. It is well known

that the Poisson’s ratio value of a solid does change under compression, but

whether the above-calculated new value is valid requires independent verification.

TiB2

0

5

10

15

20

0 10 20 30 40 50 60

Impact Stress, X (GPa)

Sh

ea

r S

tren

gth

, (

GP

a)

Eagle Picher [6,10]

Cercom [6,10]

Cercom [12]

Cercom [13]

Cercom [13]

(w/failure)Unknown [11]

Ceradyne [3]

Figure 2. Shear strength versus impact stress for TiB2 ceramics; the materials are

from the various sources indicated.

DISCUSSION OF RESULTS

The results of shock wave experiments indicate that titanium diboride exhibits

increasing shear strength with increase in impact stress. This behavior persists in

Cercom and Eagle-Picher materials to 46 and 61 GPa, respectively. The observed

work-hardening behaviour of titanium diborides is substantiated through the

results of a few shock-release and reshock experients in Ceradyne material. This

material deforms in an elastic-plastic manner under a first shock of magnitude

19.1-19.5 GPa, and maintains a shear strength of magnitude 5.6 GPa. When this

material is subjected to a second shock of the same magnitude following a

complete release from the first shock, it deforms elastically to 19 GPa. The value

of the shear strength under the second shock compression, calculated from the

offset between the shock hugoniot and the hydrodynamic compression, is 8.7

GPa. An estimate of the shear strength, using the elastic relation, Eqn. 1, with =


0.114, is 8.3 GPa. These two values for the shear strength are within the

uncertainty of the measurements.

Table III. Summary of 1st and 2nd shock response of Ceradyne TiB2 at ~19 GPa.

Experiment #: 403 406 422

1st shock:

Axial stress, X (GPa) 19.6 19.5 19.1

Particle velocity, u (km/s) 0.448 0.445 0.433

Mass density, (Mg/m3) 4.714 4.710 4.704

Shear strength, (GPa) 5.6 5.7 5.6

Lateral stress, Y (GPa) 8.4 8.1 7.9

1st release:

Free surface vel., u (km/s) 0.803 0.837 0.781

Impedance, Z (Gg/m2s) 55 50 55

Mass density, (Mg/m3) 4.571 4.554 4.564

2nd shock:

Axial stress, X (GPa) - - 19.0

Particle velocity, u (km/s) - - 0.348

Impedance, Z (Gg/m2s) - - 55.

Mass density, (Mg/m3) - - 4.700

Shear strength, (GPa) - - 8.7 (8.3)*

Lateral stress, Y (GPa) - - 1.7 (2.5)*

* see DISCUSSION text

Bourne, et al.13

observed two steps in the lateral stress profiles of Cercom

TiB2 when shocked to between 7 and 19 GPa. Figure 2 shows that the shear

strengths based on the magnitude of the first step are in reasonable agreement

with those measured by Rosenberg et al.12

and reported by Dandekar and

Benfanti6. The magnitudes of shear strength at impact stresses from 7 to 19 GPa

decrease by 16% to 28% from their respective initial values due to the second

steps in the lateral stress wave profiles, see Table II. Bourne, et al.13

attributed

this reduction of shear strength to the subsequent propagation of a failure front,

which brings about a degradation of the shear strength. However, Murray and

Proud14

showed that the observation of two-step lateral stress profiles in ceramics

is dependent upon the geometry of the experimental configuration at a given

impact stress. Thus, the observed reduction in the shear strength of TiB2 could be

simply the manifestation of the geometrical configuration of the experiments in

Ref. 13. Further, the existence of failure front propagation in a solid under plane

shock wave compression is easily verified by the presence of recompression in the


longitudinal wave profile. Therefore, such experiments must be performed on

TiB2 to independently verify the propagation of a failure front in the material, and

to corroborate the suggested degradation of the shear strength.

REFERENCES1D. Yaziv, S. Bless and Z. Rosenberg, “Study of Spall and Recompaction

Using a Double-Impact Technique”, Journal of Applied Physics, 58 [9], 3415-

3418 (1985). 2D. P. Dandekar, P. J. Gaeta and Y. Horie, “Double Shock and Release

Experiments in PMMA and Z-cut Sapphire,” pp. 281-284 in Shock Waves in

Condensed Matter - 1987, Edited by S. C. Schmidt and N. C. Holms, North-

Holland Press, New York, 1988.3D. P. Dandekar, “Response of Ceramics Under Single and Repeated Plane

Shock Wave Loading - A Case Study of Titanium Diboride,” pp. 242-253 in

Proceedings of IUTAM Symposium on Impact Dynamics, Edited by Z. Zemin,

Peking University Press, Beijing, PRC, 1994.4D. P. Dandekar, “Response of Protective Ceramics Under Single and

Multiple Impacts,” pp. 133-141 in Wave Propagation and Emerging

Technologies, AMD- 188, Edited by V. K. Kinra, R. J. Clifton and G. C. Johnson,

ASME Press, New York, 1994. 5D. P. Dandekar, “Effect of Shock-Re-Shock on Spallation of Titanium

Diboride,” pp. 487-490 in Shock Compression of Condensed Matter - 1991,

Edited by S. C. Schmidt, R. D. Dick, J. W. Forbes and D. G. Tasker, North-

Holland Press, New York, 1992. 6D. P. Dandekar and D. C. Benfanti, “Strength of Titanium Diboride Under

Shock Wave Loading,” Journal of Applied Physics, 73 [2], 673-679 (1993).7J. Frankel, A. Abbate and D. P. Dandekar, “Pressure Dependence of the

Elastic Constants of Polycrystalline Titanium Diboride,” pp. 881-884 in Recent

Trends in High Pressure Research, Edited by A. K. Singh, Oxford Press, New

Delhi, 1992. 8W. H. Gust, A. C. Holt and E. B. Royce, “Dynamic Yield, Compressional

and Elastic Parameters for Several Lightweight Intermetallic Compounds,”

Journal of Applied Physics, 44 [2] 550-560 (1973).9S. P. Marsh, p. 354 in LASL Shock Hugoniot Data, Edited by S. P. Marsh,

University of California Press, Berkeley, CA, 1980.10

D. E. Grady, “Dynamic Material Properties of Armor Ceramics,” Sandia

National Laboratories Report, SAND 91-0147, Albuquerque, NM, 1991. 11

W-D. Winkler and A. J. Stilp, “Pressure Induced Macro- and Micro-

mechanical Phenomena in Planar Impacted TiB2,” pp. 555-558 in Shock

Compression of Condensed Matter - 1991, Edited by S. C. Schmidt, R. D. Dick, J.

W. Forbes and D. G. Tasker, North-Holland Press, New York, 1992.


12Z. Rosenberg, N. S. Brar and S. J. Bless, “Shear Strength of Titanium

Diboride Under Shock Loading Measured by Transverse Manganin Gages,” pp.

471-474 in Shock Compression of Condensed Matter - 1991, Edited by S. C.

Schmidt, R. D. Dick, J. W. Forbes and D. G. Tasker, North-Holland Press, New

York, 1992. 13

N. K. Bourne, G. T. Gray III and J. C. F. Millet, “On the Failure of Shocked

Titanium Diboride,” pp. 589-592 in Shock Compression of Condensed Matter -

1999, Edited by M. D. Furnish, L. C. Chhabildas and R. S. Hixson, American

Institute of Physics, New York, 2000. 14

N. H. Murray and W. G. Proud, “Measurement of Lateral Stress and Spall

Strength in Ceramics,” pp. 151-156 in Fundamental Issues and Applications of

Shock Wave and High Strain-rate Phenomena, Edited by K. P. Staudahammer, L.

E. Murr and M. A. Meyers, Elsevier Press, New York, 2001.


DYNAMIC INDENTATION DAMAGE OF CERAMICS

Do Kyung Kim, Chul-Seung Lee, and Young-Gu Kim

Dept. of Materials Science and Engineering

Korea Advanced Institute of Science and Technology

Taejon, Korea

Chang Wook Kim, and Soon Nam Chang

Agency for Defense Development,

Taejon, Korea

ABSTRACT

A modified Kolsky bar technique with a spherical indenter was applied to

evaluate the damage behavior of armor ceramics in dynamic indentation. Also,

a small explosive detonator was used for the dynamic indentation on ceramics.

In both experiments a bonded-interface specimen was useful to analyze the

subsurface damage after the concentrated dynamic loading on ceramics. A more

extensive quasi-plastic zone was observed in the dynamic indentation than in the

quasi-static loading. Microfracture behavior of damage zone in dynamic

indentation have almost the same features as those of quasi-statically damaged

ceramics.

INTRODUCTION

Ceramics have high hardness and elastic modulus, and these properties give

ceramics high wear and impact resistances. Some ceramic materials, such as

alumina, silicon carbide and boron carbide, are primary candidates for armor

applications.[1,2] However, the dynamic responses that relate to projectile

impact are not well understood, and it is the objective of the present study to

suggest one technique for characterizing the damage behavior during impact

loading, as a basis for identifying the material parameters that primarily influence

on dynamic impact. There were some reports on the crack evolution of brittle

ceramics during dynamic impact,[3-5] showing rate-dependent hardness. But few

studies have been reported on the damage evolution during dynamic impact

because of difficulties in recovering specimens after testing.

Indentation on the polished surface of a specimen with a spherical indenter,



Figure 1. Experimental set-up for the modified-Kolsky bar dynamic indentation

experiment.

which is generally called Hertzian indentation, have been extensively studied by

Lawn and colleagues to evaluate the properties of monolithic ceramics and

currently is being extended to layered structures.[6,7] In the course of analysis

of sphere-indentation, bonded-interface specimens could provide visualization

and quantitative analysis of damaged subsurface regions.

In this study, the modified Kolsky Bar technique with a spherical indenter and

a small explosive detonator, were used for dynamic indentation of typical armor

ceramics: alumina and silicon carbide. The subsurface damage zone during

dynamic indentation was characterized by using a bonded-interface specimen.

EXPERIMENTAL PROCEDURES

Specimen Preparation

Two armor ceramics, alumina(AD85, Coors Ceramics Co.) and silicon

carbide(hot-pressed SiC, Ceradyne Co.) were used for the experiment. To

reveal the subsurface damage, a special bonded-interface configuration was

used.[8] Polished surfaces of two half-specimens(6mm by 8mm by 35mm)

were glued face to face with a thin layer of adhesive under light clamping

pressure. Indentation was made with a tungsten carbide sphere across the


Figure 2. Experimental set-up for the dynamic indentation by using small

detonator on (a) alumina and (b) silicon carbide ceramics.

interface trace. The two halves of the indented specimens were then separated

by dissolving the glue in acetone, cleaned, gold-coated, and examined by a

reflection optical microscope with Normalski interference illumination.

Dynamic indentation with modified Kolsky bar

To get the indentation with higher strain rate, compressed-gas driven Kolsky

bar equipment was used. The spherical indenter, tungsten carbide of 3.98 mm in

radius, was mounted on one side of a slender bar(100 mm by 10 mm and

placed on the interface of the bonded specimen. The other side of the slender

bar was impacted by the sabot-guided striker bar with a predetermined velocity,

which has the same length as the impact bar. The striker bar was accelerated by

a 20 mm-bore compressed-air gun. The impact was controlled by the velocity

of the striker bar, and the velocity was in the range of 5 to 15 m/s in the

experiments. Figure 1 shows the macroscopic view of experimental set-up.

Dynamic indentation with explosive detonator

With the same size of a bonded-interface specimen, the small explosive

detonator with a diameter of 5 mm was glued on the interface trace. Light vice

pressure was applied to avoid the shattering of the specimen during impact. The

separated two side of the specimen were cleaned, gold-coated, and examined by

the reflection optical microscope. Observation of the subsurface damage zone in

higher magnification was conducted by SEM.


Figure 3. (a) Optical micrograph showing surface view(top) and side view

(bottom) of dynamically indented alumina. Indentation was performed in the

modified Kolsky bar set-up. The size of indenter was tungsten carbide of

3.98mm in radius and the striker velocity was 8 m/s. (b) Vickers hardness

variation as a function of radial distance from the center of top contact area.

Two data set at typical striker velocity are shown in (b). Vertical dashed lines

represent damage zone boundaries.

Hardness Measurement

Vickers hardness measurements have also been conducted on both the

damaged and undamaged area with load P =19.8 N. At least three indentations

were performed for each area. Hardness was determined as H = P/(2a2), where

P is applied load and a is impression half-diagonal. Application of Normarski

illumination enhanced the detection of surface impressions at each indent in the

optical microscope.

RESULTS AND DISCUSSIONS

Modified Kolsky bar experiment

Figure 3(a) shows optical micrographs of the top and side view of the

damaged alumina by the modified Kolsky bar. Impression and ring crack in top

surface and the extensive subsurface damage were observed. Figure 2(b) shows

Vickers hardness as a function of the radial distance from the center of contact

area at two typical striker velocities. It is clearly shown that the hardness of the

damage zone decreased as the distance decreased and the size of the damage zone

increased as the velocity increased.


Figure 4. Side views of detonator-indented (a) alumina and (b) silicon carbide

ceramics showing quasi-plastic damage zone. Optical microscopy with Normalski

illumination highlights the detail contrast of damage zone.

Observation of Damage Zone

Figure 4 shows the subsurface side view of explosive detonator-indented (a)

alumina and (b) silicon carbide. To reveal the microstructure precisely, a mosaic

photo was made from each higher magnified micrographs. The damage zone

shows roughly semi-circular shape originated from contact area with the end of

explosive detonator. Damage scale of silicon carbide was smaller than that of

alumina. Lateral cracks were hypothesized to be generated by interference of

reflected pulses.


Macroscopically, the "quasi-plastic" deformation zone developed in the strong

shear-compression region below the contact(the contact area is 5mm in diameter

in both specimen. The classical Hertzian cone crack developed from top surface

was hardly observed at the side view. The suppression of cone crack is

considered due to that the detonator develops only high shock pulse with

minimum mass of striking object.

Figure 5. SEM micrographs showing damage evolution from the detonator-

indented dynamic fracture. Microstructure of un-indented silicon carbide

with plasma etching reveals grain structure in (a). Higher magnification of

damage area in Figure 4(b) indicates that damage occurs at the grain boundary

of silicon carbide.

Higher magnification of damage zone in silicon carbide ceramics is shown in

figure 5(b) with comparison of the original un-damaged microstructure in (a).

Only microcracks in grain boundaries are clearly shown. In polycrystalline

ceramics, Lawn[9] has documented that the generic fracture mechanical model of

the microfracture evolution within the subsurface damage zone during a full

indentation loading and unloading. Microscopically, this indentation-induced

damage is associated with the activation of discrete "shear faults", from which

microcracks initiate. Interestingly the overall features of microfracture in

damage zone during dynamic indentation were almost same as that of quasi-

statically indented specimen.

Hardness of Damage Zone

Figure 6 shows the Vickers hardness values measured in damaged and un-

damaged zone of alumina and silicon carbide specimens that were indented by the

explosive detonator. Hardness of damage zone in alumina shows 43% of

original value, and that of silicon carbide shows 52% of original one. This

indicates that the damage severity of alumina is higher that that of silicon carbide.

It is considered that these hardness changes could be used as the indication of

damage severity during dynamic impact on ceramics.


Figure 6. Vickers hardness of damaged and un-damaged region in alumina and

silicon carbide ceramics. Explosive detonator was used for the dynamic

indentation. Seven data points of radial hardness in damaged region were used to

calculate means and standard deviation.

CONCLUSIONS

Dynamic indentations on ceramics were introduced by the modified Kolsky

bar technique and the detonation of a small detonator. Subsurface damage zones

in alumina and silicon carbide ceramics were examined by using a special

bonded-interface technique. Described two techniques were suggested as a simple

and powerful technique to evaluate the damage response during dynamic impact

on ceramics. It is considered that the size and the hardness of damage zone can be

use to quantify the resistance of damage evolution during dynamic impact on

ceramics.

REFERENCES1M.L. Wilkins, C.F. Cline, and C.A. Honodel "Light Armor," UCRL-71817,

July 1969. 2R.C. Laible, Ballistic Materials and Penetration Mechanics, Edited by R.C.

Lable, Chapter 6 and 10, Elsevier Sci. Pub. Co, New York, NY. 1980. 3S.M. Wiederhorn and B.R. Lawn, "Strength Degradation of Glass Resulting

from Impact with Spheres," J. Am. Ceram. Soc, 60 [9-10] 451-58 (1977). 4A.G. Evans and T.R. Wilshaw, “Dynamic Solid Particle Damage in Brittle

Materials: An Appraisal, “ J. Mater. Sci., 12 [1] 97-116 (1977). 5D.B. Marshall, A.G. Evans, and Z. Nisenholz, “Measurement of Dynamic

Hardness by Controlled Sharp-Projectile Impact” J. Am. Ceram. Soc, 66 [8] 580-


85 (1983). 6B.R. Lawn, "Indentation of Ceramics with Spheres: A Century after Hertz,"

Journal of the American Ceramic Society, 81 [8] 1977-94 (1998). 7K. S. Lee, S. K. Lee, B. R. Lawn, and D. K. Kim, Contact Damage and

Strength Degradation in Brittle/Quasi-Plastic Silicon Nitride Bilayers, Journal of

the American Ceramic Society, 81 [9] 2394-404 (1998). 8F. Guiberteau, N.P. Padture, H. Cai and B.R. Lawn, "Indentation Fatigue: A

Simple Cycle Hertzian Test for Measuring Damage Accumulation in

Polycrystalline Ceramics," Philos. Mag. A 69 [5] 1003-16 (1993). 9B.R. Lawn, N.P. Padture, F. Guiberteau, and H. Cai, "A Model for Microcrack

initiation and Propagation Beneath Hertzian Contacts in Polycrystalline

Ceramics," Acta Metall. Mater. 42 [5] 1683-93 (1994).


TAYLOR-IMPACT EXPERIMENTS FOR BRITTLE CERAMIC

MATERIALS

L. C. Chhabildas and

W. D. Reinhart

Sandia National Laboratories

P. O. Box 5800

Albuquerque, NM 87185

D. P. Dandekar


Aberdeen Proving Ground, MD 21005-

5066

ABSTRACT

A new time-resolved test methodology is described which allows access to

loading rates that lie between split Hopkinson bar and shock-loading techniques.

Gas-gun experiments combined with velocity interferometry techniques have

been used to experimentally determine the intermediate strain-rate loading

behavior of Coors AD995 alumina, Cercom silicon-carbide and Cercom boron-

carbide rods. Graded-density materials have been used as impactors; thereby

eliminating the tension states generated by the radial stress components during the

loading phase. Results of these experiments demonstrate that the time-dependent

stress pulse generated during impact allows an efficient transition from the initial

uniaxial-strain loading to a uniaxial-stress state as the stress pulse propagates

through the rod. This allows access to intermediate loading rates over 5 x 103/s to

106/s.

INTRODUCTION

A new test methodology is described which allows access to loading rates that

lie between split Hopkinson bar and shock-loading techniques. Traditional split

Hopkinson bar techniques allow measurements on the failure stress of the

material at loading rates up to 103/s, where the definition of the failure stress is the

yield strength of the material determined under uniaxial-stress loading. In

contrast, plate impact techniques introduce uniaxial strain states at loading rates of

105/s or higher. At these very high strain rates the failure stress is defined as the

stress at which the material transitions from elastic deformation to plastic

deformation normally defined as the Hugoniot elastic limit of the material. The

experimental test methodology in each case prevents access to loading rates of the

order of 104/s. It is the purpose of this paper to report a new test methodology



that allows access to loading rates of 104/s. This is referred to, in this study, as

intermediate strain rate loading.

Taylor impact experiments consist of impacting a cylindrical rod onto a rigid

barrier [1]. Post-test observations or high-speed photography is then utilized to

determine the plastically deformed contour of the cylinder from which the

mechanical property data such as the dynamic yield stress also referred to as the

failure stress in this paper can be determined [2,3] through the measurements of

deceleration of the cylindrical rod. Most of the previous studies have been limited

to ductile specimens due to the ease with which the specimens can be recovered

for post-mortem analysis. In this test method, a rigid anvil is made to impact a

stationary sleeved-rod and its acceleration profile is used to estimate the dynamic

yield stress. The use of graded-density materials as a rigid anvil provides the time-

dependent loading profile.

A single-stage compressed gas gun combined with velocity interferometric

techniques [4] was used to experimentally determine the loading behavior of

ceramic rods. The rod dimensions are chosen so that the ratio of the length to its

diameter is at least four. Graded-density materials [5,6] were used to impact both

bare and sleeved ceramic rods [7-10] while the velocity interferometer [11] was

used to monitor the axial velocity of the free-end of the rods. This test-

methodology is well suited for brittle-ceramics because the ceramic rods will

invariably fracture during the loading process.

Results of these experiments demonstrate unique features of this novel test

methodology:

(a) a time-dependent stress pulse generated resulting from graded-density impact

allows a smooth and efficient transition from the initial uniaxial strain loading to a

uniaxial stress state as the stress pulse propagates through the rod,

(b) sleeved-rods in combination with graded-density impactors eliminate the

tension generated in the specimen during the loading stage.

(c) intermediate loading rates of 104/s obtained in this configuration lie in a

region which is not achieved easily by either split Hopkinson bar or shock-loading

techniques, and

(d) the loading rates can be varied from 104/s to 10

6/s through a combination of

increased impact velocity and different graded-density impactor design.

In this paper, only the results of experiments conducted on boron carbide rods

are reported. Results of Coors AD995 alumina rod experiments and silicon

carbide are published elsewhere [7-10].


EXPERIMENTAL TECHNIQUE

These experiments were performed on a 64-mm diameter, smooth bore,

single-stage, compressed-gas gun that is capable of achieving a maximum

velocity of 1.2 km/s. Three electrically shorting pins were used to measure the

velocity of the projectile at impact. Four similar pins were mounted flush to the

impact plane and used to monitor the planarity of impact. Projectile velocity was

measured with an accuracy of about 0.5% and the deviation from planarity of

impact was about a milliradian. The graded-density impactor assembly is

fabricated by bonding a series of thin plates in order of increasing shock

impedance from the impact surface. The series of layered materials used in these

studies were TPX-plastic, aluminum, titanium, and 4340 steel. The thickness of

each layer is controlled to tailor the (time-dependent) input stress pulse into the

silicon carbide rod. The exact dimensions of each material assembly is given in

Table 1.

TABLE 1. Summary of impact experiments on sleeved silicon carbide Test

No.

Rod

Length/Diameter

(mm)/(mm)

Impactor

Materials

Impactor Thickness Impactor

Velocity

(km/s)

B4C-1 47.165/9.418 Steel/Ti/Al/TPX 12.73/0.224/0.244/0.244 0.309

B4C -2 47.061/9.416 Steel/Ti/Al/TPX 12.75/0.234/0.241/0.244 0.419

B4C-3 47.069/9.416 Steel/Ti/Al/TPX 12.52/0.234/0.249/0.251 0.600

B4C-4 47.089/9.416 Steel/Ti/Al/TPX 12.64/0.234/0.244/0.241 0.501

B4C-5 47.066/9.416 Steel/TPX 12.79/1.450 0.941

B4C-6 47.061/9.416 Steel/Ti/Al/TPX 12.71/0.328/0.243/0.509 0.945

B4C-7 47.628/9.418 Steel/Ti/Al/TPX 12.71/0.300/0.259/0.248 0.698

B4C-8 48.242/9.418 Steel/Ti/Al/TPX 12.66/0.300/0.260/0.268 0.799

This layered material assembly is used as facing on an aluminum projectile,

which is accelerated on a gas gun to velocities from 340 m/s to about 1000 m/s

prior to impact. This provides a time-dependent loading at the impact interface

from about 6 GPa to ~ 20 GPa, which is beyond the Hugoniot Elastic Limit for

the material [12-14]. The experimental target assemblies consisted of a sleeved

born carbide rod ~ 9.5 mm in diameter. The length of the rods in this study were

nominally 48 mm and 4340 steel was chosen for the close fitting sleeve material

to provide a good shock impedance to the boron-carbide sample. The outer

diameter of the sleeve was nominally 19 mm. The experimental configuration is

shown schematically in Fig. 1.


TPX

Aluminum Ring

4340 Steel

TitaniumAluminum

Tilt Pins

VelocityPins

Sleeve:4340Steel

Boron Carbide

TungstenFoil

VISAR

Figure 1. Experimental configuration of a layered impactor and a ceramic rod target

assembly.

The boron carbide used in this study is obtained from Cercom. The density of the

material used in this investigation was 2.510 g/cm3; the longitudinal and shear wave

speed was determined to be 14.01 km/s and 8.83 km/s, respectively. This yields

9.60 km/s, 13.51 km/s, and 0.170 for the bulk-wave velocity, bar-wave velocity, and

Poisson’s ratio, respectively. Specifically, this is the same batch of material used in

previous studies on boron carbide [15] at the Army Research Laboratory.

A 0.033 mm thick tungsten reflector glued onto the free surface of the rod was used

to obtain the axial particle velocity measurements using the velocity interferometer,

VISAR having a time resolution of ~ 1 ns. The loading strain rate is varied either by

varying the impact velocity and/or by varying the thickness of the layered impactors at

the same impact velocity.

RESULTS

Eight experiments were conducted with boron carbide rods 9.5 mm in diameter and

48 mm in length with 4340-steel sleeves. The impact velocity was varied as shown in

Table 1, causing the stress and the loading rate to vary at the impact interface. Figure 2

shows the results of the experiments, B4C-1, B4C-2, B4C-3, B4C-5 and B4C-8, while

Figure 3 shows the results of experiments B4C-3, B4C-4, B4C-6, and B4C-7. The two

experiments at impact velocities below 0.6 km/s (B4C-1 and B4C-2, B4C-3, and B4C-4)

introduce stress levels that are at or below its Hugoniot elastic limit. The experiments

above 0.7 km/s are those above its Hugoniot elastic limit. The experiment B4C-3 which is

B4C


at an impact velocity of 0.60 km/s exhibits a maximum free-surface velocity

measurement suggesting a stress at or approaching its Hugoniot Elastic limit. These

experiments are displayed in Figures 2 and 3, respectively.

0.941 km/s

0.419 km/s

0.309 km/s

0.799 km/s

0.60 km/s

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

3.3 3.5 3.7 3.9 4.1 4.3 4.5-0.1fr

ee-s

urf

ace

vel

oci

ty (

km

/s)

time (microsecond)

Figure 2. Free-surface velocity measurements at the free-end of the boron-carbide

rod resulting from graded-density impacts reported in Table 1. Experiments B4C-1,

B4C-2, B4C-3, B4C-5 and B4C-8 are shown in this figure.

Effect of Loading Rate

As indicated in Figures 2 and 3, the peak free-surface velocity measurements show an

increase with increased impact velocity up to an impact velocity of 0.6 km/s. At impact

velocities beyond 0.6 km/s the peak free-surface velocity measurement at the free-end of

the rod decreases with increasing impact velocity and is indicated in Figure 4. In Figure

5, the failure stress is plotted as a function of loading strain rate. A higher peak free-

surface velocity implies a higher yield stress also defined as the failure stress. This

provides experimental evidence for the dependence of failure stress upon loading rate.

There are reports of shear-strength loss in this material above its Hugoniot elastic limit in

shock experiments [12,13]. It should also be noted that the loading or the strain-rate also

increases with increasing impact velocity. Experiments B4C-5 and B4C-6 were performed

to investigate the effects of loading rate at the same loading stress. This was

accomplished by varying the dimensions of the layered impactor at the same impact

velocity (0.941 and 0.945 km/s, respectively).The free-surface velocity measurements are

comparable in these experiments, even though the loading rates differ by a factor of two –

the measurements suggesting that the failure strength has achieved its equilibrium value

of about 10.6 GPa .


0.945km/s

0.501km/s

0.60km/s

0.698km/s

3.3 3.5 3.7 3.9 4.1 4.3 4.5-0.1

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

DISCUSSIONS

Previous studies on impact of ceramic rods have concentrated upon using a single

density impactor [16-18] to evaluate the uniaxial compressive behavior of the ceramics.

However, due to the low spall strength of ceramics [12-17] the radial stress components

will fracture the material during the loading phase, even though the mean stress of the

material indicates compression [16-18]. The technique proposed herein (i.e., using

graded-density impactors to study the uniaxial compressive behavior of the rods)

time (microsecond)

free

-su

rfa

ce v

elo

city

(k

m/s

)P

eak

fre

e-su

rfa

ce v

elo

city

, k

m/s

Impact velocity, km/s

Figure 3. Free-surface velocity measurements at the free-end of the boron carbide rod

resulting from graded-density impacts reported in Table 1. Experiments B4C-3, B4C-

4, B4C-6 and B4C-7 are shown in this figure.

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Figure 4. Peak free-surface velocity measurements vs impact velocity for the profiles

show in figures 2 and 3.


Grady

Lankford

Brar

2

4

6

8

10

12

14

16

10-4

10-3

10-2

10-1

100

101

102

103

104

105

106

Presentfa

ilu

re s

tres

s,

GP

a

strain rate, /s

Figure 5. Variation of failure strength with strain rate for boron carbide. Also shown are

results of quasi-static, Hopkinson bar and shock experiments.

Figure 6. Failure stress of Coors AD995 Alumina as a function of strainrate.

circumvents this problem by reducing the magnitude of tension generated in the ceramic

during loading [7]. A sleeved rod prevents the formation of radial tension during the

loading process [7]. The current experiments address strain rate effects in B4C at strain

rates of 104/s to over 10

5/s. This strain-rate regime is difficult to access either by

Hopkinson bar techniques or shock loading techniques. The leading edge of the axial


compression wave traverses at an elastic-wave speed (14.0 km/s), followed by a second

compression wave traveling at a bar wave speed loading the material to a final stress at

strain-rates of ~ 104/s to 10

6/s. The first compression state l is calculated using 1=

( oc1 ufs)/2, where o is the initial density, c1 the elastic-wave speed, and ufs the

incremental free surface velocity measurement associated with the longitudinal elastic-

wave. The axial compression state a, and the loading strain rates d dt associated with

the bar wave are calculated using a= ( cb ufs)/2 and d dt = ufs/(2cbt), where cb is the

bar wave velocity, and ufs the corresponding free-surface velocity measurement, and t

the time duration for loading. Results of these experiments are shown in Figure 5, and are

compared to the low strain-rate Hopkinson bar experiments at strain-rates 10 3/s [19].

The low quasi-static strain-rate experiments yield a failure stress of ~ 5 GPa and shows

evidence of an increase to ~ 6 GPa at strain rates slightly above 103/s [19]. There is

considerable experimental scatter in the experiment suggesting the variability of the

material and perhaps the difficulty of conducting these experiments. As the strain rate

varies from 104/s to 6x10

5/s in these studies, the corresponding failure stress varies from

~ 6.5 GPa to a maximum of 11.1 GPa, before it approaches it’s equilibrium value of

10.6 GPa. This implies an Hugoniot Elastic Limit of at least 14 GPa for this material

before it sustains an equilibrium value of 13.3 GPa. The results clearly indicate the

dependence of the failure stress on the loading rates and also the loss in shear stress as

indicated in shock studies. The results lead credence to the hypothesis that damage

kinetics are rate-dependent, and ultimately, shock experiments yield higher estimates of

strength because rate-dependent kinetics prevent the nucleation and growth of

flaws/defects in materials during rapid loading.

The most significant result of this study is that the use of a graded-density impactor in

combination with sleeved rods allows accessibility to intermediate strain rates. Current

results for B4C, and previous studies on alumina and silicon carbide [4-5] both suggest

that the failure stress of ceramics is strain-rate-dependent. It should be noted that this

does not preclude the dependence of failure stress on mean pressure. It appears that

loading rates of a few times 104/s to 10

6/s can be achieved by optimizing the design of the

graded density layered materials, the diameter of the bar, and the impact velocity as

indicated in this investigation. One interesting study under consideration is to use the

graded-density materials as an impactor to perform isentropic loading experiments up to

its Hugoniot elastic limit. This will achieve lower loading rates than those obtained in

single shock experiments.


2

4

6

8

10

12

14

1.E-

05

1.E-

04

1.E-

03

1.E-

02

1.E-

01

1.E

+00

1.E

+01

1.E

+02

1.E

+03

1.E

+04

1.E

+05

1.E

+06Strain Rate (S-1)

Steel Sleeved Rod - GDI

Steel Sleeved Rod - 1/2 GDI

Unsleeved Rod - GDI

Ta Sleeved Rod - GDI

Lankford

Grady, Feng, et al

106

102

100

10-2

10-4

104

Figure 7. Failure stress of Cercom SiC as a function of loading rate.

ACKNOWLEDGEMENTS

Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed

Martin Company, for the United States Department of Energy under Contract DE-AC04-

94AL85000. We acknowledge the able technical assistance provided by H. Anderson and

J. Martinez.

REFERENCES1G. I. Taylor, J. Inst. of Civil Engng. 26, pp. 486

2G. I. Taylor, Proc. R. Soc. London, Ser. A 194, pp. 289.

3J. C. Foster, Jr., M. Gilmore, L. L. Wilson, in Shock Compression of Condensed

Matter-2001, edited by M. D. Furnish, N. Thadhani, and Y. Horie, New York, AIP Press,

2002 (to be published). 4L. C. Chhabildas, and R. A. Graham, in Techniques and Theory of Stress

Measurements for Shock Wave Applications, 83, Edited by R. B. Stout, et. al., AMD,

1987 pp. 1-18. 5L. C. Chhabildas, L. N. Kmetyk, W. D. Reinhart, and C. A. Hall, Int. J. Impact

Engng. 17 (1995) pp. 183-194.6L. C. Chhabildas, J. E. Dunn, W. D. Reinhart, and J. M. Miller, Int. J. Impact Engng.

14 (1993) pp. 121-132. 7L. C. Chhabildas, M. D. Furnish, D. E. Grady, J. Phys IV FRANCE 7 (1997),

Colloque C3, (1997), pp. C3-137. 8L. C. Chhabildas, M. D. Furnish, W. D. Reinhart, D. E. Grady in Shock Compression

of Condensed Matter-1997, edited by S. C. Schmidt, D. P. Dandekar, and J. W. Forbes,

New York, AIP Press, 1998, pp. 505-508.


9K. G. Holland, L. C. Chhabildas, W. D. Reinhart, M. D. Furnish, in Shock

Compression of Condensed Matter-1999, edited by M. D. Furnish, L. C. Chhabildas, and

R. S. Hixson, New York, AIP Press, 2000 pp. 585-588. 10

L. C. Chhabildas, W. D. Reinhart, Proceedings of the U.S. Army Symposium on

Solid Mechanics, edited by S. C. Chou and K. S. Iyer, (1999), pp 233-239. 11

L. M. Barker and R. E. Hollenbach, J. Appl. Phys. 43, (1972), pp. 4669-4675. 12

D. E. Grady, Dynamic Properties of Ceramic Materials, Sandia National

Laboratories Report, SAND94-3266, February 1995. 13

M. E. Kipp and D. E. Grady, Shock- Compression and Release in High-Strength

Ceramics, Sandia National Laboratories Report, SAND89-1461, February 1989. 14

N. S. Brar, Z. Rosenberg and S. J. Bless in Shock Compression of Condensed

Matter-1991, edited by Schmidt, S. C. and Dick, R. D., Forbes, J. W., Elsevier Science

Press, 1992, pp. 467-470. 15

D. P. Dandekar, Army Research Laboratory Report, ARL-TR-2456, April 2001. 16

A. Cosculluela, J. Cagnoux, F. Collombet, Journal de Physique IV, C3 (1991) pp.

109-116.17

N. S. Brar, and S. J. Bless in Shock-Wave and High-Strain-Rate Phenomena in

Materials, Edited by M. A. Meyers et. al., 1992, pp. 1041-1049. 18

J. L. Wise, D. E. Grady, High Pressure Science and Technology--1993, AIP

Conference Proceeding 309, Edited by S. C. Schmidt et. al., 1994, pp. 733-736. 19

J. Lankford, J. Amer. Ceramic Soc., 64, C33-C34 (1981)


Analytical and Computational Modeling

HISTORICAL PERSPECTIVE ON CERAMIC MATERIALS DAMAGE

MODELS

A.M. Rajendran


ARO, RTP, NC 27709-2211

ABSTRACT

Due to their high compressive strength, ceramic materials have been

frequently employed in armor systems for vehicle and soldier protection.

Ceramics are also candidate materials for ceramic engine components (turbine

blades, etc.) and aircraft engine containment systems due to their high temperature

properties. This paper presents a history of various approaches taken by

researchers to describe the brittle fracture of ceramics from the analytical

modeling of indentation processes to the recent high fidelity computational

modeling of projectile penetration processes in ceramic plates.

INTRODUCTION

Understanding and modeling of fracture in ceramic materials began with a

detailed study on quasi-static fractures induced by indentation loading. The

indentation modeling effort focused on the microcracking that occurs due to a

contact loading. During 1970's and 1980's, a large body of research work was

performed at various institutions and universities to characterize ceramic strength

through “hardness” measurements. Rajendran and Cook [1] presented a

comprehensive review of modeling of impact damage in ceramics. Lawn and

Wilshaw [2] reviewed the indentation fracture in detail.

Hockey [3] reported dislocation networks in alumina at local indentation sites.

Shockey et al [4] and Curran et al [5] address many of the deformation

mechanisms in confined ceramics under ballistic impact loading conditions.

These studies clearly established the presence of dislocations and twinning in the

brittle ceramics due to high pressures and high strain rate loading conditions.

Espinosa et al [6] reported evidence of inelastic deformation in compression due

to microcracking at triple junctions of the grain boundaries in recovered alumina

samples at velocities below the Hugoniot Elastic Limit (HEL). The



microplasticity in a brittle solid is often attributed to the dislocation motions in the

vicinity of microflaw tip regions. In brittle solids, large-scale grain distortions are

usually absent. Ewart and Dandekar [7] conducted a detailed microstructural

study of recovered titanium diboride specimens from low-velocity spall and

reshock experiments. Their study revealed that microcracks were the primary

form of irreversible damage in shock loaded ceramics. These microscopic

investigations indicated that the various forms of inelastic strain in the brittle

ceramics under shock and high strain rate of loading were caused by dislocations

and twins, microcracking, and pore collapse.

BACKGROUND

An accurate constitutive model must explicitly describe the various inelastic

processes through appropriate governing equations/laws. For instance, the axial

splitting and faulting in brittle material due to various levels of lateral

confinement were analytically modeled by Horii and Nemat-Nasser [8] based on a

wing-crack geometry under plane stress/plane strain loading conditions.

However, micromechanical modeling of the deformation processes under a three

dimensional stress/strain state is extremely difficult. Ravichandran and Subhash

[9] presented a micromechanical model for high strain rate behavior of ceramics

based on non-interacting microcracks that are uniformly distributed in the

material. This model was developed for biaxial compressive loading based on the

sliding crack (so-called wing crack). Hazell and Iremonger [10] reported a crack

softening damage model for ceramic impact and its application within a

hydrocode.

In summary, four distinct approaches can be identified as theoretical bases for

describing the inelastic deformation and fracture in ceramics. In the first

approach, the material is assumed to be elastic and stresses are calculated in the

finite element analysis based on Hooke’s law equations. Failure is predicted using

a generalized Griffith [11] fracture criterion. Mescall and Tracey [12] used the

Griffith criterion to model the fracture response of a ceramic armor in their HEMP

simulations. In the second approach, a numerical procedure is implemented in a

finite element/difference code to relax stresses to zero when some state variable

reaches a critical value. The ceramic behavior can be assumed to be either elastic

or elastic-plastic. Wilkins [13] implemented a fracture algorithm in the

Lagrangian finite-difference wavecode HEMP to examine a 0.30 CAL AP

projectile penetration into a thin ceramic plate backed by a metal substrate.

Recently, Anderson and Walker [14] adopted Wilkins' ceramic model to examine

ceramic dwell and defeat of the AP projectile.

In the third approach, the material is assumed to behave as an elastic-plastic

solid. This approach ignores the details of crack growth and concentrates on


describing the effects of localized fracture on stress wave propagation. The

stiffness of the ceramic is not degraded due to damage, but the strength is

degraded due to plastic deformation induced damage. The concepts and equations

are the same as those derived for metal plasticity. Johnson and Holmquist [15]

modeled the strength of the ceramic as a function of pressure and strain rate

through a two-surface approach. Basically, there are two surfaces: one

corresponds to D = 0, and the other to D = 1. Steinberg [16] proposed a ceramic

damage model that is very similar to his metal fracture model. Rajendran and

Kroupa [17] modified the constitutive equations based on fragmentation [18] to

describe the shock response of silicon carbide. Recently Simha [19] proposed a

similar model.

In the fourth approach, fracture mechanics based microphysical theories and

models are employed to describe the deformation due to the compressive failure

processes. The basic idea of a microphysical model is to describe the apparent

inelastic behavior while keeping track of the microstructural evolutions under a

given set of loading conditions. For ceramics, the evolution laws for

microcracking must incorporate the fracture mechanics theories that describe the

conditions for crack growth. A statistical description of number of cracks per unit

volume as a function of position, crack size, and orientation, is an example of a

microphysical approach. The Hooke’s law based elasticity equations are

combined with an effective moduli description that relates the microstructure to

the macroscopic material properties. The models by Rajendran and Grove

[20,21], Addessio and Johnson [22], and Espinosa [23] follow this approach.

Until 1988, there were hardly any material models in hydrocodes (shock-wave

propagation based finite element/difference numerical codes) that could describe

the inelastic behaviors of brittle ceramics under shock and high strain rate loading

conditions. During the past decade, several new ceramic damage models (Partom

[24], Sadyrin, Ruzanov, and Podgornova [25], and Malaise, Tranchet, and

Collombet [26]) have been reported for hydrocode applications, each based on

one of these four approaches.

BRIEF SUMMARY OF A FEW CERAMIC MODELS

This section summarizes a few ceramic models that have been implemented

into various hydrocodes. Researchers around the world have proposed several

other models; no attempt is made to include them all in this brief summary.

Wilkins' Computational Scheme [13]

In a simplistic approach, Wilkins employed a two dimensional hydrocode

(HEMP) in a computational analysis of the impact and penetration of thin

laminate armor. He used the following simplified criteria: (1) fracture initiates on


a surface, (2) a maximum principal stress greater than 0.3 GPa in tension causes

fracture, (3) there is a time delay for the complete fracture of a zone, (4) a

fractured zone becomes a source for the fracture of a neighboring zone, and (5)

fracture occurs only within a range of distance equal to or less than the time step

times the crack velocity in ceramic.

Figure 1. Effect of fracture strength on time for fracture (left column: a fracture

strength of 3 Kbar; right column is for 8 Kbar). Time is in microseconds.


Using this numerical scheme, Wilkins modeled the evolution of the fracture

conoid in thin ceramic targets. The time delay for complete fracture is related to

the time for a crack to propagate across a computational cell. The crack speed is

assumed to be a fraction of the elastic shear wave speed. A value of 0.5 was

employed for the fraction. This implementation prevents cracks propagating from

cell to cell at an unrealistic speeds. He could successfully reproduce some of the

observed fracture patterns in the ceramic plate due a bullet penetration. Figure 1

illustrates the effects of fracture strength on time for fracture. Recently, Anderson

and Walker [14] successfully adapted Wilkins approach to model the dwell and

defeat of a 0.30-CAL AP projectile. This approach is limited to modeling thin

ceramic plates.

Liaw, Kobayashi, and Emery Computational Model [27]

A mechanically consistent model of impact damage based on elastic fractures

due to both tensile and shear loading is assumed in the simulation of dynamic

indentation for a spherical projectile on a structural ceramic. The projectile is not

explicitly modeled in the finite element modeling; instead, a transient contact load

is prescribed at the impact site. The impact is assumed to be elastic and the

surface loading conditions are derived from analytical solutions. The

implementation of their fracture algorithm includes the following steps: 1) a crack

will form at a material point perpendicular to the direction of the maximum

principal stress when this stress exceeds the ceramic’s tensile strength, 2) a set of

orthogonal cracks (parallel to the maximum shear direction) is assumed to form

when the maximum shear stress exceeds the shear strength of the ceramic, 3)

cracks are allowed to carry subsequent compressive loads according to Coulomb’s

law of dry friction, and 4) an element's stiffness across an open crack vanishes and

Figure 2. Damage patterns in a ceramic plate impacted by a hard (elastic) sphere.


returns to its initial stiffness when the crack closes. Using this procedure, they

mapped the observed crack patterns (median crack, radial cracks, cone crack, and

lateral cracks regions) in the indented samples with reasonable accuracy as shown

in Figure 2. No attempt was made to employ this modeling approach to describe

shock wave profiles.

Modified TCK Model [17,18]

Rajendran and Kroupa [17] presented a modified version of the Taylor, Chen,

and Kuszmaul (TCK) [18] model that was developed to describe the brittle

behavior of oilshale under impact loading. In the modified version, the ceramic is

assumed to flow plastically under compression and no damage is allowed under

compression. In tension, the ceramic behaves in a brittle manner without any

plastic flow. A tensile damage parameter is defined using an expression that

combines the expressions derived by Kipp and Grady for fragmentation [28] and

Budiansky and O’Connell for a cracked solid [29]. The salient equations are

summarized as follows:

Compression: Y (1))D1()lnB1(Ys

Tension: ij (2)ijijkk e)D1(G2e)D1(K3

Damage is described by:

32

IC3dd

2

c

K20

2

1a;aNC;C

21

1

9

16D (3)

Ys is the static compressive strength, B is the strain rate sensitivity parameter, K

and G are bulk and shear moduli respectively, eij are deviatoric strains, ij are total

stresses, Cd is crack density, KIC is static fracture toughness, is degraded

Poisson's ratio, is a geometric factor, a is the crack (fragment) size, c is the

sound speed, N is the number of flaws, is material density, and is the strain

rate. This model was implemented in the EPIC code [30] and an experimentally

measured shock profile for silicon carbide was successfully reproduced.

Rajendran-Grove Model [20,21]

In this model, the total strain is decomposed into elastic ( ) and plastic ( )

strains. The elastic strain consists of the elastic strain of the intact matrix material

and the strain due to crack opening/sliding. Plastic flow is assumed to occur in

eij

pij


the ceramic only under compressive loading when the applied pressure exceeds

the pressure at the Hugoniot elastic limit (HEL). Pore collapse during shock

loading is modeled using a pressure dependent yield function and the strains due

to pore collapse are assumed to be viscoplastic. The constitutive relationship for

the cracked material is given by:

(4)eklijklij M

The components of the stiffness tensor M are described by Rajendran [20]. The

microcrack damage is measured in terms of a dimensionless microcrack density .

The maximum microcrack size a is treated as an internal state variable.

Microcracks are assumed to extend when the stress state satisfies a generalized

Griffith criterion. This criterion requires the fracture toughness KIC as well as a

dynamic frictional coefficient as model parameters. The damage evolution law

is derived from a fracture mechanics based relationship for a single crack

propagation under dynamic loading conditions:

2n

I

crR1

G

G1Cna (5)

where CR is the Rayleigh wave speed, Gcr is the critical strain energy release rate

for microcrack growth, and GI is the applied strain energy release rate. The model

constants n1 and n2 are used to limit the microcrack growth rate. Under tension,

these two constants are assumed to be equal to one. The ceramic is assumed to

pulverize under compression when reaches a critical value of 0.75. The ceramic

model has six parameters to describe microcracking of the intact ceramic. This

model has been implemented in the EPIC code [30] and simulations of several

shock and impact configurations have been successfully performed.

Espinosa’s Multi-plane Model [23]

This model assumes that microcracking can occur on a discrete number of

orientations. Espinosa et al [23] selected nine orientations at intervals of 450

along three mutually perpendicular planes. The inelastic strain is entirely due to

(penny-shaped) microcrack opening/sliding of the cracks oriented normal to those

nine directions. The average inelastic strains are given by,

9

1k

)k(j

)k(i

)k(j

)k(i

)k()k(cij bnnb

2

1SN (6)


The superscript k represents the orientation, N is the number of flaws per unit

volume, S denotes the surface of the microcrack, n is the corresponding unit

normal, and b is the average displacement jump vector across the surface S. The

b have been analytically derived for normal tractions under both tension and

compression. The corresponding expressions are:

)k(i

)k(l

)k(jjl

)k(jij

)k(2

)k(i nnnn2a

)2(E3

)1(16b (7)

and)k(

i)k(

2)k(

i fa)2(E3

)1(32b , (8)

where a(k)

is the crack radius of the penny-shaped microcracks on orientation k

and f (k)

is the effective shear traction vector on orientation k. The microcrack

growth law is very similar to the one that Rajendran and Grove [20,21] employed;

the multi-plane model uses the stress intensity factor instead of the strain energy

release rate. There are two crack growth constants, n1 and n2; these two constants

can take on different values for tension and compression.

Johnson-Holmquist Model [15]

There are two versions of this model: JH1 and JH2. The differences between

these two versions are very subtle. The JH model is a phenomenological model

based on an elastic-viscoplastic approach. The strength of the ceramic is assumed

to vary with pressure, strain rate and tensile strength. Basically, there are two

surfaces; one corresponds to D = 0, and the other to D = 1. Once damage initiates,

the flow surface reduces to an intermediate state, and at D = 1 the strength lies on

the second surface. As in the Johnson-Cook fracture model for metals, the

damage (D) increases with effective plastic strain. This model has been discussed

in detail elsewhere in this volume. Recently, several impact and penetration

configurations have been successfully modeled using the JH model.

Steinberg Model [16]

Steinberg assumed that all thermomechanical behaviors could be represented

through certain macroscopic variables, such as strain, strain rate, temperature, and

pressure. He adopted the salient features of his metal model for under high strain

rate and shock loading applications. The ceramic model assumes that both the

yield strength and shear modulus vary with respect to temperature and pressure.

However, the yield strength (Y) varies also with the strain rate. The governing

equations are:


dT

dG

G

1Band

dP

dG

G

1A;)T(B

PA1G)T,P(

oo3/1oG , (9)

where is the compressive strain, is effective strain rate, and G, P, and T are

shear modulus, pressure, and temperature, respectively. Go is the shear modulus

of the ceramic before shock loading. The expression for the yield strength is:

n2ICoo

oA

n KC3D;G

GYDY , (10)

where o is the initial density of the ceramic, Co is the longitudinal wave speed,

KIC is the fracture toughness, n is a model parameter, and, according to Steinberg,

YA is a material constant which can be sample dependent, as it is a function of

purity, grain size, previous mechanical history, etc. The strain is simply

decomposed into elastic-viscoplastic, and the stress calculation in the

computational implementation follows conventional plasticity theories. Basically,

there are five model parameters: Go, YA, A, B, and D (or n). Steinberg also

employed a traditional void growth (spall) model to describe tensile cracking in

ceramics; the spall model requires three additional constants.

Addessio - Johnson Ceramic Model [22]

In this model, the inelastic strains in ceramics due to penny-shaped microcrack

growth under tension and compression are determined by integrating the

individual crack strains over a material volume, as well as all crack sizes and

orientations. By invoking several assumptions regarding the nature of crack size

and orientation, it is then possible to obtain simplistic expressions for the inelastic

strains. Addessio and Johnson [22] derived the following relationships for the

deviatoric parts of the inelastic strain components ( e ):cij

G

N

2

1

15

64where

)tension(6;)ncompressio()5(2,Sce

o

eeij

3ecij

(11)

In this equation, Sij are the stress deviators, c is the sound wave speed, is

Poisson’s ratio, and No is the number of flaws. In the deviatoric elastic stress-

strain relationship, the shear modulus is degraded through the following

expression:


3e cG1

GG , (12)

where is a material model parameter that was arbitrarily introduced into the

above expression. Addessio and Johnson obtained crack growth criteria by

considering an energy balance on isolated cracks. The crack growth rate is

described by:

smax dtanhcac , (13)

where is assumed to be the shear wave speed, “a” is a factor that will

reduce the crack speed, and d

maxc

s is a measure of the distance the state of stress

exceeds the damage surface. The main model parameters are: c , No , o , , ,

and a. In addition to these parameters, the model also requires one or two other

parameters.

Riuo-Cottenot-Boussuge Tensile Damage model [31]

This model assumes that the ceramic deforms elastically below the Huguenot

Elastic Limit. When the shock amplitudes exceed the HEL, the ceramic deforms

plastically as metals. In the model, damage initiation occurs when the principal

stress exceeds a threshold stress th. When all the principal stresses are tensile

and exceed this initial threshold value, damage initiates in the planes that are

perpendicular to the principal stress directions. Therefore, damage can initiate

and propagate when . Note that the initial value ofcri cr is th. This

threshold stress is reduced according to: . The definition of dthicr d1 i is

the ratio of total crack surface of penny-shaped cracks over the total solid surface.

The corresponding expression for di is:

2i

si2

aNd . (14)

Ns and the initial microcrack size ai are model parameters. Since damage is

assumed to increase from 0 to 1, this assumption puts a bound on the maximum

crack size. In the model, the crack extends at a constant speed, Vf . This

extension is possible only when the initiation criterion as well as the time rate of

change of the principal stress is positive. The stresses are assumed to degrade

according to (1 - di ) e, where e is the principal elastic stress. A Mohr-Columb


law is employed to describe the strength of the pulverized ceramic.

This model was used to simulate a three dimensional configuration in which a

steel cylinder (20 mm long, 11 mm in diameter) impacts a silicon carbide beam

(20 mm thick). In the experiments, stress measurements were made by

embedding a stress gauge between the back surface of the beam and a thin steel

plate, and photographs of the fracture patterns in the beam were obtained from a

high-speed camera. The authors validated their model through three-dimensional

simulations of this test configuration and compared the measured stress histories

with the computational results. The simulated crack patterns qualitatively agreed

with the observed crack patterns. The model also predicted the stress gauge

measurements well; however, an elastic analysis without damage also matched the

experiments reasonably well.

Simha’s Phenomenological Model [19]

Simha assumes that the ceramic fails at the Hugoniot Elastic Limit (HEL).

Microcracking due to sliding is assumed to be the dominant inelastic mechanism.

The Mohr-Coulomb law describes the strength of the failed ceramic. The

effective strength (Y) of the inelastic state is defined by,

)P(2

e3)P(YqsY , (15)

where P is the pressure, e is the effective deviatoric strain rate, is a parameter

that controls the contribution of the rate dependent term to the strength of the

failed material, and Yqs is defined as the rate independent part of the strength (like

a reference strength). Yqs is constant before the ceramic fails, and follows the

Mohr-Coulomb law (up to Ycap) after failure. Simha successfully used this model

to describe the shock response and penetration resistance of aluminum oxide.

VALIDATION AND VERIFICATION

The most frequently reported experimental technique to calibrate the high

strain rate and shock behaviors of ceramics is based on the plate impact test

configuration. In this configuration, a flyer plate is impacted against a target of

the same or different material at high velocity. The diagnostic measurements

include the use of either a peizo-resistive stress gauge or a velocity interferometry

system (VISAR). The measured wave profiles are often used in the calibration of

ceramic model constants. It is not possible to determine the exact nature of the

deformation processes from the measured profiles. However, microscopic studies

on the scientifically recovered targets often reveal many of the


deformation/damage processes.

Grady and Wise [32] obtained particle velocity wave profiles (VISAR Data)

for various ceramic materials, including silicon carbide (SiC), boron carbide

(B4C), and titanium diboride (TiB2). The impact velocities in those experiments

were about 1500 and 2500 m/s. Most of the models reproduced the VISAR data

very well. For example, Figure 3 shows a comparison between the plate impact

data and the computed wave profiles using the model developed by Rajendran and

Grove [20,21] for four different ceramics.

Several other impact experimental configurations are available for model

validation. Attempts have been made to match the measured profiles from a wide

AD995 – ALUMINUM OXIDE (1.943 KM/SEC)

Time ( s)

0.5 1.0 1.5 2.0 2.5 3.0

Velo

cit

y (

km

/s)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

ExperimentRG Model

S

Figure 3. Comparisons between simulations and data for four different ceramic materials.


variety of experiments with one set of model constants. Holmquist, Templeton,

and Bishnoi [33], Simha [19], and Grove and Rajendran [34] have shown

successful comparisons between their model predictions and the data. At least

one or two parameters have to be calibrated to match the measured depth of

penetration (DOP) data from the projectile penetration experiments. Since the

governing equations in the model formulation are not precisely derived to describe

the actual micro/macro damage processes in a ceramic under a wide range of

stress / strain / strain rate, it is not surprising to discover that the model parameters

calibrated solely from the shock wave experimental data are not suitable for

predicting the DOP. For completion, brief descriptions of several other

experimental configurations are discussed in the following sub-sections.

Rod-on-Rod Impact

A short ceramic rod is made to impact a long stationary ceramic rod. In this

uniaxial stress configuration, a stress gauge is typically embedded in the target rod

to record the stress history. Fracture initiates at the impact end, with several

splitting type macrocracks forming and propagating toward the gauge location.

The measured peak stress from this experiment can be used in a qualitative sense

to validate the model constants under a uniaxial stress state.

Graded-Density Plate-on-Rod Impact

Recently, Chhabildas et al [35] reported an experimental configuration in

which a ceramic rod (L/D 4, sleeved or unsleeved) was impacted by a graded-

density flyer plate consisting of extremely thin (0.1-cm thick) layers of titanium,

aluminum, and TPX bonded to a 1.9-cm thick steel plate. A VISAR was used to

record the axial particle velocity of the free end of the target rod. This test

configuration generates a time-dependent stress pulse that smoothly and

efficiently transits from the initial uniaxial strain loading to a uniaxial stress state.

Also, the intermediate loading rates obtained in this configuration are not easily

achieved by either split Hopkinson bar or conventional shock-loading techniques.

Depth of Penetration Experiment

In the projectile penetration experiment, a tungsten long rod projectile is

launched at a nominal velocity of 1.5 km/s onto a thick square ceramic tile that is

laterally confined by a steel frame; the target assembly (tile and frame) is

mechanically clamped to a thick steel backup block. The depth of penetration

(DOP) of the projectile in the backup steel plate is measured and used as a

parameter to compare in the validation and verification of a model. High speed

photographs and X-ray radiographs are also often obtained as part of the

diagnostic measurements.


SUMMARY

During the 1970's, the computational analysis of a projectile (metallic spheres

and cylinders) impacting a brittle ceramic plate was mainly performed to gain a

fundamental understanding of complex crack patterns developed due to the

impact. A combined indentation-based experimental and computational analysis

approach was employed in the evaluation of hardness and compressive strength of

ceramics. The response of the ceramic was assumed to be elastic in the

indentation analysis. During the past decade, researchers realized an urgent need

for a fully three-dimensional constitutive description of ceramic materials to

perform realistic hydrocode analyses suitable for impact-resistance applications.

Constitutive model formulations have mainly focused on incorporating the effects

of pressure, defects (pores and microcracks), and strain rate on strength and

stiffness of the ceramic. A few models have included the effects of flaw

orientation and/or microplasticity (dislocations, twins, etc.) on the degradation of

strength and stiffness. Those model parameters that cannot be directly measured

from experiments are estimated (calibrated) based on their ability to reproduce or

match the measured wave profiles. Most models use the Mohr-Coulomb law to

describe the compressive/shear loading response of the comminuted ceramic.

Generally, one or two parameters are needed for this purpose. Currently, there is

no physics-based model to accurately describe the comminuted ceramic response.

Curran et al [5] reported a micromechanical model based on the non-elastic

sliding and ride-up of fragments of comminuted particles. Their simulation of a

projectile penetration into a confined ceramic showed that the DOP is controlled

by 1) friction between the comminuted granules, 2) compressive strength of the

intact ceramic, and 3) compaction strength of the comminuted ceramic.

Current models are incapable of describing the effects of grain size and grain

boundary properties on the impact and shock resistance of ceramics. Recently,

Zavattieri and Espinosa [36] presented a grain level analysis of ceramic

microstructures subjected to impact loading. Through a two dimensional

stochastic finite element analysis, they explicitly modeled the details of grain

morphology and its effects on crack nucleation and propagation at grain

boundaries. Though we have made significant progress in modeling the ceramic

damage under shock and penetration loading conditions, there are still issues that

need attention and additional research.

REFERENCES 1A.M. Rajendran and W.H. Cook, “A Comprehensive Review of Modeling of

Impact Damage in Ceramics,” Technical Report AFATL-TR-88-143, Air Force

Armament Laboratory, Eglin Air Force Base, Florida 32542-5434 (1988).


2B.R. Lawn and T.R. Wilshaw, “Review: Indentation Fracture: Principles and

Applications,” Journal of Materials Science, 10 1049-1081 (1975). 3B. Hockey, “Plastic Deformation of Aluminum Oxide by Indentation and

Abrasion,” Journal of the American Ceramic Society, 54 [5] 223-31 (1971). 4D.A. Shockey, A.H. Marchand, S.R. Skaggs, G.E. Cort, M.W. Burkett, and

R. Parker, "Failure Phenomenology of Confined Ceramics Targets and Impacting

Rods," Int. J. Impact Engng., 9, 263-275 (1990). 5D.R. Curran, L. Seaman, T. Cooper, and D.A. Shockey, "Micromechanical

Model for Comminution and Granular Flow of Brittle Material Under High Strain

Rate Application to Penetration of Ceramic Targets,” Int. J. Imp Eng. 13, 53-83

(1993).6H.D. Espinosa, G. Raiser, R.J. Clifton, and M. Ortiz, “Experimental

Observations and Numerical Modeling of Inelasticity in Dynamically Loaded

Ceramics,” J. of Hard Materials 3, 285-313 (1992). 7L. Ewart and D.P. Dandekar, Relationship Between the Shock Response and

Microstructural Features of Titanium Diboride (TiB2),” in High-Pressure Science

and Technology - 1993, Part 2, AIP Press, 1201-1204 (1994). 8H. Horii and S. Nemat-Nasser, “Brittle Failure in Compression: Splitting,

Faulting, and Ductile-Brittle Transition,” Philosophical Transactions, Royal

Society, A319 337-374 (1986). 9G. Ravichandran and G. Subhash, “A Micromechanical Model for High

Strain Behavior of Ceramics,” Int. Journal of Solids and Structures, 32 [17] 2627-

2646 (1995). 10

P.J. Hazell and M.J. Iremonger, “Crack Softening Damage Model for

Ceramic Impact and Its Application Within a Hydrocode,” J. Appl. Physics, 82 (3)

1088-1092 (1997). 11

A.A. Griffith, “The Phenomena of Rupture and Flow in Solids,” Phil. Trans.

of Royal Soc. of London, 221, 163-198 (1920). 12

J.F. Mescall and C.A. Tracey, “Improved Modeling of Fracture in Ceramic

Armors,” an internal technical report of Army Materials Technology Laboratory,

Watertown, MA, (1986). 13

M.L. Wilkins, “Third Progress Report of Light Armor Program,” UCRL-

50460, Lawrence Livermore Laboratory, Livermore, CA (1968). 14

C.E. Anderson and J.D. Walker, “Ceramic Dwell and Defeat of the 0.30-

CAL AP Projectile,” Proceedings of the 15th

US Army Solid Mechanics

Symposium, Eds., K. Iyer and S.C. Chou, 17-28 (1999). 15

G. Johnson and T.J. Holmquist, “An Improved Computational Constitutive

Model for Brittle Materials,” in High-Pressure Science and Technology - 1993,

Part 2, AIP Press, 981-984 (1994).


16D.J. Steinberg, “Computer Studies of the Dynamic Strength of Ceramics,”

Shock Compression of Condensed Matter", Elsevier Science, 447-450 (1992). 17

A.M. Rajendran and J.L. Kroupa, “Impact Damage Model for Ceramic

Materials,” J. of Appl. Phys., 8 (66), 3560-3565 (1989). 18

L.M. Taylor, E-P. Chen, and J.S. Kuszmaul, “Microcrack-Induced Damage

Accumulation in Brittle Rock Under Dynamic Loading,” Computer Methods in

Applied Mechanics and Engineering, 55, 301-320 (1986). 19

H.M. Simha, “High Rate Loading of a High Purity Ceramic – One

Dimensional Stress Experiments and Constitutive Modeling,” Ph.D Thesis,

University of Texas at Austin, TX, 1998. 20

A.M. Rajendran, “Modeling the Impact Behavior of AD85 Ceramic Under

Multiaxial Loading,” Int. J. Impact Engng., 15 (6) 749-768 (1994). 21

A.M. Rajendran and D.J. Grove, “Modeling the Shock Response of Silicon

Carbide, Boron Carbide, and Titanium Diboride,” Int. J. Impact Engng., 18 (6)

611-631 (1996). 22

F.L. Addessio and J.N. Johnson, “A Constitutive Model for the Dynamic

Response of Brittle Materials,” J. Applied Physics, 67, (7) 3275-3286 (1990). 23

H.D. Espinosa, P.D. Zavattieri, and S.K. Dwivedi, “A Finite Deformation

Continuum/Discrete Model for the Description of Fragmentation and Damage in

Brittle Materials,” J. Mech. and Phys Solids, 46 [10] 1909-1942 (1998). 24

Partom, “Calibrating a Strength Model for AD995 Alumina from Plate

Impact VISAR Profiles,” J. de Physique IV c8-495 (1994). 25

A. Sadyrin, A. Ruzanov, and T. Podgornova, “Modeling Impact Loading and

Failure of Brittle Solids; Constitutive Damage Model for AD995,” Research

Institute of Mechanics, State University of Nizhny Novgorod, Russia (1998). 26

F. Malaise, J-Y. Tranchet, and F. Collombet, “Effects of Dynamic

Confinement on the Penetration Resistance of Ceramics Against Long Rods,”

Shock Compression of Condensed Matter-1999, Elsevier Science Publishers B.V.

1121-1124 (2000). 27

B.M. Liaw, A.S. Kobayashi, and A.F. Emery, “Theoretical Model of Impact

Damage in Structural Ceramics,” Journal of the American Ceramic Society, 67 [8]

544-549 (1984). 28

M.E. Kipp and D.E. Grady, “Shock Compression and Release in High

Strength Ceramics,” in "Shock Compression of Condensed Matter-1989, Elsevier

Science Publishers B.V. 377-380 (1990). 29

B. Budianski and R.J. O’Connell, “Elastic Moduli of a Cracked Solid,” Int.

J. Solids Structures 12, 81-97 (1976). 30

G.R. Johnson, R.A. Stryk, T.J. Holmquist, and S.R. Beissel, Numerical

Algorithms in a Lagrangian Hydrocode, Report No. WL-TR-1997-7039, Wright

Laboratory, Eglin AFB, FL (1997).


31P. Riou, C.E. Cottenot, and M. Boussauge, “Steel Rod Impact on Silicon

Carbide Beams: Experiments and Anisotropic Model of Damage,” in “Structures

under Shock and Impact,” eds., N. Jones, C.A. Brebbia, and A.J. Watson,

Computational Mechanics Publications, 533-541 (1996). 32

D.E. Grady and J.L. Wise, “Dynamic Properties of Ceramic Materials,”

Sandia Report, SAND93-0610, September (1993). 33

T.J. Holmquist, D. Templeton, and K. Bishnoi, “Constitutive Modeling of

Aluminum Nitride for Large Strain, High Strain Rate, and High Pressure

Applications,” Int J Impact Engng, 25 211-231 (2001). 34

D.J. Grove and A.M. Rajendran, “Overview of the Rajendran-Grove

Ceramic Failure Model,” a Chapter in PAC RIM IV proceedings, 2002. 35

L.C. Chhabildas, M.D. Furnish, W.D. Reinhart, and D.E. Grady, “Impact of

AD995 Alumina Rods,” Proceedings of the Shock Compression of Condensed

Matter – 1997, Eds. Schmidt, S.C., Dandekar, D.P., and Forbes, J.W., AIP, 505-

508 (1997). 36

P.D. Zavattieri and H.D. Espinosa, “Grain Level Analysis of Ceramic

Microstructures Subjected to Impact Loading,” a Chapter in PAC RIM IV

proceedings, 2002.


A COMPARISON OF CERAMIC MATERIAL MODELS

Douglas W. Templeton

U. S. Army Tank Automotive Research, Development, and Engineering Center

Warren, MI 48397-5000

Timothy J. Holmquist

Network Computing Services Inc./Army HPC Research Center

Minneapolis, MN 55415

Hubert W. Meyer, Jr., David J. Grove, and Brian Leavy



ABSTRACT

This paper presents results of a study investigating two different ceramic

models using two different computer codes and comparing their performance for

terminal ballistic problems. Computations were performed using the Johnson-

Holmquist (JH-1) and the Rajendran-Grove (RG) constitutive models for brittle

materials, for penetration problems into ceramics as implemented in both the

Eulerian CTH and the Lagrangian EPIC shock physics codes. The results of the

computations are compared to each other and experimental data, and an

assessment is made of the models’ utility for typical armor design problems.

INTRODUCTION

The spectrum of ballistic threats that may be encountered by ground vehicles

runs from small arms and low-velocity shrapnel from a variety of sources to high-

energy kinetic penetrators. Unfortunately, designing, integrating and fielding an

armor configuration for the highest order threat, to be used as the vehicle armor, is

clearly impractical from both weight and cost standpoints. As the US Army

acquires an expanded role in areas other than direct combat (such as Somalia or

Bosnia), armor packages more closely configured to match, rather than grossly

overmatch, the expected threat will be required. In order to meet this requirement

in a timely and affordable fashion, increased reliance is being placed on

simulation and modeling to replace the expensive process of build, shoot, build.



Potential armor configurations can be modeled on the computer and tested against

a large number of threats via computer simulations. In this way unacceptable or

marginally performing designs can be eliminated before committing to fabrication

and ballistic range tests, saving both time and money.

The US Army has made a major commitment to a dramatic increase in the

amount of modeling and simulation for the development of future weapon

systems. The utilization of modeling and simulation tools for end design of armor

systems is critically dependent on the accuracy of the underlying structure of such

simulations. Widespread acceptance of simulation tools hinges upon end user

trust in the predicted results. As the overall implementation of a design code can

be composed of a number of material models, it is essential that those models

accurately reflect true physical behavior. Ideally, different material models

should yield identical results, independent of the computer code used and conform

to experimental data. It is the goal of this paper to investigate the behavior of two

material models using two different computer codes.

The two ceramic models compared in this study are distinctly different: 1)

Johnson-Holmquist (JH-1) and 2) Rajendran-Grove (RG). JH-1 [1] is a

phenomenological model developed for brittle materials subjected to large strains,

high strain rates and high pressures. The equivalent strength is expressed as a

function of the pressure, strain rate, and accumulated damage; and it allows for

strength of intact and fractured material. The pressure is expressed as a function

of the volumetric strain and includes the effect of bulking for the fractured

material. JH-1 (rather than JH-2 or JH-3) was chosen for this study because it

appears to more accurately predict the SiC-B behavior. RG is a micro-crack

based constitutive model [2].

The two computer codes used in this study are the Eulerian CTH wave code

[3] and the Lagrangian EPIC hydrocode [4]. The EPIC computations were

performed with finite elements and meshless particles; the initial grids were

composed entirely of finite elements in 2D axisymmetry, and the elements were

automatically converted to particles as the elements became highly distorted [5].

The CTH computations were performed with the mix=1 option, where the yield

strength in mixed material cell is sum of volume fraction weights of individual

materials and single material cells with voids have decreased yield strength, and

the metals were modeled with Mie-Gruneisen EOS, Johnson-Cook strength and

fracture, using the same material parameters used in the EPIC simulations.

We were specifically interested in comparing computational results for two

target configurations where silicon carbide type-B ceramic is used, 1) semi-

infinite penetration in ceramics as described by Orphal and Franzen [6], and 2)

ceramic dwell as described by Lundberg, et. al [7]. These choices were made due

to the availability of the experimental data in the literature such that comparisons

could easily be made to the experimental results and because of their direct


applicability to specific Army problems. Computations are also presented for a

tungsten penetrator impacting a steel target over a large velocity range. The

primary purpose of performing these computations was to investigate the

accuracy of the two numerical schemes using well-defined material behavior. Of

particular interest was the accuracy of the EPIC computations using the particle

algorithm since this is a relatively new technique with limited evaluation.

The following sections will present the computational results for tungsten steel

(calibration) computations; ceramic dwell (Lundberg) and ceramic penetration for

high velocities (Orphal). A brief discussion will also be presented on constant

determination.

V=1000m/sV=500m/s V=1500m/s V=2500m/sV=2000m/s V=3000m/s

P=25.2mm P=51.0mmP = 3.1 mm P=74.6mm P=78.7mmP = 67.7 mm

CTH

EPIC

P=21.0mm P=47.9mmP = 2.1mm P=73.9mm P=78.2mmP = 65.7mm

0.0

0.4

0.8

1.2

1.6

2.0

0 1000 2000 3000 4000

Imp act Velo city (m/s )

P/L

Hoh ler et al.

ep ic

cth

Tungsten alloy (D17.6)

= 17.6g/ccBHN = 406

elong(%) = 10

L = 50mm

D = 5mm

HzB,A Armor Steel

= 7.85g/ccBHN = 295

120mm

80mm

Figure 1. DOP computations in Epic and CTH


CALIBRATION COMPUTATIONSTo first investigate the possible variations in computational results generated

by EPIC and CTH due to the numerics of the hydrocodes, computations of semi-

infinite penetration using the Johnson-Cook material model for strength and

fracture [8,9] were performed. These computations used the target geometry of

Hohler and Stilp [10]. Computations were performed over a velocity range from

500 m/s to 3000 m/s. The Brinell hardness for the targets ranged from 260-330; a

median value of 295 was used in the computations. The test configuration and

computational results are presented in Figure 1 and the Johnson-Cook constants

used in the computations are presented in Table I. The computations used the

Table I. Johnson-Cook strength and fracture constants

Mass/Thermal Properties

= 17600kg/m3

specific heat = 134.5 J/kg K

conductivity = 75.42 J/s m K

volume expansion coef. = 0.0000162

melt temperature = 1723 K

Elastic Constants

Shear Modulus (G) = 147 GPa

Shear velocity (Vs) = 2890 m/s

Bulk Modulus (K) = 287 GPa

Bulk velocity (Vb) = 4040m/s

Strength Model (Johnson-Cook)

C1 = 1.365 GPa

C2 = 0.1765 GPa

C3 = 0.016

N = 0.12

M = 1.0

Equation of State

Bulk sound velocity (Vb) = 4040m/s

Us-Up slope = 1.23

Gruneisin coefficient = 1.43

Max hydrostatic tension allowed = 68.95GPa

Fracture Model(Johnson-Cook)

D1 = 0.0

D2 = 0.33

D3 = -1.50

D4 = 0.0

D5 = 0.0

minimum fracture strain = 0.022

Spall stength = 6.757GPa

Tungsten

Mass/Thermal Properties

= 7850kg/m3

specific heat = 477.8 J/kg K

conductivity = 38.11 J/s m K

volume expansion coef. = 0.0000324

melt temperature = 1793 K

Elastic Constants

Shear Modulus (G) = 76.4 GPa

Shear velocity (Vs) = 3120 m/s

Bulk Modulus (K) = 165 GPa

Bulk velocity (Vb) = 4580m/s

Strength Model (Johnson-Cook)

C1 = 0.810 GPa

C2 = 0.5095 GPa

C3 = 0.014

N = 0.26

M = 1.03

Equation of State

Bulk sound velocity (Vb) = 4580m/s

Us-Up slope = 1.49

Gruneisin coefficient = 1.16

Max hydrostatic tension allowed = 68.95GPa

Fracture Model(Johnson-Cook)

D1 = -0.80

D2 = 2.10

D3 = -1.50

D4 = 0.002

D5 = 0.61

minimum fracture strain = 0.035

Spall stength = 5.723GPa

HzB, A Armor Steel

same material models, material input parameters and similar gridding. The depth

of penetration from the computations compared very well to the experimental

results, with the computed results predicting somewhat greater penetration (2%-

8%). However, maybe the most important result is that the CTH and EPIC

responses were very similar indicating that reasonable results can be obtained


using very different numerics. Figure 1 also shows the final geometry of the

penetration profile for all the computations. The penetration profiles are

remarkably similar, giving additional support to expect consistent results between

the two hydrocodes.

DETERMINATION OF CERAMIC MODEL CONSTANTS

Determination of ceramic model constants for both the JH-1 and RG models is

not a straightforward process and will not be presented in detail here. The

ceramic used for this work is a hot pressed silicon carbide known as SiC-B

produced by Cercom Inc.

The process to obtain constants for the JH-1 model is presented in detail by

Holmquist [11]. Here, the majority of the constants for the JH-1 model were

measured explicitly in laboratory experiments, although two constants were

obtained by fitting model predictions to ballistic experiments. These two

constants were obtained by matching two of the dwell/penetration experiments

performed by Lundberg et. al. The process used to get these two constants was

applied in the same manner for both EPIC and CTH. The point to stress here is

that all the JH-1 constants were the same for both EPIC and CTH with the

exception of the two constants that were determined using the computations.

The constants for the RG model were obtained by matching plate impact

experiments. The same constants were used for both the EPIC and CTH

computations. The current RG model in EPIC was unable to reproduce the

Lundberg results with the RG model SiC-B constants calibrated for the low and

high velocity plate impact tests. To achieve complete dwell in the low velocity

case, the limiting crack growth rate coefficient (n1-) for mode II/III was changed

to 0.001 (a value of 0.1 was assumed to match the high velocity plate impact

data). While this change did not affect the low velocity plate impact simulation

results, the effect on the high velocity plate impact simulation results was

significant - the slower mode II/III crack growth rates resulted in delayed and

noisy spall signals that did not match the smooth spall signals measured

experimentally.

CERAMIC DWELL COMPUTATIONS

One of the most interesting ceramic characteristics is that of ceramic dwell

and interface defeat. Dwell occurs when a high velocity projectile impacts a

ceramic target and is eroded on the surface of the ceramic with no significant

penetration. If the dwell phenomenon continues until the entire penetrator is

consumed, the event is termed interface defeat (of the penetrator). Ceramic dwell

is an important characteristic of ceramic behavior and must be reproduced

computational by ceramic models. Lundberg et al. [7] demonstrated ceramic

dwell for silicon carbide (SiC-B) in a series of ballistic experiments. Three of the


experimental results are presented in Figure 2. The two highest impact velocities

were used to get JH-1 model constants. Figure 2 shows that the JH-1 model, as

implemented in both EPIC and CTH, is capable of reproducing dwell, dwell-

penetrations transition and high velocity penetration. It should be noted that the

V = 1410m/sEPIC (JH-1) CTH (JH-1)

t = 36 s

CTH (RG)EPIC (RG)

t = 25 st = 36 st = 36 s

0

5

1 0

1 5

2 0

0 1 0 2 0 3 0 4

T im e , t ( s )

Pe

ne

tra

tio

n,

P (

m

V=2175m/s

V=1645m/s

V=1410m/s

CTH (JH-1)

0

CTH (RG)

0

5

1 0

1 5

2 0

0 1 0 2 0 3 0T im e , t ( s )

Pe

ne

tra

tio

n,

P (

4 0

m V=2175m/s

V=1645m/s

V=1410m/s

EPIC (RG)

EPIC (JH-1)

Figure 2. Comparison of experimental and computational results

two constants obtained from computations were very different for the two codes.

CTH required constants that effectively made the material softer, probably due to

the fact that the ceramic material was fixed to the confinement steel and in the


EPIC computations it was allowed to slide. When the JH-1 values obtained using

the EPIC code were used in CTH the computations produced no penetration for

the 1645m/s experiment. Failure strain and failed yield strength values were

changed to reproduce the correct dwell phenomena. No confining pressure was

included in the simulation setup. A boundary layer around the ceramic was

attempted, but not used due to a decrease in the dwell performance. Ten cells

across the penetrator diameter were used for the mesh.

The results using the RG model also demonstrated the ability to capture the

dwell, dwell-penetration transition and the high velocity penetration, but some

modifications to the model were required. No pre-loaded confining pressure was

used in the simulations. Confining pressure was not measured in the Lundberg

experiments, but we know that some (unknown) level of confinement existed.

In the CTH simulation a single cell thickness of weak confinement material

surrounded the ceramic. This layer was identical to the actual confinement in that

it had the same EOS properties (density, pressure response, etc.), but differed in

that it had no yield strength (JC strength parameters were zero or near-zero) or

fracture strength. All simulations were axi-symmetric, utilizing square cells of

size equal to 1/8th

of the penetrator radius. As can be seen in Figure 2, the high

velocity (2175 m/s) is well represented. The duration of the transition dwell

(1645 m/s) is slightly over-predicted. The CTH-RG simulation predicted that

dwell would last for 27 s. The penetration rate after dwell is well represented.

The total dwell duration of 36 s at 1410 m/s is under-predicted, with dwell

ending at about 26 s in the simulation.

SEMI-INFINITE CERAMIC PENETRATION COMPUTATIONS

Computations were also performed into semi-infinite ceramic targets as

defined by Orphal and Franzen [6]. These computations covered a wide range of

impact velocities and were in effect “validation computations” inasmuch as the

constants were not determined from the experiments. Figure 3 presents

penetration as a function of impact velocity for the experiments and the

computations. The JH-1 model, as implemented in EPIC, produced good results

at velocities up to 2000 m/s, but tended to under-predict penetration at the high

velocities (3000 – 4000 m/s). While JH-1 in CTH produced better results at the

high velocities, it still under-predicted penetration, including a larger under-

predicted penetration at 2000 m/s. The RG model in EPIC exhibited a slightly

greater under-prediction of penetration (compared to both experimental results

and JH-1 computations). RG in CTH (still under development) predicted lower

penetration depths in the simulations of the Orphal experiments than the other

model/code implementations, but still followed the general trend of the

experimental data. Both models predicted interface defeat at 1000 m/s.


V = 3000m/s

0.0

0.5

1.0

1.5

2.0

2.5

3.0

0 1000 2000 3000 4000 5000

Impact velocity (m/s)

P/L

Orphal and Franzen

EPIC (JH-1)

CTH (JH-1)

EPIC(RG)

CTH(RG)

Aluminum/ceramic interface

EPIC (RG) CTH (RG)CTH (JH-1)

t = 20 s

EPIC (JH-1)

t = 20 st = 20 s t = 13 s

Figure 3. Penetration computations and comparison to experiment

CONCLUSIONS

Computations were performed using the JH-1 and RG ceramic models as

implemented in the CTH and EPIC computer codes. Computations of a tungsten

rod into a steel target demonstrated that both CTH and EPIC produced very

similar results consistent with experimental data over a wide velocity range.

Computations were also performed of dwell, dwell-penetration transition and high

velocity penetration. The JH-1 model produced good results using both EPIC and

CTH. The RG model, after modifications, was able to reproduce ceramic dwell

behavior. However, in order to improve its ability to correctly predict dwell, it

has been proposed that the model should be modified to include a new "critical

shear stress" criterion that would be applied only when the ceramic material is

experiencing triaxial compressive loading (i.e., when all three principal stresses


are compressive). Under such loading conditions, cracking could not occur unless

the maximum shear stress exceeded the critical shear stress and the Griffith

criterion was satisfied. Finally, computations were performed into semi-infinite

ceramic targets. Both models tended to under-predict the penetration into the

ceramic, but results followed the general trend of the experimental data.

Future work will include model refinement to allow better match to

experimental data and investigations considering different computational

platforms and serial versus parallel processing.

ACKNOWLEDGEMENTS

Some of this work was sponsored by the Army High Performance Computing

Research Center under the auspices of the Department of the Army, contract

number DASW01-01-C-0015. The content does not necessarily reflect the

position or the policy of the government, and no official endorsement should be

inferred. This work was supported in part by a grant of high performance

computing HPC) time from the DoD HPC Center at APG, MD.

REFERENCES

1. G. R. Johnson and T. J. Holmquist, "A Computational Constitutive Model For

Brittle Materials Subjected To Large Strains, High Strain Rates, And High

Pressures," Proceedings of EXPLOMET Conference, San Diego, (August

1990).

2. A. M. Rajendran, “Modeling the Impact Behavior of AD85 Ceramic under

Multiaxial Loading,” International Journal of lmpact Engineering, Vol. 15, pp.

749-768, (1994).

3. J. M. Mcglaun, S. L. Thompson, and M. G. Erlick, “A Three Dimensional

Shock Wave Physics Code,” International Journal of Impact Engineering,

Vo1. 10, (1990).

4. G. R. Johnson, R. A. Stryk, T. J. Holmquist and S. R. Beissel, “Numerical

Algorithms in a Lagrangian Hydrocode,” Report No. WL-TR-1997-7039

(June 1997).

5. G. R. Johnson, R. A. Stryk, L. R. Beissel, and T. J. Holmquist, “Conversion

Of Finite Elements Into Meshless Particles For Penetration Computations

Involving Ceramic Targets,” Shock Compression of Condensed Matter-2001,

in press, (2001).

6. D.L., Orphal and R.R. Franzen, “Penetration of Confined Silicon Carbide

Targets by Tungsten Long Rods at Impact Velocities from 1.5 to 4.6 km/s,”

International Journal of Impact Engineering, Vo1. 19, No. 1, pp. 1-13, (1997).

7. P. Lundberg, R. Renstrom, and B. Lundberg, “Impact of Metallic Projectiles

on Ceramic Targets: Transition Between Interface Defeat and Penetration,”

International Journal of Impact Engineering, Vo1. 24, 259-275, (2000).


8. G. R. Johnson and W. H. Cook, “A Constitutive Model and Data for Metals

Subjected to Large Strains, High Strain Rates, and High Temperatures,”

Proceedings of Seventh International Symposium on Ballistics. The Hague,

The Netherlands, (April 1993).

9. G. R. Johnson and W. H. Cook, “Fracture Characteristics of Three Metals

Subjected to Various Strains, Strain Rates, Temperatures, and Pressures,”

Engineering Fracture Mechanics, Volume 21, (1985).

10. C. E. Anderson, Jr., B. L. Morris and D. L. Littlefield, “A Penetration

Mechanics Database,” SwRI Report 3593/001, (January 1992).

11. T. J. Holmquist and G.R. Johnson, “Response of Silicon Carbide to High

Velocity Impact,” submitted for publication, Journal of Applied Physics,

(2001).


MODELING CERAMIC DWELL AND INTERFACE DEFEAT

Timothy J. Holmquist and Gordon R. Johnson

Network CS/Army High Performance Computing Research Center

1200 Washington Avenue South

Minneapolis, MN 55415

ABSTRACT

This paper presents computational modeling of ceramic dwell and interface

defeat, using the EPIC code and the JH-1 constitutive model for ceramics.

Computations are presented for various projectiles impacting various silicon

carbide (SiC-B) targets. The computational results are shown to provide good

agreement with experimental data in the literature. Also included are the JH-1

constants for SiC-B, the procedure used to determine the constants, and a

description of some important computational features involving finite elements

and meshless particles.

INTRODUCTION

Ceramic materials are generally strong in compression, weak in tension, and

can have considerable strength after failure when they are under compression.

They have been used as armor materials for many years. More recently,

experimental data have been presented by Lundberg et al. [1] that show how

silicon carbide targets can be configured to defeat tungsten and molybdenum rods

at significant impact velocities. Other researchers have demonstrated this same

effect for other ceramics, but this paper will focus only on the silicon carbide

targets.

JH-1 CERAMIC MODEL AND CONSTANTS

The JH-1 constitutive model for ceramics, and the associated constants for

SiC-B, are shown in Figure 1. The model consists of an intact strength and a

failed strength that are functions of the pressure, the strain rate, and the damage.

Pressure, bulking and damage are other aspects of the model. This is the first of

two closely related models, JH-1 [2] and JH-2 [3], presented by Johnson and

Holmquist. One of the primary differences between the two models is that the

JH-2 model allows the strength to degrade gradually as the damage is



accumulated, rather than soften/fail instantaneously after it is fully damaged, as is

done in the JH-1 model. For SiC-B the JH-1 model appears to be better suited to

represent the strength and interface defeat characteristics of SiC-B. Apparently

the JH-2 approach, with gradual softening, does not provide the constant target

resistance required for dwell and interface defeat.

Fai

lure

str

ain,

pf

T P3

D= p/ pf

Pressure, P Volumetric strain,

Pre

ssure

, P

D<1.0

D=1.0

P

T

Density o = 3215kg/m3

Shear modulus G = 193GPa

Tensile strength T = 0.75GPa

Intact strength S1 = 7.1GPa

Intact strength P1 = 2.5GPa

Intact strength S2 = 12.2GPa

Intact strength P2 = 10.0GPa

Strain rate C = 0.009

Failed strength Sfmax= 1.3GPa

Failed strength = 0.40

Bulk modulus K1 = 220GPa

Pressure K2 = 361GPa

Pressure K3 = 0GPa

Bulking factor = 1.0

Damage = 0.012Eq

uiv

alen

t S

tres

s,

T P1 P2

S2

S1

Pressure, P

*>1.0.

*=1.0.

Intact Material (D<1.0)

Failed Material (D=1.0)

P=K1 +K22+K3

3

fmaxS

f

max

)TP/( 3

f

max

Figure 1. JH-1 model and constants for silicon carbide.

The intact strength and compressibility constants were obtained from the plate

impact data of Feng et al. [4], Grady and Moody [5], Dandekar and Bartkowski

[6], and the uniaxial compression data of Pickup and Barker [7]. A very

important characteristic of the Feng et al. data is that both longitudinal and lateral

stresses are provided, and this allows the complete stress state (both strength and

pressure) to be determined. The slope constant for the failed material, , is

provided by Klopp and Shockey [8].


The maximum strength of the failed material, cannot be obtained

directly due to the lack of experimental data, but must instead be inferred by

varying until computed penetration computations match the results of

experimental data. The damage constant,

,max

fS

fSmax

, is obtained in a similar manner. The

two experimental data points used to obtain these constants are the penetration of

a tungsten rod into a silicon carbide target at 2175 m/s, and the duration of the

dwell of a tungsten rod onto a ceramic target at 1645 m/s [1]. Additional details

concerning the model and the constants are provided elsewhere [9].

EXAMPLES

The following examples all use the silicon carbide constants for the JH-1

model as shown in Figure 1. The computations were performed with finite

elements and meshless particles; the initial grids were composed entirely of finite

elements in 2D axisymmetry, and the elements were automatically converted to

particles as the elements became highly distorted [10].

Figure 2 shows computational results compared to experimental results

provided by Lundberg et al. [1]. The tests were performed by impacting a

confined ceramic target onto a long stationary rod (of tungsten or molybdenum).

Only three test results are shown for each of the two rod materials, although

additional test results are reported by Lundberg et al. These three results

correspond to interface defeat (where the rod does not penetrate the ceramic),

dwell and penetration (where the rod dwells on the surface of the ceramic before

it begins to penetrate), and penetration (where the rod penetrates with minimum

dwell).

For the tungsten tests, the two higher impact velocities (1645 m/s and 2175

m/s) were used to determine the strength of the failed material, , and the

damage constant,

fSmax

. Although there is some coupling between these two

constants, the penetration rate/depth is most dependent on and the dwell is

most dependent on

fSmax

. It can be seen there is very good agreement between the

computed results and the experimental results for interface defeat, dwell, and

penetration. Although the tungsten experimental results were used to determine

some of the constants, the molybdenum experimental results were not used. It

should be noted that the molybdenum test data for interface defeat were for an

impact velocity of 2030 m/s, whereas the corresponding computed results are for

a slightly reduced impact velocity of 2000 m/s.

Figure 3 shows computed results for the interface defeat of the tungsten rod for

an impact velocity of 1410 m/s. For the three geometry plots in the upper portion

of the figure, the tungsten rod is represented by the darkened elements and

particles; and the steel plug, steel tube and ceramic are represented by various


colors of elements and particles. The black lines define the outlines between the

materials, and in other instances they represent an interface between the elements

and particles. For this case the ceramic remains intact, while the defeated rod

moves radially outward along the top surface of the ceramic until it is contained

by the steel case. The distribution of damage is shown in the lower right portion

of the figure, where it can be seen that most of the ceramic under the rod is only

partially damaged, and this enables the ceramic to remain intact and to defeat the

rod.

Steel plug

Silicon carbide

L = 20mm

D = 20mm

Steel plug

Hig

h s

tren

gth

ste

el t

ub

e

Tungsten or

Molybdenum

rod

L = 80mm

D = 0.5mm

0

5

10

15

20

0 10 20 30 40

0

5

10

15

20

Lundberg et al.

Tungsten rod

Molybdenum rod

V=2175m/s

V=1645m/s

V=1410m/s

V=2535m/s

V=2090m/s

V=2000m/s

Pen

etra

tion

in

to c

eram

ic(m

m)

(a) (b)

00

Computed

results

Time, t (microsecond)

Figure 2. Comparison of computational results and experimental results for tungsten

and molybdenum rods impacting a confined silicon carbide target at various velocities.

(a) Initial 2D geometry and (b) comparison of computed and experimental results.

Figure 4 shows the computed response for a slightly higher impact velocity of

1645 m/s. Here the tungsten rod dwells for a short time and then penetrates the

ceramic. When the ceramic material directly under the impacting rod becomes

completely damaged and fails, the dwell ceases and the penetration begins. The

distribution of damage for the transition between dwell and penetration is shown

in the lower right portion of the figure.

A comment should be made concerning an important advantage of converting

distorted elements into particles rather than simply eroding (or removing) the

distorted elements. When an element is eroded it introduces a void which allows


surrounding material to expand into the void and to lose pressure as it expands. If

the material strength or failure characteristics are pressure dependent (as they are

for ceramics) then the pressure drop can lead to lower strength and/or increased

damage.

Figure 3. Computed results for a tungsten rod impacting a confined silicon carbide target

at 1410m/s. (a) penetration of steel cover at 12 s, (b) dwell at 20 s, (c) dwell at 36 s,

(d) close up of material flow at 36 s, and (e) close up of material damage at 36 s.

The final example involving dwell is shown in Figure 5. Here a pointed steel

projectile impacting a thin plate composed of a silicon carbide layer (6.35 mm

thick) over a 6061-T6 aluminum layer (6.35 mm thick) is investigated. An impact

velocity of 700 m/s is shown, as it represents the computational ballistic limit.

This is only slightly higher than the experimental ballistic limit of 660 m/s

reported by Wilkins [11]. It can be seen that there is significant dwell during the

initial 10 to 20 and that the damage pattern in the ceramic forms in a conicals


pattern. The silicon carbide tested by Wilkins was not the SiC-B for which the

constants were obtained, and it is not known if the experimental results would be

significantly different for SiC-B. Nevertheless, the basic trends and mechanisms

appear to be well represented by the computations.

Figure 4. Computed results for a tungsten rod impacting a confined silicon carbide

target at 1645m/s. (a) dwell at 12 s, (b) transition from dwell to ceramic penetration at

18 s, (c) ceramic penetration at 30 s, (d) close up of material flow at 18 s, and (e)

close up of material damage at 18 s.

SUMMARY AND CONCLUSIONS

This paper has demonstrated the computational capability to simulate ceramic

dwell and interface defeat. The specific form of silicon carbide, known as SiC-B,

has been characterized from experimental data in the literature and put into the

form of the JH-1 constitutive model for ceramics. Many of the constants were

determined explicitly from the experimental data, but some of the constants for


damaged/failed material were inferred from ballistic penetration data. There was

generally very good agreement between the computational results and the

experimental results, even for the problems that were not used to determine the

constants.

Figure 5. Computed results for a steel projectile impacting a thin, layered target of

silicon carbide and aluminum at 700 m/s. (a) Material flow at 10 s, 20 s and 200 s

after projectile impact, and (b) material damage at 10 s, 20 s and 200 s after

projectile impact.

ACKNOWLEDGEMENTS

This work was sponsored by the Army High Performance Computing

Research Center under the auspices of the Department of the Army, contract

number DASW01-01-C-0015. The content does not necessarily reflect the


position or the policy of the government, and no official endorsement should be

inferred. The authors would also like to thank D. W. Templeton (Army Tank

Automotive Research, Development, and Engineering Center) for his

contributions to this work.

REFERENCES1 P. Lundberg, R. Renstrom, and B. Lundberg, “Impact of metallic projectiles

on ceramic targets: transition between interface defeat and penetration,”

International Journal of Impact Engineering, 24, p. 259, (2000). 2 G. R. Johnson and T. J. Holmquist, “A computational constitutive model for

brittle materials subjected to large strains, high strain rates, and high pressures,”

Proceedings of EXPLOMET Conference, San Diego, August 1990. 3 G. R. Johnson and T. J. Holmquist, “An improved computational constitutive

model for brittle materials,” High Pressure Science and Technology – 1993,

Edited by S. C. Schmidt, J. W. Schaner, G. A. Samara, and M. Ross, AIP 1994.

R. Feng, G. F. Raiser, and Y. M. Gupta, “Material strength and inelastic

deformation of silicon carbide under shock wave compression,” Journal of

Applied Physics, 83, p.79, (January 1998).

4

D. E. Grady and R. L. Moody, “Shock compression profiles in ceramics,”

Report No. SAND96-0551, Sandia National Laboratory, March 1996.

5

D. P. Dandekar and P. T. Bartkowski, “Tensile strengths of silicon carbide

(SiC) under shock loading,” Report No. ARL-TR-2430, Army Research

Laboratory, March 2001.

6

I. M. Pickup and A. K. Barker, “Deviatoric strength of silicon carbide subject

to shock,” Shock Compression of Condensed Matter – 1999, Edited by M. D.

Furnish, L. C. Chhabildas, and R. X. Hixon, p. 573, (AIP, 1999).

7

8 R. W. Klopp and D. A. Shockey, “The strength behavior of granulated silicon

carbide at high strains rates and confining pressure,” Journal of Applied Physics,

70, p.7318, (December 1991).

T. J. Holmquist and G. R. Johnson, “Response of silicon carbide to high

velocity impact,” Submitted for publication.

9

G. R. Johnson, R. A. Stryk, S. R. Beissel, and T. J. Holmquist, “Conversion

of finite elements into meshless particles for penetration computations involving

ceramic targets,”Shock Compression of Condensed Matter – 2001, in press.

10

M. L. Wilkins, “Fourth progress report of light armor program,” Report No.

UCRL-50694, Lawrence Radiation Laboratory, 1969.

11


3D FINITE ELEMENT ANALYSIS OF IMPACT DAMAGE IN METALLIC

AND CERAMIC TARGETS

Fenghua Zhou and Jean-Francois Molinari

Department of Mechanical Engineering, Johns Hopkins University

232 Latrobe Hall, 3400 N. Charles St., Baltimore, Maryland, 21218

ABSTRACT

A 3D explicit FEM package has been developed to analyze the deformation

and failure process of materials under impact loading. The package is constituted

of a finite deformation plastic model, a frictional contact algorithm, an adaptive

meshing capability, and a module of dynamically inserted cohesive elements to

simulate dynamic crack propagation. In this paper, we present three simulations

that investigate dynamic crack propagation in a metallic and a ceramic material.

The quantitative and qualitative agreements between experimental and simulation

results are discussed.

INTRODUCTION

The investigation of the dynamic response and the failure process of metallic

and ceramic targets under impact loading is a crucial issue to the designing of

armor and anti-armor systems. Numerical techniques are broadly applied in this

investigation to analyze and to simulate expensive experimental results. Although

many large-scale, general-purpose codes for conducting structural impact analysis

have been developed, the problem of modeling material deformation/failure

processes under intensified loading is still a challenging task. These processes are

accompanied with the mechanisms of large-deformation high strain-rate

plasticity, heat generation and conduction, and dynamic fracture and

fragmentation. Moreover, the need of using lighter and stronger composite

materials has lead to the development of complex materials such as metal-matrix

or ceramic-matrix composite materials. The combination of a variety of physical

phenomena with a range of microstructures renders the interpretation of impact

mechanisms difficult. As a design tool and as a methodology to reach a physical

understanding, a multi-physics finite element tool is of great importance.

We have developed a three-dimensional explicit finite element analysis

package. The package includes finite deformation plasticity, frictional contact,



heat generation and conduction, and adaptive meshing [1, 2]. Molinari et al. [1]

studied the impact erosion in mild-steel targets by hard-steel projectiles. The

range of impact velocity varied from 200m/s to 2000m/s and the impact angle

from glancing to normal penetration. The calculations highlighted that the friction

resistance increases with the sliding velocity for the low range of impact speeds.

Another outcome of the calculations is that at 45 degrees impact frictional heating

becomes dominant. However, crack propagation was not modeled in [1]. In this

paper, we highlight the addition of cohesive elements to the existing capability.

Cohesive zone models can explicitly describe the crack initiation and propagation

process. Several simulations of the dynamic fracture phenomena were conducted

to verify and validate the methodology. We begin with a brief description of the

relevant aspects of the material and numerical model. Then, to illustrate the

methodology, we simulate the dynamic bursting of a ceramic ring under

centrifugal force. Impact three-point-bending simulations performed on metallic

and ceramic materials constitute the core of the paper.

FINITE ELEMENT ANALYSIS AND COHESIVE MODELS

In our analysis, the volume of the structure is meshed by 10-node (quadratic)

tetrahedral elements. The simulation of the dynamic process is conducted by

using the second order accurate explicit form of Newmark’s algorithm ([Hughes,

[3]):

)(2

1)(

2

1

11

int

11

1

1

2

1

nnnn

n

ext

nn

nnnn

t

tt

aavv

FFMa

avdd

(1)

where the subscript n denotes variables at the nth

time step; t is a fraction of the

critical time step; d, v and a are nodal displacement, velocity and acceleration

vectors; M is the lumped mass matrix, Fext

and Fint

are the external and internal

nodal forces.

We use an isotropic J2 flow material model that includes thermal softening,

power-law strain hardening and strain-rate hardening (Cuitino et al., [4]):

n

p

p

refmelt

ref

y

m

pp

p

TT

TTg

Tg1

0

0

11

1),(

(2)

where is the effective Mises stress, the effective plastic strain, the

effective plastic strain rate, g the flow stress and T the temperature. The material

p p


parameters include the initial yield stress, the reference plastic strain,

the reference plastic strain rate, m the rate sensitivity exponent, n the hardening

exponent, T the reference temperature, T the melting temperature and the

thermal softening exponent.

y

2

p

0

Rigid

Smith

p

0

ref melt

g:d >0

The key issue in this paper is the simulation of dynamic crack propagation.

We follow a cohesive element type approach. The cohesive zone concept was

firstly introduced by Dugdale [5] and Barrenblatt [6], who independently assumed

that a cohesive region exist at the crack tip, where stress singularity is eliminated

by the distribution of the cohesive forces. The cohesive zone concept was

implemented into numerical analysis to explicitly simulate crack propagation. The

work of Xu and Needleman [7, 8], Camacho and Ortiz [9, 10] demonstrated

successful use of the cohesive element in 2D cases. The work of Pandolfi et. al.

[11, 12], and Ruiz et. al. [13, 14] successfully implemented cohesive elements

into 3D analysis in a range of applications. Other recent relevant numerical

investigations include the effect of microstructure on the dynamic failure process

[15, 16].

In our calculations we insert 12-nodes triangular cohesive elements between

two neighboring tetrahedral elements, Fig. 1a.

Fig.1 (a) Cohesive element between two tetrahedral elements;

(b) Two irreversible cohesive models: Smith-Ferrante law and Rigid-linear law

c

c

c

c

O penin Closing : d 0

The tractions (cohesive force) between the glued faces (S+ and S- in Fig. 1a)

are functions of their relative distance. These functions are called cohesive laws,

and they express the energy and the forces needed to open the cohesive element.

Two cohesive laws frequently used are shown in Fig. 1b. They are the irreversible

exponential decaying and the irreversible linear decaying functions [11, 12].

Irreversibility signifies that the damage in a given cohesive element cannot be

recovered. The area under the curves of Fig. 1b is the fracture energy, which is

needed to fully open a unit area of crack surface.

In the models sketched in Fig. 1b the fracture energy takes a simple form:

modellinear-

modelFerrante-

5.0cc

cc

cc

eG (3)


where Gc is the fracture energy, c is the surface energy, c is the critical opening

stress, and c is the fracture strength. Both models have their merits and demerits.

The Smith-Ferrante law is physically reasonable but inserting this element to the

structure may modify structural compliance; the rigid-linear law does not modify

the structural elastic properties, but it needs to be introduced dynamically. Its use

necessitates some computational effort. In our calculations, the simulation results

are not significantly affected by the choice of the cohesive law.

For conciseness, the remaining components of the methodology, adaptive

meshing, heat generation and conduction and frictional contact, are not described

here. A detailed description is contained in [1, 2].

DYNAMIC FRACTURE OF A ROTATING CERAMIC DISK

As an illustration of the use of cohesive

elements, we run a test simulation of the

burst of a ceramic ring under centrifugal

forces. The rotating ring experiments are a

standard test to derive mechanical

properties of materials [17]. Our model

consists of 5438 nodes and 2614 tetrahedral

elements, Fig. 2. The cohesive elements are

inserted when a critical centrifugal load is

reached. The disk is constituted of Si3N4,

whose properties are listed in Table. I. Note

that plasticity was neglected as the material

considered is brittle. The data was obtained from NIST

[http://www.ceramics.nist.gov/srd/summary/ftgsin.htm]. Two numerical tests,

with different fracture properties, Type-1 and Type-3 (Table. I), are conducted.

They reflect the variations in materials data handbooks and assess the numerical

sensitivity to materials data. In both types we keep the critical tensile stress ( c) at

a constant value of 450 MPa. However, the fracture energy (Gc) varies from 100

N/m for Type-1 to 200 N/m for Type-3.

Table I. Mechanical properties of Si3N4

Fracture Properties

(Irreversible Linear Decreasing Cohesive Law)Densi

ty

Elastic

Type-1 Type-2 Type-3 Type-4

Gc = 200 N/m Gc = 100 N/m

kg/m3

E

GPa c

Gpac

mc

GPac

mc

GPac

mc

GPac

m

3300 320 0.23 0.45 0.889 1.0 0.4 0.45 0.444 1.0 0.2

Fig. 2 Ceramic ring before burst


As the fracture stress is kept to a constant value in both tests, the rotating speeds

and the time at which the ceramic ring bursts are identical (about 47500rpm). Fig.

3 shows the initiation and crack propagation of a Type-1 ring. Clearly, multiple

cracks initiate at the inner rim, where the hoop stress is maximum. The cracks

then expand outwards and in some instances branching occurs when approaching

the outer rim. The final shapes and sizes of the fragments are shown in Fig. 4. The

two tests highlight that the fracture energy directly affects the number of

fragments. The number of cracks and fragments increases with a decreasing

fracture energy. This is qualitatively in agreement with experimental observations

[17].

DYNAMIC 3 POINT BENDING FRACTURE TEST ON SIC/AL MATERIAL

We now focus on 3-point bending

calculations in which ductile and brittle

materials are impacted by a Kolsky bar. In

this section, we simulate the dynamic

fracture process of silicon carbide particle

reinforced aluminum alloy (SiCp/Al). The

dynamic mechanical properties of this

material have been thoroughly investigated

by Li et. al. [18, 19]. The dynamic fracture

behavior of the material was also

experimentally studied using a Kolsky-bar

testing system [20]. The experimental setup is shown in Fig. 5: a 3-point bending

specimen is sandwiched between an input bar and an output tube. When a

projectile impacts the input bar, a compressive stress wave propagates along the

bar. This wave is transmitted to the specimen. The loading history on the

impacted point of the specimen can be measured by using strain-gages adhered to

the bar and tube. During the experiments, the local strains near the crack tip, and

the crack tip opening distance (CTOD) are monitored. Four tests were performed,

#6, #7, #8 and #9, where the impact speeds of the projectile were respectively

6m/s, 16.8m/s, 16.9m/s and 8.5m/s.

Fig. 3 The propagation of cracks in a

ceramic ring (Type-1 material) Fig. 4 Ceramic rings after burst (50 s):

(a) Gc=200N/m; (b) Gc=100N/m

��

��

��

C rack-Tip Strain G age

C TO D M arking Points

Incident B arProjectile O utput Tube

v0 Strain G age

Fig.5 Kolsky-bar system for 3-

point bending fracture tests


The FEM model is shown in Fig. 6a. The specimen is supported on the output

tube, which is modeled as fixed in the y-direction. The forces applied on the

specimen, (t) are dependent on the structural response of the specimen.

Therefore at the impact zone we apply a boundary condition of the form:

)()(2)( tcvtt inc (4)

where inc(t) is the input stress wave (experimental data), v(t) is the velocity of

loading point (computed), and c are the density and elastic wave speed of the

input bar. In our analysis, the incident wave is used as the input data. Eqn (4) is

introduced into the explicit algorithm to calculate the response of the specimen.

The reflective wave is an output of the calculations. This wave is compared to the

experimental data as a validation of the numerical results.

The mesh of the specimen is shown in Fig. 6b. It contains 24609 nodes and

16251 tetrahedral elements. A simulation using a finer mesh (54325 nodes, 37010

elements) was conducted. The numerical results were similar to the one obtained

with the coarser mesh. The material data used, is gathered in Table II ([18]). The

temperature is taken to be equal to Tref so that thermal softening is not considered

in our analysis.

Table II. Mechanical Properties of SiCp/Al

Den-

sity

Elastic Plastic Strain

Hardening

Strain Rate

Hardening

Fracture Properties

(Irreversible Linear)

kg/m3

E

GPa y

MPa0p n d 0

p/dt

1/s

m Gc

N/mc

GPa c

m

2738 102 0.29 210 1.556E-2 3.76 1.466E5 2.22 2306 1.02 4.52

X Y

Z

Fig.6 (a) Model for 3-point bending dynamic fracture tests; (b) Mesh


The results of the calculated and

experimental reflection waves are compared

in Fig. 7. The quantitative and qualitative

agreement is good. It is noteworthy that the

material parameters were not fitted to match

the experimental structural response.

The computed local strains, which

contain an elastic and a plastic part are

compared to the experimental results in Fig.

8. Note that in each test, the strain increases

to a maximum, at which a crack propagates,

and subsequently decreases to reach a

constant value (the irreversible plastic part).

The numerical results compare well with experiments in two aspects. First, the

times at which the crack-tip strains drop, match the experimental data, which

implies that the crack initiation time is accurately simulated. Second, the

magnitudes of the strain drop after crack propagation quantitatively agree with the

experimental data (Fig. 8a). However, the experimental peak strains are about 10-3

lower than the simulation results. The existence of residual strains at the

specimen’s crack tip may be the reason of such discrepancy. The crack tip

opening distances are compared in Fig. 8b. A quantitative agreement is observed.

An example of crack propagation is shown in Fig. 9. It can be seen that the

crack front is curved because of 3D effects. The average velocity of the main

crack propagation is shown in Fig. 10. The propagation velocity of the main crack

increases with the increase of impact velocity.

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14

0

2

4

6

8

Velo

cit

y (

m/s

)

Time (ms)

#6

#9

#7

#8

Points: Exp. Data

V_Ref #6

V_Ref #7

V_Ref #8

V_Ref #9

Fig. 7 The reflection waves from

simulations and experiments

0.00 0.02 0.04 0.06 0.08 0.10

0.0000

0.0005

0.0010

0.0015

0.0020

0.0025

0.0030

Points: Exp. Data

Str

ain

Near

Cra

ck T

ip

Time (ms)

Strain #6

Strain #7

Strain #8

Strain #9

0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.10

Points: Experimental Data

CT

OD

(m

m)

Time (ms)

CTOD #6

CTOD #7

CTOD #8

CTOD #9

Fig. 8 Comparisons of the local variables from simulations and the experiments

(a) The crack-tip strains; (b) The crack tip opening distances


It is noteworthy that the cohesive laws incorporated in the present analysis are

rate-independent. Nevertheless, the phenomenological rate-dependent dynamic

fracture process can be simulated. The reason is that cohesive fracture models

have an intrinsic time scale, which is linked to the characteristic length scale ( c)

and the wave velocity (c), Camacho et al. [9]. This property is demonstrated in

our simulations. As shown in Fig. 7 and 8, the simulations using the same set of

material data match experiments carried out at different loading rates.

DYNAMIC FRACTURE OF CERAMIC MATERIAL, VIRTUAL TEST

Having validated the implementation of cohesive elements, we now simulate

a dynamic fracture test on ceramics. No further comparison with experiments is

carried out. Thus, in essence, our following simulations constitute a virtual test,

which may be used to scale and understand experiments as well as evaluate

dynamic material properties. The specimen material is changed to silicon nitride.

We consider the material to be elastic-brittle under loading. Since an accurate

dataset of micro-cracking parameters is not available, four material types are

assumed, as listed in Table I. The specimen is loaded by the Kolsky-bar with an

incident stress wave equal to the one in Test #6 of the previous section. The

response of the specimen, and the virtual experimental measurement, are

predicted as following.

Fig. 11 shows the

reflection wave and the

loading history on the

ceramic specimen. Only

minor differences are

seen for the four types

of materials. This is

reasonable since the

stress wave

measurement contains

Fig. 9 Crack propagation in SiCp/Al specimen

(Test #8)

6 8 10 12 14 16 18

120

130

140

150

160

170

180

190

200

210

Avera

ge C

rack V

elo

cit

y (

m/s

)

Kolsky-bar Impact Velocity (m/s)

Crack Velocity

Fig. 10 Crack velocity in

SiCp/Al specimen

0.00 0.01 0.02 0.03 0.04 0.05

-10

-8

-6

-4

-2

0

2

4

6

8

Kolsky Bar Measurement

Velo

cit

y (

m/s

)

Time (ms)

Incident

Type-1 reflecion

Type-2 reflection

Type-3 reflection

Type-4 reflection

0.00 0.01 0.02 0.03

0

200

400

600

800

1000

1200

1400

1600

Lo

ad

(M

Pa)

Time (ms)

Type-1 Load

Type-2 Load

Type-3 Load

Type-4 Load

Fig. 11 Kolsky-bar recordings: (a) Incident and

reflection waveform; (b) Loading history


the response of the whole specimen structure.

On the other hand,

the recordings of crack-

tip strains and CTOD,

shown in Fig. 12,

demonstrate significant

differences for the four

types of material

properties. The reason is

that such recordings are

local to the crack tip

zone and therefore are more sensitive to the initiation of crack propagation. The

initiation time of the crack is strongly affected by the microscopic fracture

properties: the larger the fracture energy (Gc) or the larger the critical strength

( c), the later the crack initiates.

An example of crack propagation is shown in Fig. 13. It is seen that the front

of the crack is straighter in comparison to the case of SiCp/Al material (Fig. 9).

The reason is that the brittleness of the material diminishes the 3D effects.

The locations of the

crack front, and the

average crack velocity

are shown in Fig. 14.

The speed of the crack

propagation in ceramics

is about 1000m/s,

which is much higher

than for the SiCp/Al

specimen. The velocity

of the crack not only depends on the fracture energy, but also depends on the

loading history of the specimen. An accelerating-decelerating crack behavior can

be seen from the figure, as the external load on the specimen drops (see Fig. 10b).

We also simulate a case where the fracture energy is not constant, but decreases

linearly along the specimen’s width. The results are also shown in Fig. 14. In this

0.00 0.01 0.02

0.0000

0.0001

0.0002

0.0003

Cra

ck T

ip S

train

Time (ms)

Type-1 Strain

Type-2 Strain

Type-3 Strain

Type-4 Strain

0.000 0.005 0.010 0.015 0.020 0.025

0.000

0.005

0.010

0.015

0.020

Cra

ck T

ip O

pen

ing

Dis

tan

ce (

mm

)

Time (ms)

Type-1 CTOD

Type-2 CTOD

Type-3 CTOD

Type-4 CTOD

Fig. 12 Local quantity recordings: (a) Crack-tip strains;

(b) Crack-tip opening distances

Fig. 13 Crack propagation in ceramic specimen (Type-1 material)

Fig. 14 Crack front locations and crack velocities

(Type-1 and Type-3 material)


case, the velocity of the crack becomes relatively uniform. We will address the

issue of a limiting crack velocity and its practical consequences on design of

crack-resistant armors in future calculations.

CONCLUDING REMARKS

We have developed an explicit dynamic element package, which includes

large deformation plasticity, contact, adaptive meshing and dynamic insertion of

cohesive elements. This paper highlights the introduction of 3D cohesive

elements. We simulate three types of dynamic fracture phenomena: the

fragmentation of a brittle ceramic ring, crack propagation in a ductile metallic

specimen and crack propagation in a brittle ceramic specimen. The first

simulation illustrates how cohesive elements can be used to model explicitly

crack propagation and the complex associated fragmentation process. In the

second simulation, the results are compared to the experiments and quantitative

agreements are obtained. In this light, it validates the methodology. The intrinsic

time scale of the cohesive elements, which permits the reproduction of

experiments at various loading rates, is highlighted. The last simulation

constitutes a virtual test capability, which can be used to evaluate material

properties and design structures in which the crack velocity needs to be

controlled.

ACKNOWLEDGEMENT

The research is sponsored by Army Research Lab under contract

DAAD19012003. The authors would like to thank Professor K.T. Ramesh and Dr.

Y. Li of the Johns Hopkins University for the invaluable discussions.

REFERENCES 1

J.F. Molinari and M. Ortiz, “Three-Dimensional Adaptive Meshing by

Subdivision and Edge-Collapse in Finite-Deformation Dynamic-Plasticity

Problems with Application to Adiabatic Shear Banding”, International Journal

for Numerical Methods in Engineering, 53, (2002). 2

J.F. Molinari and M. Ortiz, “A Study of Solid-Particle Erosion of Metallic

Targets”, accepted at the International Journal of Impact Engineering. 3

T.L.J. Hughes, The Finite Element method: Linear Static and Dynamic

Finite Element Analysis, Prentice-Hall, Englewood Cliffs, N.J. (1987) 4

A.M. Cuitino and M. Ortiz, “A Material-independent Method for Extending

Stress Update Algorithm from Small-strain Plasticity to Finite Plasticity with

Multiplicative Kinematics”, Engineering Computations, 9, 437-451. (1992)5

D.S. Dugdale, “Yielding of Steel Sheets Containing Slits”, J. Mech. Phys.

Solids, 8, 100-104 (1960).


6G.I. Barrenblatt, “The Mathematical Theory of Equilibrium of cracks in

Brittle Fracture”, Adv. Apply. Mech., 7, 55-129 (1962). 7

X.-P. Xu and A. Needleman, “Numerical Simulations of Fast Crack Growth

in Brittle Solids”, J. Mech. Phys. Solids, 42, 1397-1434 (1994). 8

X.-P. Xu and A. Needleman, “Numerical Simulations of Dynamic Crack

Growth Along an Interface”, Int. J. Fracture, 74, 289-324 (1996). 9

G.T. Camacho and M. Ortiz, “Computational Modelling of Impact Damage

in Brittle Materials”, Int. J. Solids Structures, 33, 2899-2938 (1996). 10

G.T. Camacho and M. Ortiz, “Adaptive Lagrangian Modelling of ballistic

Penetration of Metallic Targets”, Comput. Meth. Appl. Mech. Engng, 142, 269-

301 (1997). 11

A. Pandolfi, P. Krysl and M. Ortiz, “Finite Element Simulation of Ring

Expansion and Fragmentation: The Capturing of Length and Time Scales

Through Cohesive Models of Fracture”, Int. J. Fracture, 95, 279-297 (1999). 12

A. Pandolfi, P.R. Guduru, M. Ortiz and A.J. Rosakis, “Three Dimensional

Cohesive-Elements of Dynamic Fracture in C300 Steel”, Int. J. Solids Structures,

37, 3733-3760 (2000). 13

G. Ruiz, M. Ortiz and A. Pandolfi, “Three Dimensional Finite-Element

Simulation of the Dynamic Brazilian Tests on Concrete Cylinders”, Int. J. Numer.

Meth. Engng, 48, 963-994 (2000). 14

G. Ruiz, M. Ortiz and A. Pandolfi, “Three Dimensional Cohesive Modeling

of Dynamic Mixed-Mode Fracture”, Int. J. Numer. Meth. Engng, 52, 97-120

(2001).15

J. Zhai and M. Zhou, “Finite Element Analysis of Micromechanical Failure

Modes in a Heterogeneous Ceramic Material System”, Int. J. Fracture, 101, 161-

180 (2000). 16

P.D. Zavattieri and H.D. Espinosa, “Grain Level Analysis of Crack

Initiation And Propagation in Brittle Materials”, Acta. Mater. 49, 4291-4311

(2001).17

R. Hashimoto, A. Ogawa, T. Morimoto and M. Yonaiyama, “Spin Tests of

Silicon Nitride Disk”, Proceedings of 74th

Japan Society of Mechanical

Engineering Fall Annual Meeting, Vol.2, 441-442 (1996) 18

Y. Li, K.T. Ramesh and E.S.C. Chin, “The Compressive Viscoplastic

Response of an A359/SiCp Metal-Matrix Composite and of the A359 Aluminum

Alloy Matrix”, Int. J. Solids Structures, 37, 7547-7562 (2000) 19

Y. Li, K.T. Ramesh and E.S.C. Chin, “Viscoplastic Deformations and

Compressive damage in an A359/SiCp Metal-Matrix Composite”, Acta Mater.,

48, 1563-1573 (2000) 20

Y. Li, K.T. Ramesh and E.S.C. Chin, “A Simple Approach to the

Measurement of Dynamic Fracture Toughness”, to be published


A NUMERICAL INVESTIGATION OF MICROCRACKING DIFFUSION IN SANDWICHED GLASS PLATES

Z. Chen and L. Shen G.I. Kanel and S.V. Razorenov Dept. of Civil and Environmental Engr. Inst. for Chemical Physics Research University of Missouri-Columbia Russian Academy of Sciences Columbia, MO 65211-2200, USA Moscow Region, Russia

ABSTRACT Based on the previous research on modeling and simulation of the failure wave phenomenon as observed in shocked glasses, a numerical investigation is conducted here to simulate microcracking diffusion in sandwiched glass plates. The essential assumptions made are that the deviatoric elastic strain energy in the intact material is converted into the volumetric potential energy in the comminuted and dilated material during the time- and space-dependent microcracking diffusion process, and that each surface of the glass plates introduces additional microcracking sources. The simulation results appear to match the available experimental data very well in the loading phase. Future work is discussed based on the current results.

INTRODUCTION

Under plate impact, some brittle solids may undergo elastic deformations at the shock wave front, and fail catastrophically at a distinctly later time if the shock stress is near but below the apparent Hugoniot elastic limit (HEL). The phenomenon has therefore been interpreted as the result of a slowly propagating failure wave in the shocked solids. Since Brar et al.1 and Kanel et al.2 reported the formation and propagation of failure waves in shocked glasses, continued efforts have been made to explore this interesting physical phenomenon.3-16 However, no consensus can be made at the moment on the exact physics behind this failure wave phenomenon. Especially, there is a lack of consistent experimental data for developing a three-dimensional constitutive model that predicts the essential feature of failure wave, and that could be easily implemented into a numerical code for large-scale computer simulation. Also, the relationships among different wave structures are still not clear.



The shock response of glasses beyond the apparent HEL often displays a distinctive two-wave structure in wave profile. The trailing longitudinal stress wave is referred as the inelastic shock wave. However, in the shock wave experiments reported so far, no obvious jump in the longitudinal stress history has been detected at the failure wave front.1, 2, 4, 9 One interpretation is that the apparent HEL may not be a true elastic limit, rather the manifestation of a transition in failure mechanisms. A possible transition is the one from a delayed kinetic-controlled failure process below the HEL to a prompt stress-controlled failure process above the HEL.12 Another possibility is that the HEL may represent the stress level above which bulk glass undergoes permanent densification.9 From existing experimental data, however, it appears that the signature feature that separates the failure wave from the usual inelastic shock wave is that only the lateral stress history is changed significantly while the longitudinal stress history remains almost constant at failed (due to the loss of shear strength) material particles. In other words, the propagation of a “failure wave” might not be the result of momentum balance. The underlying mechanism might be a process governed by a field equation other than the stress wave equation. Recent experiments conducted by Dandekar7 revealed that the longitudinal stress measured on the impact surface of a shocked glass plate is different from the stress measured at some distance from the impact surface during the propagation of the failure wave. From this observation, therefore, the formation and propagation of failure waves appear to depend not only on the local state, but also on the information in the domain of influence, which is similar to localization problems. In other words, a nonlocal approach should be considered to describe the failure wave. It has been demonstrated that the initiation and evolution of localized material failure can be related to the transition between governing field equations of different types.17, 18 Representing a hyperbolic-to-elliptic transition with a parabolic (diffusion) equation and using a local elastoplasticity model, Xin and Chen19 obtained an analytical solution for a dynamic softening bar. A diffusing failure front could be simulated via the jump forms of conservation laws, together with a local elastodamage model.5 Changes of governing equation type also arise in many thermal and fluid mechanics problems. For example, depending on the ratio of thermal diffusivity to relaxation time, heat may propagate at a finite speed as a thermal wave or at an infinite speed (in the absence of relaxation) as a thermal diffusion.20, 21 Two different elliptic equations may hold respectively inside and outside of a turbulence domain.22 Hence, multi-physics as reflected through the transition between governing equations is not unusual. From the available experimental results on the failure wave phenomenon, an attempt has been made to construct a micromechanics-based picture for the evolution of failure waves.11 It has been proposed that under plane shock wave


loading, the material failure below the HEL occurs through simultaneousprocesses of heterogeneous microcracking, shear dilatancy and void collapsingunder high confining stresses, which result in an increase in the mean stress and a decrease in the deviatoric stress while all the longitudinal field variables remainunchanged. This particular form of failure initiates at the impact surface where thesurface defects and transient loading conditions are conducive for such a process,and propagates into the material bulk through progressive multiplication of microcracks, i.e., a percolation process with a certain threshold.

To develop an effective simulation procedure, a three-dimensional isotropiccontinuum damage model has been proposed based on the above micromechanics-based picture.23 The progressive percolation of micro-damage is described as a nonlinear diffusion process lagging behind the shock compression.Material dilatancy induced by shear microcracking is assumed and used toquantify the average intensity of damage. A unique feature of the proposed model is the postulation that the deviatoric elastic strain energy in the intact material is converted into the volumetric potential energy in the comminuted and dilated material during the time- and space-dependent failure evolution process. The twofield equations governing the elastic shock wave and the trailing damage diffusionare solved numerically via a staggered manner in a single computational domain.It appears that the simulations based on the proposed model and solutionalgorithm can predict the essential features of the stress histories associated withthe failure wave phenomenon as observed in plane shock wave experiments onsingle glass plates, with an assumed threshold. However, there is a lack ofunderstanding of the multi-physics and multi-scale effects on the initiation andevolution of dynamic structural failure. Especially, model parameters need becalibrated via consistent experimental data, and the change in the longitudinalstress profile, as observed in the experiments conducted by Dandekar,7 must be considered in the failure wave modeling.

Based on the recent experimental data of an aluminum target impacting on sandwiched glass plates, a numerical study is conducted here to simulate thelongitudinal stress histories measured at the copper-glass interface and the glass-glass interface, respectively.

CONSTITUTIVE MODELING AND DAMAGE DIFFUSIONFor the purpose of simplicity, a nonlinear elastic–perfectly plastic model is

used for metals, with an associated flow rule. The yield surface takes the form of

with 0sJ3f 2y2

pddss :

2

1J 2 , denoting the deviatoric stress tensor,

and being the yield strength. It can then be found that in the deviatoric space

ds

ys


ddd

de

ss

ssPs d:

:G2

dddd (1)

with Pd denoting the deviatoric orthogonal projector, and ed being the deviatoricstrain tensor. Based on the shock physics, the pressure-dependent shear modulusis given by

200

m0

c

b41GG (2)

in which G is the original shear modulus, represents the initial mass density,

and b and can be determined via the relationship between the shock wave

velocity and particle velocity U , i.e., U . In the volumetric

space, the mean compressive stress is related to the current specific volume V

through the following equation:

0

c

sU

0

0

p p0s bUc

14exp4 0

0200

V

VVb

b

cm (3)

with V0 being the initial specific volume. The material parameters for copper havethe following values: 0=8924kg/m

3, b=1.51, c0=3910m/s, sy=60Mpa, and G0=49GPa, while for aluminum the values are 0=2703kg/m

3, b=1.34,c0=5350m/s, sy=40Mpa, and G0=25GPa.

To be complete, the essential ideas of the previous constitutive model forfailure waves in shocked glasses 23 are summarized here, with an emphasis on the modifications made. The diffusion equation governing damage variable Vd in the3-D space x with time t can be written as

dd Vtt

V,xD (4)

where D(x,t) denotes the second order damage diffusivity tensor. If themicroscopic details of percolation in different orientations are not pursued, it isreasonable to let with i being the second order identity tensorand

ixxD tDt ,,


0

0

orif0

andif0,x

ddTHD

FHEL

F

ddTHD

VVYYYY

YYd

VVYY

tD (5)

where d is the diffusion coefficient, Y is the 2nd invariant of deviatoric stress, andsubscripts “F”, “HEL” and “THD” denote the values of the stress deviatorvariable in the completely failed but compressed material, at the HEL, and at thethreshold for initiating the failure process, respectively. Note that Y<YF during unloading.

It is assumed that the initial distribution of isolated microdamage sites near theimpact surface is sufficiently uniform and planar. As a result, the evolution offailure in a uniaxially shock-compressed material can be considered as a pseudo-3D process, in which the propagation of failure is described by longitudinaldiffusion supplemented with a time-dependent evolution function accounting forthe much faster lateral percolation of microdamage. Thus, we may approximateEqs. (4) and (5) by using

txQx

VtxD

xt

V dd ,, with 0,

, 0

d

dd

T

VV

d

txDtxQ (6)

where is the threshold below which is inactive, and T is the

characteristic time for the lateral evolution of microdamage at a givenlongitudinal location. The volumetric and deviatoric responses are modified to be

0dV txQ , d

m=45.36 e-137.0 e2 +208.3 e

3 and G=G0/[1+( x/ G)2], respectively, in unit of

GPa, with e=(V0+Vd)/V-1, x being the longitudinal stress, and G=5.0GPa and G0=30.43GPa for the glass material considered here.

SIMULATION AND DISCUSSIONThe experimental arrangement is shown in Fig.1. The tested material is a soda

lime glass with the density of 2450kg/m3 and the longitudinal sound speed of5.58km/s. The shock compression pulse was created by the impact of analuminum flyer plate which was launched by an explosive facility with a velocityof 1.17 0.05km/s. A series of numerical calculations based on the above materialmodels have been carried out to determine the material parameters that result inthe best simulation of the failure evolution characteristics as observed in theshocked sandwiched glass plates. Using Y=|s11-s22| as the stress deviator measure


Fig. 1. Configuration of impact experiment.

Str

ess

(GP

a)

Time (µs)

simulated

measured

Fig. 2. Time histories of simulated and measured longitudinal stresses at copper-glass and glass-glass interfaces, respectively.

Time (µs)

lateral

longitudinal

S

tres

s (G

Pa)

Fig. 3. Time histories of simulated longitudinal and lateral stresses atcopper-glass and glass-glass interfaces, respectively.


leads to the following values of the material parameters: YHEL=4.529GPa, YTHD=4.34GPa, and YF=0.75GPa for glass plate-1, and YHEL=4.529GPa, YTHD=3.76GPa, and YF=0.75GPa for glass plate-2. The parameters for damage diffusion in glass are of the following values: Vd0=5.0x10-7m3/kg, d=0.4m2/s, and Td=20ns. By assuming that each surface of the glass plates introduces additional microcracking sources as reflected through YTHD, the simulation results appear to match the available experimental data very well in the loading phase. However, the interactions among different kinds of waves in the unloading phase are still not clear from the experimental data available. An integrated experimental, analytical and computational study is needed to better understand the physics behind the impact failure responses of both single and composite materials.

ACKNOWLEDGMENT

This work was sponsored in part by National Science Foundation.

REFERENCES 1Brar, N.S., Bless, S.J., Rosenberg, Z., Impact-Induced Failure Waves in

Glass Bars and Plates. Applied Physics Letter 59, 3396-3398, 1991. 2Kanel, G.I., Rasorenov, S.V., Fortov, V.E., The Failure Waves and

Spallations in Homogeneous Brittle Materials. Shock Compression of Condensed

Matter–1991 (Edited by Schmidt, S.C., Dick, R.D., Forbes, J.W., Tasker, D.G.), Elsevier, 451-454, 1992.

3Bless, S.J., Brar, N.S., Impact Induced Fracture of Glass Bars. High-Pressure

Science and Technology (Edited by S.C. Schmidt, J.W. Shaner, G.A. Samana and M. Ross). AIP, New York, NY, 1813-1816, 1994.

4Bourne, N.K., Rosenberg, Z., Field, J.E., High-Speed Photography of Compressive Failure Waves in Glasses. Journal of Applied Physics 78, 3736-3739, 1995.

5Chen, Z., Xin, X., An Analytical and Numerical Study of Failure Waves. International Journal of Solids and Structures 36, 3977-3991, 1999.

6Clifton, R.J., Analysis of Failure Waves in Glasses. Applied Mechanics

Reviews 46, 540-546, 1993. 7Dandekar, D.P., Index of Refraction and Mechanical Behavior of Soda Lime

Glass under Shock and Release Wave Propagation. Journal of Applied Physics

84, 6614-6622, 1998. 8Dandekar, D.P., Beaulieu, P.A., Failure Wave under Shock Wave

Compression in Soda Lime Glass. Metallurgical and Material Applications of

Shock-Wave and High-Strain-Rate Phenomena (Edited by L.E. Murr, K.P. Staudhammer and M.A. Meyers). Elsevier Science B.V., 211-218, 1995.


9Espinosa, H.D., Xu, Y., Brar, N.S., Micromechanics of Failure Waves in Glass: Experiments. Journal of the American Ceramic Society 80, 2061-2073, 1997a.

10Espinosa, H.D., Xu, Y., Brar, N.S., Micromechanics of Failure Waves in Glass: Modeling. Journal of the American Ceramic Society 80, 2074-2085, 1997b.

11Feng, R., Formation and Propagation of Failure in Shocked Glasses. Journal

of Applied Physics 87, 1693-1700, 2000. 12Grady, D.E., Shock-Wave Properties of Brittle Solids. Shock Compression

of Condensed Matter–1995 (Edited by Schmidt, S.C. and Tao, W.C), AIP, 9-20, 1996.

13Partom, Y., Modeling Failure Waves in Glass. International Journal of

Impact Engineering, 21, 791-799, 1998. 14Raiser, G., Clifton, R.J., Failure Waves in Uniaxial Compression of an

Aluminosilicate Glass. High-Pressure Science and Technology (Edited by Schmidt, S.C., Shaner, J.W., Samana, G.A., and Ross M.), AIP, 1039-1042, 1994.

15Raiser, G.F., Wise, J.L., Clifton, R.J., Grady, D.E., Cox, D.E., Plate Impact Response of Ceramics and Glasses. Journal of Applied Physics 75, 3862-3869, 1994.

16Rosenberg, Z., Bourne, N.K., Millett, J.C.F., Direct Measurements of Strain in Shock-Loaded Glass Specimens. Journal of Applied Physics 79, 3971-3974, 1996.

17Chen, Z., Continuous and Discontinuous Failure Modes. Journal of

Engineering Mechanics 122, 80-82, 1996.18Chen, Z., Sulsky, D., A Partitioned-Modeling Approach with Moving Jump

Conditions for Localization. International Journal of Solids and Structures 32,1893-1905, 1995.

19Xin, X., Chen, Z., An Analytical Solution with Local Elastoplastic Models for the Evolution of Dynamic Softening. Int. J. Solids and Structures 37, 5855-5872, 2000.

20Tzou, D.Y., On the Thermal Shock Wave Induced by a Moving Heat Source. ASME Journal of Heat Transfer 111, 232-238, 1989.

21Tzou, D.Y., Macro- to Microscale Heat Transfer: The Lagging Behavior,Taylor & Francis, Washington, DC, 1997.

22Chen, Z., Clark, T., Some Remarks on Domain-Transition Problems. Archives of Mechanics 47, 499-512, 1995.

23Chen, Z., Feng, R., Xin, X., Shen, L., A Computational Model for Impact Failure with Shear-Induced Dilatancy. Submitted for publication in International

Journal for Numerical Methods and Engineering, 2001.


GRAIN LEVEL ANALYSIS OF CERAMIC MICROSTRUCTURES

SUBJECTED TO IMPACT LOADING

Pablo D. Zavattieri and Horacio D. Espinosa

Mechanical Engineering

Northwestern University

Evanston, IL 60208

ABSTRACT

A study on the accuracy of cohesive models for capturing dynamic

fragmentation of ceramic microstructures is presented. The investigation consists

of a combined experimental/numerical approach in which microcracking and

damage kinetics are examined by means of plate impact recovery experiments.

The numerical analysis is based on a 2-D micromechanical stochastic finite

element analysis. The model incorporates a cohesive law to capture microcrack

initiation and evolution as a natural outcome of the calculated material response.

Normal plate impact velocity histories are used not only to identify model

parameters, but also to determine under what conditions the model captures

failure mechanisms experimentally observed. The analyses show that in order to

capture damage kinetics a particular distribution of grain boundary strength and

detailed modeling of grain morphology are required.

INTRODUCTION

Critical elements in the development of a physically-based model of the

dynamic deformation and failure of ceramics requires experiments specically

designed to examine inelasticity. For instance, to study the initiation and

evolution of microcracks in ceramics, an experiment that can cause controlled

microcracking, under well defined stress conditions, was developed by Clifton

and co-workers [1, 2]. These investigators performed plate impact soft recovery

experiments by subjecting the central region of a square ceramic specimen to

known and controllable stress pulses. Microcracking resulted yet the specimens

were recovered intact for microscopic analysis.

A large portion of the microcracks was found to originate at triple points and

both inelasticity in compression and tension was interferometrically measured. In

the tension dominated region, several microcracks linked together to form a spall



plane perpendicular to the impact direction. In spite of these contributions to the

field of damage, lack of consensus on the mechanisms responsible for ceramic

failure under multi-axial dynamic loading still remains. Attempts have been made

to model the inelastic constitutive behavior of ceramics in the presence of cracks,

and to validate the models through simulation of plate and rod impact

experiments. Available models for the failure of ceramics are continuum damage

theories which are based on homogenizing the cracked solid and finding its

response by degrading the elasticity of the material, and discrete approachs [3, 4],

able to nucleate cracks, and follow their propagation and coalescence during the

deformation process, the influence of microscopic heterogeneities on the overall

material behavior, which depends on morphological characteristics such as size,

shape, lattice orientation and spatial distribution of grains, is not accounted for.

In order to provide powerful tools to understand the mechanisms that lead to

macroscopic failure and, at the same time, refine the theories of damage utilized

in continuum or continuum/discrete models, a 2-D micromechanical model is

presented to assess intergranular microcrack initiation and evolution. A

representative volume element (RVE) of an actual microstructure, subjected to

multi-axial dynamic loading, is considered for the different analyses. A large

deformation elastic-anisotropic visco-plasticity model for the grains,

incorporating grain anisotropy by randomly generating principal material

directions, is included. Cohesive interface elements are embedded along grain

boundaries to simulate intergranular fracture through microcrack initiation and

evolution. Their interaction and coalescence are a natural outcome of the

calculated material response.

This micromechanical model provides explicit account for arbitrary

microstructural morphologies and microscopic fracture patterns making it easier

to identify and design microstructural configurations that enhance fracture

toughness, and therefore lead to improvements in the manufacturing of ceramic

materials. A detailed study of the damage initiation and kinetics in soft-recovery

experiments is carried out.

The objective of this work is to provide tools and means to understand the

macroscopic inelastic response of ceramics when subjected to dynamic multi-

axial loading at the micron scale. This bridging between scales is achieved by a

micro-mechanical stochastic finite element model. Experiments are not only used

to examine and validate the micromechanical model but also to explain the

different failure mechanisms.

MICROMECHANICAL MODEL

The finite element analysis of the initial boundary value problem is performed

using a total Lagrangian continuum approach with a large deformation elastic and

thermal anisotropic visco-plastic model [5, 6]. The elastic and thermal anisotropic


model is used to describe grains' single crystal anisotropic behavior. Each grain is

assumed to be elastic orthotropic and the orientation of the principal material

directions differs from grain to grain.

(a)

(b)

Figure 1: (a)Schematics of microcracking at grain boundaries using an irreversible

interface cohesive law.w (b) Soft-recovery normal impact configuration.


A multi-body contact-interface algorithm is used to describe the kinematics at

the grain boundaries and to simulate crack initiation and propagation. Figure 1

describes the contact model, integrated with interface elements to simulate

microcracking at the grain boundaries and subsequent large sliding, opening and

closing of the interface. The tensile and shear tractions in the zero thickness

interface elements, embedded along grain boundaries, are calculated from the

interface cohesive law. The interface cohesive law describes the evolution of

these tractions in terms of both normal and tangential displacement jumps. Within

the framework of cohesive interface elements the two most noteworthy cohesive

failure models available in the literature are the potential-based law [7], and the

linear law [3]. More detail on the cohesive model used in this work can be found

in the following references [5, 6].

SOFT-RECOVERY IMPACT EXPERIMENTS

The “soft-recovery” plate impact experiment has been described in detail by

Raiser et al. [1], and Espinosa et al. [2]. The experiment uses an eight pointed

start-shaped flyer plate that impacts a square ceramic specimen, subjecting the

central octagonal region to a plane pulse. Figure 1(b), shows this soft-recovery

normal impact configuration. A tensile pulse is originated from a gap between the

specimen and the momentum trap upon reflection of the compressive pulse. The

velocity-time profiles recorded at the rear surface of the momentum trap plate

provide information on microcrack initiation and evolution.

Let x denote the distance from the front surface of the specimen measured in

the direction of impact, and let Ls denote the thickness of the specimen, Lf the

thickness of the flyer and LMT the thickness of the momentum trap. The particle

velocity induced in the rear surface of the momentum trap is measured as a

function of time by a normal displacement interferometer (NDI) and a normal

velocity interferometer (NVI).

In the case of brittle materials readily damaged in tension, the tensile region

becomes the likely site of substantial damage called spall region. When spallation

initiates, the release waves emitted from the newly created free surfaces

completely change the pattern of waves inside the specimen. The shape of the

pull-back signal and second compressive pulse reflects both microcracking, under

the tensile pulse itself, and attenuation while traveling through material already

damaged. The above one-dimensional analysis is valid in the central region of the

specimen, where the effects of diffracted waves from the corners and the edges of

the flyer are minimized [2]. The experimental findings suggested that the

modeling of crack nucleation and growth requires consideration not only of the

amplitude of the applied stress but also of its time dependence [2]. Several

successful tests have been conducted using this experimental design by Espinosa


et al. [2] and Raiser et al. [1]. A summary of the shots used for comparisons with

the proposed numerical model can be found in Zavattieri and Espinosa, 2001 [8].

STOCHASTIC FEM SIMULATIONS

A representative volume element of an actual microstructure is considered for

the analysis. Although the exact grain geometry can be taken from a digital

micrograph, it is well established that the grain structure in polycrystalline

materials can be simulated by a Voronoi tessellation [6]. We followed the last

approach to generate enough statistical data.

Figure 2(a) shows a strip of the various plates used in the experimental

configuration, only the flyer, momentum trap and specimen are considered in the

analysis and due to the limited spread of tensile damage observed experimentally,

only a small portion of the ceramic in the spall region is simulated. The top and

bottom boundaries of the cell are modeled using viscous boundary conditions

which represent the exact elastic wave solution along characteristic lines. Details

on the boundary conditions and convergence can be found in [6].

ANALYSIS OF THE SOFT-RECOVERY IMPACT EXPERIMENTS

In order ot simulate these experiments, two important features were

incorporated in the simulation of the experiments, namely, (1) a Weibull

distribution of the interfacial strength and fracture toughness along the grain

facets. (1) Realistic microstructures considering grain morphology and size

distributions.

As discussed in [5, 6, 8], it is physically incorrect to select a uniform Tmax and

KIC for all grain facets. Not only that grain misorientation affect the interfacial

strength, but also it affect the presence of glassy phase, glass pockets, and other

impurities that modify grain boundary properties. In the following analyses, the

interfacial strength parameters will be described by a Weibull distribution.


(a)

(b)

Figure 2. (a) Schematics of the computational cell used for the analyses. (b)

Experimental particle velocity versus time for one (Shot 88-04) of the

experiments performed by Espinosa et al. [2].


Simulation of experiment 88-04

Figure 2(b) shows the experimental particle velocity vs time for shot 8804

performed by Espinosa et al. [2]. The impact velocity used in shot 88-04 was V0 =

48.4m/s. Figure 2(b) shows the experimental velocity history for this experiment.

The elastic solution is also shown in the same figure. The most significant

features of this experiment is the pullback signal (almost 30% of the maximum

stress) and the spreading of the second compressive pulse. Numerical simulations

using the microstructure shown in Figure 2(a) result in a pull-back signal with a

maximum stress equal to the first compressive pulse, which is well above the pull-

back signal measured experimentally. Once microcracks nucleate, they grow at

rates such that a major crack from side to side of the RVE develops. It is worth

noticing that even for the case in which there is only one nucleation site every

200µm, the crack has to propagate to the other side of the RVE, in more than 67

nanoseconds (tensile pulse duration), in order to have a pullback signal below

100% of the compressive pulse. This would require a crack speed of less than

50% of the Rayleigh wave speed, which for alumina is 3 mm/µs, or a delay in the

decohesion process produced by rate effects. From the SEM Micrographs [2], it is

observed that the microcracks need to follow grain boundaries, with large

variations in grain size. The net effect is that crack propagation speed on a

projected horizontal plane is reduced to a fraction of the Rayleigh wave speed.

We closely examine this feature in conjunction with the observation of possible

nucleation sites as a function of overstress from the threshold level.

Two microstructures are considered in this analysis. Both meshes have a

width of 300 µm such that if there is only one nucleation site, the crack will have a

total time equal to the pulse duration to coalesce into a main crack. The main idea

of this analysis is to compare vis-à-vis the crack propagation for two different

types of microstructures: Microstructure A, with a non-uniform distribution of

grain sizes and shapes (motivated from the micrographs), and microstructure B

with a uniform distribution of grains (all with the same size and similar shape).

Figure 3(a) shows in detail the pullback signal for simulations considering

microstructure A. Microstructures A and B are shown in Figure 4. In these

simulations three different Weibull distributions have been considered. The best

fit is obtained for a Weibull distribution with T = 5 GPa, = 2 MPa · m0

max

0

ICK1/2

and m =3. This distribution contains interface elements with Tmax = 0.5 GPa and

Tmax 10GPa. The same distributions have been considered for microstructure B,

see Figure 3(b), and the pullback signals are much more pronounced than those

obtained with microstructure A. An explanation can be inferred by examining the

evolution of crack patterns as shown in Figure 4. The evolution of the

microcracks for T = 5 GPa, = 2 MPa · m0

max

0

ICK1/2

and m =3 using mesh A is

shown in Figure 4(a); the grain morphology is shown in the first frame. In this


(a)

(b)

Figure 3: (a) Comparison between three different Weibull distribution for shot 88-

04 using mesh A. (b) Comparison between the velocity history using meshes A

and B.


(a)

(b)

Figure 4: (a) Evolution of crack pattern for the case with T = 5GPa and m =3

using mesh A (b) Evolution of crack pattern for the case with T = 5GPa and

0

max

0

max

m =3, using mesh B.


case, it can be observed that the microcracks need to go around the large grains at

the center of the RVE. The time that it takes the crack to surround the large grains

is similar to the pulse duration and then the pullback signal is significantly lower

than for cases where the crack propagates, from one side to the other of the RVE,

at uniform speed. Figure 4(b) shows the crack evolution for the case with

microstructure B. The crack initiates almost in the middle of the RVE and

propagates at constant speed until it coalesces into a main crack just before the

tensile pulse vanishes. As a result, the pullback signal for this case is much higher

than that for the case where the crack is forced to follow a path around large

grains.

Higher impact velocities

In this subsection we examine an experiment with higher impact velocity. The

experiment (shot 92-11) has been reported by Raiser et. al [1], the initial velocity

was V0 = 92.3m/s. The main variation in this experiment is the average grain size

of the ceramic, Coors AD-999, of approximately 3 µm. For this analysis an RVE

of 200 x 200µm is considered and two type of microstructures, uniform and bi-

modal grain sizes, are analyzed. The main motivation for examining two different

microstructures is to study the effect of grain morphology and how this affect the

crack path and crack speed along the spall plane. Although the second

microstructure with a bi-modal distribution of grain sizes may not be totally

representative of the tested ceramic, it is used to evidence the effect of grain

morphology.

An analysis with three different Weibull distribution on the RVE with uniform

grain size has been carried out; weak interface case: T = 3GPa and m =3; the

case considered in previous experiments, i.e., T = 5GPa and m =3; and a strong

interface case: T = 10GPa and m =10. The intention of this analysis is not to

study parametrically the effect of m, or T

0

max

0

max

0

max

0

maxT

max. In all cases KIC= 2 Mpam1/2

. Figure 5

shows the crack pattern for each one of these cases; the grain morphology is

shown in the first frame. In the weak interface case, the ceramic fails from side to

side right after the tensile pulse is generated at the spall plane. Crack nucleation

occurs basically at a large percentage of triple points and coalescence of

microcracks occurs before the end of the tensile pulse. For the case with T =

5GPa and m =3 the crack start propagating from the center to the borders and

crack branching in the form of a “funnel” is observed. As it is expected, the

strongest case ( = 10GPa and m =10) shows less branching and microcrack

density. The energy to create new surfaces is higher so that branching is inhibited.

0

max


Figure 5: Velocity history for Shot 92-11 using a microstructure with a uniform

distribution of grain size and three different Weibull distributions.

DISCUSSION

The micromechanical analyses, together with the experimental velocity

profiles and SEM observations, have demonstrated that there are two factors to be

taken into account to capture the right damage kinetics occurring during the

experiments. In view that not all grain facets have the same interface strength and

local fracture toughness, it is important to consider Weibull distributions of Tmax

and KIC. Similarly, since the ceramic microstructures interrogated in these

experiments do not contain grains with the same shape and size, microstructures

with non-uniform distributions of grain size and shape must be considered. On the

other hand, microstructures with non-uniform distribution of grain size and shape

strongly affect crack speed along the spall plane.

From a computational standpoint, simulations of ballistic penetration, vehicle

crash analysis, manufacturing processes, etc. cannot be conducted at the grain

level. Hence, this fundamental study of brittle failure provides insight into the


utilization of cohesive laws at other size scales. Our simulations clearly show that

the scale at which simulations are performed plays an important role in the

selection of cohesive models. The calculations in this work make assumptions that

limited the degree of achievable accuracy. For instance, the model is two-

dimensional and crack interaction is stronger than in the 3-D case and therefore,

the computed rate of crack coalescence may be thought of as an upper bound.

Despite these limitations, the numerical results obtained with this model were not

only in good agreement with the experiments, but also were used to explain

several microscopic failure mechanisms that have never been quantified before

through other mathematical models.

REFERENCES

1. Raiser G., Wise J.L., Clifton R.J., Grady D.E., and Cox D.E., “Plate impact

response of ceramics and glasses”, J. Appl. Phys., 75(8):3862-69,1994.

2. Espinosa H.D., Raiser G., Clifton R.J., and Ortiz M., “Experimental

observations and numerical modeling of inelasticity in dynamically loaded

ceramics”, J. Hard. Mat., 3:285-313, 1993.

3. Camacho G.T. and Ortiz M. “Computational modeling of impact damage in

brittle materials”, Int. J. Sol. Str., 33: 2899-2938, 1996.

4. Espinosa H.D., Zavattieri P.D., and Dwivedi S., “A finite deformation

continuum/discrete model for the description of fragmentation and damage in

brittle materials”, J. of the Mechanics and Physics of Solids, 46(10): 1909-1942,

1998.

5. Zavattieri P.D., Raghuram P.V., and Espinosa H.D., “A computational model

of ceramic microstructures subjected to multi-axial dynamic loading”, J. of the

Mechanics and Physics of Solids, 49(1): 27-68, 2001.

6. H.D. Espinosa and P.D. Zavattieri, “A grain level model for the study of

dynamic failure of polycrystalline materials. Part I: Theory and numerical

implementation”, Submitted to Mechanics of Materials, 2001.

7. Xu X-P and Needleman A., “Numerical simulation of dynamic interfacial

crack growth allowing for crack growth away from the bond line”, Int. J. Fra.,

74:253-275, 1995.

8. P. D. Zavattieri and H. D. Espinosa, “Grain level analysis of crack initiation

and propagation in brittle materials”, In press Acta Materialia, 2001.


ANALYSIS AND MODELLING OF CERAMIC ARMOUR PENETRATION

S.J. Cimpoeru and R.L. Woodward†

DSTO Aeronautical and Maritime Research Laboratory,

P.O. Box 4331, Melbourne, 3001, Australia.

ABSTRACT

This paper summarises fragmentation and energy absorption studies

conducted on a wide range of ceramic armour materials. Aspects of ceramic

armour depth of penetration tests are examined and how such data can be

interpreted and depend on test configuration. The Woodward one-dimensional

momentum balance model is also outlined and is used to represent some of the

characteristics of depth of penetration tests, including the importance of the

destruction of the penetrator nose, erosion of the remainder of the projectile and

the derivation of effective ceramic strength values.

FRAGMENTATION AND ENERGY ABSORPTION STUDIES

Quantitative fragmentation studies and ballistic performance measurements

were made on ceramics of a wide range of hardnesses and toughnesses, including

soda lime glass and zirconia, and on ceramics with similar measured mechanical

and physical properties but different ballistic performances. An inverse

correlation was found between between the volume of fragments produced and

fracture toughness [1,2]. It is of interest that any variation in fragmentation

behaviour due to small differences in toughness was masked by the shot to shot

inconsistency in the results and that significant variations in fragmentation

occurred despite similar residual depths of penetration. However, no correlation

was established between toughness and ballistic performance, which was to be

expected as measurements of the surface area of fractured ceramics and

calculations of fracture work [3,1] demonstrated that very little of the initial

projectile kinetic energy goes into fracturing the ceramic. A large proportion of

this energy simply ends up as the residual kinetic energy of the ejected ceramic

debris [3,1].

A marked difference was found in the fragmentation behaviour of blunt and

pointed projectiles, but this depended on whether the hardness of the ceramic was

†Deceased



sufficient to blunt the projectile upon impact [1]. For pointed projectiles

impacting soft ceramics, where the projectile remains undeformed, the ballistic

performance was found to increase with ceramic hardness. For hard ceramics the

residual penetration depths were generally similar due to the ability of these

ceramics to blunt pointed projectiles. Thus the ballistic performance of the hardest

ceramics is not simply related to hardness, but more related to the velocity

reduction and erosive mass loss incurred by the blunted projectile.

A SIMPLE ONE-DIMENSIONAL APPROACH TO MODELLING CERAMIC

COMPOSITE ARMOUR DEFEAT

Woodward [4,5] developed two one-dimensional momentum balance models

to highlight the essential physical processes of ceramic armour defeat. One model

was for the perforation of targets with thin backings which under the influence of

the ceramic fracture conoid deform by dishing, i.e. stretching and bending

deformation. A second model, a simplification of the thin backing model, was

developed for thick backings where the backing remains stationary whilst the

ceramic is eroded, allowing direct analysis of depth of penetration test results.

A analysis of the early work of Wilkins [6] concluded that the resistance to

penetrator motion was initially determined by the inertia of the ceramic and

backing that was bounded by the conoidal cracking of the ceramic and dishing of

the backing plate [4], i.e. a target acceleration stage. A second failure stage of

perforation was also identified where the penetrator and target material bounded

by the cone, moved forward at a common velocity until they are either slowed to

zero velocity or the backing plate ruptures via biaxial tensile failure [4]. An

additional failure case is also where the ceramic is eroded to zero thickness and

the backing plate is perforated according to a simple failure criterion [7].

Figure 1 shows the lumped mass model for the thin backing model [4]. In any

time step, t, a mass MP and a mass MC are eroded from the projectile and

ceramic, respectively, if the interface forces exceed the strength of the projectile

or ceramic, i.e. FI either exceeds FP or FC or both. Once ceramic erosion has

occurred, the area of the backing that is loaded is reduced to account for the

reduced mass distribution of ceramic in the fractured conoid, with a consequent

reduction in energy absorption.

Figure 1: Basic concepts of the lumped mass model [4].


The model predictions were originally validated against the experimental results

of a number of other workers [4,5] and further validations have occurred in recent

years, e.g. [8].

INTERPRETATION OF DEPTH OF PENETRATION TEST DATA

Depth of penetration testing has become a popular low-cost means of

evaluating the ballistic performance of the wide range of ceramic options that are

now available. However, there are some disadvantages to this form of testing,

including the interpretation of the results in terms of the fundamentals of

penetration mechanics and a clear understanding of what material characteristics

are being measured. There are also a number of commonly used weight and space

merit ratings; understanding the meaning and therefore the limitations of the

ratings is as important as the values of the ratings themselves [9].

Depth of penetration test results are configuration dependent. Table I lists

residual depth of penetration results and derived ballistic and penetration

efficiency values for tungsten alloy and hardened steel projectiles, impacting

laterally confined alumina tiles that have different backing materials [10].

Table I:Depths of penetration and efficiency values for 99.5% alumina-faced targets [10].

WP—pointed tungsten alloy; WB—blunt tungsten alloy; SP—pointed steel; SB—blunt

steel.

ProjectileTarget

Backing

Reference

Depth

(mm)

Residual

Depth

(mm)

Ballistic

Efficiency [11]

Penetration

Efficiency [12]

WP AL 265 66 16.9 3.4

WB AL 75 57 1.5 1.1

SP AL 85 28 4.8 2.1

SB AL 62 20 3.6 2.0

WP RHA 34 20 3.4 1.4

WB RHA 23.5 14 2.3 1.3

SP RHA 30 2.5 6.7 4.6

SB RHA 10 3 1.7 1.4

WP HHS 23.5 13.7 2.4 1.3

WB HHS 23 13 2.5 1.3

SP HHS 19 1.5 4.3 3.4

SB HHS 11.5 3 2.1 1.6

Figure 2(a) shows the most stark of the comparisons of Table I in the form

suggested by Rosenberg and Yesherun [11] for the determination of ballistic

efficiency values, which equates to the slope of such plots. The standard tungsten

round penetrates the aluminium reference target without deforming, whereas the

truncated round of similar velocity and mass is a much less efficient penetrator


because it deforms and erodes during its penetration into the reference target.

While a marked difference is seen between the reference penetration depths of

pointed and blunt penetrators, the penetration depth of both rounds is similar

when fired through ceramic tiles, because the immediate effect of impact onto the

hard ceramic is the destruction of the nose of the standard, pointed round. Thus in

both cases a blunt penetrator of similar geometry effectively impacts the ceramic,

in one case the blunt penetrator is pre-formed, and in the other case it is formed

by impact. The use of both pointed and blunt projectiles therefore allows the

magnitude of the nose fracture effect to be separated from velocity reduction and

erosive mass-loss effects.

Figure 2: Plots of AlhAl against chc: (a) for tungsten alloy projectiles, pointed,

, and blunt, ■, fired into 99.5% alumina-faced, aluminium-backed targets [10]

and (b) for hardened steel projectiles fired into AD85 alumina-faced, aluminium-

backed targets [11]. The two calculated intercepts in (b) are joined to obtain a

straight, dashed line to compare with the empirical result .

(b)

AlhAl

(kgm-2

)AlhAl

(kgm-2

)

chc (kgm-2

)

Experimental

(a)

chc (kgm-2

)

Some progress has been made in understanding behaviour by exercising models

of penetration mechanics in comparison with the experiments. Figure 2(b)

compares the data of Rosenberg and Yesherun [11] with calculations which

required three models. The penetration through the ceramic was calculated using

the Woodward momentum balance model [4,5]. A separate deep penetration

model for deformable, blunt penetrators was used to calculate the small residual

penetration into the backing material [13,14], which also established the intercept

on the abscissa. The ceramic strength was taken as 5.6 GPa, which is less than the


Vickers Hardness, 8.8 GPa, but chosen to give a reasonable correspondence with

the results of Rosenberg and Yesherun [11].

The depth of penetration into the reference target was calculated using a

model for ductile hole formation by a non-deforming projectile [7]. An alternative

calculation for reference penetration was made using the deep penetration model,

which assumes a blunt projectile, and as shown it gives a much smaller reference

penetration. This illustrates the large effect on results of blunting the projectile tip,

particularly when an aluminium backing is used, and a comparison between

Figures 2(a) and 2(b) shows that the magnitude of the blunting effect observed in

experiments can be predicted with simple models.

Figure 2(b) illustrates that care should be exercised in using the slopes of such

curves to estimate ballistic efficiencies. The straight lines are usually constructed

from one point for the reference target and a few points from ceramic faced

targets that have little residual penetration into the backing armour with no

substantial data set in between. The relationship is not necessarily linear,

particularly at applique areal densities that are near to preventing penetration into

the reference target.

Table I also shows why ballistic efficiency values should be critically

examined and understood. For example, the ballistic efficiency value of 16.9

obtained for pointed tungsten alloy projectiles against an aluminium backing,

should be compared to the corresponding value of 1.5 for a blunt projectile. While

there is a significant difference between these merit ratings, the real effect of the

applique is almost identical in terms of residual penetration capability. Another

example is that given that the relative density of steel and alumina is 2.0, it

follows that a ballistic efficiency close to this value would imply similar

resistance per unit thickness for steel and alumina targets as also shown by Senf,

et al. [12]. It is seen that a number of ballistic efficiencies are close to this value

for the steel targets listed in Table I. Clearly the ceramic has not achieved the full

potential of its strength, despite its much greater hardness.

DERIVED VALUES FOR CERAMIC STRENGTH

Perhaps the greatest uncertainty with the application of one-dimensional

models to ceramic armour penetration problems is the difficulty in obtaining a

unique easily measured parameter, related to material strength, that indicates

resistance to penetration. While the indentation hardness of the ceramic is not

only easily measurable but also a physically meaningful strength parameter, other

measures of strength may be more suitable. Sternberg [15], for instance, indicated

that the penetration resistance is initially governed by the indentation hardness but

then drops to some lower value when cracking precedes penetration, the strength

parameter possibly being affected by ceramic toughness and confinement. It has

also been suggested that the appropriate strength value is a function of velocity


[16], being low at low velocities where cracking can preceed penetration, but

higher as velocity increases closer to the rate of propagation of the damage front

[17].

The momentum balance model developed by Woodward [4,5], and

independently for finite thickness targets by den Reijer [18], bears a close

similarity to the modified hydrodynamic theory of penetration [19-21] which is

given by the equations

(1a)p p

Lu

(1b)L p u

p p c cu p p

1

2

2 1

2

2 (1c)

where p and c are the penetrator and target densities, respectively, L is the

penetrator length, p the depth of penetration, u the penetrator velocity, and p and

c are the strengths of penetrator and target, respectively. Formulating the

Woodward momentum balance model in the same way leaves equations 1(a) and

1(b) as they are and the third equation becomes

(2) p p c c

u p u p2

The similarity between equations (1c) and (2) is such that the methods give

similar results with different (but proportional) values for target strength.

Figure 3: Derived ceramic strength data for various impact velocities and tile

thicknesses: 20 mm, ◆; 25 mm, ■; 30 mm, ; 40 mm, ; and 50 mm, .

0

1

2

3

4

5

0 5 10 15 20

Momentum Balance (GPa)

Modif

ied H

ydro

dynam

ic (

GP

a)

1250 ms-1

1000 ms-1

1500 ms-1


Despite the success in using such one-dimensional models in Figure 2(b), it is still

informative to analyse the velocity dependent data of Senf et al. [12] on the

penetration of 99.5% alumina-faced semi-infinite RHA targets. Figure 3 compares

ceramic strength values required by the models to match the data at impact

velocities between 1000 and 1500 ms-1

, with the deep penetration model [13,14]

being used to calculate the residual penetration into the backing armour. The

strength values for the two models are proportional to each other over a range of

tile thicknesses, but should be compared to the hardness of the alumina, 12.6 GPa

[12], and its HEL, 6.67 GPa [22]. Importantly, the proportionality is also found to

depend on impact velocity. Calculations such as this indicate that the models are

to some extent incomplete and as such should be applied with care.

CONCLUSION

A general inverse correlation has been found between the degree of ceramic

fragmentation and fracture toughness. It has been found also that a negligible

proportion of projectile kinetic energy is converted into surface area while a large

proportion of this energy ends up as the residual kinetic energy of the ejected

ceramic debris. Depth of Penetration (DOP) testing and the associated efficiency

ratings do not measure a fundamental armour property because they are test

configuration dependent and are usually strongly influenced by the fracture of the

projectile nose. The use of DOP data for real armour designs therefore requires a

close correspondence between the test set-up and the expected service

configuration. Current one-dimensional analytical models are able to represent

some of the characteristics of DOP tests such as the approximate calculation of

ballistic efficiencies. However, accurate prediction of performance over a range of

impact conditions requires a better understanding of ceramic strength effects and

so at present such models should be used with care.

REFERENCES

1. R.L. Woodward, W.A. Gooch, Jr, R.G. O’Donnell, W.J. Perciballi, B.J.

Baxter and S.D. Pattie, “A Study of Fragmentation in the Ballistic Impact of

Ceramics,” Int. J. Impact Engng, 15, 605-618 (1994).

2. R.G. O’Donnell, “An Investigation of the Fragmentation Behaviour of

Impacted Ceramics,” J. Mat. Sci. Lett., 10, 685-688 (1991).

3. R.L. Woodward, R.G. O’Donnell, B.J. Baxter, B. Nicol and S.D. Pattie,

“Energy Absorption in the Failure of Ceramic Composite Armours,” Materials

Forum, 13, 174-181 (1989).

4. R.L. Woodward, “A Simple One-Dimensional Approach to Modelling

Ceramic Composite Armour Defeat,” Int. J. Impact Engng, 9, 455-474 (1990).


5. R.L. Woodward, “A Basis for Modelling Ceramic Composite Armour

Defeat,” Materials Research Laboratory, Australia, Research Report MRL-RR-3-

89 (1989).

6. M.L. Wilkins, “Computer Simulation of Penetration Phenomena”; pp. 225-

252 in Ballistic Materials and Penetration Mechanics. Edited by R.C. Laible.

Elsevier, Amsterdam, 1980.

7. R.L. Woodward, “The Penetration of Targets by Conical Projectiles,” Int. J.

Mech. Sci., 20, 349-359 (1978).

8. R. Zaera and V. Sánchez-Gálvez, “Analytical Modelling of Normal and

Oblique Ballistic Impact on Ceramic/Metal Lightweight Armours,” Int. J. Impact

Engng, 21, 133-148 (1998).

9. S.J. Cimpoeru, B.J. Baxter and R.L. Woodward, “Some Extensions of

Simplified Ballistic Test Procedures to Comparative Protection Analysis,” Proc.

16th Int. Symp. on Ballistics, ADPA, San Francisco, CA, USA. 23-27 Sept. 1996,

3, 17-26 (1996).

10. R.L. Woodward and B.J. Baxter, “Ballistic Evaluation of Ceramics:

Influence of Test Conditions,” Int. J. Impact Engng, 15, 119-124 (1994).

11. Z. Rosenberg and Y. Yesherun, “The Relation Between Ballistic

Efficiency and Compressive Strength of Ceramic Tiles,” Int. J. Impact Engng, 7,

357-362 (1988).

12. H. Senf, E. Strassburger, H. Rothenhäusler, W.A. Gooch and M.S.

Burkins “Ballistic Resistance of AD995 Al2O3 Ceramics against Short Projectiles

at Impact Velocities Between 1000 and 2000 m/s,” Proc. 15th Int. Symp. on

Ballistics, ADPA, Jerusalem, Israel. 21-24 May 1995, 1, 361-376 (1995).

13. R.L. Woodward, “Penetration of Semi-Infinite Metal Targets by

Deforming Projectiles,” Int. J. Mech. Sci., 24, 73-87 (1982).

14. R.L. Woodward, “Modelling Penetration by Slender High Kinetic Energy

Penetrators,” Materials Research Laboratory, Australia, Report MRL-R-811

(1981).

15. J. Sternberg, “Materials Properties Determining the Resistance of

Ceramics to High Velocity Penetration,” J. Appl. Phys., 65, 3417-3424 (1989).

16. Y. Partom and D. Littlefield, “Dependence of Ceramic Armor Resistance

on Projectile Velocity”, Proc. 14th Int. Symp. on Ballistics, ADPA, Québec,

Canada. 26-29 Sept. 1993, 2, 563-572 (1993).

17. Hornemann, H. Rothenhäusler, H. Senf, J.F. Kalthoff and S. Winkler,

“Experimental Investigation of Wave and Fracture Propagation in Glass Slabs

Loaded by Steel Cylinders at High Impact Velocities”; pp. 291-298 in Mechanical

Properties at High Rates of Strain. Edited by J. Harding. Institute of Physics

Conference Series No. 70, Institute of Physics, Bristol and London, 1984.

18. P.C. den Reijer, “Impact on Ceramic Faced Armour,” Doctoral Thesis,

Delft University of Technology, The Netherlands, (1994).


19. V.P. Alekseevkii, “Penetration of a Rod into a Target at High Velocity,”

Fiz. Goren. Vzryva, 2, 99-106 (1966).

20. A Tate, “A Theory for the Deceleration of Long Rods After Impact,” J.

Mech. Phys. Sol., 15, 387-399 (1967).

21. A. Tate, “Further Results in the Theory of Long Rod Penetration,” J.

Mech. Phys. Sol., 17, 141-150 (1969).

22. D.P. Dandekar and P. Bartkowski, “Strength of AD995 Alumina under

Impact Loading,” 1994 Army Science Conf., Orlando, Florida, June 1994.


OVERVIEW OF THE RAJENDRAN-GROVE CERAMIC FAILURE MODEL

D. J. Grove A. M. Rajendran

U. S. Army Research Laboratory U. S. Army Research Laboratory

APG, MD 21005-5067 ARO, RTP, NC 27709-2211

ABSTRACT

This paper provides an up-to-date detailed description of the Rajendran-Grove

(RG) ceramic failure model. Damage initiates and evolves when the stress state

satisfies either a generalized Griffith criterion or a spall criterion. The ceramic

material's stiffness decreases as the microcrack damage increases, and microcrack

coalescence is assumed to occur at a critical crack density. The RG model has

been permanently implemented in the latest version of the EPIC code. EPIC

simulations were performed to evaluate the predictive capabilities of the model

for a benchmark suite of dynamic impact experiments. Results from these

calculations are presented and discussed in this paper.

INTRODUCTION

Historically, constitutive damage models have not demonstrated the predictive

capabilities necessary to justify their widespread use in the design of armor/anti-

armor systems. Empirical models are relatively easy to use (i.e., small number of

constants, etc.) and computationally efficient, but they tend to be applicable only

for a limited set of loading conditions. Three-dimensional fracture mechanics

based microphysical models have more complex formulations that tend to require

larger numbers of model constants, some of which may be difficult to determine

from experimental measurements. In addition, since their solution algorithms

may include iterative procedures to produce accurate results, these models tend to

require significantly more computing resources to solve a problem. However, a

model formulation which addresses the microphysics of the damage evolution

process offers the greatest potential for achieving an accurate predictive

capability. This paper describes the Rajendran-Grove (RG) microphysical

ceramic failure model in detail, and then demonstrates its ability to reproduce and

predict experimental measurements from a variety of impact experiments on

99.5% pure aluminum oxide (AD995) ceramic targets.



MODEL DESCRIPTION

The Rajendran-Grove (RG) ceramic failure model (Rajendran [1], Rajendran

and Grove [2]) assumes the following: 1) preexisting randomly distributed flaws,

2) plastic flow and pore collapse when shocked above the Hugoniot elastic limit

(HEL), 3) no plastic flow in tension, 4) degradation of elastic moduli under both

compression and tension due to microcracking, and 5) pulverization at a critical

crack density. A strain-rate-dependent strength relationship is employed to

describe the response of the ceramic material due to inelastic (plastic) deformation

under high-compression pressure. The deviatoric stresses are calculated using a

conventional radial return approach that is often used in viscoplasticity models.

The total strain is decomposed into elastic and plastic strains. The elastic strains,

which consist of the elastic strains in the intact matrix material and the strains due

to crack opening/sliding, are obtained by subtracting the plastic strains from the

total strains. After damage initiates, the unloading and reloading paths follow the

degraded elastic modulus (secant modulus), thus allowing full recovery of the

strains due to microcracking. The elastic stress-strain relationship for the

damaged aggregate material is given by

pklklijklij M , (1)

where is the total stress rate, is the total strain rate, is the plastic

strain rate due to viscoplastic flow and pore collapse, the strain rate difference

is the elastic strain rate, and the components of the stiffness tensor M

are determined by analytically inverting the compliance tensor.

ij

)pkl

klpkl

( kl

For non-interacting, penny-shaped microcracks of various sizes and random

orientations, Margolin [3] provided the following expression for the elements of

an isotropic compliance tensor C:

C , (2)klijjkiljlikijkl ccc 321

whereG

Bo4

11c ,

GDo

4

12c , and

G12Ac o3 . (3)

In the above equations, is the Kronecker delta, G and are the shear modulus

and Poisson's ratio, respectively, of the undamaged material, and Ao, Bo, and Do

are parameters whose values depend on the principal stress state and the extent of

damage. Before damage initiates, Ao = Bo = Do = 0. For stress states in which all

three principal stresses are compressive, Margolin proposed that


, , and , (4)*o 1A *

o 14B oo AD

whereE45

16* and . (5)3*

o aN

In the above equations,* is a microcrack density parameter, is the microcrack

density (a dimensionless measure of damage), E is the Young's modulus of the

undamaged material, (a model constant) is the number of microcracks per

unit volume, and a is the maximum crack length in the exponential distribution of

microcracks. Upon substituting Equations (4) into Equations (3) and inverting the

compliance tensor C, we obtain the following expressions for the shear and bulk

moduli (

*oN

cG and cK , respectively) of the damaged ceramic material under

compression:

GB32

G2

ocG (6)

and213

1G2Kc . (7)

The above equations result in degradation of the shear modulus, but not of the

bulk modulus, because under compression only crack movement of the closed

microcracks (modes II and III) is permitted. During tensile loading, the RG model

assumes that both the bulk and shear moduli rapidly degrade as a result of

microcrack opening (mode I). Based on the damaged stiffness solution proposed

by Budiansky and O'Connell [4] for randomly oriented, non-interacting

microcracks under tensile loading, we employ the following expressions for the

shear and bulk moduli ( tG and tK , respectively) of the damaged ceramic material

under tension:

t

ttt

2

51

45

321GG (8)

andt

ttt

213

1G2K , (9)

where9

161t . (10)


The above equations (valid only for tension), limit the microcrack density to

9/16; at this critical value, the ceramic material experiences a complete loss of

stiffness ( tK = tG = t = 0) and is subsequently unable to carry any tensile load.

Microcrack growth/extension (increasing a) causes the microcrack density

to increase (see Equations (5)), which results in the relaxation of stresses in the

cracked ceramic material. The crack growth (damage evolution) law, derived

from a fracture mechanics based relationship for a single crack propagating under

dynamic loading conditions, is described by

2n

I

crR1

G

G1Cna , (11)

where C is the Rayleigh wave speed, G is the critical strain energy release rate

for microcracking, are the applied strain energy release rates, and and n

are model constants that are used to limit the microcrack growth/extension rates.

The "+" superscript corresponds to microcrack opening under tension (mode I),

while the "-" superscript relates to microcrack extension under compression

(modes II/III). The model constants , , and are always assumed to be

equal to 1.0, while n is generally assumed to be equal to 0.1. In the model,

microcracks nucleate and grow/extend (i.e., > 0) when the stress state satisfies a

generalized Griffith criterion [5] developed by Margolin [6] and Dienes [7]. For

this criterion, , G , and G are calculated as:

R cr

1n n

IG

1

I

1n 2

2

a

2n

crG I

E

1K 22IC

crG , (12)

kji,2

2

E

a14G

2jk

2ik2

kk

2

I , (13)

and kji,2E

a18 2

kk2jk

2ik

2

IG , (14)

where (a model constant) is the fracture toughness of the undamaged

material,

ICK

E and are the degraded Young's modulus and Poisson's ratio,


respectively, of the damaged (cracked) material, and (a model constant) is the

dynamic friction coefficient. Note that G is computed from the undamaged

(virgin) material properties. Microcrack opening occurs when G exceeds ,

or microcrack extension occurs when exceeds .

cr

I crG

IG crG

lnC1

Y2

P3

m

f2f 2

Under high tri-axial tensile stress loading conditions, the following damage

processes often occur in brittle materials: 1) debonding of the hard carbide and

oxide particles from the matrix, and 2) non-spherical pore (planar crack) growth at

triple point grain boundaries. To capture the effects of damage due to these

processes, the RG model employs a critical stress based spall criterion (in addition

to the Griffith criterion). This spall criterion assumes initiation and growth of

damage when all three principal stresses are tensile and the maximum principal

stress exceeds a critical spall threshold stress, s (a model constant). The damage

rate in this case is assumed to be simply proportional to the Rayleigh wave speed

(i.e., ). The spall damage criterion is only applied to tensile stress states

that fail to satisfy the generalized Griffith criterion for crack opening/extension.

Consequently, the microcrack density is accumulated in a continuous manner

due to either microcracking or spall damage.

RCa

The RG model considers the material to be in a comminuted (pulverized) state

when the microcrack density ( ) exceeds p (a model constant) during compressive

loading. Generally p is set to 0.75, based on the assumption that pulverization

occurs when the microcracks coalesce [8].

Prior to pulverization, the compressive strength Y of the matrix (void-free)

ceramic material is described by the following strain rate dependent relationship:

peffAY , (15)

where A is the quasi-static maximum strength, C is the strain rate sensitivity

parameter, and is the normalized (dimensionless) equivalent plastic strain

rate; A and C are model constants. The model assumes that pore collapse may

occur during compressive loading above the HEL due to local microscopic plastic

flow in the matrix material surrounding the pores. The pore collapse strain

components are derived from Gurson's pressure dependent yield surface [9]:

peff

0cosh1YJ3 2m2 , (16)

where J2 is the second invariant of the deviatoric stress in the porous (void-


containing) aggregate material, Ym is the effective stress in the matrix (void-free)

material, P is the compressive pressure in the porous aggregate material, and f is

the porosity (void volume fraction). Note that in the absence of porosity (i.e.,

when f = 0), Equation (16) reduces to the von Mises yield condition. If pore

collapse occurs, the effective shear and bulk moduli of the damaged aggregate

material are defined as follows using a modified form of Mackenzie's relationship

[10,11]:

oo

eff

fG8K9

G12K61f1

fG8K9

G12K61f1

GG (17)

and

fG4

K31f1

fG4

K31f1

KK

o

o

eff , (18)

where . (19)pvo ef11f

In the above expressions, and are the effective shear and bulk moduli

(respectively) of the aggregate material,

effG effK

G is the degraded shear modulus of the

cracked matrix material (either cG or tG ), K is the degraded bulk modulus of

the cracked matrix material (either cK or tK ), fo (a model constant) is the initial

porosity, (= ) is the plastic volumetric strain due to pore

collapse, f is the porosity, and G and K are the shear and bulk moduli

(respectively) of the virgin material.

pv

p33

p22

p11

The RG model employs the following modified Mie-Gruneisen relationship to

compute the pressure P in the aggregate material prior to pulverization:

0,1E5.01bK/K

0,1E5.01bbbK/KP

s1eff

s3

32

21eff

, (20)

where is the elastic volumetric compressive strain ( = / o - 1, where and o

are the current and initial densities, respectively), is the Gruneisen coefficient,


b1, b2, and b3 are empirical constants used for a cubic fit to the Hugoniot curve for

the virgin material, Es is the internal energy per initial volume, K is the bulk

modulus of the virgin material, and is the effective bulk modulus of the

aggregate material (see Equation (18)).

effK

Once pulverization has occurred (i.e., p during compressive loading), the

RG model assumes that pore collapse and crack growth/extension no longer occur

(i.e., ). Also, the model assumes that the pulverized material is unable

to carry a tensile load, so that

0af

effK

ij = P = 0 in tension. In compression, however, the

stresses and pressure are computed using the effective shear and bulk moduli

( and ) corresponding to the values of and f at the time of

pulverization. The pressure P and strength Y of the comminuted material are

described by,

effG

(21)

0,0

0,KP

ev

ev

eveff

and

0P,0

0P,Y,Pmin maxpp

Y , (22)

where is the elastic volumetric strain, (a model constant) is the dynamic

friction coefficient for granular motion, and Y (a model constant) is the upper

limit on the compressive strength of the pulverized ceramic material. Since

experimental data for the fractured strength is generally either unavailable or

difficult to interpret, we usually set to "1" and calibrate Y to match the

measured penetration depths from projectile penetration experiments.

ev p

maxp

pmaxp

Generally, seven of the RG ceramic model constants require some calibration

with experimental data: strain rate sensitivity parameter (C), initial crack size

( ), microcrack number density ( ), dynamic friction coefficient ( ),

coefficient to limit the mode II/III crack extension rate ( ), critical spall stress

(

oa *oN

1n

s), and the maximum compressive strength of the pulverized material (Y ).

We employed the following set of model constants to describe the dynamic

response of AD995 ceramic in this study:

maxp

o = 3.89 g/cm3, G = 156 GPa, K = 231

GPa, b1 = 231 GPa, b2 = -160 GPa, b3 = 2774 GPa, = 2.3, A = 2.3 GPa, C = 0.2,

KIC = 3 MPa m , fo = 0.0, a = 1.5 x 10o-6

m, = 2.0 x 10*oN

maxp

-11 m

-3, = 0.60,

= 0.1, 1n s = 0.5 GPa, p = 0.75, = 1.0, and Y = 4.5 GPa. p


MODEL RESULTS FOR AD995 CERAMIC

We verified the generality of the RG model constants for 99.5% pure

aluminum oxide (AD995) through computer simulations of the following four

impact configurations: 1) plate impact, 2) rod-on-rod impact, 3) graded-density

plate-on-rod impact, and 4) projectile penetration. The details of these

experimental configurations are provided in a companion chapter, "Historical

Perspective on Ceramic Materials Damage Models," by A. M. Rajendran.

The simulations were performed using the EPIC finite element code, modified

to include the RG ceramic failure model. EPIC is a well-established three-

dimensional production code that was initially developed in the early 1970's to

describe the response of solid materials to dynamic impact loading. Johnson,

Stryk, Holmquist, and Beissel [12] have described the details of this explicit

Lagrangian finite element code. To maintain the stability of the explicit finite

element solution, an iterative scheme based on a second-order diagonally implicit

Runge-Kutta method was employed in the RG model solution algorithm.

Plate Impact

Using EPIC's one-dimensional (1D) strain option, we obtained an initial

calibration of the model constants through simulations of two AD995 plate impact

tests: 1) a low velocity test (flyer thickness: 4 mm, target thickness: 8 mm, impact

velocity: 83 m/s) performed by Dandekar and Bartkowski [13], and 2) a high

velocity test (flyer thickness: 5 mm, target thickness: 10 mm, impact velocity:

1943 m/s) reported by Grady and Moody [14]. Figure 1 indicates that the

computed spall signals (profiles beyond point S) agree with the measured profiles.

Time ( s)

0.5 1.0 1.5 2.0 2.5

Axia

l S

tres

s (G

Pa)

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Experiment

Model

S

(a) Impact velocity = 83 m/s

Time ( s)

0.5 1.0 1.5 2.0 2.5 3.0

Vel

oci

ty (

km

/s)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Experiment

Model

S

(b) Impact velocity = 1943 m/s

Figure 1. Comparison of computational results with plate impact data.


Rod-on-Rod Impact

Using the two-dimensional (2D) axisymmetric geometry option in EPIC, the

rod-on-rod impact configuration was simulated for two different impact velocities

(0.175 km/s and 0.278 km/s). The AD995 striker rod was 5 cm long and 1.25 cm

in diameter (L/D = 4), while the AD995 target rod was 10 cm long and 1.25 cm in

diameter (L/D =8). Figure 2 compares the model-predicted stress histories with

those measured experimentally by Simha [15]. While the model does not exactly

match the data, the peak stress levels are reasonably close to the measurements.

Time ( s)

6 7 8 9 10

Str

ess

(GP

a)

0

1

2

3

4

5

Experiment

Model

(a) Impact velocity = 175 m/s

Time ( s)

6 7 8 9 10

Str

ess

(GP

a)

0

1

2

3

4

5

Experiment

Model

(b) Impact velocity = 278 m/s

Figure 2. Comparison of computational results with rod-on-rod impact data.

Graded-Density Plate-on-Rod Impact

Two-dimensional axisymmetric simulations of the graded-density plate-on-rod

impact configuration were performed for both unsleeved (bare) and sleeved

AD995 ceramic target rods. In both cases, the impact velocities were around

0.300 km/s. The flyer plate was modeled as a layered circular disk (diameter = 5

cm, thickness = 2.2 cm); a continuous finite element grid was employed in the

flyer plate to simulate a "perfect" bond between adjacent layers of material. The

target was modeled as a solid rod (diameter = 1.9 cm, length = 7.4 cm), while the

steel sleeve was modeled as a hollow rod (inner diameter = 1.9 cm, outer diameter

= 3.8 cm, length = 7.4 cm). Frictionless sliding was permitted between the inner

surface of the sleeve and the outer surface of the ceramic rod. Figure 3 compares

the model-predicted velocity histories with those measured experimentally by

Chhabildas, Furnish, Reinhart, and Grady [16]. As the figure indicates, the model

does an excellent job of predicting the peak velocity levels, as well as the

constant-velocity behavior (due to spallation of the ceramic rod near the free end).


(a) Bare target rod,

impact velocity = 300 m/s

(b) Sleeved target rod,

impact velocity = 321 m/s

Figure 3. Comparison of computational results with graded-density plate-on-rod

impact data.

Time ( s)5 10 15 2

Vel

oci

ty (

km

/s)

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Experiment

Model

Time ( s)5 10 15 2

Vel

oci

ty (

km

/s)

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Experiment

Model

00

Projectile Penetration

For the two-dimensional axisymmetric simulations of the projectile

penetration configuration, the target was assumed to be an AD995 ceramic disk

(diameter = 15.24 cm) backed by a thick steel cylinder (diameter = 20.32 cm,

thickness = 12.7 cm); the ceramic disk was also radially confined by a steel ring

(inner diameter = 15.24 cm, outer diameter = 20.32 cm, thickness = 5.08 cm) that

was fixed to the surface of the steel cylinder. The projectile was modeled as a

tungsten rod (diameter = 0.787 cm, length = 7.87 cm) with an impact velocity of

1.5 km/s. Simulations of this configuration were performed for seven different

thicknesses (between 1.02 and 5.08 cm) of AD995 ceramic disks. The penetration

process was modeled through EPIC's erosion algorithm (using an erosion strain of

150%). Figure 4 compares the measured [17] and computed residual depths of

penetration (DOP) into the backup steel block versus the areal densities (mass per

unit area) of the ceramic disks. The straight line in this figure is a linear least-

squares fit to the experimental data. As Figure 4 indicates, the model-predicted

depths of penetration are consistent with the experimental measurements.

SUMMARY

The governing equations for the RG ceramic failure model were described in

detail, and a set of model constants for AD995 ceramic was proposed. This set of

constants was then employed in a series of finite element simulations for the

following benchmark suite of experimental impact configurations: plate impact,


Areal Density (g/cm2)

0 5 10 15 20 25

Res

idual

DO

P (

cm)

0

1

2

3

4

5

6

7

8

Experiments

Model

Figure 4. Comparison of computational results with projectile penetration data.

rod-on-rod impact, graded-density plate-on-rod impact, and projectile penetration.

The simulation results demonstrated the model's ability to reproduce the

experimentally measured stress and velocity histories, as well as the DOP data.

While these results are very encouraging, it is important to continue evaluating the

model's predictive capability through simulations of more complex ceramic armor

impact configurations.

ACKNOWLEDGEMENTS

The authors greatly appreciate the funding support of Dr. Doug Templeton and

Krishan Bishnoi of TARDEC, Warren, MI. This work was supported in part by a

grant of HPC time from the DoD HPC Center at Aberdeen Proving Ground, MD.

REFERENCES1A.M. Rajendran, "Modeling the Impact Behavior of AD85 Ceramic Under

Multiaxial Loading," Int. J. Impact Engng., 15 (6) 749-768 (1994). 2A.M. Rajendran and D.J. Grove, "Modeling the Shock Response of Silicon

Carbide, Boron Carbide, and Titanium Diboride," Int. J. Impact Engng., 18 (6)

611-631 (1996). 3L.G. Margolin, "Elastic Moduli of a Cracked Body," Int. J. of Fracture, 22,

65-79 (1983). 4B. Budiansky and R.J. O'Connell, "Elastic Moduli of a Cracked Solid," Int. J.

of Solids and Structures, 12, 81-97 (1976).


5A.A. Griffith, "The Phenomena of Rupture and Flow in Solids," Phil. Trans.

of Royal Soc. London, 221, 163-198 (1920). 6L.G. Margolin, "A Generalized Griffith Criterion for Crack Propagation,"

Eng. Fracture Mechanics, 19 (3), 539-543 (1984). 7J.K. Dienes, "Comments on 'A Generalized Griffith Criterion for Crack

Propagation', by L.G. Margolin," Eng. Fracture Mechanics, 23 (3), 615-617

(1986).8A.M. Rajendran, High Strain Rate Behavior of Metals, Ceramics, and

Concrete, Air Force Report WL-TR-92-4006, Wright-Patterson Air Force Base,

OH 45433-6533, April 1992. 9A.L. Gurson, "Continuum Theory of Ductile Rupture by Void Nucleation and

Growth; Part I: Yield Criteria and Flow Rules for Porous Ductile Materials," J.

Engr. Mat. Tech., 99, 2-15 (1977). 10

J.K. Mackenzie, "The Elastic Constants of a Solid Containing Spherical

Holes," Proc. Phys. Soc., 2, 63, (1950). 11

J.N. Johnson, "Dynamic Fracture and Spallation in Ductile Solids," J. Appl.

Phys., 52 (4), 2812 (1981). 12

G.R. Johnson, R.A. Stryk, T.J. Holmquist, and S.R. Beissel, Numerical

Algorithms in a Lagrangian Hydrocode, Report No. WL-TR-1997-7039, Wright

Laboratory, Eglin AFB, FL (1997). 13

D.P. Dandekar and P. Bartkowski, "Shock Response of AD995 Alumina";

pp. 733-736 in High-Pressure Science and Technology - 1993, Part 1. Edited by

S.C. Schmidt, J.W. Shaner, G.A. Samara, and M. Ross. AIP Press, New York,

1994.14

D.E. Grady and R.L. Moody, Shock Compression Profiles in Ceramics,

Sandia Report No. SAND96-0551, Sandia National Laboratory, Albuquerque,

NM. (1996). 15

C.H.M. Simha, High Rate Loading of a High Purity Ceramic - One

Dimensional Stress Experiments and Constitutive Modeling, Ph.D Thesis,

University of Texas, Austin, Texas. (1998). 16

L.C. Chhabildas, M. D. Furnish, W.D. Reinhart, and D.E Grady, "Impact of

AD995 Alumina Rods"; pp. 505-508 in Shock Compression of Condensed Matter

- 1997. Edited by S.C. Schimdt, D.P. Dandekar, and J.W. Forbes. AIP Press,

1998.17

P. Woolsey, "Residual Penetration Ballistic Testing of Armor Ceramics,"

Unpublished Work, U. S. Army Materials Technology Laboratory, Watertown,

MA (1991).


Damage Evolution and Micromechanisms

FAILURE PHENOMENOLOGY OF CONFINED CERAMIC TARGETS AND

IMPACTING RODS

Donald A. Shockey and A.H. Marchand

SRI International

333 Ravenswood Avenue

Menlo Park, CA 94025

S.R. Skaggs, G.E. Cort, M.W. Burkett, and R. Parker

Los Alamos National Laboratory

Los Alamos, NM 87545

ABSTRACT

The mechanism by which a long rod penetrates a steel-encased ceramic block

was sought by performing impact experiments at a range of velocities, and

examining the fracture and deformation in the recovered targets and impactors.

The key processes are the crushing of a small volume of ceramic adjacent to the

leading surface of the advancing penetrator, and the subsequent flow of the fine

fragments lateral to and then opposite the direction of attack. The results suggest

that nonconventional material properties such as the dynamic compressive failure

energy and the friction, flow and abrasive properties of the finely fragmented

material govern the penetration resistance of confined ceramics. This

understanding of penetration mechanisms can be used to guide development of

specialized tests and failure models to measure pertinent material properties and

to predict penetration behavior, respectively.

INTRODUCTION

When ceramic plates are used as overlays or incorporated as a layer within

conventional monolithic steel armor, the ballistic protection is significantly

enhanced [1]. This finding has encouraged the use of ceramics as a component of

advanced armor structures and has motivated researchers to identify materials and

structural configurations that maximize ballistic performance.

The design of current ceramic armors is based predominantly on empirical

ballistic performance data. Test firings using the threat of interest (long rod,

shaped charge jet, or small caliber ammunition) are conducted against armor

specimens in which ceramic component parameters, such as type of ceramic,


Reprinted from International Journal of Impact Engineering, Vol 9, No 3, Shockey et al, “Failure Phenomenology of Confined CeramicTargets and Impacting Rods”, pp 263-275, copyright1990, with permission from Elsevier Science.

thickness, and spacing of plates, are varied independently and systematically. The

combination of parameters that produce maximum ballistic protection is

determined from test results and used to design the armor package. This

procedure is lengthy and expensive, and because the number of potentially

influential material and geometry variables is large, a comprehensive test matrix is

not practical and so it is doubtful that armor packages affording optimal protection

are attained.

A more efficient procedure is to combine experiments with computational

simulations of experiments. Computations of the ballistic behavior for various

impact conditions and target geometries indicate which target parameters are

important and suggest combinations of target parameters that will give favorable

performance. A limited matrix of ballistic experiments based on these guidelines

are performed. The results are used to modify the models and the code; then a

second generation of computational simulations is conducted and used to design a

second set of test firings. This procedure is repeated until an armor package with

acceptable performance is obtained. The reduction in the number of required test

firings reduces substantially the expense and time required to attain a suitable

armor, and the understanding gained promotes optimum armor design.

Implementation of the iterative computational/experimental procedure,

however, requires reliable models for the microfailure behavior of ceramics and

penetrators under penetration conditions. These models should be based on an

understanding of the failure phenomenology during penetration. In particular, the

material properties governing penetration resistance must be known.

Unfortunately, conventional material properties such as fracture toughness,

strength, and hardness correlate poorly with penetration behavior [2, 3],

suggesting that under the complex, high-rate, multiaxial load produced by the

penetrator one or more nonconventional material properties control penetration.

The goal of the work reported here was to establish the failure phenomenology

of confined ceramic targets and impacting long rods during penetration, and to

deduce the ceramic properties governing penetration resistance. Our approach was

to perform impact experiments on confined ceramic specimens at several

velocities to produce damage ranging from incipient to severe. Very early stages

of damage were studied by performing experiments with low velocity spherical

particles. From fractographic and metallographic examination of the targets and

rods after impact, we inferred the failure mechanisms and speculate on the

properties controlling penetration behavior. The study was aimed at ceramics as a

class of materials and sought a qualitative understanding of penetration

phenomenology to provide the basis for a computational model; therefore,

experiments were performed on a variety of ceramics and details of microstructure

and mechanical properties of the individual materials are not presented.


ROD IMPACT EXPERIMENTS

Tungsten-nickel-iron rods (7.70 mm in diameter and 77.0 mm long), having

hemispherical noses, were accelerated in a powder gun to desired velocities in the

0.8 to 1.4 km/s range and caused to impact a steel-encased block of ceramic at

approximately zero degrees obliquity (Fig. 1). Ceramics investigated included

Al2O3, SiC, B4C, and TiB2.

After the impact event, the target assemblies were removed carefully from the

mounting fixture and placed on the floor of the bunker with the impact surface

facing up. A two-component epoxy was poured into the hole on the front cover

plate to fill the crater and run into the cracks in the ceramic. This procedure was

intended to strengthen the fractured ceramic block so that the confining steel

could be removed and the ceramic block could be sectioned without crumbling of

the fractured ceramic. We found in later experiments that impacted targets were

often strong enough to be disassembled, sectioned, and even wafered without

infiltration of epoxy.

Fig. 1. Arrangement for rod impact experiments.

The front surface of the ceramic block and the inside surface of the steel cover

plate showed a starburst pattern of linear markings radiating outward from the


impact site. These markings were produced by ceramic and tungsten rod

fragments originating near the leading edge of the penetrator.

All ceramics tested exhibited qualitatively similar cracking patterns; however,

the numbers and sizes of cracks generally differed. The crack patterns on the rear

surfaces of a B4C and a TiB2 specimen impacted at 0.8 km/s are shown in

Fig. 2(a). Two main types of cracks are evident: circular cracks and radial cracks.

Three to four dominant circular cracks were observed in both ceramics; however,

approximately 50% more radial cracks were produced in the B4C (23 cracks) than

in the TiB2 (15 cracks). Fracture damage was heaviest in both materials in a

region directly beneath the impact site.

A diamond saw was used to cut through the crater centers on a plane

containing the impact direction. Cutting the ceramic block was difficult. The

diamond wheel wore out quickly and stopped cutting about halfway through the

block. The wheel needed to be redressed several times before the sectioning was

completed.

Orthogonal views of the craters and crack patterns were obtained on the as-

sawn, unpolished section surfaces. These surfaces for a B4C and a TiB2 specimen

are shown in Fig. 2(b). Crater size was greater in the B4C. Material directly

beneath the crater in the TiB2 specimen remained intact, probably because of the

triaxial compressive stress state and the higher compressive yield strength of

TiB2. At higher velocities resulting in deeper rod penetration, the material in

advance of the tip of the penetrator was crushed to a fine powder. Similar damage

patterns were observed in Al2O3 and SiC targets.

The views in Fig. 2(b) show that the circular cracks in Fig. 2(a) are traces of

cracks that ran outward at an angle from the impact site. The resulting cone

configuration corresponds to the Hertzian cracks observed in the particle impact

experiments reported in the following section and observed by others in ceramics

and glasses under static indentation and particle impact [4, 5]. The lateral cracks

lying roughly parallel to the surface apparently formed after the cone cracks, since

they are discontinuous across the cone cracks. Thus, lateral cracks may be

produced by tensile stress waves reflecting from the specimen boundaries or by

later unloading of the target. Radial cracks are not revealed on cross sections

containing the impact direction.

In several instances tungsten fragments were observed lodged between faces

of Hertzian cone cracks. That these fragments were not moved into the cracks by

the sectioning operation was confirmed by computed tomography results that

showed fragments in cracks in unsectioned specimens. Such observations suggest

that debris emanating from the eroding end of the penetrating rod can have a

significant forward velocity component. These observations support the premise

of Hauver [6], who observed tungsten fragments in advance of the penetrator in

x-radiographs of ceramic blocks during impact by tungsten rods.


Fig. 2. Rear surface crack pattern (a) and crack patterns on cross sections (b) in

targets of B4C and TiB2.

Loose fragments produced in the impact experiments were collected for

examination. The size distribution of the collected fragments was determined by

a sieve analysis (placing the recovered fragments on the topmost of a stack of

successively finer screens and vibrating the stack for an hour). We separated

ceramic fragments from penetrator fragments by passing a strong magnet over the

sieved fragments and extracting the slightly magnetic tungsten alloy fragments

from the ceramic debris.

Fig. 3 shows the fragments of SiC retained on screens with various mesh

openings. Fragment shape did not vary substantially with fragment size; aspect

ratios ranged from 1 to about 3. Fracture was predominantly transgranular rather

than intergranular. Differences in fragment size distributions for the four ceramic


materials were small over the entire size range. Fig. 4 compares the distributions

for three ceramics in the 2- to 40- m size range.

In higher velocity experiments in which the rod penetrated 60 to 100 mm, the

crushed ceramic material produced at the leading edge of the rod flowed around

and behind the rod, closing the hole made by the rod (Fig. 5). So well

consolidated were these fine fragments that no fragments were loosened during

sectioning and individual fragments were not easily discernable by high

magnification examination. Hardness and scratch tests indicated strengths of the

compacted powder comparable to that of the unimpacted material. In Fig. 5, the

penetrator has stopped just short of the rear confinement plate.

Fig. 3. Fragments from a SiC target retained on screens with different mesh

openings.


Fig. 4. Distribution of fragments of B4C, SiC, and TiB2 in the 2- to 40- m size

range.

Fig. 5. Cross section through shot line of 100-mm-thick confined B4C target

impacted at 1.6 km/s showing cracking pattern, compacted ceramic fragments in

cavity produced by penetrator, and embedded penetrator fragments.

The debris from several targets was searched for distal portions of the

penetrator. Distal portions ranging in length from 3-20 mm were found [Fig. 6(a)].


Proximal ends had either a mushroom shape or a sharpened-pencil shape. We

speculate that a penetrator tip may alternate shapes between that of a mushroom

and a pencil-point several times during the penetration process. Initially, we

expect plastic deformation of the leading edge to produce a mushroom shape. The

mushroom zone then shears away on roughly a 45° conus, producing a pencil

point. The pencil point then deforms plastically and the tip acquires a mushroom

shape again; this mushroom shape becomes unwieldy and shears to a pencil point.

This alternating shape change continues until penetration ceases. The proximal

end surfaces of all recovered penetrators were faceted and gouged, suggestive of

shear failure. The lateral surfaces were unscored.

Fig, 6. Distal portions of tungsten alloy rods recovered from ballistic experiments

(a) and polished and etched cross section showing deformation of the

microstructure (b).

Polished and etched cross sections on planes containing the rod axis revealed

that the tungsten particles in this sintered alloy retained their original roughly

spherical shape everywhere except near the proximal failure surface. Adjacent to

the failure surface, the tungsten particles were greatly elongated, often to aspect

ratios of 5 or more. Particle distortion decreased with distance from the failure


surface, rather gradually (over a distance of about 4 mm) in the mushroomed

region. Thus, the distribution of deformed tungsten particles provides a map of

the plastic strain field in a penetrator.

Tungsten fragments extracted magnetically from the debris ejected from the

impact surface were also examined with a scanning electron microscope. Failure

surfaces and etched cross sections suggested that fragment formation was by

localized shearing of the microstructure, in accord with observations on distal

penetrator ends. Tungsten particle distortion in the fragments, however, decreased

abruptly (usually within about 300 m) from the surface of fragments [Fig. 6(b)].

We computationally simulated the experiment depicted in Fig. 1 using a two-

dimensional, Lagrangian finite difference code. The results provided an estimate

of the distribution and time variation of the stresses and strains produced in the

ceramic target by the impacting long rod before failure occurred, and assisted in

the interpretation of the fractographic observations.

PARTICLE IMPACT EXPERIMENTS

Low-velocity particle impact experiments were performed to study incipient

stages of impact damage. The evolution of fracture damage was established in hot

pressed (HP) silicon nitride by accelerating single solid spheres of tungsten

carbide (WC) or steel onto the polished surfaces of small plate specimens of HP

Si3N4 at a 90°-angle [7]. Particles were accelerated to velocities from 16-368 m/s

by filling the gun chamber with nitrogen gas to various pressures, then suddenly

releasing the nitrogen by rupturing a disk. The diameters of the WC spheres were

1.6 and 2.4 mm; the steel spheres were 2.4 mm in diameter. Impact and rebound

velocities were recorded with photomultipliers. Photomultiplier records also

showed whether particles remained intact or fragmented after impact. The

specimen fracture damage was studied by optical and scanning electron

microscopy of impact surfaces and polished cross sections normal to the impacted

surfaces.

The impact tests caused several kinds of cracks, small craters, and

fragmentation in the target plates and eventually plastic deformation or fracture of

the impacting spheres. Targets sustained no damage at impact velocities below

17 m/s, at which point ring cracks appeared. As impact velocity increased, the

damage progressed to cone cracks, an inelastic impression, radial cracks, lateral

cracks, and median-vent cracks. Ring cracks, as shown in Fig. 7, are

circumferential cracks that extend less than a millimeter beneath the surface. As

the impact velocity increased, more and longer ring cracks formed [Fig. 7(b)].

The ring cracks are similar to the Hertzian ring cracks formed under quasi-static

loading [8]. The surface ring cracks that start approximately normal to the

specimen surface veer outward at various angles up to about 75° from the vertical

to become Hertzian cone cracks [9]. As the velocity increases, additional cone


cracks form both inside and outside the existing damage umbrella, and the

innermost cone grows several millimeters in depth.

An inelastic impression and radial cracks [Fig. 7(c)] seemed to form at the

same time in the failure sequence. As the impression deepened with increasing

velocity, the radial cracks grew in both size and number, although only a small

number (8 or 9) of the radial cracks grew to several millimeters [Fig. 7(d)].

Fig. 7. Cracks on the surface of HP Si3N4 caused by impact of 2.4-mm-diameter

tungsten carbide spheres at velocities of (a) 19.5 m/s, (b) 46.2 m/s, (c) 97.7 m/s,

and (d) 159 m/s.

Fig. 8 shows the internal damage and the extent of growth of the various

cracks below the specimen surface. The nucleation and growth sequence of the

ring/cone cracks is illustrated in Figs. 8(a) and 8(b). Under increasingly severe

impacts, cone cracks seemed to cease growing; instead, two new types of cracks


were created, as shown in Fig. 8(c). Lateral cracks nucleated internally near the

contact center and ran approximately parallel to, and eventually veered toward, the

impact surface of the specimen. Vertical cracks initiated internally in the region

within the innermost cone crack. These latter penny-shaped cracks are similar to

the median-vent cracks observed by Evans and Wilshaw in quasi-static

indentation experiments on ZnS [8]. Observations with polarized light showed

that a zone of densely microcracked material, approximately spherical in shape,

was formed beneath the contact area. Zinc sulfide impacted by 0.4-mm and 0.8-

mm WC spheres exhibited a similar microcracked zone [10].

Impacting steel spheres, which are softer than tungsten carbide, caused only

ring and cone cracks and introduced little additional damage above 300 m/s, at

which velocity the particle failed by plastic deformation. This limit on the damage

inflicted on the ceramic occurs because the particle cannot exert a pressure on the

ceramic greater than the particle's yield strength. Since the yield strength of the

steel is less than the pressure required for inelastic deformation of the ceramic

surface, higher velocity impacts only result in more deformation of the steel

sphere.


Fig. 8. Sectional views of subsurface cracking pattern in HP Si3N4 impacted by

2.4-mm-diameter steel spheres at velocities of 56.4 m/s (a) and 231 m/s (b) and by

a 2.4-mm-diameter tungsten carbide sphere at 231 m/s (c).


FAILURE PHENOMENOLOGY OF THE PENETRATION PROCESS

The picture of the penetration process that begins to emerge from these

observations and consideration of the initial stress history is as follows.

Calculations simulations of a tungsten alloy rod impacting a target as in Fig. 1

at 1600 m/s show that at the instant of impact, a shock wave with an amplitude of

several hundred kbars is generated at the impact site. Radial divergence and

plastic flow and fracture in the steel cover plate quickly and drastically reduce the

stress so that the strength of the shock that passes into the ceramic is below the

Hugoniot elastic limit. Thus, the initial shock wave is not expected to condition

the ceramic. A steady-state ramp wave follows the shock, loading the ceramic

material at the tip of the penetrator to a maximum compressive stress of about 50-

60 kbars. The ceramic initially resists the stress in the ramp wave and exerts large

stresses on the tungsten rod, which may deform, fracture, or be deflected.

Ceramics are substantially stronger in compression than in tension, and

consequently, the tensile strength of the ceramic is exceeded at the impact surface

near the rod periphery and tensile fracture begins to occur soon after impact. The

stress fields in the ceramic are initially elastic, and the largest tensile stresses are

in the radial direction (the Boussinesq stress field). Therefore, the cracks that form

(normal to the direction of maximum principal stress) are ring cracks concentric

about the impact site. These cracks are shallow cracks, extending initially only a

millimeter or so beneath the ceramic surface. Upon continued loading, however,

several ring cracks continue to grow and, following the paths normal to the

direction of the principal tensile stress, assume angled trajectories 25-75° outward

from the initial normal-to-the-surface direction. Thus, several large Hertzian cone

cracks extend through the ceramic block, intersect the specimen surfaces, and

cause structural failure of the target.

Up to this point, the stress fields and the fracture response are elastic. But as

the rod continues to advance, the compressive strength is exceeded in material

directly beneath the penetrator. Microcracking occurs in a shallow zone near the

penetrator tip, and the stress field changes in character. The principal tensile

stresses are now in the circumferential direction, and a new type of tensile crack is

invoked six to twelve large radial cracks run outward from the impact site like

spokes from a hub [9]. These cracks intersect the impact surface and may extend

to all specimen boundaries, resulting in strength degradation and eventual

structural failure of the target.

A third crack type, lateral cracks, form beneath the impact surface and

propagate roughly parallel to it, probably during unloading. These cracks intersect

cone cracks and radial cracks, thereby providing the orthogonal surfaces necessary

for fragment formation. Cratering results when these large fragments are liberated

from the vicinity of the impact site.


The tensile cone, radial, and lateral cracks do not provide an easy path for

penetrator advance and hence do not assist the penetration process directly. Intact

and laterally confined ceramic remains in the penetrator path despite the presence

of these tensile cracks, and this material must be moved out the way for the

penetrator to advance. This occurs by pulverization of the ceramic material in a

shallow zone immediately ahead of the penetrator and the subsequent flow of this

material laterally and opposite to the impact direction, processes that occur under

large compressive and shear stresses.

Thus, the development of a densely microcracked zone in a ceramic directly

ahead of the impactor is a prerequisite for penetration. Insight into how this zone

forms can be gleaned from the observations of Hagan and coworkers [11, 12] of

damage zones in soda-lime glass and zinc sulfide produced by quasistatic

indentation. These workers observed a curvilinear grid of narrow, fault-like flow

lines beneath indentations, and voids and microcracks at many of the nodes in the

grid. Flow in the polycrystalline ZnS occurred by slip and twinning within the

grains and by grain boundary sliding; voids formed when grain boundary

displacements became large either along the flow lines or at flow line

intersections.

The finely fragmented material at the leading edge of the penetrator wants to

occupy a larger volume (i.e., dilation), but expansion is resisted by the

confinement of the steel encased ceramic block. The resulting increase in pressure

makes fragment flow more difficult and adds to the resistance exerted on the

penetrator. The tensile cracks may assist penetration indirectly by reducing the

level of constraint on the pulverized material, thereby allowing easier flow of the

material out and away from the advancing penetrator, but the main resistance to

penetration is probably coupled to the flow characteristics of highly comminuted

ceramic powder. The cracking pattern in the ceramic target envisioned during the

steady-state phase of the penetration process is depicted in Fig. 9.

As the ceramic particles flow across the leading surface of the penetrator, they

erode the rod, shortening and eventually consuming it as the rod moves through

the ceramic. No scoring or erosion of the sides of the penetrator results from

particles flowing opposite the direction of penetration. Fragments of the penetrator

fretted from the leading surface generally have an initial forward velocity

component and may travel into open cone and radial cracks ahead of the tip of the

penetrator. Other penetrator fragments mix with and flow with the ceramic

powder, becoming part of the front surface ejecta. The eroded tungsten fragments

exhibit greatly elongated grains close to the fragment surfaces, indicative of heavy

localized plastic flow.


Fig. 9. Cracking pattern in ceramic targets during the steady-state phase of the

penetration process.

CONCLUSIONS

According to this concept of penetration phenomenology, the properties of a

ceramic that govern penetration resistance include the compressive strength and

hardness, the pulverization characteristics, the frictional flow characteristics of

fine fragments, and fragment abrasiveness. These properties are consistent with

those suggested by Mescall [13, 14].

Initial resistance to penetration is provided by the compressive strength or

hardness of a ceramic. High compressive strength is desirable to deform, fracture,

and deflect an impacting body. Projectiles with low aspect ratios can be defeated

if the strength of the ceramic exceeds the strength of the projectile. High aspect

ratio projectiles such as long rods may suffer heavy deformation and fracture

damage to the proximal end, but the intact distal portion will continue to advance

and penetrate the ceramic. Thus, a high ceramic compressive strength can resist

penetration only to a certain extent.

The stresses exerted by a long impacting rod will eventually pulverize the

ceramic material in a small zone immediately ahead of the leading surface of the

penetrator. As explained in the following paragraph, production of a pulverized

zone is a necessary condition for a penetrator to advance in a confined ceramic

target. Thus, resistance to comminution is desirable for penetration resistance.

Although the specific fracture surface energy for most ceramics is small in

tension, the energy required to produce a unit of failure surface area under large

dynamic compressive and shear forces may be significantly greater. Thus, the

energy absorbed in creating the surface area of the powder may be a significant

ceramic property for penetration resistance.

A penetrator can only advance if the material in its path is pushed ahead of it

or to the side. Because of heavy rear confinement, the crushed ceramic cannot be

pushed ahead and out the rear surface in the way that metallic armor plates fail by

plugging. And if the ceramic is nonporous and snugly confined laterally, the


pulverized material cannot be pushed to the side. Indeed, the only recourse is for

the powder to flow opposite the penetration direction along the cavity being

produced by the penetrator. Thus, the pulverized and dilated ceramic must flow

under high pressure, and so the frictional flow property of the comminuted

ceramic should influence penetration resistance. We expect this property to

depend on pressure, strain rate, and size and shape of the fragments, and to be

describable by a Mohr-Coulomb type curve. For thick, highly confined ceramic

blocks, the friction-flow property of ceramic fragments may be the most important

material property for penetration resistance.

Finally, the ability of a ceramic to erode a penetrating rod is a desirable

property for penetration resistance. Whereas erosion may be by gross local plastic

flow of the leading rod surface, ceramic fragments that gouge, score, shear, or

otherwise abrade the rod material may reduce the incoming mass and terminate

the penetration earlier than nonabrasive target materials. Wear and erosion can

occur by a number of mechanisms depending on penetrator and target material,

fragment geometry and size, pressure, temperature, and flow rate. Thus, optimal

erosive behavior might be achievable by matching the abrasive characteristics of a

ceramic material to the threat.

The fractographic observations and the deduced penetration phenomenology

reported here can also be used to identify properties governing the penetration

capability of rods. To be effective as a penetrator, a material should have high

density to produce high stresses in the target; a high yield strength to resist

mushrooming at the leading edge; a high work hardening rate to suppress the

tendency to shear band and fret; a high fracture toughness to resist the propensity

for rod shaft failure; and high abrasion resistance to resist scoring and erosion by

ceramic particles.

In future work, this understanding of penetration phenomenology will be used

to develop tests that measure dynamic shear strength and flow resistance of intact

and fragmented ceramic material under high confining pressure, and to develop

computational models of penetration that can be used to assist in designing

ceramic armor.

ACKNOWLEDGMENTS

Financial support provided by the Defense Advanced Research Projects

Agency and the Army Research Office (Contract DAAL03-88-K-0200), and by

Los Alamos National Laboratory (Contract 9-X69-3295X-1). The authors

gratefully acknowledge the interest and encouragement of Drs. Andrew Crowson,

Eugene Farnum, Francis W. Patten, and William Snowden. Mr. Thomas Cooper

of SRI performed the computational simulations of the rod impact experiments.


REFERENCES1F.S. Mascianica, "Ballistic Technology of Lightweight Armor Materials,"

U.S. Army Materials Research Agency, AMRA MS 64-07, Sept 1964 (updated in

1981), AMMRC Report 81-20, Army Materials and Technology Laboratory,

Watertown, MA. 2D. Viechnicki, W. Blumenthal, M. Slavin, C. Tracy and H. Skeele, "Armor

Ceramic-1987," Proceedings of The Third TACOM Armor Coordinating

Conference, February 17-19, 1987, Monterey, CA. 3W. Rafianello, B. Brubaker and R. Hoffman, "Evaluation of a New Low-Cost

Aluminum Nitride Armor," Proceedings of The Fifth TACOM Armor

Coordinating Conference, March 7-9, 1989, Monterey, CA. 4B. Lawn and T.R. Wilshaw, "Review of Indentation Fracture: Principles and

Applications," J. Mater. Sci. 10, 1049-1081 (1975). 5A.G. Evans, "Impact Damage in Ceramics"; p. 302 in Fracture Mechanics of

Ceramics, Vol. 3, Edited by R.C. Bradt, D.P.H. Hasselmann and F.F. Lange,

Plenum Press, New York, 1978. 6G. Hauver, U.S. Army Ballistic Research Lab, personal communication. 7K.C. Dao, D.A. Shockey, L. Seaman, D.R. Curran and D.J. Rowcliffe,

"Particle Impact Damage in Silicon Nitride," SRI Annual Report, Part III, to the

Office of Naval Research, Arlington, VA, N00014-76-C-0657 (1979). 8A.G. Evans and T.R. Wilshaw, "Quasi-Static Solid Particle Damage in Brittle

Solids - I. Observations, Analyses and Implications," Acta Metallurgica 24, 939-

956 (1976). 9A.G. Evans, M.E. Gulden and M. Rosenblatt, "Impact Damage in Brittle

Materials in the Elastic-Plastic Response Régime," Proc. R. Soc. Lond. A 361,

343 (1978). 10D.A. Shockey, K.C. Dao, L. Seaman and D.R. Curran, "Nucleation and

Growth of Cracks in CVD ZnS Under Particle Impact," SRI Annual Report, Part

II, to the Office of Naval Research, Arlington, VA, N00014-76-C-0657 (1979). 11J.T. Hagan, "Shear Deformation Under Pyramidal Indentations in Soda-Lime

Glass," J. Mater. Sci. 15, 1417-1424 (1980). 12S. Van der Zwaag, J.T. Hagan and J.E. Field, "Studies of Contact Damage in

Polycrystalline Zinc Sulphide," J. Mater. Sci. 15, 2965-2972 (1980). 13J. Mescall and C. Tracy, "Improved Modeling of Fracture in Ceramic

Armor," Proceedings of the 1986 Army Science Conference, U.S. Military

Academy, West Point, June 17-20, 1986. 14J. Mescall and V. Weiss, "Materials Behavior Under High Stress and Ultra-

high Loading Rates-Part II," Proceedings of the 29th Sagamore Army Conference,

Army Materials and Mechanics Research Center, Watertown, MA (1984).

Reprinted from International Journal of Impact Engineering, Vol. 9 (3), D.A.

Shockey, A.H. Marchand, S.R. Skaggs, G.E. Cort, M.W. Burkett and R. Parker,


"Failure Phenomenology of Confined Ceramic Targets and Impacting Rods,"

pp. 263-275, 1990, with permission from Elsevier Science.


MICRO-MECHANISMS OF COMPRESSION FAILURE

Sia Nemat –Nasser and Sai Sarva

Center of Excellence for Advanced Materials

Department of Mechanical and Aerospace Engineering

University of California, San Diego

La Jolla, CA 92093-0416

ABSTRACT

Materials such as rocks, concrete and ceramics fail under different modes

ranging from brittle to plastic failure depending on the deformation conditions.

The under pinning micro-mechanisms of dynamic brittle failure in compression

are examined over a broad range of deformation rates, from quasi-static to strain

rates encountered in ballistic experiments. An overview of recent advances in

novel experimental techniques to study the dynamic behavior of brittle materials

is presented. Recent data on damage initiation and evolution in ceramic armor

materials are considered with a view toward deciphering the essential feature of

failure phenomena. It is observed that under moderate confining pressures and at

moderate deformation rates, brittle failure involves initiation of micro-cracks at

dominant micro-flaws and pre-existing micro-cracks and their subsequent

interactive growth, leading to axial splitting, faulting or a mixture of brittle-

ductile failure. Recent data on SiC is compared to a wing-crack array model,

which describes the influence of microstructure on the dynamic behavior of

ceramics.

Under great confining pressures, common in ballistic impact on the other hand,

classical crack-growth models seem inadequate for representing the actual failure

initiation and evolution. Computational simulations of the early stages of impact

response of ceramic armor show development of stress states involving extremely

high shear stresses within the target ahead of the projectile. This suggests a region

conducive for pulverization. Transmission electron microscopy examination of

recovered Al2O3 powder from a confined sample impact-penetrated by W (X21-

alloy) at high velocity shows extensive twinning with sub-micron spacing.

Corresponding author: [email protected] (858) 534-4914, Fax: (858) 534 2727



INTRODUCTION AND BACKGROUND

Ceramics, rock and concrete are characterized by brittle failure under

compression. These materials find varied applications based on their mechanical

properties. SiC, Al2O3 and TiB2 find extensive usage in high velocity impact

applications such as multifunctional armor. It is important to understand the

micro-mechanisms of compressive failure so as to help design improved structural

elements. Under quasi-static compressive loading conditions, micro-structural

factors such as mismatches in elastic compliance between adjacent grains and

inherently present processing flaws (e.g. pores and inclusions) create local tensile

stresses. Tensile micro-cracks originate at pre-existing flaws and grow unstably in

the direction of maximum compressive load. Failure occurs by fragmentation

caused by formation and coalescence of these tensile micro-cracks. This mode of

failure is termed as axial splitting. Micro-mechanical models based on pre-

existing flaws, which include frictional and cohesive resistance, have been

presented to describe the failure process. Brace and Bombalakis1 present a sliding

crack model, also termed as wing-crack. The corresponding failure process has

been quantified analytically by Nemat-Nasser and Horii.2 At higher strain rates it

is observed that a many more micro-cracks are nucleated resulting in finer

fragment size. The Hopkinson bar has been modified and extensively used to

study the dynamic behavior of ceramics.3-6

Sarva and Nemat-Nasser7 have

studied the dynamic compressive strength of SiC under uniaxial compression. It

has been observed that the compressive strength of SiC, drastically increases at

strain-rates higher than 1000 s-1

. Some researchers have focussed on the

energetics of nucleation and growth of these microcracks.8,9

It has also been observed that the dynamic properties of brittle materials are

sensitive to confinement.10

In the presence of moderate confining pressure, failure

is inhibited resulting in an improvement in mechanical properties. Failure

eventually occurs by faulting due to a preferential growth of micro-cracks. Horii

and Nemat-Nasser11,12

have suggested that faulting may be a result of unstable

growth of tension cracks at suitable sets of interacting flaws. Chen and

Ravichandran13

have studied the dynamic compressive behavior of a soft ceramic

Macor under confinement.

When confining pressure is extremely large, a transition from brittle failure by

faulting to a ductile response by overall plastic flow takes place. Horii and Nemat-

Nasser12

include possible zones of plastically deformed materials at high shear

stress regions around pre-existing flaws to model the transition from brittle failure

to ductile flow under very high confining pressures. The use of Hopkinson bar to

study the dynamic response under very high confinement pressures is difficult.

However extremely high inertial confinement can be induced in the ceramics by

high velocity impact.


Deng and Nemat-Nasser14

have proposed a two dimensional model to simulate

dynamic damage evolution in uni-axial compression. To study the strain rate

effect at moderate strain rates and confinements, Nemat-Nasser and Deng15

have

developed a simple model of an array of interacting wing cracks to describe the

influence of microstructure on behavior of brittle materials at high strain rates. A

brief summary of the wing-crack array model is presented below.

Wing-Crack Array Model

Nemat-Nasser and Deng consider an infinite array of interacting, dynamically

growing wing-cracks.15

The wing-crack array is subjected to a dynamic farfield

compressive load. See Fig.1(a). A bi-axial compressive field is considered to

include lateral confinement. The tensile cracks emanating from the tips of wing-

cracks grow unstably in the direction of the maximum compressive load at limited

speeds. Coalescence of these tensile cracks results in failure. The model is

simplified to an array of collinear cracks as shown in Fig. 1(b). The dynamic

stress intensity factor is calculated by superposition of the solution for a crack

array under pairs of concentrated forces applied at the crack centers and the

solution for a crack array under uniform farfield stresses. The solution to this

collinear crack array displays the influence of the microstructure and the loading

conditions on the dynamic behavior. The microstructure is described by the flaw-

size c and spacing w. See Appendix for the mathematical solution.

Figure 1 (a) and (b). Echelon of wing cracks and collinear crack array model


The model helps study the effects of strain-rate, microstructure and lateral

confinement on the compression failure. It predicts that the length scale at which

the material heterogeneities interact with each other leading to micro-cracking and

possible pulverization is dependent on magnitude of compression and strain rate.

At low level of pressure, large defects interact and lead to failure. As the

amplitude of compression is increased, the length scale at which the defects

interact diminishes. The effect of length scale is illustrated in Fig. 2. The model

also predicts that the compressive strength increases with strain-rate and lateral

confinement.


Figure 2. The effect of flaw spacing and confinement on the failure stress as

predicted by the Wing-crack array model15


Uniaxial compression

The Hopkinson bar has been extensively used to study the dynamic behavior

of brittle materials at moderate strain-rates and confinement. Ceramics have very

high compressive strengths and low failure strains. Hence, the classical

Hopkinson bar is modified.16,17,18

Since, the behavior is essentially linear elastic,

a pulse shaper, in form of a copper cushion is placed at the front end of the

incident bar to generate a ramp loading and hence maintain uniform strain-rate.

Strain gages are attached to the sample to help measure strains accurately.

Impedance matched platens increase specimen-bar interface area and help reduce

stress concentrations. Previously reported dynamic tests7 of uniaxial compression

of SiC were performed on a 12.7 mm Hopkinson bar. Elastic wave propagation

relations used to calculate the constitutive behavior are valid at the specimen-bar

interface. However, the strain gages that measure the wave pulses are located


mid-bar length away from the specimen interface. This shift causes perturbations.

To eliminate the error induced due to these perturbations, the wave pulses are

corrected for dispersion. However, dispersion effects limit the accuracy at high

strain rates (>1500 s-1

) for a 12.7mm Hopkinson bar.

Hence, to attain higher strain rates, a 4.76 mm diameter Hopkinson bar is

used. Further experiments were conducted using cylindrical samples of hot-

pressed SiC. The samples were of 2.03 mm diameter and 3.05 mm length. The

samples were polished parallel to a tolerance of less than 3 m. A striker bar of

38.11 mm length was used to attain a strain rate of nearly 9000 s-1

. Impedance

matched platens were used to prevent bar damage. The W4C platens were

confined in Kovar, to improve their strength. The failure stress was calculated

from the transmitted pulse. Due to the miniature size of the samples, it was not

possible to attach strain-gages onto the sample for accurate measurement of strain.

Hence the strain-rate was inferred from the transmitted pulse and the Young’s

modulus of SiC (470 GPa), assuming that the behavior is linear elastic until

failure. To confirm the validity of the above procedure, experiments were

conducted on a 12.7 mm Hopkinson bar, with strain-gages attached to a SiC

sample. Results indicate that the strain-rate calculations made using the above

technique matched well with the strain-rate measured by the strain gauge attached

to the sample.

For other brittle materials such as concrete, mortar, rock and other relatively

coarse microstructures materials, a 76 mm Hopkinson bar can be used. Strain

rates of nearly 1000 s-1

can be attained.

Moderate confinement

Interference fit technique: An interference-fit, maraging steel sleeve can be used

to achieve a static lateral confinement on the SiC samples. Maraging steel has a

Young’s modulus of 200 GPa and a yield strength of more that 2.34 GPa. This

method consists of fitting a sleeve over the sample, with the sleeve’s inner

diameter smaller than the outer diameter of the SiC sample. Due to the radial

misfit, the sleeve exerts lateral confinement on the SiC sample. Recently,

experiments were conducted to study the compressive strength of SiC under

confinement, using this method. See Fig. 3 for the confinement assembly design.

The radial misfit was about 0.01 mm. The sample was mechanically forced into

the sleeve. Alternatively, they can be shrunk-fit. The confining pressure was

calculated by using the solution for a linear elastic axisymmetric boundary value

problem19

. It was calculated to be approximately 300 MPa.

When the sample-sleeve assembly is axially loaded, it undergoes

compressive strain in the direction of loading. However, it expands in the radial

direction. The Poisson's ratio for the maraging steel is much higher than that of

SiC. Hence, it can be expected that as the axial loading increases, the sleeve


expands a larger amount than the SiC sample. This results in the reduction of

lateral confining pressure provided by the sleeve. To counter this, another

maraging steel sleeve, slightly smaller in length was used as an additional sleeve.

The second sleeve was 0.25 mm shorter in length. The inner diameter of this

additional sleeve was chosen to be the same as the outer diameter of the first

sleeve. It helps retain the confining pressure on SiC sample, without inducing any

additional confining pressure. The smaller length of the second sleeve prevents it

from being affected during elastic loading of the sample. Chen and

Ravichandran13

have used a similar technique to laterally confine Macor samples.

The 12.7 mm Hopkinson bar was used to study the dynamic compressive

strength of the confined SiC samples. Since the attachment of strain gauge onto

the sample is not feasible, the strain rate is inferred from the transmitted pulse.

The failure stress was calculated from the transmitted signal. The failure stress

calculation for the confined ceramic is corrected for the inclusion of the metal

sleeve. This adjustment is made by deducting the elastic energy of the metal

sleeve during deformation. The Young’s modulus of maraging steel is 200 GPa.

The sleeve is assumed to be in its elastic regime until the sample fails. This gives

the approximate failure stress for SiC sample.

Figure 3. Sample and confinement design

Pneumatic confinement: For large size samples of materials such as concrete,

rock and polymeric composites, confinement can also be attained pneumatically.20

Fig. 4 shows the 76 mm Hopkinson bar and the pneumatic confinement assembly.

A large diameter pressure vessel, constructed such that it encompasses the entire

sample, provides confinement. The pressure vessel is placed onto the incident and

transmission bar. The proper assembly of pressure vessel is important to ensure

safety and good results. Concrete samples can be tested at confining pressures of a

few hundred MPa and up to a strain-rate of 1000 s-1

, using this system.


Figure 4. Pneumatic confining techniques for a 76mm Hopkinson bar20

Figure 5. Tri-axial test configuration21


Triaxial tests: The Hopkinson bar can be modified to simultaneously load the

sample in radial and axial directions.21

It consists of larger (27.1 mm) and smaller

(19.1 mm) incident bars and transmission bars as seen in Fig. 5. Incident and

transmission tubes which encompass the smaller incident and transmission bars,

but move independently of them, help load a Teflon sleeve. The Teflon sleeve

surrounds the sample and is restricted by an aluminum sleeve on the outside.

During the test, a large hydrostatic stress is induced in the Teflon sleeve, which in

turn exerts a large radial stress on the sample. The radial stress increases during

the test, as the incident and transmission bars axially load the sample and the

Teflon sleeve. The radial stress is estimated by measuring the hoop strain on the

outer surface of the aluminum sleeve. Strain rates of several thousand s-1

and

radial stresses of several hundred MPa can be attained with this system.

Huge inertial confinement

Experimentally it is very difficult to achieve extremely high confining

pressures or strain rates using the Hopkinson bar. However, large lateral

confinements can be attained by high velocity planar or projectile impact, due to

mass inertia. Nemat-Nasser et al.22

have studied the effect of high velocity impact

by Tungsten X21 alloy on the microscopic failure mechanisms of Al2O3. A 2.54

cm thick Al2O3 tile confined in a steel casing was impact penetrated by W (X21-

alloy) projectile (dia. = 4.8 mm) at about 900 ms-1

. A single stage gas-gun was

used to propel the projectile. During the very initial stages of impact, the sample

is shock loaded and a stress-state similar to that of uni-axial strain exists ahead of

the projectile head. However, as penetration progresses, the emanating stress

waves result in a much more complex state of stress. The ceramic fragments from

the pulverized zone were recovered and examined microscopically by TEM.

RECENT EXPERIMENTAL RESULTS AND FAILURE MODES

Uniaxial compression

As can be seen from Fig. 6, ultra-high strain rate tests indicate that the

uniaxial compressive strength of SiC is approximately 8.5 GPa at a strain rate of

9000 s-1

. Fig. 6 also includes previously reported7 uniaxial compressive strength

data for comparison. Similar to previously reported results, failure occurred by

fragmentation as a result of axial splitting.


2

3

4

5

6

7

8

9

10

11

0.00001 0.0001 0.001 0.01 0.1 1 10 100 1000 10000 100000Strain-rate

Com

pressiv

e s

tren

gth

(G

Pa)

Model (confinement = 300 MPa)

Model (unconfined)

Figure 6. Experimental results and comparison to Nemat-Nasser – Deng Model

Moderate confinement

Lateral confinement results in a substantial improvement in the ceramic

strength. The quasi-static failure strength, as measured on an Instron test machine

is 7 GPa. The dynamic strength at a strain rate of 1100 s-1

is 9 GPa. It is seen that

the compressive strength improves by 2 GPa, for a lateral confining pressure of

300 MPa. The strain rate sensitivity of the failure strength is maintained. For the

laterally confined samples, it was observed that the failure is by fault formation

rather than by axial splitting. Fig. 7 indicates the top and bottom view of the

recovered samples. The faults formed were conical in nature. The apex of the

fault cone can be observed in the top view. It runs diagonally across the top face.

The samples were mounted in epoxy resin and ground. The cross section was

observed at regular intervals to examine the failure mode. Fig. 8 is the schematic

of comparison for quasi-static and dynamic failure. It was observed that, for the

same amount of strain, the faults formed for the dynamic test were wider. Also

considerably more microcracking was observed in the dynamic failure as

compared to the quasi-static case.


Static Dynamic

Top view

Bottom view

Figure 7. Top and bottom view of the failed samples under moderate confinement

Figure 8. Schematic of faulting for moderately confined samples


Large confinement

Extremely high compressive stress, lateral confinement and temperatures are

attained due to shock loading during impact-penetration. The ceramic samples fail

by a combination of failure processes, which include pulverization and

fragmentation due to radial and circumferential cracking. In the pulverized zone,

fragments smaller than the grain size are formed. Transmission electron

microscopy of Al2O3 powder recovered from the pulverized zone indicates

extensive localized plasticity.22

Deformation twins in the sub micron scale were

observed. See Fig. 9 (a). A solitary deformation twin has been isolated and its

electron diffraction pattern studied in Fig. 9 (b). The electron diffraction pattern

displays mirror images of a single pattern super-imposed on each other indicating

that the twin is a ‘reflective-twin’. The characteristic axis of the twin is given by

‘m-m’. Part of the atomic lattice crystal has symmetrically re-oriented itself such

that its structure is a mirror image of the parent matrix lattice. Though Al2O3 has

a HCP crystal structure and a very high degree of symmetry, it is not commonly

known to exhibit twinning under moderate loading conditions. Under extreme

loading conditions, twinning can accommodate extensive plastic deformation with

very little volume change. It is expected that micro-cracks resulting in the

eventual pulverization of the ceramic, accompanies twinning.

Numerical simulations23,24

on DYNA 2D ( a two-dimensional hydrodynamic

finite element code) indicate that during the initial stages of impact, release waves

emanating from the edge of the projectile, produce a zone of high shear stress and

low pressure ahead of the projectile at a distance of the order of its diameter. The

state of stress in this zone lasts only a fraction of microsecond, but is sufficient to

produce pulverization of ceramic.

(a)


(b)

Figure 9. Deformation twinning in Al2O3 recovered after impact-penetration by W

(X21-alloy) projectile22

COMPARISON TO NEMAT-NASSER – DENG MODEL

The model has been compared to compressive failure strength data of SiC

for both unconfined and moderate-confinement tests. See Fig. 6. Since the axial

load is substantially larger than the lateral confining pressure, the multi-axial

loading is approximated to bi-axial loading. The model includes the effect of the

microstructure in terms of micro-flaw size and micro-flaw spacing. For

comparison purposes, the model is plotted for pre-existing micro-flaw size of 90

m and flaw spacing of 950 m.

The material constants of SiC are taken as: Young’s Modulus E = 470

GPa; fracture toughness K Ic = 4.5 MPa m ; and the effective Rayleigh wave

speed ms2000Rc-1

. Rayleigh wave speed is the limiting crack velocity. It is

expected that the damage caused by wing-cracks reduces the effective Rayleigh

wave speed of the material. Hence a wave speed, which is approximately third of

an intact material, is chosen. For multi-axial loading, the model is plotted for a


confining pressure of 300 MPa. Other parameters defining the wing cracks are

taken as: co-efficient of friction = 0.4; and preferred orientation of micro-flaws

= 72o.

CONCLUSIONS

It is observed that the compressive strength of SiC improves from a quasi-

static strength of 4.2 GPa to 8.5 GPa at a strain rate of 9000 s-1

. Moderate

confining pressures (300 MPa) can be achieved with a ‘two-sleeve interference fit

technique’. The compressive strength of SiC sharply increases under lateral

confinement. A lateral confinement of 300 MPa improves the compressive

strength by about 2 GPa. The strain-rate sensitivity of the compressive strength is

maintained. Under lateral confinement, the failure mode changes from axial

splitting to fault formation. For cylindrical samples, it was observed that conical

faults were formed. Preferential crack growth results in fault formation. An

increased amount of micro-cracking during a dynamic test results in wider faults

during dynamic tests. The Nemat-Nasser – Deng model gives a quantitative

description of the high strain rate behavior of ceramics. It correlates well with

experimental data obtained from the unconfined and moderately confined tests.

Under high amplitude shock compression, the interaction length scale

diminishes to grain size and eventually to nano dimensions. Classical crack-

growth models are no longer applicable under such conditions. In such a regime

even very brittle solids may deform plastically. Thus, comminution may occur

under great confinement through coupling between shear stress, low pressure and

material microstructure. The resulting failure stress will depend on confinement

as well as length scale at which micro-defects interact. TEM results of Al2O3

samples recovered after impact-penetration by W (X21-alloy) show plastic

deformation in form of extensive deformation twinning.

ACKNOWLEDGEMENT

The US Army Research Office supported this project under Contract No.

DAAH04-96-1-0376, to the University of California at San Diego.

REFERENCES1W.F. Brace and E.G. Bombalakis, “A note on brittle crack growth in

compression,” J. Geophys. Res., 68 3709-3713 (1963)2S. Nemat-Nasser and H. Horii, “Compression induced non-planar crack

extension with application to splitting, exfoliation and rockburst,” J Geo. Phys.

Res., 87 6805-6821 (1982)


3J. Lankford, “Temperature-strain rate dependence of compressive strength

and damage mechanisms in aluminum oxide,” J. Mater. Sci., 16 1567-1578

(1981)4J. Lankford, “Mechanisms responsible for strain-rate dependent compressive

strength in ceramics,” J. Am. Ceram. Soc., 64[2] pC33 –34 (1981) 5G. Subhash and G. Ravichandran, “Mechanical behavior of hot-pressed

aluminum nitride under uni-axial compression,” J. Mater. Sci., 33 1933-1939

(1998)6C.J. Shih, M.A. Meyers, V.F. Nesterenko and S.J.Chen, “Damage evolution

in dynamic deformation of silicon carbide,” Acta. Mater., 48 2399-2420 (2000) 7S. Sarva and S. Nemat-Nasser, “Dynamic compressive strength of SiC under

uni-axial compression,” Mat. Sci. Engng, A317 140-144 (2001) 8J. Lankford and C. R. Blanchard, “Fragmentation of brittle materials at high

rates of loading”. J. Mater. Sci. 26[11] 3067-3072 (1991) 9D.E. Grady, “Local inertial effects in dynamic fragmentation,” J. Appl. Phys.,

53[1] p825 (1982) 10

J. Lankford, “Dynamic compressive failure of brittle materials under

hydrostatic confinement,” Experimental techniques in the dynamics of

deformable solids, AMD 165 1-11 (1993) 11

H. Horii and S. Nemat-Nasser, “ Compression induced microcrack growth in

brittle solids: axial splitting and shear failure,” J Geophys. Res., 90 [B4] 3105-

3125 (1985) 12

H. Horii and S. Nemat-Nasser, “ Brittle failure in compression: Splitting,

faulting and brittle-ductile transition,” Phil. Trans. R. Soc. Lond. A319 337-374

(1986)13

W. Chen and G. Ravichandran, “Dynamic compressive failure of a glass

ceramic under lateral confinement,” J. Mech. Phys. Solids, 45[8] 1303-1327

(1998)14

H. Deng and S. Nemat-Nasser, “Dynamic damage evolution in brittle

solids”, Mech. Mater., 14 83-103 (1992) 15

S. Nemat-Nasser and H. Deng, “Strain-rate effect on brittle failure in

compression,” Acta Metall. Mater., 42[3] 1013-1024 (1994) 16

S. Nemat-Nasser, J.B. Isaacs and J.E. Starrett, “Hopkinson techniques for

dynamic recovery experiments,” Proc. R. Soc. Lond., A435 371-391 (1991) 17

W.P. Rogers and S. Nemat-Nasser, “Transformation plasticity at high strain

rate in Mg-PSZ” J. Am. Ceram. Soc., 73 136-139 (1990) 18

V. Sharma, S. Nemat-Nasser and K.S. Vecchio, “Dynamic-compression

fatigue of hot pressed silicon nitride,” Expt. Mech. 34[4] 315-323(1994) 19

I. H. Shames and F. A. Cozzarelli, “ Elastic and inelastic stress analysis,”

Taylor and Francis Publishing ltd., p539 (1997)


20J. Rome, J.B. Isaacs and S. Nemat-Nasser, “Hopkinson techniques for

dynamic triaxial compression tests,” to appear in Proceedings of Symposium in

honor of I.M. Daniel, Kluwer Academic Publishers, 2002. 21

S. Nemat-Nasser, J.B. Isaacs, and J. Rome, “Triaxial Hopkinson

Techniques,” ASM 8, Mechanical Testing and Evaluation Handbook, 516-518

(2000).22

S. Nemat-Nasser, J.B. Isaacs and B. Kad, unpublished results23

S. Nemat-Nasser and J. Zhang, unpublished results24

S. Nemat-Nasser, S. Sarva, J. B. Isaacs and D.W. Lischer, “Novel ideas in

multi-functional ceramic armor design,” PACRIM IV Conference Proceedings

Maui Nov 4-8, 2001, this volume.

APPENDIX

The Mode I dynamic stress intensity factor at the crack tips in a crack array under

both concentrated and uniform loads is given by

1 21 2 222 2

2

array array array arrayIs IsId Is Is Is

lK k ( l )K k ( l )K , K w tan

w,

(1)

1 2

1

2022 2

2

array array

Is Is

( l l ) lK F cos( ) w sin , K w tan

w w. (1)

As seen from Fig. 1, w is the crack spacing and c is the flaw size. The functions

and , which represent inertia effect during dynamic crack growth in

the stress intensity, are approximately given by

)(1Is lk )(

2Is lk

k (1Is l )

0 75

R

R

c l

c - . l, k (

2Is l )0 5

R

R

c l

c . l

,

(2)

where is the Rayleigh wave speed. where is the driving shear

stress acting on the pre-existing flaw, given byRc 2F c

11 22 11 22 11 22

1 12 2

2 2c( )sin ( )cos , (3)


where is the frictional co-efficient, is the cohesive stress describing the wing

crack. The small length is introduced to make the model applicable at = 0.

µ c

0l l

For fracture criterion, it is assumed that the dynamic stress intensity factor

does not exceed a constant fracture toughness, i.e. . The common

growth speed of the compression induced tension cracks is now

arrayIcIdK K

1 2

1 2

1 5 1 75 1 25

1 5 0 75

array arrayIcIs Is

R array arrayIcIs Is

. K . K . K Xl c

K . K . K (4)

where

(5) 1 2

1 2 1 2

2

1 5 1 75 1 25

4 0 5 0 75 0 375

array arrayIcIs Is

array array array arrayIc IcIs Is Is Is

X . K . K . K

K K - K ( . K . K . K ).

The crack length is obtained by integrating the crack tip speed until failure or

complete unloading occurs. To obtain failure stress for a given stress pulse and

material microstructure, the time to failure is calculated from the crack speed and

the length to which the cracks must grow for coalescence to occur. The failure

stress is then defined by the value of the applied axial compressive stress at crack

coalescence.


DAMAGE MITIGATION IN CERAMICS: HISTORICAL DEVELOPMENTS

AND FUTURE DIRECTIONS IN ARMY RESEARCH

D.M. Stepp

U.S. Army Research Office

P.O. Box 12211

Research Triangle Park, NC 27709-2211

ABSTRACT

U.S. Army Research in materials science continues to address the need for

high performance ceramic materials capable of providing superior protection for

the soldier and vehicles in the battlefield; critical to this need is the elucidation of

physical mechanisms that govern the initiation, propagation, and mitigation of

deformation-induced damage. Examples of significant research advances in this

area include such topics as advanced processing, novel characterization tools,

biomimetics, and the altering of comminuted ceramic flow. The U.S. Army’s

objective force, which requires that ceramic armor materials be extremely

lightweight and affordable, serves to further underscore the critical need for

illumination of these governing mechanisms in order to improve ceramic armor

design with robust methods of damage mitigation.

INTRODUCTION

The U.S. Army Research Office has focused on advanced materials since its

earliest days. In 1968, the Metallurgy and Ceramics division summarized the

needs in one of its primary research focus areas as, “…materials having an

appropriate ratio of strength to density, and showing reliability in inhospitable

environments.” Advanced materials were expected to be developed through

investigating the relationships between the phases within a material and its

properties, and between the properties of a material and the principles and

mechanisms that govern them. By 1982, the newly named Materials Science

division identified primary thrust areas in mechanical behavior and synthesis and

processing. Microstructure-property and processing-performance relationships

were expected to yield the future of advanced materials. Today, with the call for

transformation of U.S. Army forces and specific objectives such as the future



combat system of systems and the objective force warrior, advanced materials

remain at the forefront of Army research.

Although many of the terms and some of the methods have changed, the

underlying need for lightweight materials with superior properties and the

underlying scientific philosophy of developing advanced materials via processing-

structure-property relationships both remain essentially unchanged. These

observations give reason to pause and consider the lessons learned during over

thirty years of ceramics research, that optimal future research directions might be

more clearly identified. Of particular interest is damage mitigation in ceramic

materials, a topic that defines the ultimate figure of merit for armor materials.

Much more than merely penetration resistance, damage mitigation addresses the

need for armor materials to limit the effect of damage on the properties of the

damaged material. As such, proper attention to this capability is essential for a

successful armor materials by design strategy.

BACKGROUND

Processing

The fact that the mechanical properties of ceramics are heavily dependent on

the concentration of agglomerates, inclusions and pores has provided considerable

motivation for exploring optimal and robust methods for processing ceramic

materials. Even early work exploring the mechanical behavior of spinel included

a considerable effort to develop higher purity starting materials and increased

densification [1]. Significant advances in the areas of powder production (purity,

control of particle size distribution, etc.) and processing commercialization

(sintering, slip casting, etc.) have since greatly increased the properties and

availability of a wide variety of ceramics. Although a considerable number of

innovative and promising processing methods have also been developed, they

have generally been limited by cost effectiveness and have not enabled ceramic

armor materials that are vastly superior to those that are available commercially,

particularly in terms of their damage mitigation capabilities.

Mechanical Properties and Characterization

Despite excellent hardness and compressive strengths, the inability to provide

ceramics with substantial toughness has limited their widespread use in many

applications. Although ceramic materials research efforts have demonstrated

substantial strength improvements with additives [2, 3], and microstructural

reinforcements [4, 5], most have not been found to increase the toughness

substantially. Research exploring degradation in the mechanical properties of

ceramics due to such factors as moisture, fatigue loading, and residual stress

concentrations has demonstrated the majority of these mechanisms to ultimately

be governed by the formation and propagation of microcracks [6]. Similarly,


thermal shock studies have shown the strength of fractured ceramics is inversely

proportional to the number of cracks that have propagated [7]. It is therefore not

surprising that ceramics incorporating layers, gradients, or confinement, each of

which enhances the ability to inhibit crack formation and impede or deflect crack

propagation, have consistently demonstrated significant property improvements.

In the case of armor, the variations in ballistic performance, inability to

tolerate significant plastic deformation, and propensity for brittle fracture have

severely diminished efforts to develop rugged ceramic armor materials. The

difficulty in achieving this goal is complicated further by the fact that there exists

a significant lack of correlation between quasi-static and high strain-rate

mechanical properties; therefore, even the most substantial improvements in

quasi-static fracture or strength properties tend to provide only marginal, if any,

improvements in ballistic performance. One notable exception to this observation

is confined ceramics, which appear to offer good potential for both structural and

armor applications; however, the cost associated with processing these materials,

particularly with the very large confinement stresses required for ballistic

performance, have been prohibitive to date.

This lack of correlation has provided considerable motivation to develop

experimental tools and characterization techniques to understand the fundamental

mechanisms that govern impact behavior in order to identify and improve upon

salient materials properties and enable the design of superior ceramic armor

materials. The fundamental understanding of the mechanisms by which ceramic

materials deform during impact loading has been advanced tremendously by the

development, modification, and augmentation of such techniques as Taylor

impact, Kolsky and split-Hopkinson bar, plate impact, explosive cylinder, and

spherical cavity expansion. However, depth of penetration (DOP) testing remains

by far the most accepted and predictive method for deriving the ballistic

performance of materials [8]. Nonetheless, accuracy and validity of DOP testing

is difficult to assure, and the technique is often limited and misinterpreted by a

lack of appropriate statistical analysis.

Modeling and Simulation

Due to the difficulties associated with developing robust and valid

experimental tools for predicting ceramic armor material performance, and the

extraordinary costs associated with full-scale armor testing, it is not surprising

that efforts to develop computational predictive models have been numerous. A

wealth of models and simulations have been developed, in many cases

significantly advancing the state-of-the-art in computational theory in order to

address the complexities of penetration behavior in a physically meaningful

manner. At the same time, the desire for increased accuracy has also resulted in

models with required parameters that are so numerous or nonphysical that the


results are difficult to interpret. While many simulations have been validated by

comparisons with experimental results, none has emerged with the predictive

capabilities necessary to substantially influence armor design. Further, although

computational models and simulations, particularly when combined with some of

the advanced experimental characterization techniques discussed in the previous

section, enable an enormously detailed analysis of the prevalent

micromechanisms during the penetration process, they have provided no

substantial improvements in damage mitigation capabilities.

CERAMICS IN NATURE

Although no biological organism or system is known to possess substantial

ballistic protection or damage mitigation capabilities, numbers of the robust and

adaptive materials systems found in nature exhibit considerable tolerance for

other forms of deformation damage. For these reasons, the potential exists, albeit

a tremendous challenge, to illumine the mechanisms by which these biological

materials respond and apply them, in a manner appropriate for modern ceramic

materials and high-strain rate response, in order to develop improved ceramic

armor materials capable of mitigating significant damage.

Damage Accumulation in Biological Systems and Ductile Carbides

Analysis of biological ceramic systems has demonstrated an intriguing affinity

for damage accumulation. Both Strombus gigas (Conch) [9] and Haliotis

rufescens (Abalone) [10] shells have been characterized with quasi-static and

dynamic compression and three-point bending tests. In both cases, significant

orientation and strain-rate dependencies were observed. Failure strengths at high

strain-rates (i.e., >103) were measured to be approximately 50% greater than those

measured at quasi-static rates. The materials also exhibited an extraordinary

affinity for damage accumulation prior to failure. Crack deflection, delocalization

of damage, plastic microbuckling (kinking), and viscoplastic deformation were

determined to be the significant mechanisms governing the mechanical response

and enabling the observed damage accumulation.

Recent characterization of bulk Ti3SiC2 has shown a truly remarkable

similarity to the aforementioned biological ceramic systems. These ductile

carbides are characterized by a “layered” unit cell consisting of planer Si layers

connected by TiC octahedra. Mechanical characterization of these materials has

shown significant plastic deformation and the ability to tolerate a considerable

degree of damage. In microstructural analysis, crack deflection, diffuse

microcracking, buckling, delamination, crack deflection, grain push-out, and grain

pull-out were determined to be the significant mechanisms governing the

mechanical response and enabling the damage tolerance. [11, 12]


The similarities between the observed deformation mechanisms for these

materials and their ability to tolerate considerable mechanical deformation while

retaining structural integrity are quite surprising. The potential for further

development of layered materials with increased damage accumulation

capabilities, particularly at the scale of the unit cell, appears very strong. Further

work in this area is expected to provide considerable benefit for damage tolerant

structural applications, and possibly for advanced armor materials, if the impact

response of the material can be made to delocalize rapidly enough so as to

mitigate the local stresses and strains within the material.

Lustrin fibers and self-healing

Additional characterization of the abalone shell has led to the discovery of

lustrin, an elastomeric adhesive protein that binds the aragonite plates [13, 14].

Careful analysis of this protein found it to be a modular polymer consisting of

repetitive modular domains. When a strain is applied, these domains enable a

sequential, and reversible, unfolding that provides a “self-healing” response for

the bulk material via sacrificial bonds within the protein, thereby providing

additional fracture resistance and damage mitigation. Although a direct

application of this approach is expected to be of only minimal advantage to an

armor material, the concept is worthy of consideration in order to identify an

appropriate analogue that would enable ceramic materials to substantially mitigate

ballistic damage.

COMMINUTED CERAMIC FLOW

Analysis of penetrated and partially penetrated ceramic materials has led to

the observation of a comminuted zone, also referred to as the Mescall zone, near

the leading edge of the penetrator. Both the resistance to comminution and the

ability of the penetrator to move through the resulting comminuted ceramic

particles have been identified as significant factors governing the ballistic

performance of a ceramic material. In fact, it is primarily on the basis of these

factors that the enhanced performance of confined ceramics has been explained.

In light of these experimental observations, the FRAGBED models were

developed to simulate penetration solely as a combination of fracture,

comminution, compaction, and fragment flow [15, 16]. The model is similar in

principle to atomic dislocation theory, allowing blocky fragments to glide in

discrete increments along fixed material planes, at speeds determined by the local

stresses acting upon them. In addition, fragments are allowed to reduce further in

size by the incorporation of a comminution rate law. This approach to penetration

modeling penetration phenomena demonstrated good agreement with ballistic

testing results, thereby identifying comminuted fragment flow as, potentially, a


highly significant mechanism in governing the ballistic performance of ceramic

materials.

More recently, high speed photography and high speed X-ray analysis have

demonstrated that even membrane confinement (e.g., fiber-reinforced packing

tape) of ceramic tiles can substantially alter the shape of the ejecta plume, which

appears to have a significant effect on the ballistic performance of the ceramic

[17]. With minimal front-surface confinement, the ejecta plume has been found

to become less divergent; this focusing of the ejected material appears to have the

effect of further eroding the penetrator, thereby increasing the ballistic

performance of the ceramic. Although this work is preliminary, it provides what

appears to be a novel direction for future ceramic armor research, namely the

mitigation of damage by directing the comminuted ejecta plume against the

penetrator in an optimal manner.

CONCLUSIONS

For more than thirty years, U.S. Army research has investigated and

developed high performance ceramic materials capable of providing superior

armor protection for both the soldier and vehicles. Significant research advances

have been made in low-cost reliable processing and in the characterization and

computational simulation of ceramics during impact. As the Army’s need for

lightweight rugged armor materials becomes increasingly rigorous, efforts to

develop new ceramic materials with significant damage mitigation capabilities

become ever more important.

Despite the numerous advanced processing techniques that have been

developed, cost effectiveness and processing variations continue to severely limit

their applicability to advanced armor materials. Wherever possible, future

processing research efforts should utilize existing commercial processes and focus

on the problems pertaining to the most promising materials solutions (e.g.,

confined ceramics).

While considerable achievements have been made in both experimental and

computational characterizations of ceramic materials, these must be fused to

enable robust predictive armor design tools. One concept of particular interest in

this area is to provide physically-based predictions of the ultimate performance

potential of a wide range of ceramic armor materials and designs, thereby

enabling criteria for eliminating inadequate armor solutions and stimulating

optimum design and processing efforts.

Extraordinary examples of self-healing and damage tolerance capabilities

have been found in biological ceramic systems. Although ductile carbides have

recently been shown to exhibit very similar damage tolerance behavior,

tremendous challenges still remain in elucidating the governing mechanisms in

biological systems and applying them properly; the goal is to obtain a similar


level of relative improvement in damage mitigation, but with modern ceramic

armor materials, and during ballistic loading. The potential value of such an

accomplishment would be truly unprecedented.

Finally, recent work has demonstrated the ability to alter the comminuted

ejecta plume and thereby influence the ballistic performance of the ceramic.

Establishing more rigorous control of the ejection process and the potential for

disrupting penetrator performance, including perturbing the path and orientation

of the penetrator, appear to constitute a highly innovative direction for future

ceramic armor research with excellent potential for damage mitigation.

REFERENCES 1H. Palmour III and D.R. Johnson, “Phenomenological Model for Rate

Controlled Sintering,” in Sintering and Related Phenomena, G.C. Kuczynski,

N.A. Hooton, and C.F. Gibbon, eds., Gordon and Breach, New York, 1967. 2R.M. Spriggs, L. Atteraas, and S.K. Dutta, “Strengthening of

Thermomechanically-Processed Polycrystalline Magnesia by Alloying,” in

Structural Ceramics and Design With Brittle Materials, S. Acquaviva and S.

Bortz, eds., Gordon and Breach , New York, 1969. 3J.M. Marder, T.E. Mitchell and A.H. Heuer, “Precipitation From Cubic ZrO2

Solid Solutions,” Acta Metallurgica 31 387 (1983). 4H. Palmour III, “Multiple Slip Processes in Magnesium Aluminate at High

Temperatures,” Proceedings of the British Ceramic Society 6 209-224 (1966). 5J. Hong et al., “Directional Solidification of SiC-B4C Eutectic,” Materials

Research Bulletin 14 775 (1979). 6A Venkateswaran and D.P.H. Hasselman, “Elastic Creep of Stressed Solids

Due to Time-Dependent Changes in Elastic Properties,” Journal of Materials

Science 16 1627-32 (1981). 7B.E. Bertsch, D.R. Larson, and D.P.H. Hassleman, “Effect of Crack Density

on Strength Loss of Polycrystalline Alumina Subjected to Severe Thermal

Shock,” Journal of the American Ceramic Society 57 (6) 235-36 (1974). 8Z. Rosenberg et al., “On the Relation Between the Penetration Resistance eof

Ceramics and Their Dynamic Properties,” Proceedings of the 6th

International

Conference on Mechanical Behavior of Material, ICM 6, 29 July – 2 August

1991, Pergamon Press, 1991. 9R. Menig, M.H. Meyers, M.A. Meyers, and K.S. Vecchio, “Quasi-static and

Dynamic Mechanical Response of Strombus gigas (Conch) Shells,” Materials

Science and Engineering A – Structural Materials Properties Microstructure and

Processing 297 [1-2] 203-211 (2001).


10R. Menig, M.H. Meyers, M.A. Meyers, and K.S. Vecchio, “Quasi-static and

Dynamic Mechanical Response of Haliotis rufescens (Abalone) Shells,” Acta

Materialia 48 [9] 2383-2398 (2000). 11

T. El-Raghy, A. Zavaliangos, M.W. Barsoum, and S.R. Kalidindi, “Damage

Mechanisms around Hardness Indentations in Ti3SiC2,” Journal of the American

Ceramic Society 80 [2] 513-516 (1997). 12

M.W. Barsoum and T. El-Raghy, “Room-Temperature Ductile Carbides,”

Metallurgical and Materials Transactions A 30A 363-369 (1999). 13

B.L. Smith et al., “Molecular Mechanistic Origin of the Toughness of

Natural Adhesives, Fibers and Composites,” Letters to Nature 399 [6738] 761-

763.14

X. Shen et al., “Molecular Cloning and Characterization of Lustrin A, A

Matrix Protein From Shell and Pearl Nacre of Haliotis Rufescens,” Journal of

Biological Chemistry 272 [51] 32472-32481.” 15

D.R. Curran, L. Seaman, T. Cooper, and D.A. Shockey, “Micromechanical

Model for Comminution and Granular Flow of Brittle Material under High Strain

Rate Application to Penetration of Ceramic Targets,” International Journal of

Impact Engineering 13 53-83 (1993) 16

J.T. McGinn, R.W. Klopp, and D.A. Shockey, “Deformation and

Comminution of Shock Loaded -Al2O3 in the Mescall Zone of Ceramic Armor,”

Materials Research Society Proceedings 362 61-66 (1995). 17

S. Nemat-Nasser, “Experiments and Modeling,” presented at the Fourth

Annual Review of the Damage Tolerant Lightweight Armor Materials MURI

Programs, Aberdeen Proving Ground, MD, June 27, 2000.


INDENTATION DAMAGE BEHAVIOR OF ARMOR CERAMICS

Do Kyung Kim and Chul-Seung Lee

Dept. of Materials Science and Engineering

Korea Advanced Institute of Science and Technology

Taejon, Korea

Chang Wook Kim and Soon Nam Chang

Agency for Defense Development

Taejon, Korea.

ABSTRACT

Hertzian indentation technique has been suggested to analyze the damage

response of ceramic materials under the concentrated loading. In the sense that

the impact loading is the specific case of indentation loading, typical armor

ceramics were analyzed by the indentation technique. Experimental indentation

along with numerical calculation was performed to evaluate elastic property, the

yielding stress, and quasi-plastic property. Bonded-interface technique could

provide the observation of subsurface damage pattern after indentation. It is

suggested that quasi-plastic property as well as elastic property is closely related

with the resistance against the ballistic penetration.

INTRODUCTION

Ceramics shows an excellent performance as an armor materials due to its

high hardness and Young’s modulus but low Poisson’s ratio and density. However,

it has not been fully understood which physical, chemical and mechanical



Figure 1. Hertzian contact of sphere on flat ceramic specimen. Beyond elastic

limit, contact initiates cone fracture (“brittle mode”) or subsurface (“quasiplastic

mode”) [4].

properties influence the armor characteristics of ceramics. Fracture toughness,

strength and hardness data do not provide the enough information to correlate

material properties and armor performance. Even though the dynamic hardness of

ceramics showed some relation with the impact resistance,[1] it is not well known

about the dominant interaction of projectile or shaped charge jet with ceramics.

When a projectile impacts on the ceramics, the stress and the damage

distribution of ceramics are similar to the case of Hertzian indentation, the sphere

indentation on flat ceramics[2, 3]. The sphere indentation on ceramics can provide

the indentation stress-strain relation of ceramics over the wide range of strain. The

compressive stress below the contact area reaches to tens of GPa even with the

normal mechanical testing machine[4, 5], which is comparable to HEL (Hugoniot

Elastic Limit) of the material. It has been suggested that the ballistic efficiency of

armor material is proportional to the average of compressive yield stress and the

HEL of the material[6]. Therefore the characterization of sub-surface microscopic

change of ceramics at a high compressive stress would be helpful to correlate the

materials parameters and the ballistic resistance.


In this study the basic mechanical properties of ceramics such as elastic

modulus, Poisson’s ratio, strength, hardness and toughness were characterized.

From the Hertzian indentation method, the microscopic changes were observed

and the indentation stress-strain curves were measured. Yield stress and

strainhardening coefficient were estimated from the numerical analysis based on

the experimental data.

EXPERIMENTAL

In this experiment, well-known eight armor ceramics were evaluated, two

aluminas (AD85, AD90), three silicon carbides (reaction bonded, solid state

sintered and hot pressed), aluminum nitride, boron carbide, and titanium diboride.

The microstructural and mechanical properties are shown in Table I. Specimens

were cut into 3 × 4 × 35 mm and polished. Microstructures were observed using

SEM. Elastic modulus and Poisson’s ratio were measured by pulse-echo method.

Strength was measured by 4-point flexure test and Vickers indentation was used

to measure the hardness and toughness[7].

Figure 1 shows the schematic configuration of Hertzian indentation test. A

spherical ball of radius r is pressed over the flat polished specimen. Beyond a

critical load, either a Hertzian cone crack (“brittle solid”) or a subsurface

deformation zone (“plastic solid”) initiates[4]. At normal load P, the contact

radius a is given by

')'1()1(

16

9,

3

4 223

E

Evvkwhere

E

rkPa (1)

The prime notation denotes the indenter material. The contact radius a defines the

spatial scale of the contact field. The mean contact pressure,

2/ aPpo (2)

defines the intensity of the contact field. From equation (1) and (2), the useful

indentation stress strain relation is defined by


r

a

k

Epo

4

3 (3)

Equation (3) means a linear relationship between p0, “indentation stress,” and a/r,

“indentation strain,” leading to a procedure for obtaining basic stress-strain

information. From the contact radius a and load P, the indentation stress and

strain can be experimentally obtained.

Applied load was in the range of 500 and 2000N. Indentation was also made

on “bonded specimen” which was made and polished from two polished

specimens bonded with the glue[8]. Spherical ball was loaded at the exact

position of bonded interface and after detaching the sample using acetone, the

subsurface damage mode was observed. All damage behaviors were observed

using optical microscopy with Nomarski interference after the specimen was

coated with gold.

Finite element computations in this study of elastic-plastic contacts were

carried out using a commercial package (Strand7, G+D Computing, Sydney,

Australia). The configuration is modeled as a sphere of specified radius in

axisymmetric frictionless contact with the flat surface of a half-space, 4 × 4 × 4

mm. Plastic deformation in the test material is governed in our calculations by a

critical shear stress criterion with linear strain-hardening. By imposing a generic

uniaxial compression, stress-strain response of specimen can be described as,

)()(

)(

Y-YEY

YE (4)

where Y is the yielding stress and is the strain-hardening coefficient in the range

0 1 ( = 0, fully plastic; = 1, fully elastic). From this uniaxial

compression, result of numerical indentation stress-strain curves were measured

and compared with the experimental curve. Maximum 50 times of calculations

were iterated to converge the strain-hardening coefficient .


Table I. Characterization of microstructure, density and basic mechanical

properties of each specimen

Abbrev. Sample Condition Grain

shape

Grain size

( m)

Density

(g/cm3)

AD85 AD85 Mixed 5 3.44

AD90

Al2O3

AD90 Mixed 2 3.59

AlN AlN Equi-axed 5 3.37

RBSC Reaction

bonded

Bimodal 40,

4

3.08

S-SiC Solid

sintered

Equi-axed 5 3.17

HP-SiC

SiC

Hot

pressed

Equi-axed 2-10 3.22

B4C B4C Hot

pressed

Equi-axed 2-8 2.5

TiB2 TiB2 Hot

pressed

Equi-axed 5-20 4.48

Abbrev. Sample Modulus

E (GPa)

Poisson

ratio,

Strength

(MPa)

Hardness

H (GPa)

Toughness

T

(MPa.m

1/2)

AD85 236 0.230 266 9.2 3.23

AD90

Al2O3

278 0.229 309 12.8 3.19

AlN AlN 327 0.231 288 11.2 2.49

RBSC 394 0.175 440 18.6 3.69

S-SiC 440 0.168 553 29.1 2.46

HP-SiC

SiC

442 0.174 525 19.5 3.75

B4C B4C 456 0.167 390 27.3 3.66

TiB2 TiB2 564 0.081 293 20.6 4.38


Figure 2. Half-surface and side views of Hertzian contact damage in (a) AD85,

(b) AD90 (c) AlN (d) RBSC from the WC sphere of radius r = 1.98 mm at P =

1500 N. Reflection optical micrographs of bonded-interface specimens in

Nomarski illumination.

RESULT AND DISCUSSION

Materials Characteristics

Table I shows the result of microscopic characterization and measurement of

basic mechanical properties. TiB2 shows the highest density and B4C shows the

lowest value but the others are similar. In case of Young’s modulus, TiB2 also

shows the highest value. Three kinds of silicon carbides and B4C are the second

highest group. AlN and two kinds of alumina show the lowest modulus, which

means lowest E/d ratio.

Microstructures can be divided as three groups. AlN, S-SiC, HP-SiC, B4C,

and TiB2 have the equi-axed shape and broad size distribution. On the contrary,

RBSC has a bimodal distribution. In case of alumina elongated and equi-axed

grain shape are mixed but the grain size of AD85 is about twice larger than that of


AD90. S-SiC and B4C shows the highest hardness. B4C has the highest strength to

weight ratio due to its lowest density.

Figure.2 (continued). Half-surface and side views of Hertzian contact damage in

(e) S-SiC (f) HP-SiC (g) B4C, (h) TiB2 from the WC sphere of radius r =

1.98mm at P = 1500 N. Nomarski optical micrographs of bonded-interface

specimens.

Contact Damage Behavior

Figure 2(a)-(h) compares the section views of the bonded interface including

contact damage of top and side view from the indentation using a WC ball with a

radius of 1.98 mm at 1500 N. Subsurface is observed using bonded specimen

method. Two kinds of opposite behavior are obvious; cone crack mode, and

quasiplastic mode. Ring cracks on surface are connected to subsurface cone

cracks. AD85, AD90, and AlN show the typical damage zone behavior. On the


contrary RBSC, S-SiC, and B4C show the typical cone crack damage. HP-SiC and

TiB2 have both kinds of damage characteristics. AD85 – grain size is about twice

larger than that of AD90 – shows larger damage zone than that of AD90. It is

expected that from the analysis of damaged zone the ballistic penetration can be

evaluated.

Figure 3. Side view of damaged zone and FEM-generated stress contours of HP-

SiC. Indentation with WC sphere, radius r = 1.98 mm, load P = 1500 N; (a) side

view of bonded specimen, (b) contour of maximum principal shear stresses with

yielding zone shaded (4.7 GPa at the boundary of shaded zone).

Figure 3 shows the side view of damaged zone with the FEM-generated

contours of maximum shear stress in HP-SiC. The area of stress contour over 4.7

GPa is similar with that of damaged zone. All material constant used for this

calculation are represented in Table II. Yield stress (Y) is defined as the deviation

from the Hertzian elastic limit and can be estimated by experimental data. Strain-

hardening coefficient ( ) is calculated through the iteration of FEM results.

Materials with low show the damaged zone behavior. On the other hand

specimens with high are observed to have the brittle cone cracking behavior.

Figure 4 represents those two opposite characteristics using the indentation

stress-strain relation: (a) in the case of high Y and , (b) in the case of low Y and

. If of a material is zero, it shows fully plastic behavior. And if is 1, it is

considered as a fully elastic material. The result of AD85 is inserted to both (a)

and (b) as a reference date. In figure 4 (a), S-SiC, B4C, and HP-SiC are a group of

showing high Y ( 10 GPa) and high (=0.7-0.8). TiB2 shows a flat graph after

yielding owing to small ( 0.4). In figure 4 (b), RBSC, AD90, AlN, and AD85

show low Y and lower . Each specimen has low yield stress about from 6 to 10

GPa, which is considered to be the limit of elastic regime during impact loading.


From the results, indentation stress-stress curve is thought to have some

relationship with impact resistance.

Table II. Elastic and yield parameters for materials used in finite element

modeling

Abbrev. E (GPa) H (GPa) Y (GPa)

AD85 0.230 236 9.2 6.11 0.5

AD90 0.229 278 12.8 7.05 0.6

AlN 0.231 327 11.2 6.58 0.3

RBSC 0.175 394 18.6 6.58 0.6

S-SiC 0.168 440 29.1 9.4 0.7

HP-SiC 0.174 442 19.5 8.93 0.8

B4C 0.167 456 27.3 10.34 0.7

TiB2 0.081 564 20.6 8.46 0.4

WC 0.22 614 19.0 6.00 0.1


Figure 4. Hertzian indentation stress-strain curves are plotted for each

specimen. In both (a) and (b), AD85 is inserted as a reference. Data points are

from the experimental measurements. Solid curves are FEM fit as the value of

indicated in Table II.

CONCLUSION

Sphere-indentation technique has been suggested to analyze the damage

response of armor ceramic materials. A special bonded-interface specimen could

provide the observation of subsurface damage pattern after indentation.

Indentation stress-stain curves of each ceramic in elastic-plastic range could be

constructed by the experiments along with the numerical calculations. It is

suggested that quasi-plastic property as well as elastic property is closely related

to the resistance against impact loading.

REFERENCES1D. B. Marshall and A. G. Evans, “Measurement of Dynamic Hardness by

Controlled Sharp-Projectile Impact,” J. Am. Ceram. Soc., 66[8] 580-585 (1983). 2A. G. Evans and T. R. Wilshaw, “Quasi-Static Solid Particle Damage in

Brittle Solids- I. Observations, Analysis and Implications,” Acta Metall., 24, 939-

56 (1976). 3A. G. Evans and T. R. Wilshaw, “Dynamic solid particle damage in brittle

materials: an appraisal,” J. Mater. Sci., 12, 97-166 (1977).


4B. R. Lawn, "Indentation of Ceramics With Spheres: A Century After

Hertz," J. Am. Ceram. Soc., 81 [8] 1977-94 (1998). 5B. R. Lawn and T. R. Wilshaw, "Indentation Fracture: Principles and

Applications," J. Mater. Sci., 10 [6] 1049-81 (1975). 6G.R. Anstis, P. Chantikul, B.R. Lawn, and D.B Marshall, “A Critical

Evaluation of Indentation Techniques for Measuring Fracture Toughness: I, Direct

Crack Measurement,” J. Am. Ceram, Soc., 64 [9] 533-538 (1981). 7H. Chai, M. A. S. Kalceff and B. R. Lawn, “Deformation and Fracture of

Mica-Containing glass-Ceramics in Hertzian Contats,” J. Mater. Res., 9 [3] 762-

770 (1994). 8B. R. Lawn, N. P. Padture, H. Cai and F. Guiberteau, "Making Ceramics

"Ductile"," Science, 263 1114-16 (1994). 9A. C. Fischer-Cripps and B. R. Lawn, "Stress Analysis of Contact

Deformation in Quasi-Plastic Ceramics," J. Am. Ceram. Soc., 79 [10] 2609-18

(1996).


PROGRESS IN THE 3-D VISUALIZATION OF INTERIOR BALLISTIC DAMAGE IN ARMOR CERAMICS

Joseph M. Wells*, Nevin L. Rupert, and William H. Green U.S. Army Research Laboratory, Weapons and Materials Research Division Bldg 4600, APG, MD 21005-5069

ABSTRACT The authors present an overview of their research results utilizing X-ray

Computed Tomography (XCT) techniques to nondestructively reveal the internal meso-scale damage morphology within encapsulated armor ceramic targets of TiC, TiB2, Al2O3, and SiC. Examples of the physical damage observed in situ include traditional conical, radial and laminar cracking in impacted samples both with and without penetration. Additional observations reveal instances of outer edge radial cracks, periodic through-thickness laminar cracks, and concentric inplane circular cracking "beach-marks". Examples of asymmetric mixed-type cracking damage isolation and of the spatial distribution of residual tungsten alloy penetrator material are also presented for improved 3-D visualization of complex internal damage conditions. Finally, the authors discuss the premise that this observed meso-scale cracking contributes significantly to the onset conditions for penetration.

INTRODUCTION The physical damage resulting from a high velocity impact of a sub-scale long

rod penetrator with the surface of an armor ceramic is of significant interest to the armor materials community. Even in the case of complete dwell and destruction of the penetrator at the ceramic front surface, i.e. interface defeat, there is considerable damage internal to the ceramic target. With penetration, this damage increases in addition to the growth of a penetration cavity and the deposition of residual penetrator material. Such damage may consist of micro-scale cracking and deformation twinning in a comminuted region immediately under the impact location and of meso-scale cracking in the surrounding elastic ceramic. It is the premise of the authors that the extent and morphology of the latter meso-scale cracking and its resultant structural degradation have a major influence on the cessation of interface dwell and the initiation of penetration.



Under constraint conditions where an impacted ceramic remains substantiallyintact, it is desirable to characterize the internal damage to understand the nature and location of that damage. Such characterization has been neither effective nor easy to conduct nondestructively. Hence, most prior characterization has been conducted by selective and destructive sectioning and polishing. To better understand the meso-scale (>200µm) details of damage behavior and failure of opaque armor ceramic materials, the authors have applied the nondestructivemethod of X-ray Computed Tomography, XCT. A brief overview of the XCT techniques utilized and several results are presented to demonstrate the innovative and powerful capabilities of this method in the 2-D and 3-D visualization ofinternal damage. Ceramic specimens examined include TiC, TiB2, Al2O3, and SiC. The in situ damage presented occurred predominantly by high velocityballistic impact except for the as-fabricated Al2O3 encapsulated assembly.

IMPACT DAMAGE IN TiCAn evaluation of impact damage in a titanium carbide, TiC, armor ceramic tile

was conducted with details reported elsewhere [1,2]. This sample was impactedwhile confined in a heavy steel encasement that was disassembled prior tosectioning and then a sample half-disk nondestructively scanned for XCTanalysis. A particular feature of this sample is that it was not penetrated, but rather supported extensive dwell or interface defeat of the tungsten alloy penetrator at the impact surface. Nevertheless, appreciable meso-cracking damage was

(a)

(b) (c)

Figure 1. A 3-D Solid Visualization of meso-cracking observed on vertical(b) and horizontal (c) virtual sections of a TiC half-disk sample after interfacedefeat ballistic impact on the front face [1,2].


25mm

observed in the interior of this sample. Two quite different types of XCT reconstruction images were prepared to assist in the 3-D visualization of thisdamage. The first is a 3-D solid visualization of the sample half-disk (see figure1a) which has two virtual sections revealing the interior damage on these sections.

The second XCT visualization mode (shown in figure 2) is a reconstructionknown as a 3-D "point cloud" wherein only the selected XCT data related to cracking location, orientation and size are shown. All of the XCT data relating to the opaque ceramic itself has been removed thus allowing the defect cracking to be more easily visualized in isolation. Characteristic features of radial, laminarand conical meso-cracking are clearly observed in the overall asymmetricaldamage condition shown in the front, top and side views of figure 2.

Bifurcated CracksRadial Cracks

Periodic Laminar Cracks

Cone Cracks

Figure 2. X-ray CT Point Cloud virtual representations of 3-D meso-crackingdamage morphology in a TiC ceramic target resulting from ballistic impact. Notethe cracking images are isolated in space without the opacity effects of the TiCmaterial. The outlines of the original sample are superimposed for clarity. [2]

IMPACT DAMAGE IN TiB2

An evaluation of impact damage in titanium diboride, TiB2, armor ceramicdisks was conducted with details reported elsewhere [3,4]. These samples, 72 mm diameter by 25 mm thick, were encapsulated in a titanium alloy, Ti-6Al-4V,welded case and impacted by an L/D=10 tungsten alloy penetrator. Among themore interesting results of this work are:(1) with penetration, the residual


penetrator debris was observed in 3-D in a through thickness columnarconfiguration with some dispersion along the side branching cracks (see figures3a & b), (2) considerable meso-cracking damage is observed radially outwardfrom the impact cavity to the outer circumference of the samples (see figure b), (3) radial cracking originating from both the penetration cavity and the outer diameter of the disk was also observed (see figure c) and (4) circumferential orcircular "beach-mark" cracking is also observed in figure 3c.

(a)

(c)

(b)

Figure 3. Virtual 3-D solid sections through the 72 mm dia. x 25 mm thick TiB2

disk (a & b) with penetrator residue and meso-cracking. XCT scan (c) near impactsurface of TiB2 sample showing radial and circular cracks [3,4].

Fabrication Damage in Encapsulated Al2O3

An XCT evaluation of the internal damage in a titanium alloy, Ti-6Al-4V,encapsulated sample with both aluminum oxide, Al2O3, and silicon carbide, SiC,ceramic tiles was conducted in the as-fabricated condition. It is desirable to determine the existence and nature of any initial baseline damage existing prior to ballistic impact.

As shown in figure 4(a), significant damage was revealed in the two left mosttiles in the digital radiograph, DR, taken of the encapsulated sample on edge.Cracking is visible in both of the lower density Al2O3 tiles on the left but not inthe third higher density SiC tile on the right hand side. A XCT scan (see figure


4b) taken through the center Al2O3 tile clearly reveals both corner cracks as wellas radial cracking starting in the center and extending outward to the 9 o'clockposition. Such information is useful in the modification of either the targetarchitecture design and/or the fabrication-processing conditions.

(a) (b)

Figure 4. Digital radiograph (a) and X-ray CT scan (b) showing as-fabricatedmeso-cracking damage in Al2O3 tile encapsulated within Ti-6Al-4V sample.

Impact Damage in Encapsulated SiC A preliminary XCT evaluation of the internal damage in a titanium alloy, Ti-

6Al-4V, encapsulated silicon carbide, SiC, ceramic tile sample was conducted in the impacted condition. Partial penetration was experienced with significant lossof ceramic material in the form of a concave cavity. The front face cavity wasfilled with an epoxy resin prior to machining the sample to its reduced size (seefigure 5a). The original sample was too large for XCT with the in-house unit and consequently was machined to 4.75 in square x 0.9 in thick from the originalencapsulation target. As shown in Figure 5b, the damage revealed in the XCT scan at the 13 mm height level (~10 mm from front impact face) in the SiC tileconsists of large asymmetric voids (missing material) and spiral or circular meso-cracking damage rings with several connecting cracks between them. The secondscan image, only 3 mm from the rear face of the tile, is quite different from thefirst. The two scan images are at different depths with a scan thickness of 0.5 mm and a mean separation of 10 mm in the direction of ballistic impact.


(a)

(b) (c)

Figure 5. Macro-photograph of epoxy coated Ti-6Al-4V/ SiC impacted sample (a)and two XCT scans: (b) at 13 mm from rear face and (c) at 3 mm from rear faceshowing interior meso-scale damage in 100 cm square SiC tile.

Premise of Critical Damage LevelA general schematic of the types of meso-scale impact damage observed via

XCT in the encapsulated ceramics is represented in figure 6. Damage is definedhere simply as one or more forms of detectable cracking. In simplest terms, threecracking forms observed can be distinguished as conical (or cone cracks), radial(originating from either the center or the outer edge of the target ceramic) andlaminar (parallel to the impact face of the ceramic tile). There is considerableoverlap and merging of these cracking forms in the regions of higher damagedensity and the damage is not necessarily symmetric. The resolution of this XCT technique does not permit the discrimination of the micro-scale damage featuresin either the center comminuted zone or in the surrounding elastic ceramic.


With complete dwell and little, if any, penetration, considerable interiordamage can exist in the comminuted region and in the surrounding elastic ceramicregion. For penetration to be prevented by the comminuted region, the structuralsupport (i.e. dynamic confinement pressure) of this region, which is provided and maintained by the surrounding elastic ceramic bulk, must be maintained. Withincreased impact energy of the incoming penetrator, the meso-scale crackingincreases in degree and extent and the structural support provided to the comminuted region decreases. Thus as the meso-scale damage increases, anonunique but "critical" meso-scale damage configuration may be attainedwherein the support to the comminuted region is no longer adequate to maintain its strength to sustain dwell and consequently penetration advances.

Confining Resistance of BulkSurrounding Elastic Ceramic

Ceramic Target Disc LaminarCracks

Cone Cracks

Radial Cracks - ID Radial Cracks - OD

Comminuted Zone

penetrator

Figure 6. Schematic of postulated structural support of the comminuted ceramiczone by surrounding bulk elastic ceramic that allows the comminuted zone toresist penetration

SUMMARY AND CONCLUSIONSX-ray computed tomography has been introduced as a novel and effective

nondestructive methodology to characterize the interior meso-scale damage in penetrator armor ceramics. Damage characterizations described above are post-mortem and are not obtained in real time during the ballistic event. It is thus not possible to directly establish the time sequence of the development of theobserved damage with this technique. It is prudent to conduct baseline DR and XCT scan procedures before, as well as after, impact to capture pre-existingdamage caused by fabrication and handling of target assemblies.


Ballistic damage observed with the XCT has included traditional conical, radial and laminar cracking, as well as less-reported spiral or circular beach-mark cracking and outer edge radial and periodic laminar cracking. Resolution of the XCT equipment available for large sample volume sizes of interest prevents the use of XCT for the assessment of micro-scale damage at present. Technology improvements currently under contract are anticipated to provide better resolution by a factor of at least two within present sample size limitations.

It is emphasized that the meso-scale damage observed is postulated to be quite significant in affecting the ballistic performance of the ceramic. Structural support degradation relating to increased meso-cracking surrounding the comminuted zone may be critical to the onset of penetration into the comminuted zone. Future ceramic processing methods introduced to limit or inhibit the meso-scale damage may have a significant benefit in improving the ballistic performance of next generation armor ceramics.

REFERENCES 1.

2.

3.

4.

W.H. Green, and Joseph M. Wells, "Characterization of Impact Damage in Metallic /Nonmetallic Composites Using X-ray Computed Tomography Imaging," pp622-629 in AIP Conference Proceedings 497,1999. J. M. Wells, W.H. Green, and N.L. Rupert, "Nondestructive 3-D Visualization of Ballistic Impact Damage in a TiC Ceramic Target Material," pp159-165 in Proceedings MSMS2001, 2nd Intn'l Conf. on Mechanics of Structures, Materials and Systems, 14-16 February 2001, University of Wollongong, Wollongong, NSW, Australia. W.H. Green, K.J. Doherty, N.L. Rupert, and J.M. Wells, "Damage Assessment in TiB2 Ceramic Armor Targets; Part I - X-ray CT and SEM Analyses," pp130-136 in Proceedings MSMS2001, 2nd Intn'l Conf. On Mechanics of Structures, Materials and Systems, 14-16 February 2001, University of Wollongong, Wollongong, NSW, Australia. N.L. Rupert, W.H. Green, K.J. Doherty, and J.M. Wells, "Damage Assessment in TiB2 Ceramic Armor Targets; Part II - Radial Cracking," pp137-143 in Proceedings MSMS2001, 2nd Intn'l Conf. on Mechanics of Structures, Materials and Systems, 14 - 16 February 2001, University of Wollongong, Wollongong, NSW, Australia.

(Form-CC by Gov Employees - not subject to copyight)


Processing and Manufacturing

AN ASSESSMENT OF LOW COST MANUFACTURING TECHNOLOGY

FOR ADVANCED STRUCTURAL CERAMICS AND ITS IMPACT ON

CERAMIC ARMOR

Richard E. Tressler

Department of Materials Science and Engineering

and the Materials Research Institute

The Pennsylvania State University

118A Steidle Building

University Park, PA 16802

ABSTRACT

The state-of-the-art in manufacturing of advanced structural ceramics,

particularly nonoxides, was recently assessed through visits to several companies

and institutes in Europe and the U.S. Cost of production is a barrier to

widespread application unless the performance is so superior that a cost/benefit

analysis results in favorable economics. The costs of the elements of the

production process are reviewed for specific production processes relevant to

armor production. Target areas for cost reduction for ceramic armor are clear

from this assessment.

INTRODUCTION

The commercialization of new processes and products in the general area of

advanced materials is truly a global enterprise. Technology that is developed

anywhere in the world eventually spreads by licensing, by worldwide marketing,

or by morphing into a variant of the original development. In the areas of

advanced structural ceramics and ceramic matrix composites, the Far East

(particularly Japan), the U.S., and Europe have all been active in developing and

commercializing new technologies. However, the growth in sales of these new

products has not met the expectations of forecasters. Also, the normal price

reductions with increased volume of production have not materialized, further

retarding the realization of mass markets. The military establishment has been

impacted by the high cost of structural ceramic products slowing the widespread

use of ceramics in armor and energy conversion systems. The weakness of market



pull for these materials and components has had the impact of curtailing

development efforts, particularly in the U.S.

With the slowing of the pace of development and commercialization these

areas in the U.S., the U.S. Army Research Office decided to commission an

assessment of "The State-of-the-Art in Low Cost Manufacturing Technologies for

Advanced Structural Ceramics and Ceramic Matrix Composites" starting in

Europe where some of the leading developments were thought to have occurred in

recent years. Ceramics and ceramic matrix composites have demonstrated great

promise in the production of complex structures with exceptionally high stiffness-

to-weight ratios, chemical stability, impact resistance, and high temperature

capability which are leading to substantial improvements in weight critical and

temperature critical applications. However, the relatively high cost of raw

materials and the complexity of the manufacturing process have created a barrier

to their widespread usage in industrial and defense applications.

This report outlines the findings of this assessment which was conducted in a

two week visit to industrial, academic, government sites in France and Germany,

culminating in a panel discussion at the International Ceramics in Engines

conference held June 19-21, 2000 in Goslar, Germany. Of particular interest for

this assessment were developments in non-oxide ceramics (carbides and nitrides)

and ceramic-matrix composites with both oxide and nonoxide constituents. To

benchmark the findings against comparable industrial activities in the U.S., visits

to five industrial organizations were conducted after the European trip. These

visits focused on industrial firms since the evaluation team was well aware of

research and development efforts underway in academia and government

laboratories through technical meetings and recent NMAB activities (see for

example Reference 1).

The details of the assessment on monolithic structural ceramics are presented

here since this class of materials is of direct interest for ceramic armor. Covalently

bonded, nonoxide ceramics are of special interest because of their low density and

high hardness

OVERVIEW OF MONOLITHIC STRUCTURAL CERAMIC

MANUFACTURING PROCESSES

Most of the manufacturing processes for monolithic structural ceramics start

with powders or powder precursors (sols or gels) and the final densification is

achieved by a sintering or pressure assisted sintering process (hot-pressing or hot

isostatic pressing). Chemical vapor deposition processes have been used to form

polycrystalline ceramic components. Only in the fabrication of certain high purity

semiconductor processing equipment or in preparing coatings are these processes

widely practiced. Some processes involve melt processing (reaction sintering, for


example) but generally a powder preform is the starting point for these processes

as well.

Therefore, powder preparation for structural ceramics is the natural starting

point for assessing the manufacturing technologies used for monolithic structural

ceramics.

Powders and Processes for Making Them

The oxides that are important for monolithic structural components are Al2O3

and ZrO2 – alumina because it is still the dominant structural ceramic for a wide

variety of wear, erosion, and impact resistant applications including ceramic

armor, and zirconia because in its toughened form it is an important wear, erosion,

and impact resistant material for use in dies, cutting edges, slitting tools, etc.

Alumina powders for advanced structural ceramics are priced from 1/10 to ¼ of

the price of the least expensive fine grain SiC powder (Carborundum) while

zirconia powders for fine grain advanced ceramics are priced comparably to

submicron SiC powders.

The important nonoxide ceramics are silicon carbide, silicon nitride, boron

carbide, and titanium diboride, and they are presented in order of decreasing

tonnage produced per year. Aluminum nitride is poised to join the top four as it is

becoming more widely used as a thermal management ceramic for electronic

packaging and a erosion/corrosion resistant material for integrated circuit

processing equipment (etchers, for example). Molybdenum disilicide is also

emerging for certain niches such as resistance heating elements. Sheppard has

recently tabulated the various suppliers of fine ceramic powders for advanced

structural ceramics along with the process used to synthesize the powders and the

approximate annual tonnages produced (2).

Silicon carbide is the clear leader in terms of tonnage produced primarily

because there is a large market for abrasive grit. Also, silicon carbide ceramics in

the form of reaction bonded or reaction sintered silicon carbide (where the grains

are bonded with a second phase or phases) have been used for many decades as

specialty refractories for blast furnaces, for metal melting, for porous filters, etc.

Siliconized silicon carbide which is fabricated by infiltrating molten silicon into a

silicon carbide plus carbon preform is widely used for process tubes (for example,

in the semiconductor industry and the metal heat treating industry).

The -silicon carbide powders are made by carbothermal reduction of SiO2

and reaction with carbon. The Acheson furnace which was invented more than

one century ago is still used by all of the major producers of -SiC. Global

production capacity is estimated to be 1 million tons per year (2) with production

levels at about 75% of that figure. The fine sinterable powders are produced

primarily by Norton/Carborundum with Exolon-ESK in second place. H.C. Starck

is a supplier of sinterable powder as is Superior Graphite Co. Both of these latter


two companies also market a fine grain -SiC powder but so far the markets for

them have been limited.

Carborundum Structural Ceramics group markets a sinterable premix to a

large number of fabrications. Their production in the calendar year 2000

approached 1 million pounds of premix. They are also a major consumer of this

premix for their Hexoloy family of sintered silicon carbides. Their premix sells

for $11-14/lb depending on the quantity produced.

Silicon carbide powder production is not a roadblock for low cost armor. The

production capacity is larger than the demand worldwide and the price is low

because the fine powder is a side stream of a much larger production of SiC for

abrasive grain and as an additive to metals during melting and refining

Silicon nitride powder is produced by a variety of methods, the most common

being nitridation of silicon powder. Reduction of silicon imide is also used to

make large quantities of sinterable silicon nitride powder. Global production of

silicon nitride is estimated to be several hundred tons. Prices range from $30/kg to

$150/kg depending on particle characteristics (purity, grain size, surface area) and

volume. The major suppliers are Japanese companies (Ubé is a major source) and

European (H.C. Starck).

Boron carbide is produced in a carbothermal reduction process similar to the

Acheson furnace process. The advanced ceramics manufacturers in this country

purchase their powder from ESK. In fact, H.C. Starck resells ESK produced

powder for the advanced ceramics market (including armor). According to

Sheppard's survey there are other suppliers of B4C, which is also used as a

specialty abrasive and as control rod material in nuclear reactors (2). The total

worldwide production is ~100 metric tons. According to ESK the price of B4C

could be comparable to that of SiC if the volume were similar. As it stands now,

hot-pressable powder sells for ~$35-40/kg.

Titanium diboride powders are synthesized by carbothermic reduction of TiO2

and B2O3. There are a number of suppliers (2). The annual production is ~120

metric tons and, thus, the price per kg is high compared to SiC, ranging from $35

to $65/kg.

"Green" Forming of Ceramics

The forming processes employed in the fabrication of parts depends primarily

on the complexity of the shape. Uniaxial dry pressing is used for simple shapes

in volume production because of its reproducibility and use of automation. Free

flowing granules of the powder plus binder, plasticizer, and lubricant are usually

formed by spray drying. Much of the production of armor tiles, wear tiles, pump

seals, etc., use dry pressing for green forming. Dimensional tolerances can be held

accurately; uniformity of composition and, thus, the final microstructure is easy to

achieve, and automatic presses can produce large volumes of parts.


Cold isostatic pressing is used for larger diameter cylinders and hollow

shapes. In dry bag pressing the operation is similar to uniaxial dry pressing except

the pressure is applied radially by pressurizing a liquid against a flexible mold

with a rigid shell. At ESK large diameter (~3-4" OD) closed end tubes of Si3N4

were formed this way and green machined to make molten metal handling

equipment.

For complex shapes such as turbine rotors or turbocharger rotors injection

molding is the forming method of choice borrowing from injection molded

plastic technology. However, pressure gel casting, in which a powder-liquids

slurry (slip) is injected into a porous mold under pressure, has been demonstrated

to be amenable to automation and easier to debind by Allied-Signal (Honeywell).

Traditional slip casting is used for large complex shapes with internal

surfaces, particularly in low volume production.

Extrusion is used for constant cross-section products such as tubes and rods.

Carborundum structural ceramics group uses this method to fabricate SiC tubes

for heat exchangers.

Tape casting is used primarily in the electroceramics industry, but is also

used to build up B4C shapes for hot-pressed armor at Ceradyne.

Hot-pressing is clearly an important forming-densification process for

ceramic armor components and for the large more or less flat shapes required for

the semiconductor equipment manufacturers. In most cases a dry pressed or tape

lay-up preform is used to fill the hot-press die cavity. But for single part pressings

it can be filled with free flowing powder. (More on hot-pressing in the next

section.)

Densification

The densification process for the oxide ceramics is straightforward – firing in

air usually with natural gas fired kilns. The densification of the nonoxide ceramics

has been developed in the last 10-15 years so that pressureless sintering of SiC

and Si3N4 based ceramics to 99+% of theoretical density is accomplished

routinely albeit by relatively few companies.

The sintering of silicon nitride ceramics is usually accomplished by liquid

phase sintering. The additives are generally rare earth sesquioxides, often with

Al2O3, and they combine with the SiO2 that is present in all high surface area

powders to form a silicate phase at temperatures of 1750-1900°C. Generally, a

nitrogen overpressure of a couple of atmospheres is required to prevent

dissociation of the Si3N4 and nitrogen loss during sintering.

The sintering of silicon carbide was first commercialized by Carborundum

who used B and C additives which many think altered the surface energetics such

that the material would densify rather than just form necks between adjoining

particles. However, some researchers have also speculated that there is a fugitive


eutectic liquid that forms and allows liquid phase sintering. However, the final

product (after ~2000°C firing in inert atmosphere or vacuum) is a 99+ dense –

SiC with some carbon and B4C inclusions. The patents by Carborundum

revolutionized SiC ceramic technology which previously could only be fully

densified by hot-pressing or reaction sintering with excess Si in the final ceramic.

Carborundum and others also experimented with liquid phase sintering of SiC

where Al2O3 and Y2O3 were added, and the firing temperature was above the

eutectic temperature of the oxides. The resulting microstructure contained yttrium

aluminum garnet second phase which resulted in a toughening of the material

although it is not as hard. Carborundum commercialized their product (Hexoloy-

SX) but discontinued it. ESK produces a similar product which is used as wear

plates, primarily in the paper industry.

Hot-pressed silicon nitride and silicon carbide are both being produced by

a number of companies (Carborundum, Ceradyne, Cercom, Kyocera, ESK)

primarily for very high performance applications and where no porosity can be

tolerated (semiconductor processing machinery). Graphite dies are used and

controlled atmosphere or vacuum are required to achieve full density. For the

semiconductor process equipment makers hot-pressing appears to be the only

consistent way to make large parts (18"-20" in diameter) with uniform

microstructure throughout. The other reason for hot-pressing parts for this

industry is that the very high purities required can be achieved since no additives

are needed to achieve densification.

Hot-pressed SiC is also used for armor when very high performance is

required. In general, the hot-pressed products perform better and more

consistently in ballistic tests than the sintered products although it is not clear to

this writer that the latest sintered products have been tested. The Carborundum

enhanced Hexoloy-SA and ESK liquid phase sintered SiC are examples.

The markets for B4C and TiB2 are more limited than SiC and Si3N4. Titanium

diboride is hot-pressed by ESK, primarily for evaporation boats for aluminizing

polymer films and other products. Boron carbide in hot-pressed form is a

lightweight armor material. Tiles of B4C for personnel vests are hot-pressed in

stacks of 20+ at Cercom and 50+ at Ceradyne in large hot-press furnaces that

cycle through the press so that the press is in use essentially full time.

Sintered B4C with metallic additives has been studied for years but no suitable

armor material has evolved from these studies. Carborundum uses a sinter-HIP

(hot isostatic press) process to make complex shapes of B4C and is studying the

process to make B4C helmet armor.

Hot-pressed products in the final machined state cost 2 ½ to 3 times the cost

of sintered products in the final machined state (per Cercom). Much of this

additional cost is the cost of graphite hot-pressing tooling. Machining is a larger

part of the final cost in hot-pressed products. Cheaper hot-pressing is possible


(according to Cercom) if a semi-continuous process were developed. For the

armor market none of the manufacturers are willing to invest the capital required

to make the costs lower because the armor contracts are all relatively short term

compared to the time period to recover the additional equipment costs. Thus, to

reduce costs and take much of the manual labor out of the process, much larger

contracts are required or the DOD could fund a Mantech initiative in automated

hot-pressing, or the DOD could develop and own an automated line. No one

armor product volume is large enough to justify the capital costs for a

manufacturer to develop an automated line.

Considerable research is going on now to develop reaction sintering processes

which have the potential to be cheaper and more flexible in terms of incorporating

second phase particles or fibers. Probably the best known process is the Reaction

Bonding of Aluminum Oxide (RBAO) pioneered by Professor Nils Claussen

(now at the University of Hannover). In this process aluminum is incorporated

with aluminum oxide powder and the preform is fired at 1000-1200°C during

which the aluminum melts and oxidizes to form Al2O3. The product can be fully

dense, the forming temperatures are lower than required for Al2O3, and the

process can be net shape through the precise control of volume fractions of Al and

Al2O3.

The process has been applied to Al2O3 with second phase particles such as

SiC; it has been used to make mullite, and researchers around the world are

extending the concept to novel ceramic composites. It has not found widespread

commercial use at this time.

APPLICATIONS

Monolithic Advanced Structural Ceramics

Oxide advanced structural ceramics are used in a wide variety of niches,

primarily where the wear, erosion, and corrosion resistant properties are

important. Coors is probably the largest U.S. producer of oxide structural

ceramics, and they characterize their business as a large number of ~$5M niches.

In general, the wear and erosion resistance of oxide based ceramics is not as good

as SiC and Si3N4 ceramics unless an oxidizing, high temperature environment is

present. In general, SiC and Si3N4 based ceramics have better thermal shock

resistance than oxides with Si3N4 being superior. Silicon carbide ceramics are

better thermal conductors than oxides and most nonoxides with the exception of

AlN. Thus, the oxide ceramics are the low cost choice in many cases while the

nonoxides are the high performance choice. In the case of ceramic armor

materials, alumina is used because of low cost even though it is ~25% more dense

than SiC.

The primary applications for nonoxide ceramics are well-summarized in Table

I. Add to this list the applications identified by ESK (molten metal handling,


pump components, evaporation boats) and pursued by Honeywell (nozzle blades

for APUs, aeroengine seals and other components for aeroengines) and one has

most of the niches for application of these materials. Turbocharger rotors have

been in production in Japan but the market is not growing, and diesel engine parts,

particularly the injector parts, are important.

Table I. Primary applications for nonoxide structural ceramics.

Sintered Silicon Carbide

Industrial Seals Heat Exchanger tubes

Auto Seals Semiconductor Equipment Components

Pump Bearings Armor Tile

Wear Tile Kiln Furniture

Silicon Nitride

Bearing Balls Semiconductor Equipment Components

Roller Bearings Wear Parts

Paper Making Equipment Parts Nuclear Seals

Boron Carbide

Nozzles Dressing Sticks

Armor Tile Wear Components

Aluminum Nitride

Electronic Substrates Semiconductor Equipment Components

Source: Saint-Gobain Industrial Ceramics, Structural Ceramics Group, Niagara

Falls, NY

The semiconductor process equipment components are an important market

because the application can tolerate high cost. The hot-pressed, high purity

ceramics are the material of last resort for this class of applications. The total

sales of the major vendors in this market are about $200 million/year which is

probably the largest segment although the segment is composed of many different

parts and configurations.

Meanwhile, investments in processing equipment for the semiconductor

component market is resulting in better processing capability that spills over into

the hot-pressed ceramic armor market. In other words, a commercial market is

subsidizing capital equipment which is used for the ceramic armor market.

Other trends to watch:

The market for seals and bearings is becoming large.


Heat exchanger tubes have not developed into a major market, although

there is continuing activity in production of ceramic heat exchangers.

Armor tile is a continuing market, particularly where hot-pressed B4C is

concerned, and components with similar geometry include wear tile for

the paper industry.

Dry-pressed and sintered tiles are in reasonably wide spread production,

and large presses are available or being purchased so that large scale,

semi-automated production of pressureless sintered SiC tiles up to

14"x14" is possible.

There is no semi-automated or semi-continuous hot-pressing of ceramics

at this time, so if hot-pressed armor tile is required by the military, it will

be made in labor intensive, batch process lines. This implies much higher

cost than dry-pressed and sintered material.

Technical and Economic Issues in Manufacturing of

Advanced Structural Ceramics

Advanced ceramic components costs are still too high for widespread

application unless the performance is so superior that a cost/lifetime benefit

analysis results in a favorable economic situation for the ceramics. Elements of

the high cost of production include powder costs, machining costs, and firing

costs. The forming costs are similar to that of powder metals except when hot-

pressing is used, which combines forming and firing.

Powder Costs

For all of the nonoxide ceramics except SiC the powder costs are high. In the

case of SiC the large volume of SiC produced for abrasive, grinding wheels, and

primary metal additives results in lower costs than otherwise would be the case in

view of the relatively small volume of silicon carbide advanced ceramics

produced. The primary cause of high costs of powders is the low volumes

produced. A secondary effect is the stringency placed on powder characteristics –

the higher the purity requirement and the higher the particle size distribution

controls the higher the price (Figure 1). Thus, if the component fabricator can

meet the component specifications with a less stringently controlled powder, costs

can drop. In silicon nitride component production, less expensive powders are

routinely used for wear and erosion components that do not have load bearing

requirements at high temperature.


Figure 1. Schematic of powder costs as a function of purity and particle size

control (3).

Machining Costs

Finished machining (fired shapes) must be done with diamond tooling, and all

of the nonoxide ceramics are hard, which means that machining is laborous and

expensive. The key to cutting machining costs is to fire to net-shape so that only

surfaces that require a high finish are machined. There are successes in cutting

machining costs such as pump seal manufacture and ceramic valve machining

where the final machining time was cut to 30 seconds.

Final machining of hot-pressed parts is more expensive than sintered parts

because of the difficulty in holding tolerance during the hot-pressing. Final

machining can contribute 50% to the final component cost for hot-pressed parts.

Firing Costs

Firing of nonoxide ceramics requires temperatures of 1700-2100 C in an inert

atmosphere or vacuum, which is intrinsically more expensive than firing oxides in

air. Continuous kilns have been effective in cutting costs but sufficient volume

must be produced to warrant the continuous operation of kilns. Thus, scale of

production is one of the key factors in firing costs.

Nonuniformity of microstructure across large area parts is difficult to achieve

during sintering due to nonuniformities in green density and binder content and

due to differential rates of heating from edge to center. Additional development

efforts are required to reproducibly produce reliable, sintered armor tiles that will

perform near the level of hot-pressed tiles.

For costs to be cut for hot-pressed parts automated, semi-continuous

processing methods must be developed which requires longer term contracts to

the vendors or direct government investment in automated, semi-continuous lines.


Higher Purity/PSD Control

Tota

l Pow

der

Cos

t

The erratic nature of ceramic armor contracts makes it economically unwise for

the vendors to make large capital investments in these types of facilities. Graphite

tooling for hot-pressing is a significant part of the manufacturing cost; there is no

obvious way to substantially reduce these costs.

Reaction based processing of advanced structural ceramics and, particularly,

ceramic matrix composites, holds the promise of reducing costs by reducing firing

temperatures, using cheaper raw materials, shortening processing cycles and

providing near net shape capability. More research and development in this area is

required to commercialize this approach. Examples include reaction sintering of

silicon carbide and liquid silicon infiltration of C/C preforms, and C/SiC preforms

for CMCs. Liquid polymer infiltration as a method to process CMCs warrants

further development.

SUMMARY

Powder costs are high for nonoxide structural ceramics compared to those for

oxide ceramics with SiC being the cheapest of the nonoxides (4-10X the price for

Al2O3). The only way to decrease cost is to increase volume. Machining costs for

these ceramics are on the order of 50% of the total costs except for simple shapes

which can be used with as-fired surfaces. Automated production lines (as in

automobile water pump seals) are necessary for low machining costs.

Firing costs are intrinsically higher for nonoxide ceramics compared to oxides

with temperatures of 1700-1200 C in inert atmospheres or vacuum. Continuous

kilns have been effective in cutting costs but sufficient volume must be produced

to warrant the continuous operation of kilns. Hot-pressing is the process of choice

when uniformity of microstructure across large area parts is required (as in

armor). The cost is 2-3 times the cost of sintered parts. For costs to be cut

automated, semi-continuous hot-pressing methods must be developed which

requires longer term contracts to the vendors or direct investment in such lines by

the customer.

Pressureless sintering of SiC plates has been developed by a few companies to

the state where large area plates can be produced by semi-automated methods

with uniform properties (Weibull moduli approaching 30) and low cost. These

products should be investigated for armor tile and modifications made to attempt

to use these production lines for low cost SiC armor.

ACKNOWLEDGMENT

This work was supported by the U.S. Army Research Office through Batelle

Scientific Services Agreement. The contributions to this assessment by Dr.

Andrew Crowson (ARO) and Dr. James McCauley (ARL) are gratefully

acknowledged.


REFERENCES 1Committee on Advanced Fibers for High Temperature Ceramic Composites,

"Ceramic Fibers and Coatings: Advanced Materials for the Twenty-first Century,"

National Materials Advisory Board, National Research Council, NMAB-494,

National Academy Press, Washington, DC, 1998. 2Laurel M. Sheppard, "Global Assessment of High Performance Ceramics for

Armor," report submitted to the Army Research Laboratory, Aberdeen Proving

Ground, MD, 2000. 3D. A. Lathrop, "Non-oxide Powders for Advanced Engineered Ceramics,"

presented at Advanced Ceramics for the New Millenium, March 10-12, 1998,

Atlanta, GA.


HIGH-PURITY SUBMICRON -AL2O3 ARMOR CERAMICS

DESIGN, MANUFACTURE, AND BALLISTIC PERFORMANCE

Andreas Krell Elmar Strassburger

Fraunhofer Institut für Keramische Tech- Fraunhofer Institut für Kurzzeit-

nologien und Sinterwerkstoffe (IKTS) dynamik (EMI)

D – 01277 Dresden D – 79588 Efringen-Kirchen

Germany Germany

ABSTRACT

New grades of sintered corundum armor ( -Al2O3) have been designed here to

obtain a ballistic mass efficiency close to SiC and, preferentially, to exhibit a high

optical in-line transmission by associating (i) a small sub- m grain size with (ii)

a very high density and (iii) purity, and (iv) a microstructure that is free of flaws.

Different ceramic technologies like dry (cold isostatic) pressing and casting

approaches (with the option of free shaping) are investigated with respect to these

objectives. Results of ballistic tests give evidence of a strong correlation of pro-

tective efficiency and rising hardness in fine grained sintered Al2O3.

INTRODUCTION

Structural ceramics which associate a high hardness with a low density are suc-

cessfully used as ballistic armor when a high protective power is required at a low

weight. Rankings of the ballistic resistance of different grades of Al2O3, SiC, B4C,

and TiB2 have been established by means of Depths of Penetration (DOP) tests.

However, there is still a lack of fundamental knowledge about the correlation

between the real microstructure of ceramics and their ballistic resistance.

A first systematic study of the influence of materials properties was focused

on alumina ceramics in 1995 and exemplifies the typical difficulties of such in-

vestigations1: the study comprised about twenty commercial Al2O3 ceramics with

different grain size, purity, porosity, and glassy phases, and it was impossible to

analyze suggested influences of individual microstructural parameters (e.g. grain

size) when porosity and glass phase concentration were not constant. Also, the

results showed little correlation between the Hugoniot elastic limit (HEL), the

spalling strength and the ballistic mass efficiency Em. Therefore, only high purity

ceramics with relative densities > 98.5 % should be used in future studies investi-



gating the influences of grain size or hardness on the ballistic performance.

Previous tests revealed indications that the ballistic resistance of ceramics in-

creases with increasing hardness,2 and it is well known that in polycrystalline ce-

ramics glassy sintering additives reduce the hardness which, on the other hand,

increases with decreasing grain size.3 Starting from these results, it was the ob-

jective of the present work to develop pure Al2O3 ceramics with sub-µm grain

size and to investigate their ballistic performance.

DESIGN OF NEW GRADES OF CORUNDUM ARMOR

To assume a high hardness as a most powerful tool for obtaining a high protec-

tive power seems the more justified as it is commonly agreed that a penetrating

projectile looses a major part of its kinetic energy by deformation and wear inter-

action with the hard armor. On the other hand, it was suggested that on wear there

is a specific hierarchic order of microstructural influences in a way that wear is

more affected by direct influences of the grain size on interface properties (e.g.

reducing pull-out effects by smaller grain sizes) than by the associated hardness.4

In analogy, it may be speculated that a smaller grain size may give some benefit

for an improved ballistic efficiency even when the hardness is not maximum.

Therefore, tests with ultrafine alumina ceramics (grain size < 400 nm) were de-

signed to investigate this issue.

On penetration, the microstructure of the armor collapses within a few micro-

seconds. Therefore, the significance of the usually measured strength of the ce-

ramics is not clear and was addressed here from an empirical point of view.

Among today’s technical Al2O3 ceramics, commercial alumina armors are repre-

sentatives of lower strength grades, often with a 4-point bending strength < 400

MPa. Therefore, the technological efforts of the present investigation were fo-

cused to associate a high macro-hardness close to 20 GPa (at testing load 10 kgf)

with a minimum of flaws in the sintered sub-µm Al2O3 ceramics to enable a high

strength of 500-700 MPa (4-point bending).

On the other hand, these extremely fine grained (sub-µm) corundum micro-

structures that are highly pure and free of defects are also expected to exhibit a

high in-line transmittance of unscattered light (increasing at smaller grain sizes);

the smallest grain size for dense samples was 0.7-0.8 m in these early investiga-

tions with an in-line transmission < 46 % for 1 mm thick disks.5,6

Whereas cubic

materials like spinel can become transparent (“clear”) even with larger grain sizes

as far as the residual porosity is small enough (< 0.05 % requested!), sub-µm

grains are imperative to obtain transparency in hard sintered corundum. Fig. 1

shows the high real in-line transmittance (RIT) obtained now at IKTS Dresden by

eliminating the last residual porosity from the new sub-µm grades of armor.

Contrary to known developments of corundum ceramics that become translu-

cent by a reduced number of grain boundaries per volume (i.e. by grain coarsen-


ing associated with a decrease of hardness, protective power, and strength), the

(a) (b)

(c)

Fig. 1. 30-mm discs of sub- m Al2O3 with RIT = 45 % (a, b) and 100x100m2 tile with RIT = 52 %

(all samples 0.8 mm thick; grain size 0.5 m, relative density > 99.9 %; bending strength

650 MPa 4-point - 850 MPa 3-point ). In contrast to translucent armor, transparency is dem-

onstrated here by Fig. 1c and by comparing a polished plate placed (a) immediately on and

(b) in a distance over the printed paper.

sub- m design provides the advantage of combining a greatly improved mechani-

cal performance (cp. data to Fig. 1) with a transition from translucence to a

transparent appearance. Fig. 2 shows the strong increase in the real in-line trans-

mission in a perfect agreement of experimental results and the physical model.7

0

10

20

30

40

50

60

70

80

90

100

0 1 2 3 4 5 6

Experimental data

Calculated model (R. Apetz)

Average grain size (µm)

Real in

-lin

e tra

nsm

issio

n (%

)

7


Fig. 2. Influence of grain size on the in-line transmission of sintered Al2O3 (0.8 mm thick samples,

= 640 nm) with a relative density close to 100 %. Physical model7 and measured data.

The real importance of the state of grain boundaries for the ballistic and optical

performance is not finally clear at present. It is, however, worth to note that high

resolution TEM gives evidence that all interfaces of the high-purity alumina ce-

ramics developed here are free of even thinnest amorphous films (Fig. 3).8

Fig. 3. Typical HREM image of a grain boundary in high-purity (>99.9 %) corundum. The

boundary is free of amorphous material or crystalline precipitates.8

PREPARATION OF TILES FOR BALLISTIC TESTS

Ground tiles with a lateral dimension of 100x100 mm2 and with different

thickness (5-15 mm) were prepared from 99.99 % pure Taimicron TM-DAR co-

rundum powder (Boehringer Ingelheim Chemicals, Japan) by

(i) an approach of spray drying and cold isostatic pressing9 or by

(ii) advanced gelcasting8,10

followed by sintering in air; the casting approach offers the additional advantage

of free shaping.10

The samples were prepared without doping additives and by

pressureless sintering if not stated otherwise in Tab. 1.


Table 1. Sintered corundum ( -Al2O3) for ballistic investigations. S-samples were pre-

pared by spray drying and cold isostatic pressing, G denotes gelcast materials;

D-samples were provided by Dornier (Friedrichshafen, Germany).

Relative density

(%)

Grain size

( m)

Hardness HV10

(GPa)

Strength4-pt bend

(MPa)

Comments

S-0.3

S-0.5

G-0.6

S-0.7

D-0.9

AD-995

92.5

99.3

100

99.5

98.7

98.8

0.32

0.53

0.57

0.71

0.92

10 - 20

15.0

19.3

20.2

19.1

16.5

12.3

not determined

203 16

644 70

526 55

244 41

350 25

MgO doped; + HIP

Improved non-

aqueous process

Sintering temperatures at 2 hours isothermal hold are about 1420 C for dry

pressed samples and 1260 C after gelcasting to obtain a relative density of 99.5

% (Fig. 4); transparent microstructures require additional hot isostatic pressing.

Fig. 4. Typical microstructure of 99.5 % dense sintered -Al2O3 with 0.54 m grain size.

Table 1 comprises the characteristic data of the samples prepared by different

approaches for ballistic investigations. AD-995 supplied by Coors (Golden, Colo-

rado) was used as a reference which exhibited the highest mass efficiency among

previously tested commercial alumina grades.

BALLISTIC INVESTIGATIONS

Testing set-up and definitions

The different grades of alumina ceramics were tested in a DOP-configuration

(depths of penetration) with a RHA backing (rolled homogeneous steel armor)

of 100 mm thickness and a hardness of HV30 = 3 GPa. The DOP-method was

chosen because it is well established for many years as a method for ranking the

ballistic performance of ceramics,11

and a large body of DOP data exists which


can be referred to. Both the surfaces of the ceramic tiles and the steel backing

were ground in order to guarantee reproducible conditions in all experiments.

A tungsten alloy penetrator was selected because this type of projectile is con-

sumed continuously by abrasion. Thus, the scatter of the DOP results is much

smaller than with hard core projectiles which can break or shatter during penetra-

tion. Moreover, a large number of DOP results with that particular projec-

tile/target configuration are available at EMI. In the present investigations, cylin-

drical projectiles with a hemispherical nose (diameter 10 mm, length 32 mm,

mass 44 g) were fired from a 20 mm smoothbore gun by means of plastic sabot

comprising of four petals, an obturator and a steel pusher plate. The impact veloc-

ity was 1250 m/s nominally.

The figure-of-merit for ballistic performance was the ballistic mass efficiency

Em, determined from the residual penetration PR, the penetration into the reference

steel target Pref, the thickness of the ceramic TCer and the densities St, Cer of the

steel and the ceramic. Fig. 5 shows the test configuration and the definition of Em.

RStCerCer

refStm

PT

PE

Fig. 5. DOP configuration and definition of the measured mass efficiency Em

According to Fig. 5, the residual penetration PR observed for a specific ceramic

armor will depend on the thickness Tcer of the ceramic tile and on the densities

Cer and St of ceramic and steel. Usually, a linear decrease of the residual pene-

tration is observed when the ceramic thickness increases resulting in a linear in-

crease of the mass efficiency Em with increasing values of Tcer. From such plots, a

linear extrapolation of Em to a ceramic thickness which would stop the projectile

just at the ceramic-steel interface defines the maximum mass efficiency Em,max as a

characteristic materials parameter.

Experimental Results

The maximum ballistic (protective) mass efficiency Em,max was obtained from

penetration experiments with Al2O3 tiles of different thickness (5-15 mm).

Whereas tests with samples G-0.6 are still in progress, results for the correlation


between Em,max and the hardness are given by Fig. 6 for the other grades; a com-

panion paper12

discusses the DOP plots in more detail. Em,max of the commercial

reference AD995 was 2.1 in the projectile/target combination considered here,

typical values for silicon carbide (SiC) are in the range of 3.

Fig. 6 shows significantly higher Em,max values of the fine grained, harder

grades compared with AD995. Plots where Em data for a thickness of 20 mm were

obtained from monolithic tiles or from 10 mm + 10 mm composites yielded

Em,max 2.6 at a hardness of about 18-19 GPa; an even higher value of Em,max =

2.9 was obtained with a 5 mm / 15 mm configuration.

Note that the extremely fine-grained but porous grade S-0.3 with Em,max = 2.3

still exhibits a higher protective power than AD995 - a clear merit of its hardness

which compared to AD995 was increased by the small grain size despite the high

porosity of 7.5 % (cp. Tab. 1).

10 12 14 16 18 20 22 24

Vickers hardness HV10 (GPa)

2.0

2.5

3.0

3.5

no influence of

flaws / strength

Ba

llist

ic m

ass

eff

icie

nc

y E

m,m

ax

AD995

S-0.3

D-0.9S-0.5

S-0.7

G-0.6(expected)

Fig. 6. Influence of the hardness on the maximum ballistic mass efficiency.

The position of the S-0.3 result right on the linear fit of hardness and Em,max in

Fig. 6 excludes any separate influence of the grain size on Em,max beyond the

hardness effect. Hence, Em,max of S-0.3 is lower than that of the coarser but dense

ceramics with grain sizes of 0.5-0.9 µm because here the detrimental effect of the

high residual porosity on the hardness is not balanced by the smaller grain size.

It is important to note that no deviations from the “usual” linear fit in Fig. 6

are observed at constant hardness neither for grades with a low bending strength

(caused by flaws emerging from hard spray-dried granules in S-0.5, cp. Tab. 1)

nor due to different grain sizes: for Em,max it is unimportant whether a high hard-

ness is obtained by a smaller grain size in spite of some residual porosity (S-0.5)

or with a slightly coarser grain size at a higher density (S-0.7).

As to strength effects, however, an influence seems probable for test configu-

rations without the confinement used in the present study (Fig. 5).


CONCLUSIONS

The protective power of sintered Al2O3 armor is linearly related to hardness;

there is no separate influence of the grain size or of flaws beyond their impact on

the hardness. The design of new armor ceramics should thus be focused on

- smallest grain sizes significantly below 1 µm,

- high relative density (i.e. minimized residual porosity), and

- high purity.

Additionally, a high strength (e.g. 200 % of today’s commercial Al2O3 armor)

will be beneficial for the general mechanical performance in the technical applica-

tion.

These microstructural conditions are close to the design of new high-strength

transparent sub-µm -Al2O3 (armor) ceramics with a high in-line transmittance.

The new sub-µm alumina grades exhibit significantly higher mass efficiencies

(close to SiC) than commercial corundum armor tested under the same conditions.

Free shaping of these armor components is enabled by new casting approaches.10

ACKNOWLEDGEMENTS IKTS Dresden gratefully acknowledges the cooperation with Dr. R. Apetz and Dr. M.

van Bruggen at Philips NatLab (Eindhoven, NL) within the STARELIGHT project

funded by the European Commission (“Growth” program, contract G5RD-CT-1999-

00088).

REFERENCES 1B. James, “The influence of the material properties of alumina on ballistic

performance,” pp. 3-9 in Proceedings of the 15th

International Symposium on Bal-

listics (Jerusalem/Israel, 1995 published by the Organizing Committee).2I. Faber, K. Seifert and L.W. Meyer, “Correlation between the mechanical

data of ceramics and their protective power against impact loading” (in German),

Final Report EB 6/95 (part 3), Technical University Chemnitz-Zwickau, Depart-

ment of Engineering Materials, 1995. 3A. Krell and P. Blank, “Grain Size Dependence of Hardness in Dense Submi-

crometer Alumina,“ J. Am. Ceram. Soc. 78 4 1118-20 (1995).4A. Krell, “Improved hardness and hierarchic influences on wear in submicron

sintered alumina,“ Mater. Sci. Eng. A 209 1-2 156-63 (1996). 5K. Hayashi, O. Kobayashi, S. Toyoda and K. Morinaga, “Transmission optical

properties of polycrystalline alumina with submicron grains,“ Materials Transac-

tions (JIM) 32 11 1024-29 (1991). 6H.Mizuta, K. Oda, Y. Shibasaki, M. Maeda, M. Machida and K. Ohshima,

“Preparation of high-strength and translucent alumina by hot isostatic pressing,“

J. Am. Ceram. So.) 75 2 469-73 (1992).


7R. Apetz and M. van Bruggen, “Transparent Alumina: a Light Scattering

Model,” submitted to J. Am. Ceram. Soc.8A. Krell, E. Pippel, J. Woltersdorf and W. Burger, “Subcritical crack growth

in sub-µm Al2O3,” J.Europ. Ceram. Soc. (in press - 2002). 9E. Strassburger, H. Senf, H. Rothenhäusler, B. Lexow and A. Krell, “Influ-

ence of grain size and microstructure on the ballistic resistance of alumina,” pp.

1216-23 in Proceedings of the 18th

International Symposium on Ballistics (San

Antonio/TX, 1999), Technomic Publishing Co., Lancaster/PA, 1999. 10

A. Krell, „High-strength Al2O3 joint prostheses of complex shape,“

http://www.

ikts.fhg.de/business/strukturkeramik/basiswerkstoffe/oxidkeramik/al2o3_bio_eng.

html.11

Z. Rosenberg, S. Bless, Y. Yeshurun and K. Okajina,“A new definition of

ballistic efficiency of brittle materials based on the use of thick backing plates”,

pp. 491-98 in Impact Loading and Dynamic Behaviour of Materials (Proc. Impact

87 Conf., Bremen, Germany, 1987), DGM Informationsgesellschaft, Oberursel,

1988.12

E. Strassburger , A. Krell, B. Lexow, „Ceramic Armor with Submicron Alu-

mina against AP Projectiles,“ pp. Xx-xx in Proceedings of PAC RIM IV, Ceramic

Armor Materials by Design (Wailea, Maui, Hawaii, 2001).


SOLID FREEFORM FABRICATION OF ADVANCED ARMOR CONCEPTS:

OPPORTUNITIES FOR DESIGN AND MANUFACTURE

R.C. McCuiston, S.C. Danforth, M.J. Matthewson, and D.E. Niesz

Malcolm G. McLaren Center for Ceramic Research

Rutgers, The State University of New Jersey

607 Taylor Road

Piscataway, NJ 08854

ABSTRACT

There is tremendous interest in advanced armor concepts. Fortunately, there

are novel manufacturing methods available for such systems, referred to as Solid

Freeform Fabrication (SFF), or Layered Manufacturing (LM). These methods are

generally free of the normal constraints imposed by traditional manufacturing.

Designers now have the capacity to optimize design for performance in ways

never before possible. Using SFF or LM technologies, one can manufacture

components out of several different materials to achieve multi-functionality. This

is accomplished by controlling the spatial distribution of materials by a computer

driven material deposition system. A brief review of our SFF method, Fused

Deposition of Ceramics (FDC) will be given along with possible novel armor

design concepts.

INTRODUCTION

In what seems to be an ever-shrinking world, the need to travel around it on a

moments notice is all the more important. This is especially true in times of war

and conflict. One of the expressed goals of the United States Army is to be able to

deploy to anywhere in the world from a multitude of dispersed sites in a matter of

days. Unfortunately with the increasing lethality of today’s weapons, designers of

vehicle platforms have had to compensate by adding increasingly thicker armor,

typically dual hardness steel. As a result air transportation of these increasingly

heavier vehicles is very difficult. The added weight also increases fuel

consumption, decreases maneuverability and tests the limits of portable bridges.

[1]

To help realize the goals of an easily deployable force, research has been

conducted on novel armor concepts such as confined ceramic tiles [2], confined



multi-layered ceramic metal systems [3,4] as well as unconfined multi-layered

ceramic metal composites. [5] In the first two examples, it has been shown with

depth of penetration tests (DOP), that an external confinement, in which either a

lateral, or hydrostatic pressure is applied via a metal phase, provides improved

ballistic performance over unconfined armor systems. It is known that a highly

cracked region of ceramic directly in front of a projectile is needed in order for a

penetration event to initiate. [4] This highly cracked region will develop into a

comminuted zone and the projectile will penetrate by forcing comminuted

fragments to flow around the advancing projectile and thus becoming ejected

from the impact site. Under external confinement however, this flow, and

subsequent ejection of comminuted fragments is hindered, allowing the

comminution zone to aide in projectile erosion. Unfortunately, it is impractical to

use these externally confined armor systems on a large scale. The edges of these

armor systems are pure metal, creating unprotected areas when a single layer of

tiles is applied.

In the past, metal matrix composites (MMC) have been shown to have

improved ballistic performance, this being attributed to dynamic work hardening.

[1] The work hardening of the metal was limited due to microstructural damage

created by shockwave interactions. It was thought that creating a multi-layered

ceramic metal composite, where each layer would contain different percentages of

ceramic, might further improve ballistic performance. [5] Further research is still

required, however, as many fundamental questions, such as what layering design

and what size scale is critical for optimal shockwave mitigation.

It is increasingly apparent that new concepts in armor design, as well as new

methods to rapidly create them to allow for multiple design iterations is needed.

This paper will discuss several new armor design concepts, as well as SFF

manufacturing methods for them. Some preliminary results are presented which

show feasibility for fabricating these new concepts by FDC.

ARMOR DESIGN CONCEPTS

If a level of confinement is to be provided to the ceramic phase, without using

an external method, some form of internal confinement via a reinforcing phase

must be applied. Infiltrating a porous ceramic perform with a molten metal,

creating a metal matrix composite, might provide some degree of internal

confinement. Evans et. al [6] has shown that periodic metal structures, when

designed properly, have improved properties over that of stochastic metal

structures. It stands to reason then, that a purposely-designed internal

reinforcement phase should provide improved properties and thus performance,

over that of randomly created reinforcement phase.

Work by Rödel et. al [7] and Claussen et. al [8] on alumina / aluminum

composites has shown that alumina reinforced with fibers of aluminum had the


same fracture toughness as alumina performs infiltrated with molten aluminum.

However the fiber reinforced alumina only required 13 volume % aluminum

whereas the infiltrated alumina contained 25 volume % aluminum. The fiber

reinforced system obviously allows for a much better improvement in properties,

while using less reinforcing phase.

Figure 1 is a schematic of a fiber reinforced armor composite concept. It

should be noted that to allow for easy visualization, the impact face has been

placed towards the bottom. The light gray regions are the ceramic phase and the

dark gray regions are the metal fiber reinforcement. Design flexibility to enable

testing of multiple designs is quite large. The diameter and placement of the metal

fibers, as well as their volume fraction can all be tailored for optimal properties,

when using FDC.

Figure 1.) Schematic of an internally reinforced ceramic metal armor

composite.

Figure 2 is a schematic of another possible armor composite that would utilize

shockwave mitigation as a means of improved performance. [9] The light gray

region is a continuous phase, while the darker spheres are a discontinuous phase.

Chin et. al [5] have stated that macroscopic interfaces in layered armor

composites are extremely important. It is thought that these interfaces will play a

role in controlling the reflection and refraction of shock waves during impact

events and could, if designed properly, be used to essentially steer the stress

waves and improve performance. Figure 2 provides many size scales of

interfaces, to enable the control of various frequency shock waves, and is easily

tailorable by changing the size of the spheres, their stacking order, and volume %.


Figure 2.) Schematic of an armor composite showing many scales of

macroscopic interfaces.

FUSED DEPOSITION OF CERAMICS

To fabricate and test multiple design iterations of the armor composites shown

in Figures 1 and 2, both rapidly and accurately, one would ideally want to use

some form of a solid freeform fabrication (SFF) technique. There are several SFF

techniques capable of producing functional ceramic components, among them are,

Stereolithography [10], 3-Dimensional Printing [11], Selective Laser Sintering

[12], Robocasting [13], and Fused Deposition of Ceramics (FDC) [14].

It has been shown through prior work with ISR-Si3N4, that FDC is capable of

producing functional components. [14] An average four point bend strength of

908 MPa and a chevron notch fracture toughness of 8.5MPam1/2

were measured

on FDC Si3N4 bars, which is comparable to commercial Si3N4. Moreover, the

bend strength and fracture toughness were not statistically different when

measured parallel and perpendicular to the build layers, indicating that FDC

produces nearly homogenous parts. Due to its extrusion based technology

however, FDC can be used to introduce crystallographic texture. By adding -

Si3N4 seeds to filament feedstock, preferred grain orientation was observed in

FDC Si3N4. [15] Figure 3 is a schematic of the FDC process.


Figure 3.) Diagram of the Fused Deposition of Ceramics process.

The FDC process works by extruding a ceramic loaded thermoplastic

filament, through a fine nozzle. The roads are laid down in the x-y plane in a

controlled fashion until a single layer is completed. A z-stage then indexes down

one layer thickness and another layer is built on top of the previous layer. A

complete description of the FDC process is given elsewhere. [16-18]

Subsequently, after FDC part fabrication, binder removal and sintering steps are

performed.

An advantage to using FDC as a fabrication technique for new armor design

concepts is that it can spatially distribute material in the x, y and z planes. With a

multiple extrusion head FDC system, one can also spatially distribute multiple

materials in all three planes, lending another tool to the design of these new

concepts.

MODEL MATERIAL SYSTEM

Research has been initiated to study the effect that reinforcing metal fibers

have on the impact performance of ceramic metal armor systems. A model system

of alumina and copper has been selected to allow for relatively easy fabrication,

and thus rapid design iteration. FDC has been used to fabricate several alumina

scaffolds containing designed channels for molten metal infiltration. These

scaffolds will then be spontaneously infiltrated with a wetting copper-oxygen

alloy to create a confining fiber phase. [19,20].


Initial work has been done using FDC filaments containing 55 volume %

Alcoa 152-SG alumina. Figure 4 is an SEM image showing the cross section of a

sintered alumina scaffold produced by FDC. This scaffold was designed with a

volume fraction gradient through the thickness and it is apparent that the channel

volume is relatively uniform in each layer.

Further work was done to show that infiltration of a sintered alumina scaffold

was feasible. Figure 5 is a light optical image of a sintered alumina scaffold

spontaneously infiltrated with copper. This sample was produced by filling the

sintered scaffold with copper powder, and then melting it under static air. The

copper alloyed with oxygen in the air and subsequently wet and infiltrated the

scaffold. While this is by no means an ideal method of infiltration, it does show

that ceramic metal reinforced armor composites can be fabricated using a

combination of FDC and spontaneous infiltration.

Figure 4.) SEM image showing a cross section of a sintered alumina scaffold.


Figure 5.) Light optical image showing an infiltrated alumina scaffold. The

dark ovals are the alumina scaffold, while the lighter phase in between is the

copper-oxygen alloy.

SUMMARY

New armor design concepts are needed to help solve the externally confined

ceramic armor problem as well as improve upon armor performance by

shockwave mitigation. It is thought that providing a purposely designed, internal

reinforcement phase might provide a degree of internal confinement. It is further

thought that tailoring of macroscopic interfaces in armor composites to mitigate

stress waves is another approach. The use of FDC, along with metal infiltration

has been shown to be feasible way to rapidly design iterate and fabricate novel

internally reinforced ceramic armor composites.

ACKNOWLEDGEMENTS

The authors would like to thank the U.S Army Research Laboratory for

funding under cooperative agreement number DAAD19-01-2-0004, as well the

CCMC for additional support. We would also like the thank Dr. McCauley, Dr.

Adams, and Dr. Chin of the ARL for technical input.

REFERENCES1E.S.C. Chin, “Army focused research team on functionally graded armor

composites,” Materials Science and Engineering A, 259 [2] 155-61 (1999).


2C.E. Anderson Jr. and S.A. Royal-Timmons, “Ballistic performance of

confined 99.5%-Al2O3 ceramic tiles,” International Journal of Impact

Engineering, 19 [8] 703-13 (1997) 3H.D. Espinosa, N.S. Brar, G. Yuan, Y. Xu, and V. Arrieta, “Enhanced

ballistic performance of confined multi-layered ceramic targets against long rod

penetrators through interface defeat,” International Journal of Solids and

Structures, 37 [36] 4893-4913 (2000). 4H.D. Espinosa, S. Dwivedi, P.D. Zavattieri, and G.Yuan, “A numerical

investigation of penetration in multilayered material/structure systems,”

International Journal of Solids and Structures, 35 [22] 2975-3001 (1998). 5Y. Li, K.T. Ramesh, and E.S.C. Chin, “Dynamic characterization of layered

and graded structures under impulsive loading,” International Journal of Solids

and Structures, 38 [34-35] 6045-61 (2001).6A.G. Evans, J. W. Hutchinson, N. A. Fleck, M. F. Ashby and H. N. G.

Wadley, “The topological design of multifunctional cellular metals,” Progress in

Materials Science, 47 [3-4] 309-27 (2001) 7H. Prielipp, M.. Knechtel, N. Claussen, S.K. Streiffer, H. Müllejans, M.

Rühle, and J. Rödel, “Strength and fracture toughness of aluminum/alumina

composites with interpenetrating networks,” Materials Science and Engineering

A, 197 [1] 19-30 (1995). 8O. Raddatz, G.A. Schneider, W. Mackens, H.Voß, and N. Claussen,

“Bridging stresses and R-curves in ceramic/metal composites,” Journal of the

European Ceramic Society, 20 [13] 2261-73 (2000). 9E.S.C. Chin, Private Communication

10M.L. Griffith and J.W. Halloran, “Freeform Fabrication of Ceramics via

Stereolithography,” Journal of the American Ceramic Society, 79 [10] 2601-608

(1996)11

E. Sachs, M.J. Cima and J.Cornie, “Three-Dimensional Printing: Rapid

Tooling and Prototypes Directly from CAD Representation”; pp. 27-47 in Solid

Freeform Fabrication Proceedings, Vol. 1. Edited by J.J. Beamen, H.L. Marcus,

D.L. Bourell, R.H. Crawford, and J.W. Barlow. University of Texas, Austin, TX,

199012

D.L. Bourell, H.L. Marcus, J.W. Barlow and J.J. Beamen, “Selective Laser

Sintering of Metals and Ceramics,” International Journal of Powder Metallurgy

Technology, 28 [4] 369-80 (1992) 13

J. Cesarano, “Review of Robocasting Technology,” in Proceedings of the

1998 MRS Fall Meeting, Symposium V, Solid Freeform and Additive

Fabrication, edited by D. Dimos, S.C. Danforth, and M.J. Cima, Boston, MA, pp.

133-39 (1998) 14

S. Ranngarajan, J. McIntosh, A. Bandyopadhyay, R.C. McCuiston, N.

Langrana, A. Safari, S. C. Danforth, M. Bertoldi, S. Guceri, R. B. Clancy, V.


Jamalabad, C. Gasdaska and P. J. Whalen, “Functional Si3N4 Ceramics by Fused

Deposition: Microstructure and Mechanical Properties,” To be Submitted to

Journal of Materials Research. 15

R.C. McCuiston, B.L. Harper, S. Rangarajan, W.E. Mayo, S.C. Danforth

and C. Gasdaska, “Generation of Texture in Si3N4 made by Fused Deposition of

Ceramics (FDC) through use of -Silicon Nitride Seeds” to be submitted Journal of the American Ceramic Society

16C. Dai, G. Qi, S. Rangarajan, S. Wu, N. Langrana, A. Safari, and S. C.

Danforth, “High Quality, Fully Dense Ceramic Components Manufactured Using

Fused Deposition of Ceramics,” pp. 411-20 in Proceedings of the 7th

Solid

Freeform Fabrication Symposium, edited by D. L. Bourell, J. J. Beaman, R.H.

Crawford, H. L. Marcus and J. W. Barlow. University of Texas, Austin, TX, 1997 17

S. Rangarajan, G. Qi, N. Venkataraman, A. Safari, and S.C. Danforth,

“Powder processing, rheology, and mechanical properties of feedstock for fused

deposition of Si3N4,” Journal of the American Ceramic Society, 83 [7] 1663-

1669 (2000) 18

N. Venkataraman, S. Rangarajan, M. J. Matthewson, B. Harper, A. Safari, S.

C. Danforth, G. Wu, N. Langrana, S. Guceri, and A.Yardimci, “Feedstock

Material Property – Process Relationships in Fused Deposition of Ceramics

(FDC),” Rapid Prototyping Journal, 6 [4] 244-52 (2000) 19

E.J. Gonzalez and K.P. Trumble, “Spontaneous infiltration of alumina by

copper-oxygen alloys,” Journal of the American Ceramic Society, 79 [1] 114-20

(1996)20

N.A. Travitzky, and A. Shlayen, “Microstructure and mechanical properties

of Al2O3/Cu-O composites fabricated by pressureless infiltration technique,”

Materials Science and Engineering A, 244 [2] 154-60 (1998)


Ultra-Lightweight and Novel Concepts

DEVELOPING AN ULTRA-LIGHTWEIGHT ARMOR CONCEPT

Charles E. Anderson, Jr.

Southwest Research Institute

P.O. Drawer 28510

San Antonio, TX 78228-0510

ABSTRACT

Significant reductions in armor weight have been realized over the past 30

years by the introduction of non-metallic materials (e.g., ceramics, composites,

fabrics) into armor designs. Further reduction in state-of-the-art lightweight

armors, so as to have an ultra-lightweight armor system, is a daunting challenge,

and most probably can be accomplished only by the use of materials and

geometries in novel arrangements. The process of identifying possible defeat

mechanisms and then how to exploit these mechanisms, including the

development of materials with enhanced properties, is explored.

INTRODUCTION

Armor is a synergy of mechanics and materials. I will use the term “defeat

mechanism” to denote the mechanics that the armor designer invokes to achieve a

desired affect on the threat, which for light armor is typically a small arms (rifle-

fired) bullet. A threat is characterized by its velocity, mass (inertia), geometry

(length, diameter, nose shape), and strength (flow stress and some measure of

failure, such as stain to failure). For the purposes of this paper, where we are

considering light armor, the threat is defined as the 7.62-mm armor-piercing

(APM2) bullet, shown in Fig. 1; and the 0.30-cal monolithic steel bullet

developed by Wilkins [1]. Wilkins developed the 0.30-cal bullet as a surrogate

projectile for the APM2 bullet, largely to decrease the scatter in experimental data

that resulted from fracturing of the hard steel core in the APM2 bullet. Muzzle

velocity for the bullets is 820-850 m/s. The physical characteristics of these two

bullets are summarized in Table I. The surrogate bullet has a penetration

performance that is similar to that for the APM2 bullet into hard targets.

Defeat mechanisms that might be used against an armor-piercing (AP) bullet

are shown in Table II. These defeat mechanisms are not all inclusive, and they

are often used in combination with each other. For example, tipping/rotating the



bullet is usually used with spaced elements with the objective of spreading the

load of the bullet onto a subsequent element.

.7840cm

3.53cm

2.74cm

.6172cm

Jacket

Point Fi l lerBase Fi l ler

Core

Fig. 1. Schematic of 7.62-mm APM2 Bullet

Table I. Physical Properties of the 7.62-mm AP Bullets

7.62-mm APM2 Bullet 7.62-mm Surrogate AP Bullet

Mass: 10.74 g Mass: 8.32 g

Length: 3.53 cm Length: 2.81 cm

Core Mass: 5.25 g Nose: 55 cone

Core Length: 2.74 cm Hardness: Rc 55

Core Hardness: Rc 62-65

Table II. Defeat Mechanisms

Deceleration (retarding force) Erosion

Obliquity Stripping the jacket

Tipping or rotating Spreading the load

Projectile fracture Blunting the nose

Spaced elements Structural response (holding the

load through a distance)

As stated in the first paragraph, armor is a synergy of mechanics and

materials. Materials are used to amplify the performance of the mechanics. And

since weight is always an issue with armor, we demand the “ultimate”

performance out of materials. The materials are pushed to their limit, that is,

failure. As Wilkins states: “The application of materials to light armor is unusual

because material properties are utilized in the region of material failure, i.e., if the

armor doesn’t fail for a given ballistic threat, it could be made lighter” [2].

These observations set up an alternative title for the paper: Why is it so

difficult to decrease the weight of a lightweight armor system? In the remainder

of the article, I will show how invoking different defeat mechanisms (often

through a change of materials) can lead to weight reductions of an armor, and also

show the difficulties inherent in achieving significant weight reductions through

evolutionary improvements in material properties.


TRADITIONAL ARMOR

The conventional role of armor is to decelerate the projectile until it stops, i.e.,

it is defeated. The depths of penetration (DOP) as a function of impact velocity

for the AP bullet into 6061-T6 aluminum and armor steel are shown in Fig. 2.

The filled circles denote experimental data for the APM2 bullet into an aluminum

target. The lines are predictions using the Walker-Anderson penetration model

[3]. For metallic targets, semi-infinite penetration data can be used to estimate the

thickness of armor required to stop the bullet, at a specified impact velocity, by

adding approximately one bullet diameter to the semi-infinite DOP.

The bullet penetrates considerably less into armor steel than into 6061-T6

aluminum. However, the armor designer is concerned about the weight of an

armor system. The figure of merit, instead of depth of penetration (or thickness of

the target), is areal density, which is the product of the armor thickness and the

material density. The areal densities of 6061-T6 aluminum and armor steel

required to defeat the AP bullet are shown in Fig. 3. Although the bullet

penetrates considerably less into steel than into aluminum, the decrease in

penetration is not sufficient to compensate for the differences in density.

Velocity (m/s)

0 200 400 600 800 1000

Pen

etr

ati

on

De

pth

(m

m)

0

10

20

30

40

50

60

Alu

min

um

Armor S

teel

Armor Steel (eroding)

Fig. 2. DOP vs. velocity for several

metallic targets.

Are

al

De

ns

ity (

g/c

m2)

0.0

5.0

10.0

15.0

20.0

Al-

60

61-T

6

Arm

or

Ste

el

Ero

din

g S

teel

B4C

/Al

Fig. 3. Areal density required to defeat

the AP bullet at ~820 m/s.

Steels come in different strengths, and if a steel of a different strength is

substituted for the armor steel, then the penetration is changed. For example, if

mild steel is used, the areal density to stop the AP threat is approximately

22 g/cm2; if a high-hard steel is used, the areal density to stop the threat is

approximately 10 g/cm2. In general, stronger materials provide higher

decelerating forces to the penetrator. However, since there is usually a trade off


of strength versus ductility, there is generally a limit to the strength that can be

realized. Effectively, the advantages of the increased strength are not realized

through the entire thickness of the armor element because the material fails,

generally through some damage localization process (typically involving shearing

out of an intact plug). This is the reason dual-hardness armor steel is fabricated;

the front side is made very hard, but the backside of the armor element is less

strong, but considerably more ductile.

AP bullets are very hard, and they penetrate into metallic targets in the rigid-

body penetration mode; that is, the bullets do not deform during penetration. If

the target material could be made stronger so that the bullet deforms, penetration

will decrease. If the hard steel penetrator can be made to erode—the turning of

projectile material so that there is radial flow (mushrooming), to such an extent

that the induced strains exceed the ability of the material to remain cohesive,

thereby resulting in particulation of projectile material and, as a consequence,

mass loss—then the depth of penetration is considerably reduced, as denoted by

the short dashed line in Fig. 2. Eroding penetration results in a significant

reduction in areal density, as shown in Fig. 3. To achieve erosion, a material is

required that is “harder” than the penetrator material (and so is harder than armor

steel), but is lighter than steel (so that the areal density is favorable). Ceramics

are such a material; they have very large compressive strengths, and have

densities less than that of steel. Wilkins determined that a 7.62-mm boron carbide

(B4C) ceramic tile glued to 6.35-mm 6061-T6 aluminum substrate could defeat

the AP surrogate bullet at an impact velocity of 820 m/s [4]. The areal density of

this armor is 3.62 g/cm2, which is also plotted in Fig. 3.

The response of an AP bullet against a B4C ceramic tile glued to an aluminum

(6061-T6) substrate is shown in Fig. 4. The front view shows the damage to the

ceramic, and the side view shows the deformation of the aluminum substrate

plate. Horizontal lines were drawn on the back of the substrate plate to assist in

visualizing the deformation. As can be seen, the substrate plate absorbs some of

the kinetic energy through deformation. An estimate of the kinetic energy that is

absorbed by the plate can be obtained by examining the results of VS-VR

experiments against a bare aluminum plate, where VS is the striking (impact)

velocity of the bullet, and VR is the residual velocity of the bullet after plate

perforation. The results of a number of experiments with the APM2 bullet are

shown in Fig. 5. These same data are plotted as a function of the impacting

kinetic energy (instead of VS) in Fig. 6. It is seen that the substrate material can

absorbed approximately 0.5kJ of kinetic energy.

Thus, there are requirements for armor elements with different material

properties. A hard element is needed to erode and decelerate the bullet. A ductile

element is required to capture the remnants of the eroded bullet. Materials with

different properties need to be assembled in the most advantageous way.


(a) (b)

Fig. 4. Post-test photograph of impact of AP bullet against ceramic/aluminum

target: (a) front view of ceramic element; (b) side view of target showing

deformation of aluminum element.

VS (m/s)

0 200 400 600 800

VR (

m/s

)

0

200

400

600

800

Experiment: Normal

Experiment: Reverse

Computation: Normal

Computation: Reverse

Fig. 5. VR vs. VS for 6.35-mm-thick

6061-T6 aluminum plate.

Kinetic Energy (kJ)

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

VR (

m/s

)

0

200

400

600

800

Experiment: Normal

Experiment: Reverse

Computation: Normal

Computation: Reverse

Fig. 6. Residual velocity vs. striking

kinetic energy (from Fig. 5).

FURTHER WEIGHT REDUCTION

Now we want to decrease the weight of the armor further. Four possible ways

include: 1) make the front ceramic element thinner; 2) make the substrate thinner;


3) change the substrate material; and 4) improved material properties. Each of

these will be discussed, with an emphasis on the last item.

Make the Front Ceramic Element

Thinner: Figure 7 shows the results of

decreasing the thickness of the ceramic

element. The threat defeats the target

easily, and with relatively high residual

velocity. So unless something can be

done to enhance the properties of the

ceramic (which is the fourth item),

decreasing the ceramic thickness is not

a viable option for decreasing the

weight of the system.

Make the Substrate Thinner: The

substrate material must absorb the ki-

netic energy of the residual bullet after

being decelerated and eroded by the

ceramic element. Wilkins determined

the ballistic limit, VBL, for an AD85/Al

VS

(m/s)

500 600 700 800 900 1000

VR

(m/s

)

0

100

200

300

400

500

600

700

800

5.08 mm

6.35 mm

7.62 mm

Fig. 7. Experimental VR vs. VS for AP

bullet.

substrate system as a function of ceramic thickness ( ) and substrate thickness ( )

[1]; the results for this experimental parametric study are shown in Fig. 8.

Wilkins found a significant decrease in ballistic performance for a

ceramic/aluminum substrate system when the aluminum thickness dropped below

~6 mm. He determined that this result is a consequence of the failure mode for

the substrate changing from shear plugging to petalling at 6 mm. Therefore,

the substrate cannot be made much, if any, thinner.

Change the Substrate Material: Current, state-of-the-art, lightweight armors

use a composite material in place of the aluminum substrate. Such composites

consist of Kevlar™ and polyethylene composites. Two such materials, for

example, are Gold Shield™ and Spectra Shield™, which consist of Kevlar™

fibers and polyethylene fibers, respectively, embedded in a polyethylene matrix.

The thicknesses of these composite substrate materials are considerably greater

than that of the aluminum for comparable ballistic performance. However,

because the density of the composites is considerably less than that of aluminum

(~0.90-1.2 g/cm3 vs. 2.7 g/cm

3), the overall areal density of the armor system is

decreased. In effect, this is the reverse of the aluminum-steel trade-off of density

versus strength described earlier. The areal density can be decreased by

approximately 15% using composite substrates instead of aluminum.

Improved Material Properties: Improvements in material properties can lead

to increased ballistic performance. It is not unusual to have material scientists

claim that the improvement in a material property will “naturally” result in better


ballistic performance of the material. Since there are significant costs associated

with developing a material with enhanced properties, it is desirable to have an

estimate of the gains in ballistic performance that might be realized from such an

improvement. This is the advantage of having models, which then can be used to

make such projections. The remainder of the article will focus on the use of

models to enhance our understanding of experimental observations, and to

quantify the improvement in ballistic performance with an enhanced material

property.

Thickness, (mm)

0.0 2.5 5.0 7.5 10.0 12.5

Ba

llis

tic

Lim

it,

VB

L (

m/s

)

0

200

400

600

800

1000 = 8.64 mm

= 7.87 mm

= 6.35 mm

= 4.06 mm

Fig. 8. Ballistic limit velocity as a function of ceramic and substrate

thickness for AD85 (Al2O3)/6061-T6 Al (from Wilkins [1]).

ANALYTICAL AND COMPUTATIONAL MODELING

We would like to use the results of modeling to guide armor development. In

particular, we would like to investigate, and quantify, the advantage of improved

material properties. In order for modeling to be useful for this endeavor, it must

be demonstrated that the modeling captures the essential features of observed

phenomena, and that the modeling provides reasonable agreement with

experimental data. It is not necessary for the model to reproduce exactly

experimental results, but it is necessary that the model be sufficiently accurate so

that it can predict the correct trends. This is why the first requirement is

necessary—that the model captures the essential features of observed

phenomena—because model parameters can be tuned to provide good agreement

with experimental results, but not have the correct mechanics and physics.

Figure 9 shows flash radiographs of the APM2 bullet, 15.3 s and 20.7 s

after impact, against a 7.62-mm-thick B4C tile backed by a nominal 6.35-mm-


thick, aluminum (6061-T6) substrate plate. The impact velocity was approxi-

mately 820 m/s for these experiments. The bullet is “dwelling” at the surface of

the ceramic (not penetrating) in the first image; by approximately 20 s, the

integral strength of the ceramic no longer can support dwell, and the bullet is

penetrating (the right image).

(a) 15.3 s (b) 20.7 s

Fig. 9. Flash radiographs of the APM2 bullet impacting a B4C/Al target.

A simple analytical model of dwell has

been developed; the idealized model is

shown in Fig. 10. The governing equations

are shown below the figure, where p is the

projectile density, v is the tail velocity, is

the current length of the projectile, Y

λp is the

projectile flow stress, u is the penetration

velocity, and t denotes time. The first equa-

tion describes deceleration of the tail, and the

second equation describes the shortening

(and thus mass loss) of the projectile. The

third equation is the statement of the assump-

tion that the penetration velocity, u, is zero.

These equations can be solved explicitly.

The solutions for the surrogate AP projectile,

at an impact velocity of 820 m/s, are shown

��

VV

0

)v(

vpp

u

udt

d

Ydt

d

λ

λ

Fig. 10. Analytical model for

dwell.

in Figs. 11-13. The results of deceleration of the bullet as a function of time are

shown in Fig. 11. The length of the bullet decreases because of erosion, with an

attendant loss of mass, Fig. 12. Mass loss, initially, is quite small because of the

pointed noise.

The kinetic energy of the bullet as a function of time is plotted in Fig. 13. The

percentage of kinetic energy lost to erosion, and that lost to deceleration, can be


separated. It is seen that each of the “defeat” mechanisms contributes to a

significant loss in kinetic energy of the AP bullet. Although the model provides

an idealized description of dwell, it permits a quantification of the advantages if

dwell can be extended for a few additional microseconds. For example, the flash

radiograph in Fig. 9(a) was taken at 15.3 s after impact. At this time, the bullet

has lost 44% of its initial kinetic energy. If dwell could have been made to extend

to 20.7 s, Fig. 9(b), then the kinetic energy would have decayed to

approximately 23%, a significant decrease in kinetic energy.

Time ( s)

0 5 10 15 20 25 30

% K

ineti

c E

nerg

y

0

10

20

30

40

50

60

70

80

90

100K

ineti

c E

nerg

y (

kJ)

0.0

0.5

1.0

1.5

2.0

2.5

KE with Mass Loss

(Erosion Only)

Erosion & Deceleration

KE lost due

to erosion

KE lost due

to velocity decay

Time ( s)

0 5 10 15 20 25 30

Ve

loc

ity (

m/s

)

0

200

400

600

800

1000

Fig. 11. Velocity vs. time for

dwelling AP bullet.Fig. 12. Length and mass vs. time for

dwelling AP bullet.

Time ( s)

0 5 10 15 20 25 30

Mass (

g)

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

Len

gth

(m

m)

0.0

5.0

10.0

15.0

20.0

25.0

30.0

Fig. 13. Kinetic energy vs. time for dwelling AP bullet.


The analytical dwell model is useful in quantifying the effects of dwell, but it

cannot predict if dwell will occur, and if it does, for how long it will last. To

make this prediction, we need to turn to a computational model. Anderson and

Walker [5] modified a computational ceramics model developed by Wilkins [1-2].

The modified model, implemented into the wavecode CTH [6], has 5 model

constants: intact (compressive) strength of the ceramic, tensile strength of the

ceramic, the slope and cap of a Drucker-Prager yield surface for the damaged

(comminuted) ceramic, and a constant that governs the speed of damage from

intact to comminuted ceramic (a fraction of the shear wave velocity). A damage

parameter, f, is defined: 0f implies intact ceramic; 1f denotes

completely failed ceramic within a computational cell. Failure of a computational

cell is initiated when the calculated tensile stress exceeds the material tensile

strength, subject to the condition that a cell is next to a cell that has failed, 1f ,

or is next to a material interface or free surface. Once damage is initiated, the

strength of the computational cell goes from that of intact material to that of the

comminuted material at the prescribed damage rate (the fifth parameter). All

parameters but the last are determined from independent laboratory experiments;

the last parameter was calibrated to achieve the correct residual length (LR) of

recovered cores from the APM2 bullet. (Thus, the intact ceramic is modeled

elastic-plastic until failure; thereafter, the failed or comminuted ceramic strength

follows a Drucker-Prager constitutive relationship. The metallic elements—

projectile and substrate materials—are modeled as elastic-plastic, with strain

hardening, strain rate and temperature effects. All materials are considered

isotropic.) The modified model reproduces quite accurately a wide variety of

experimental results of impact into thin ceramic tiles, including the phenomenon

of dwell.

The nose and tail velocities of one such simulation, of the AP surrogate bullet

into 7.62-mm B4C/6.35-mm 6061-T6 Al, is shown in Fig. 14. Wilkins showed

that this armor configuration stopped the bullet; the simulation is in agreement

with experiment. The simulation results indicate that dwell lasts for

approximately 20 s, and that the bullet penetrates as a rigid body (nose and tail

velocities the same) after 26 s. It is interesting to note that the kinetic energy

remaining after 23 s of dwell (Fig. 13), is approximately 0.5 kJ, the same energy

that can be absorbed by 6.35 mm of aluminum (Fig. 6).

The analytical dwell model and the computational model provide an

explanation for the “sudden” drop off in performance of our ceramic armor

system as the ceramic tile is made thinner, Fig. 7. If it is assumed that dwell does

not last as long if the ceramic tile is made thinner, then there is less deceleration

of the bullet, less erosion of the bullet, and the resulting kinetic energy of the

“remnant” bullet when it reaches the substrate is considerably higher than 0.5 kJ

(see Figs. 11-13). Evidence of less erosion as the ceramic tile is made thinner is


seen in recovered cores, Fig. 15. However, it turns out to be even worse than

simply a decrease in dwell time. According to the computational model, dwell

hardly persists for the thinner tiles, Fig. 16. There is a pseudo-dwell period where

the penetration velocity is relatively small, but an almost zero penetration velocity

is not predicted. The reason for this will be discussed a little later.

Time ( s)

0 10 20 30 40 50 60 70 80 90 100

Vel

oci

ty (

m/s

)

0

100

200

300

400

500

600

700

800

900

Tail

Nose

Fig. 14. Nose and tail velocities vs. time from numerical simulation of the

surrogate bullet against 7.62-mm B4C/6.35-mm 6061-T6 Al.

VS (m/s)

600 700 800 900 1000

LR

(m

m)

0

5

10

15

20

25

5.08 mm

6.35 mm

7.62 mm

Experiment--5.1 mm

Experiment--6.4 mm

Calculation--5.08 mm

Calculation--6.35 mm

Experiment--7.62 mm

Fig. 15. Residual length of AP cores.

Time ( s)

0 10 20 30 40 50 60 70

Vel

oci

ty (

m/s

)

0

100

200

300

400

500

600

700

800

900

5.08 mm

6.35 mm

7.62 mm

Fig. 16. Nose and tail velocities for

different ceramic tile thickness.


IMPROVED MATERIAL PROPERTIES

The models will now be used to quantify the gains that might be expected if a

boron carbide ceramic could be made with improved material properties. Within

the context of the computational model, the two parameters that might be

improved through changes in processing are the compressive and tensile

strengths. Simulations indicate that increasing the compressive strength of the

ceramic does not substantially change the ballistic performance of the ceramic.

This implies that the ceramic is already sufficiently hard to erode the AP bullet. It

might be expected, however, that since the model is a tensile-failure-initiation

model, that improving the tensile strength of the ceramic will improve ballistic

performance.

Computational results, using the

measured tensile strength of B4C ( f =

0.3 GPa), are compared to experi-

mental results in Fig. 17 for a tile

thickness of 5.08 mm (over a 6.35-mm

Al substrate). Good agreement is seen

for VR vs. VS. So additional simula-

tions were performed where the tensile

strength was increased by a factor of 3

and 5, to 0.9 and 1.5 GPa, respectively.

The results are shown for a 5.08-mm-

thick tile in Fig. 18. Even with a five-

fold increase in the tensile strength, the

residual velocity for an 850-m/s-impact

velocity is over 300 m/s. A strength of

1.5 GPa is equivalent to the flow stress

of a hard armor steel, so it is not clear

VS (m/s)

500 600 700 800 900

VR (

m/s

)

0

100

200

300

400

500

600

700

f = 0.3 GPa

Fig. 17. Computational and

experimental VR vs. VS (5.08-mm B4C)

that such a ceramic can even be fabricated. Even if it such a “new” ceramic could

be fabricated, the overall areal density would change from 3.62 g/cm2 (7.62-mm

B4C) to something slightly greater than 3.0 g/cm2 (5.08-mm B4C), a change of

only 18% for a fivefold increase in the tensile strength!

The simulation results were analyzed to determine why such a dramatic

increase in a physical property has so little influence on ballistic performance.

The minimum principle stress throughout the ceramic tile was plotted at a number

of times after impact. These plots show that tensile stresses within the ceramic

element exceed 1.0 GPa in the entire volume under the penetrator during the first

few microseconds, with some areas having tensile stresses in excess of 1.5 GPa

(for the 5.08-mm-thick tile). Thus, the problem is that the impact event is so

severe that the material is simply “overwhelmed” by the dynamics. As the

ceramic tile is made thinner, the ability to resist tensile stresses decreases


nonlinearly (bending stiffness is proportional to the thickness cubed). Therefore,

increased tensile strengths of 3 to 5 times the current material property value are

not sufficient to compensate for the increased tensile stresses generated from

impact.

VS (m/s)

500 600 700 800 900

VR (

m/s

)

0

100

200

300

400

500

600

700

f = 0.3 GPa

f = 0.9 GPa

f = 1.5 GPa

Fig. 18. Effects of increased tensile

strength: 5.08-mm B4C

VS (m/s)

500 600 700 800 900

VR (

m/s

)

0

100

200

300

400

500

600

700

f = 0.3 GPa

f = 0.9 GPa

5.08 mm

6.35 mm

Fig. 19. Effect of increased tensile

strength: 5.08-mm & 6.35-mm B4C

Nevertheless, failure time through the ceramic element is increased by the

increased f, as can be inferred from the decrease in residual velocity. That is, the

armor system can be made lighter using the “improved” material. Predictions of

VR vs. VS for a 6.35-mm-thick tile are shown in Fig. 19. VR, for an impact

velocity of 850 m/s, is approximately 200 m/s for f = 0.9 GPa. Experience has

shown that when VR’s are ~200 m/s or lower—because of the steepness of the

VR-VS curve near the ballistic limit—the armor system is approximately at the V50

design. So the increased fracture strength does have an effect on ballistic

performance, but the effect is not nearly as large as one might have thought based

on the significant increase in f. The increase of f from 0.3 GPa to 0.9 GPa

results in a decrease in the weight of the armor system from 3.62 g/cm2 to

approximately 3.30 g/cm2. Unfortunately, it has taken a significant improvement

in a material property to realize a 9% decrease in areal density.

SUMMARY AND CONCLUSIONS

Light armor is a synergy of mechanics and materials. Because the armor

designer is demanding the ultimate performance out of the materials that are being

used, the performance of lightweight armor is “precipitous,” i.e., a very small

change in geometry (for example, a small decrease in the thickness of an armor


element) and the armor is defeated quite easily. This “precipitous” behavior

makes it difficult to decrease the weight unless the operative mechanics (defeat

mechanisms) are changed (e.g., adding erosion to deceleration), or unless a

material is changed (e.g., changing the substrate from aluminum to a composite).

Thus, as shown by our example of an improved ceramic (an increase in the

tensile strength of the ceramic), evolutionary changes in material properties result

in incremental changes in ballistic performance, and incremental decrease in

weight. A significant increase in ballistic performance (i.e., a significant decrease

in weight) requires an advance defeat mechanism (or invoking several defeat

mechanisms), and/or a revolutionary advance in materials.

ACKNOWLEDGEMENT

The author would like to thank Dr. Steve Wax of DARPA and Mrs. Janet

Ward of the U. S. Army Soldier Systems Command for their support and

suggestions in the preparation of this paper. The author would also like to thank

Mr. Dick Sharron (SwRI) for his assistance in running of the numerical

simulations. This work was funded under contract DAAD16-00-C-9260.

REFERENCES 1M. L. Wilkins, “Mechanics of Penetration and Perforation,” Int. J. Engng.

Sci., 16(11), 793-807, 1978. 2M. L. Wilkins, “Third Progress Report of Light Armor Program,” UCRL-

50460, Lawrence Livermore Laboratory, Livermore, CA, July 1968. 3J. D. Walker and C. E. Anderson, Jr., “A Time-Dependent Model for Long-

Rod Penetration,” Int. J. Impact Engng., 16(1), 19-48, 1995. 4M. L. Wilkins, R. L. Landingham, and C. A. Honodel, “Fifth Progress Report

of Light Armor Program,” UCRL-50980, Lawrence Livermore Laboratory,

Livermore, CA, 1970. 5C. E. Anderson, Jr. and J. D. Walker, “Ceramic Dwell and Defeat of the 0.30-

Cal AP Projectile,” 15th

U.S. Army Symp. on Solid Mech., Myrtle Beach, SC,

April 12-14, 1999. 6J. M. McGlaun, S. L. Thompson, and M. G. Elrick, “CTH: A Three-

Dimensional Shock Wave Physics Code,” Int. J. Impact Engng., 10, 351-360,

1990.


NOVEL IDEAS IN MULTI-FUNCTIONAL CERAMIC ARMOR DESIGN

Sia Nemat –Nasser*, Sai Sarva, Jon B Isaacs and David W Lischer

Center of Excellence for Advanced Materials

Department of Mechanical and Aerospace Engineering

University of California, San Diego

La Jolla, CA 92093-0416

ABSTRACT

Ceramics such as Al2O3, SiC, TiB2 and B4C have been used in integrated

armor for over a decade and are an excellent prospect for the next generation

multi-functional armor systems. It is necessary to incorporate novel ideas in

ceramic armor design so as to develop improved armor with minimal added mass.

Preliminary research has demonstrated that the defeat capability of ceramic armor

tiles could be considerably improved by tightly wrapping them in a thin

membrane of suitable tensile strength. In the present paper we present some

recent experimental results relating to the effect of thin membranes attached to the

front face of Al2O3 armor tiles, on their ballistic performance. The experiments

were conducted to study the comparative effect of several front-face materials,

such as glass-fiber tape, E-glass/epoxy pre-preg, Carbon-fiber/epoxy pre-preg and

Ti-3%Al-2.5%V alloy. Tungsten heavy alloy was used as the projectile material.

It was observed that confinement by a thin layer of E-glass/epoxy pre-preg

resulted in a nearly 20% improvement in the ballistic efficiency for a mere 2.5%

increase in areal density. The improvement in ballistic efficiency is accompanied

by an altering of the failure mechanisms. High-speed photography and flash

radiography techniques have been used to gain insight into the mechanisms that

may be responsible for this improvement.

INTRODUCTION

The next generation armor systems require integration of several attributes

within hybrid structures, which can be accomplished through introduction of

novel concepts in the materials-structural design. These attributes may include

great agility, effective communication, and controlled signature. New

Corresponding author: [email protected] (858) 534-4914, Fax: (858) 534 2727



materials/structures must therefore be created in such a manner that they are light-

weight, impact resistive, have structural integrity and at the same time can have

signature management and controlled communication capabilities. This can be

achieved by incorporating periodic arrays of thin conductor wires1, exhibiting the

desired electro-magnetic response into high strength, low-density hybrid

composites. Extensive research has been conducted to increase the ballistic

efficiency to areal density ratio of ceramics through various techniques. A number

of researchers have studied the effect of confinement of ceramics on their ballistic

performance and failure modes. Shockey et al.2 studied the failure

phenomenology of confined ceramics under rod impact. They concluded that the

key processes are crushing of the ceramic and the subsequent flow of fine

fragments lateral to and opposite to the direction of impact. Woodward et al.3

studied the perforation of confined and unconfined ceramic targets by pointed and

blunt projectiles. It was observed that front confinement of ceramic results in a

greater overall fragmentation. However, their experiments suggest that less

amount of very fine ceramic powder may form in the confined case as compared

to the unconfined target. Anderson and Morris4 have studied the effect of

projectile diameter on its erosion for thick (~ 4 cm) Al2O3 tiles under lateral and

rear confinement. They also observed that for constant-mass projectiles, longer

rods erode more than shorter rods for the same ceramic thickness.

When projectiles impact ceramic targets, a pulverized zone is formed ahead of

the projectile head due to intense stress conditions. Understanding of the failure

mechanisms resulting in this pulverization is important for developing improved

models and for designing better armor systems. Curran et al.5 present a

micromechanical model for comminution and granular flow of ceramics under

impact. Cortes et al.6 have numerically modeled the impact of ceramic-composite

armor. They present a constitutive model for finely pulverized ceramic taking into

account internal friction and volumetric expansion. Grace and Rupert7 have

incorporated models of Curran et al.5 and Cortes et al.

6 to analyze long rods

penetrating ceramic targets at high velocities. McGinn7

et al.8 have

microscopically studied the deformation and comminution of shock loaded Al2O3

to understand the failure mechanisms that produce this pulverized zone, often

referred to as the ‘Mescall zone’.

Recently, McGee et al.9 studied the effect of thin membrane wrapping on the

defeat capability of Al2O3 and SiC ceramic armor tiles. It was observed that

tightly hand-wrapping the tiles in commercially available Scotch glass fiber tape

improves the ballistic efficiency by nearly 20%. It was also observed that this

improvement is mainly a result of impact-face constraint that the tape provides,

and that the back-face constraint had little (if any) effect on the ballistic

efficiency.


Following these observations, further experiments were conducted to

investigate the effect of the front-face-attached membrane on the failure

mechanisms and the projectile-target interaction. Also, the material and the

thickness of the front-face-attached membrane were varied to observe the

resulting effects on the ballistic efficiency. The experimental results are discussed

in the present paper together with some numerical simulations, leading to some

tentative conclusions on the potential factors that may be involved in this process.


Gas-Gun

A single stage gas-gun is used to launch the projectile. Helium is the

driving gas. The barrel diameter is 2.54 cm and its length is 4.8 m. Two velocity

sensors at the muzzle end of the barrel are used to measure the intial velocity of

the projectile. The sensors also trigger the high-speed camera and flash X-ray

heads. The gas-gun can launch a 17 gm sabot-projectile assembly at up to about

1100 m/s. The gas gun is operated with two different configurations of target

assembly depending on the nature of data of interest.

The stripped-sabot configuration: An Aluminum sabot carries the projectile

through the barrel. Prior to impact, the sabot is stripped by means of a maraging

steel stripper. After penetration, the projectile erodes and its velocity reduces. The

residual velocity is measured by means of residual velocity sensors. The

projectile is recovered from paper stacks, which act as momentum dump and the

residual mass is measured. Ballistic performance is evaluated by comparing the

kinetic energy of the residual rods. See Fig. 1.

The unstripped-sabot configuration: Using a sabot-stripper creates sabot debris

during the stripping process. This debris interferes with high-speed photography.

Hence, tests were also conducted without the stripper and the residual velocity

sensors. This configuration provides immaculate imagery of the initial stages of

the impact phenomenon and helps study ejecta characteristics. However, the time

window of data acquisition is limited to until the sabot interferes with the

penetration process.


Figure 1. Stripped-sabot configuration for ballistic tests

High-speed Photography

The Hadland Imacon 200 high-speed image acquisition system was used to

study the ultra high-speed phenomenon of ballistic penetration. The camera can

be programmed to record a sequence of separate images at prescribed time

intervals. A sixteen-channel camera was used. Images were acquired from a point

of view normal to the path of the projectile.

Flash Radiography Procedures

During ceramic penetration, fine pulverized ceramic powder is ejected from

the front and rear surfaces of the tile. This obscures the view of projectile-target

interaction and the flow of eroded particles. An experimental set-up for flash

radiography provides dynamic, real time images of the projectile penetrating the

ceramic. Two 100 kV heads were used. Two configurations were used. In the

inclined X-ray configuration, as seen in Fig. 2, the X-ray heads are placed

inclined to the path of the projectile. This reduces the ceramic cross-section that is

pierced by the X-rays. This configuration helps study the interior of the target and

hence the target-projectile interaction during penetration. In the edge-on


configuration, the X-ray heads (see Fig. 2) are moved so that they are orthogonal

to the path of the projectile. Since, the target thickness is large this configuration

does not reveal the interior. It helps study the flow of rod erosion products

emerging from the front surface.

Inclined X-ray Edge-on X-ray

Figure 2. Flash radiography configurations

Target Material

Coors Al2O3 AD995 CAP3 armor grade tiles were used. These are 99.5%

purity tiles of 10.16 cm 10.16 cm 1.27 cm dimensions. The areal density of

the ceramic tiles is 4.98 gm/cm2.

2.5 Projectile Material

WHA (93% W, ~5% Ni, ~2% Fe) manufactured by Hogen Industries was

used. The projectiles were flat-ended cylinders of diameter 6.14 mm and length

20.86 mm. Also, WHA (93%W, ~5%Ni, ~2%Fe) procured from ARL was used

for flash radiography studies.

Membrane application techniques

Scotch 893 Glass Fiber tape: Commercially available Scotch 893 glass fiber tape

was used to hand-wrap the ceramic tiles. Scotch fiber tape has a tensile strength of

525 N/cm. It is 0.15 mm thick. Elongation is approximately 4.5%. Eight layers of

fiber tape were hand-wrapped on tiles and then the back-face tape was cut out so

that only the front-face and edges of the ceramic tiles were taped. The glass fibers

on the cellophane tape run uni-directionally. Hence, the orientation of the tape


was alternated after every two layers (02/902/02/902). Taping the ceramic tile

increases its areal density from 4.98 to 5.31 g/cm2.

Ti-3%Al/2.5%V sheets: Ti-3/2.5 alloy sheets of 0.127 mm, 0.254 mm and 0.508

mm were used. The sheets were bonded to the front-face of the ceramic tiles

using Loctite 312 super glue. The tensile strength of Ti-3/2.5 is approximately

620 MPa. Elongation is approximately 15%. The areal densities of ceramic tiles

with 0.127mm, 0.254 mm and 0.508 mm Ti-3/2.5 sheets are 5.036 gm/cm2, 5.093

gm/cm2 and 5.207 gm/cm

2 respectively.

E-glass/Epoxy pre-preg: E-glass/Epoxy pre-preg (BT-250E-1) manufactured by

Bryte technologies Inc. was used. The E-glass reinforcement has a cross weave

and the overall tensile strength is 434 MPa. The pre-preg was pressed onto the

front surface of the ceramic tiles and cured at 250o F (121

o C), in a hot press.

Samples with one and three layers of pre-preg were prepared. The areal densities

are 5.019 gm/cm2 and 5.099 gm/cm

2 respectively.

Carbon fiber/Epoxy pre-preg: Carbon-fiber/Epoxy pre-preg (BT-250E-1), also

manufactured by Bryte technologies was used. The Carbon (Graphite)

reinforcement also has a cross weave and the overall tensile strength is 669 MPa.

Samples were prepared using techniques similar to those used for E-glass/Epoxy

prepreg. Samples with one and three layers of pre-preg have areal densities of

5.017 gm/cm2 and 5.083 gm/cm

2 respectively.

EXPERIMENTAL RESULTS

WHA projectiles were used to impact Al2O3 tiles at 900 m/s. The velocity was

well above the ballistic limit (V50) of the Al2O3 tiles and was maintained the same

for all the tests. Bare tiles and tiles with front-face fiberglass tape, Ti-3/2.5,

Carbon fiber/Epoxy pre-preg, or E-glass/ Epoxy pre-preg membrane of various

thicknesses, were studied.

Ballistic performance

Tests conducted with the stripped sabot configuration help to understand the

effect of impact-face constraint on the ballistic performance. The projectiles

weighed about 10.6 gm. The initial velocity of the projectile, measured by the

velocity sensors was used to calculate the initial kinetic energy. The residual

velocity sensors measured the exit velocity of the projectiles, after penetration.

The eroded projectiles were recovered from the paper stacks and weighed. The

residual kinetic energy was calculated. The ballistic performance was evaluated

by determining the kinetic energy fraction, defined by fKE = residual kinetic

energy/ initial kinetic energy. Some of the results are shown in Fig. 3.


As can be seen, the fKE for bare tiles is approximately 0.35. From the test

results for the E-glass/Epoxy, Carbon-fiber/Epoxy, and Ti-3/2.5, it is observed

that fKE tends to diminish with increasing thickness of the membrane layer. The

fKE for a three layered E-glass/Epoxy sample is approximately 0.12, about the

third of that for the bare tiles. This is a nearly 23% improvement in the ballistic

efficiency for a mere 2.5% increase in the areal density. It is also observed that

glass-fiber tape improves the ballistic efficiency substantially. However, the areal

density is increased by nearly 7%, mainly as a result of the cellophane content. It

is expected that after a certain critical thickness for the front-face membrane,

there will be a gradual reduction in the resulting improvement due to the

constraint effect.

0.05

0.12

0.19

0.26

0.33

4.95 5 5.05 5.1 5.15 5.2 5.25 5.3 5.35

Areal density ( gm/cm^2 )

Kin

eti

c E

nerg

y F

racti

on

( Unconfined Alumina )

( 5 mil Ti )

(10 mil Ti)

(20 mil Ti)

(Glass Fiber Tape)

8 layers - 48 mil

E Glass/Epoxy prepreg

1 layer - 8 mil

E Glass/Epoxy prepreg

3 layers - 24 mil

Carbon fiber/Epoxy

1 layer

Carbon fiber/Epoxy

3 layers

Figure 3. The effect of front-face constraint on the ballistic performance of

allumina tiles

Table 1. lists the residual velocity and residual mass measurements of the

projectiles. As can be seen from the Ti-3/2.5, Carbon-fiber/Epoxy, and E-

glass/Epoxy tests, the residual velocity is decreased by more than 100 m/s for the

front-face-constraint samples. However, no strong correlation is yet observed

between the residual velocity and increasing thickness of the membrane. It is

observed that increasing the thickness of the membrane results in an increase in

erosion. Hence, preliminary observations suggest that increasing thickness and

hence the ensuing increase in tensile strength of the impact-face membrane,

increases the erosion of the projectile. Further tests are needed to isolate the


effects of key material properties such as tensile strength, stiffness, and elongation

of the front-face membrane on the residual velocity and erosion of the projectiles.

Constraning

membrane material

Initial

velocity (m/s)

Initial

Mass (gms)

Residual

velocity (m/s)

Residual

mass (gms)

Unconfined 903.9 10.708 682.0 6.421

Unconfined 900.7 10.658 671.0 6.489

Glass fiber tape 897.5 10.582 545.5 5.706

Glass fiber tape 900.7 10.582 563.7 4.955

Ti – 0.127 mm 887.5 10.668 624.4 6.309

Ti – 0.127 mm 912.0 10.662 584.7 5.008

Ti – 0.254 mm 900.7 10.570 636.7 3.567

Ti – 0.254 mm 894.4 10.582 561.7 4.429

Ti – 0.508 mm 891.2 10.634 616.0 2.328

Carbon – 1 lyr 894.4 10.647 632.5 6.791

Carbon – 1 lyr 905.5 10.671 633.6 5.846

Carbon – 3 lyrs 892.3 10.660 540.8 4.805

Carbon – 3 lyrs 891.2 10.663 538.0 4.339

E-glass – 1 lyr 892.7 10.622 593.4 5.721

E-glass – 1 lyr 900.7 10.610 527.5 4.813

E-glass – 3 lyrs 907.1 10.656 517.0 3.808

E-glass – 3 lyrs 864.1 10.615 532.1 4.253

Table 1. The effect of various constraining materials on the residual velocity and

the residual mass of the projectile

High speed photography results

Front face: Fig. 4 shows the initial stages of an unstripped-sabot test for a bare

sample. The ceramic ejecta can be seen ejecting from the front surface. Soon

after impact, a pulverized zone (the Mescall zone) is formed ahead of the

projectile due to intense stress conditions. The ejection process clears the

pulverized ceramic away to accommodate the penetration of the projectile. A

significant portion of the kinetic energy of the projectile is transferred to the

ejecta. As can be seen, the ejecta for a bare sample is radially disperse and conical

in shape. Fig. 5 shows the initial stages of an unstripped-sabot test for a

constrained sample. The flow of ejecta particles is much more acute and


cylindrical in nature. The displacement measurement tools of the Imacon 200

software were used to calculate the velocities of these ejecta. It was observed that

during the initial stages, the ejecta velocity for samples with front-face membrane

was nearly 40% higher than that of the corresponding bare samples. The higher

kinetic energy associated with ejecta signifies reduced residual kinetic energy for

the projectile.

(1 ) (5 )s s

(9 ) (13 )s s

Figure 4. Initial stages of impact of a bare tile


(0 ) (4 )s s

(8 ) (12 )s s

Figure 5. Initial stages of impact of E-glass/Epoxy constrained tile

(10 ) (15 )s s

(20 ) (25 )s s

Figure 6. Back face displacement of a bare tile


(10 ) (15 )s s

(20 ) (25 )s s

Figure 7. Back-face displacement of an E-glass/Epoxy constrained sample

Back face displacement: The projectile’s travel velocity and the rate of its

erosion govern its penetration rate. The back-face displacement gives a good

indication of the penetration rate. Figs. 6 and 7 show the back-face displacement

of a bare and a front-face constrained sample, respectively. It can be seen that the

back-face displacement is delayed by nearly 15 for the constrained sample.

This implies increased erosion and/or reduction in velocity.

s

Flash Radiography

Fig. 8 compares the X-ray images for bare and front-face-membrane

constrained samples. The edge-on X-rays indicate that the eroded projectile

particles for the constrained sample, exhibit a more oblique flow as compared to

that for the bare sample. The inclined X-rays provide a view of the interior during

penetration. The projectiles deform by mushrooming and shearing of its tip,

indicating ductile nature of its failure. The projectile for the constrained sample

exhibits a larger mushroom head. This confirms the increased penetration

resistance and erosion of the projectile. The more oblique flow of the eroded

projectile particles for a constrained sample, observed in the edge-on X-rays, is a

result of greater mushrooming.


(7 ) (15 ) (8 )s s s

Edge-on X-rays Inclined X-raysBare tile

(7 ) (15 ) (9 )s s s

Edge-on X-rays Inclined X-raysConstrained tile

Figure 8. Flash radiography results

CONCLUSIONS

Al2O3 tiles when impacted by WHA projectiles fail through a complex

combination of processes resulting from the shock-wave propagation and

reflection. These processes include fragmentation and formation of radial and

circumferential macro-cracks, pulverization of the ceramic into fine powder, and

ejection of the fine powder from front and rear surfaces. The WHA projectile

undergoes deformation and erosion. It is expected that the morphology of the

pulverized ceramic fragments and its flow characteristics govern the penetration

resistance of the ceramic tiles. Hence, it is important to understand the underlying

mechanisms producing the pulverization of ceramics. Preliminary numerical

simulations10

on DYNA2D (a two-dimensional hydrodynamic finite element

code) indicate that release waves emanating from the projectile edges reduce the

pressure and increase the shear stress at a distance equal to the projectile diameter,

ahead of the projectile. The resulting stress condition is highly conducive to the

pulverization of ceramic11, 12

, See Fig. 9.


Figure 9(a) Computational grid displaying the geometry

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.40

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

1.2

x (cm)

y (c

m)

Maximum Shear Stress (GPa)

15.8755

14.4

323

12.9

891

12.9891 11.545

8

10.1

026

8.65

94

7.216

2

5.772

9

4.3297

2.8865

1.4432

4.32

97

5.7729


0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.40

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

1.2

x (cm)

y (c

m)

Pressure (GPa)

30.39

01

22.31

14

14.232

68.

8468

6.15

393.

461 0.76811

0.76

811

3.461

6.15

39

0.76811

Figure 9 (b) and (c). Contours of maximum shear stress and constant ressure in

an Al2O3 tile 0.83 s after being impacted by a 6.35 mm WHA projectile 9

High-speed photographs indicate that the front face confinement of Al2O3 tiles

vastly alters the flow of the pulverized ceramic that is ejected out. The front-face

ejecta from a bare tile is radially disperse and conical. For a constrained tile the

ejecta flow is more acute and cylindrical. Also the ejecta velocity for a

constrained tile is nearly 40% higher. Flash radiography indicates that

constraining the ceramic tile results in a much greater mushrooming and erosion

of the projectile. The greater erosion and reduced velocity of the projectile are

also manifested in the form of a significant delay in the back-face displacement of

the ceramic tile. Experiments indicate that impact-face restraint by fiber

reinforced polymer results in a substantial improvement in the ballistic efficiency.

Thin layers of E-glass/Epoxy improve the ballistic efficiency by nearly 20% for

an increase in areal density of 2.5%. Further research is needed to study the effect

of front-face membrane of other materials, so as to isolate the key material

properties governing the improvement in ballistic efficiency.

ACKNOWLEDGEMENT

The reported work was supported by US Army Research Office under contract

No ARO DAAH04-96-1-0376, to University of California at San Diego


REFERENCES1D.R. Smith, D.C. Vier, W. Padilla, S. C. Nemat-Nasser and S. Schultz, “Loop

wire medium for investigating plasmons at microwave frequencies,” Appld. Phys.

letters 75[10], 1425-1427, (1999)2D.A. Shockey, A.H. Marchand, S.R. Skaggs, G.E. Cort, M.W. Burkett and R.

Parker, “Failure phenomenology of confined ceramic targets and impacting rods,”

Int. J. Impact Engng. 9[3], 263-275 (1990) 3R.L. Woodward, W.A. Gooch Jr, R.G. O’Donnell, W.J. Perciball , B.J.

Baxter and S.D. Pattie, “A study of fragmentation in the ballistic impact of

ceramics,” Int. J. Impact Engng. 15[5], 605-618 (1994)4C.E. Anderson, Jr and B.L. Morris, “The ballistic performance of confined

Al2O3 ceramic tiles,” Int. J. Impact Engng. 12[2], 167-187 (1992) 5D.R. Curran, L. Seaman, T. Cooper and D.A. Shockey, “Micromechanical

model for comminution and granular flow of brittle material under high strain rate

application to penetration of ceramic targets,” Int. J. Impact Engng. 13[1], 53-83,

(1993)6R. Cortes, C. Navarro, M.A. Martinez, J. Rodriguez and V. Sanchez-Galvez,

“Numerical modelling of normal impact on ceramic composite armors,” Int. J.

Impact Engng. 12[4], 639-651, (1992) 7 F. I. Grace, and N. L. Rupert, “Analysis of long rods impacting ceramic

targets at high velocity,” Int. J. Impact Engng. 20, 281-292 (1997),8 J.T. McGinn, R.W. Klopp and D.A Shockey, “Deformation and

comminution of Al2O3 in the Mescall zone of ceramic armor,” Ma.t Res. Soc.

Symp. Proc. 362, 61-66 (1995)9J.D. McGee, S. Nemat-Nasser, and J.B. Isaacs, “Ballistic performance of

ceramic tiles with thin membrane confinement,” submitted for publication10

S. Nemat-Nasser and J.Zhang, unpublished results 11

H. Horii and S. Nemat-Nasser, “Brittle failure in compression: Splitting,

faulting and brittle ductile transition,” Phil. Trans. Roy. Soc. Lond. 319[1549],

337-374 (1986) 12

S. Nemat-Nasser and H. Deng, “Strain-rate effect on brittle failure in

compression,” Acta Metall. Mater. 42[3], 1013-1024 (1994)


A NEW FAMILY OF REACTION BONDED CERAMICS FOR ARMOR

APPLICATIONS

M. K. Aghajanian, B. N. Morgan, J. R. Singh J. Mears and R. A. Wolffe

M Cubed Technologies, Inc. Simula Safety Systems, Inc.

1 Tralee Industrial Park 7822 South 46th

Street

Newark, DE 19711 Phoenix, AZ 85044

ABSTRACT

Reaction bonded SiC has existed for many years. It is produced by reactively

infiltrating a preform consisting of SiC and carbon with molten Si. During the

infiltration process, the Si and carbon react to form SiC. Thus, the finished body

consists of the original SiC, reaction-formed SiC, and residual Si. Historically,

reaction bonded SiC processes are designed such that high levels of reaction-

formed SiC are produced. With high levels of reaction formed SiC, the resultant

microstructure has a fully interconnected (coarse) SiC phase that provides good

performance in traditional ceramic applications (e.g., wear, corrosion, high

temperature, creep).

Within the past few years, new applications for ceramics have emerged in the

semiconductor industry (e.g., wafer chucks, wafer handling arms, process

chambers). These applications have different requirements than those for which

reaction bonded SiC was previously developed. For instance, the semiconductor

applications do not require wear or creep resistance, but do require excellent net

shape processing characteristics and a fine microstructure suited to machining of

minute details to high tolerance. To this end, a novel approach to reaction bonded

SiC was taken. Preforms that possessed a very high content of less than 50

micron SiC particles and a low carbon content were produced with a resin

molding process. Upon infiltration, little reaction occurred. This resulted in

minimal process shrinkage or distortion; and a microstructure with little

coarsening and low levels of residual stress. Such a material was found to be well

suited to near net shape production and machining to extremely precise tolerances

for semiconductor applications.

This novel reaction bonded SiC ceramic was evaluated for utility in armor

applications. The product was shown to possess good ballistic properties and the

ability to be produced to the desired tolerances without the need for machining.



Subsequently, a variation to the material was made. For applications requiring

greater hardness and lower density, a reaction bonded B4C ceramic was

formulated.

Herein, the processing, microstructure, properties, and ballistic performance

of these novel materials are presented and discussed.

INTRODUCTION

Reaction bonded SiC was first developed in the 1950’s 1,2,3

. Other terms for

the process include ‘reaction sintered’ and ‘self bonded’ 4. Conventionally, the

process consists of Si infiltration (liquid or vapor) into preforms of SiC + carbon.

During the infiltration step, the Si and carbon react to form SiC. Typically, all

carbon is consumed, yielding a product of porous SiC (vapor infiltration) or dense

Si/SiC (liquid infiltration). The maximum SiC particle size used in the production

of such bodies is generally in excess of a few hundred microns 1,2

.

A major advantage of the process is that the volume of the reaction-formed

SiC is 2.3 times larger than the volume of the reacted carbon. Thus, by

infiltrating Si into preforms that contain high carbon contents, ceramic bodies rich

in SiC can be produced.

Variations to the process have been studied. For example, Taylor and Palicka5

produced preforms of B4C and B4C + carbon and subsequently reactively

infiltrated the preforms with molten Si. The process conditions were selected to

encourage partial reaction between the Si and B4C, thus forming SiC (and

presumably SiBx). The resultant ceramic bodies contained B4C, SiC and Si. The

presence of B4C, which has a much lower density than SiC (2,540 vs. 3,210

kg/m3), yielded a ceramic body of low mass. To maximize the B4C content in the

components, a particle size distribution was utilized. A maximum B4C particle

size in the distribution of nominally 300 microns was chosen.

The reaction bonding process for SiC ceramics has several advantages relative

to traditional SiC processes (e.g., hot pressing, sintering). First and foremost,

volume change during processing is very low (generally well less than 1%),

which provides very good dimensional tolerance control. In addition, the process

requires relatively low process temperatures and no applied pressure, which

reduces capital and operating costs. Moreover, fine reactive powders capable of

being densified are not required, which reduces raw material cost. Finally, unlike

most monolithic SiC materials, Si/SiC is typically electrically conductive. This

allows EDM machining and assists in sensitive applications where static

discharge is required.

However, the vast majority of commercial reaction bonded SiC ceramics have

coarse microstructures. This is a due to the use of large SiC particles in the

preforms and the fact that many of these materials are made using high levels of

carbon in the preform. As the carbon reacts in an expansive manner with the Si to


form SiC, the SiC particles in the preform are networked together to form large

SiC clusters. Since the strength of a ceramic is controlled by the largest flaw

within the stressed volume, a coarse grained material will tend to have low

strength. Therefore, reaction bonded SiC ceramics are traditionally used for high

temperature, creep, corrosion and wear sensitive applications, but not structural

(strength critical) applications.

The present work expanded on the aforementioned prior art with the goal of

producing optimized reaction bonded ceramic materials for room temperature

structural applications, such as semiconductor capital equipment components and

armor tiles. In particular, the activities focused on the production of components

with relatively fine-grained microstructures. Two different material types were

studied, namely reaction bonded SiC and reaction bonded B4C.


All of the reaction bonded ceramics described herein were produced with

nominally the same process steps. First, a preform was fabricated by mixing

ceramic particles with a resin binder, and casting the mixture into a mold. Next,

the mixture was cured, demolded, and exposed to about 600°C in an inert

atmosphere to pyrolyze the binder. Finally, the resultant carbon-bound preform

was contacted with a molten Si-containing alloy in a vacuum atmosphere, thus

allowing reactive infiltration to occur. Less than 0.5% volume change occurred

during the infiltration process.

After the fabrication step, various mechanical and physical properties of the

materials were measured. Density was determined by the water immersion

technique in accordance with ASTM Standard B 311. Elastic properties were

measured by an ultrasonic pulse echo technique following ASTM Standard D

2845. Hardness was measured on the Vicker’s scale with a 2 kg load per ASTM

Standard E 92. Flexural strength in four-point bending was determined following

MIL-STD-1942A. Fracture toughness was measured using a four-point-bend-

chevron-notch technique and a screw-driven Sintech model CITS-2000 universal

testing machine under displacement control at a crosshead speed of 1mm/min.

Specimens measuring 6 x 4.8 x 50 mm were tested with the loading direction

parallel to the 6 mm dimension and with inner and outer loading spans of 20 and

40 mm, respectively. The chevron notch, which was cut with a 0.3 mm wide

diamond blade, has an included angle of 60° and was located at the midlength of

each specimen. The dimensions of the specimen were chosen to minimize

analytical differences between two calculation methods according to the analyses

of Munz et al.6.

Microstructure was characterized in two manners. Polished sections were

examined using a Nikon Microphot-FX optical microscope. Fracture surfaces

were studied with a Jeol 840 scanning electron microscope (SEM).


Advanced light armor designs typically consist of a ceramic tile to blunt or

break projectiles and a second layer (e.g., fiber-reinforced polymer composite) to

catch or stop the remains. For the present work, ballistic testing was conducted

using a simple configuration that simulated a typical light armor design.

Specifically, 100 mm x 100 mm ceramic tiles were bonded to 300 mm x 300 mm

fiber-reinforced polymer plates, and then were tested versus ballistic projectiles.

Ballistic resistance of the samples was determined by the procedures described in

MIL-STD-662F. Three materials were evaluated, namely reaction bonded SiC,

reaction bonded B4C, and commercial hot pressed B4C (control). In one series of

tests, the reaction bonded SiC and commercial hot pressed B4C were tested versus

ball rounds; and in a second set of tests the reaction bonded B4C and hot pressed

B4C were tested versus armor piercing (AP) rounds.


Fabrication of Reaction Bonded SiC

The reaction bonded SiC ceramic material was produced in three basic steps.

First, a preform of SiC particles and organic resin was fabricated. Second, the

resin was pyrolyzed (converted to carbon). Finally, the preform was reactively

infiltrated with molten Si at nominally 1600°C. The final product was 100%

dense and consisted of the original SiC, reaction formed SiC (Si + carbon), and

remaining Si.

The goal of the reaction bonded SiC process development activities was to

produce a relatively fine grained ceramic for structural applications. To achieve

such a microstructure, the work utilized preforms with relatively small SiC

particles and low carbon content. The small particles led to a fine structure and

the low carbon content resulted in minimal reaction-formed SiC that would

cluster the small particles together. Specifically, a SiC particle size blend was

used to maximize particle packing. A maximum particle size of nominally 45

microns was used in the blend. The preforms produced with the blend contained

75 vol. % SiC and 4 vol. % carbon (pyrolyzed binder). After infiltration with

molten Si, the resultant bodies consisted of 84 vol. % SiC (75 original and 9

reaction formed) and 16 vol. % Si (i.e., an Si/SiC composite). A typical

microstructure (optical photomicrograph) of the material is shown in Figure 1.

In the optical photomicrograph, it is not possible to differentiate between the

original SiC and the reaction formed SiC. Nonetheless, it is clearly evident that

by the use of the relatively low carbon content little growth and interlocking of

the SiC particles has occurred, thus allowing a relatively fine microstructure to be

maintained.


Fabrication of Reaction Bonded B4C

The reaction bonded B4C was produced in a nearly identical manner to the

reaction bonded SiC. A B4C particle blend with a maximum particle size of

nominally 45 microns was prepared. Preforms were then made using this blend.

The preforms consisted of nominally 75 vol. % B4C and 4 vol. % carbon. After

infiltration, the ceramic material contained nominally 75 vol. % B4C, 9 vol. %

reaction-formed SiC, and 16 vol. % remaining Si (i.e., an Si/SiC/B4C composite).

An optical photomicrograph of the material is shown in Figure 2.

Figure 1. Optical Photomicrograph of Reaction Bonded SiC

Figure 2. Optical Photomicrograph of Reaction Bonded B4C

As with the reaction bonded SiC, the reaction bonded B4C ceramic shown in

Figure 2 displays little interlocking and clustering of the particles. Also, the

photomicrograph shows little visible reaction between the Si and B4C as a result


of the infiltration process. This was achieved by using process conditions

specifically designed to minimize reaction, including low process temperature,

short process time, and B-doping of the Si infiltrant. If Si-B4C reaction is allowed

to occur, as was the case in some previous work5, the microstructure significantly

coarsens. A coarse microstructure leads to a ceramic with a larger flaw size, and

thus lower strength. In Figure 3, a typical microstructure is shown were Si-B4C

reaction has occurred. Coarsening of the structure (i.e., large ceramic clusters

within the Si matrix) is clearly evident.

Figure 3. Optical Photomicrograph of Reaction Bonded B4C

with Unwanted Si-B4C Reaction

Mechanical and Physical Properties

Results of density, Young’s modulus, flexural strength and fracture toughness

are provided in Table I. When appropriate, the results are provided as a mean +/-

one standard deviation.

Table I. Properties of Reaction Bonded Ceramics

Property Reaction

Bonded SiC

Reaction

Bonded B4C

Density (kg/m3) 3060 2570

Young’s Modulus (GPa) 384 +/- 2 382 +/- 6

Flexural Strength (MPa) 284 +/- 14 278 +/- 14

Fracture Toughness (MPa-m1/2

) 3.9 +/- 0.5 5.0 +/- 0.4

The density of the SiC-based material is about 6% lower than monolithic SiC

due to the presence of the Si phase, which has relatively low density. This

reduced density is important for applications, such as armor, that are weight


specific. The B4C-based material has very low density and is similar to that of

monolithic B4C.

The Young’s moduli of the reaction bonded SiC and reaction bonded B4C

ceramics are essentially the same, and compare favorably with other high

performance ceramic materials. The specific results are as predicted based on the

handbook Young’s modulus values for dense SiC, B4C and Si of ~450, ~450 and

120 GPa, respectively7. In particular, on a weight specific basis, the reaction

bonded B4C has a very high Young’s modulus.

The fracture toughness of the reaction bonded SiC of nominally 4 MPa-m1/2

,

is consistent with most SiC-based ceramics7. Surprisingly, the reaction bonded

B4C ceramic shows a 28% increase in toughness relative to the SiC material,

despite the fact that no ductile phase was added. A possible explanation for this

increased toughness was found by examining fracture surfaces, as is explained in

the next section.

Hardness is a very important parameter for armor materials. Previous work

has demonstrated that high mass efficiencies are only obtained versus hard armor

piercing projectiles when the projectiles are fractured, and that to effectively

fracture the projectile, an armor must have high hardness8,9

. However, it is

difficult to compare the many hardness data in the open literature because results

can be highly dependent on test method and technique. Therefore, for the present

work many different commercial materials were obtained. Hardness

measurements were then made on both the commercial materials and the new

reaction bonded ceramics in an identical manner so that true comparisons could

be made. The results are provided in Table II.

Table II. Results of Hardness Measurements

Material Vicker’s Hardness with

2 kg Load (kg/mm2)

7.62 mm M2 AP Bullet (Tool Steel) 926 +/- 26

14.5 mm BS-41 Bullet (WC/Co) 1644 +/- 30

Sintered AlN 1044 +/- 63

Pure Si 1243 +/- 21

90% Sintered Al2O3 1250 +/- 89

Hot Pressed AlN 1262 +/- 51

99.5% Sintered Al2O3 1499 +/- 74

Hot Pressed Al2O3 2057 +/- 82

Hot Pressed TiB2 2412 +/- 135

Hot Pressed TiC 2474 +/- 188

Hot Pressed SiC 2640 +/- 182

Hot Pressed B4C 3375 +/- 212


Figure 4. SEM Fractographs of Reaction Bonded SiC (A)

and Reaction Bonded B4C (B)

Reaction Bonded SiC 2228 +/- 274

Reaction Bonded B4C 2807 +/- 54

The reaction bonded SiC and B4C ceramics have very high hardnesses that are

well in excess of both tool steel and WC/Co projectiles. In both cases, the Si/SiC

and Si/SiC/B4C composites have hardnesses that more-or-less reflect the weighted

average hardness of the constituents. In particular, because of the very high

hardness of monolithic B4C, the reaction bonded B4C has a very high hardness

value.

Analysis of Fracture Surfaces

The relatively high fracture toughness of the reaction bonded B4C ceramic

was unexpected. To gain an understanding for this result, the fracture surfaces of

the reaction bonded SiC and reaction bonded B4C ceramics were studied and

compared. The SEM fractographs for the two materials are provided in Figure 4.


A significant difference between the two fracture surfaces is seen. The

reaction bonded SiC ceramic shows brittle, transgranular fracture of the SiC

particles. Also, brittle fracture of the Si matrix is seen. In addition some

indications of interfacial cracking between the Si and SiC are seen. The reaction

bonded B4C ceramic shows brittle, transgranular fracture of the B4C particles.

However, the Si matrix shows some highly unexpected ductile behavior with the

characteristic chisel-like rupture pattern. In addition, no evidence of failure at the

interfaces between the particles and matrix is seen. It is felt that the observed

semi-ductile failure of the Si phase is contributing to the relatively high toughness

of the reaction bonded B4C ceramic (Table I).

A review of the literature10-13

finds that Si undergoes a brittle to ductile

transition in the 500°C temperature range. The transition temperature decreases

as the dislocation density in the Si increases. In one study11

, more surface

dislocations were introduced to the surface of a sample by grinding, which

reduced the brittle to ductile transition temperature.

In the reaction bonded SiC system, little stress will be induced in the Si phase

on cooling from the processing temperature because both Si and SiC have CTEs

of nominally 4 ppm/K14

. Thus, the dislocation density in the Si should be low.

However, the situation is very different in the reaction bonded B4C ceramic.

Upon cooling from the process temperature, the B4C and Si will shrink at

different rates (B4C has a CTE of about 5.6 ppm/K14

). Thus, the Si will become

highly stressed and thus will have a high dislocation density. It is postulated that

this high dislocation density leads to the semi-ductile behavior of the Si in the

reaction bonded B4C ceramic at room temperature. More study of this

phenomenon is needed.

Ballistic Properties

The results of ballistic testing are provided in Tables III and IV. In Table III,

test results versus a 7.62 mm M80 ball round for reaction bonded SiC and

commercial hot pressed B4C (control) are provided. In Table IV, test results

versus a 7.62 mm AP M2 round for reaction bonded B4C and commercial hot

pressed B4C are provided. In each case, the tables provide the areal density of the

system, the mass efficiency of the target, and the normalized mass efficiency

relative to the hot pressed B4C control. The mass efficiencies in the tables were

determined based on available data for rolled homogeneous steel armor (RHA)

versus the same threats. Specifically the mass efficiency was calculated as the

areal density of RHA required to give the same performance divided by the areal

density of the tested targets.


The ballistic results are very encouraging. They show that the armor designs

employing lower cost reaction bonded ceramics had mass efficiencies equivalent

to armors of the same design using hot pressed ceramics. This has enabled the

production of cost effective armor products for various applications, as is

discussed in the next section.

Table III. Ballistic Test Results versus 7.62 M80 Ball Threat

Armor System

Areal Density

kg/m2 (psf)

Mass Efficiency

(RHA Equivalent)

Normalized Mass

Efficiency

Hot Pressed B4C

(control)

23.5 (4.82) 4.56 1.00

Reaction Bonded

SiC

23.9 (4.89) 5.11 1.12

Table IV. Ballistic Test Results versus 7.62 AP M2 Threat

Armor System

Areal Density

kg/m2 (psf)

Mass Efficiency

(RHA Equivalent)

Normalized Mass

Efficiency

Hot Pressed B4C

(control)

29.0 (5.95) 4.53 1.00

Reaction Bonded

B4C

30.2 (6.18) 4.85 1.07

Examples of Products

Numerous armor products have been fabricated and tested using the novel

reaction bonded ceramic materials. Examples are provided in Figure 5. Key

process elements are that large components can be fabricated (no pressure

required), high tolerances can be obtained without machining (< 0.5% process

shrinkage), and costs are relatively low (no fine reactive powders, relatively low

fabrication temperatures). In Figure 5, the aircraft armor and personnel armor

tiles are fabricated from reaction bonded SiC. The vehicle armor plate is

fabricated from reaction bonded B4C.

Presently, reaction bonded SiC personnel armor plates are being manufactured

in very high volumes for the US Marine / US Army Interceptor program. In

addition, various aircraft and vehicle armor components are being produced in

lower volumes for both standard (reaction bonded SiC) and AP (reaction bonded

B4C) armor applications.


Figure 5. Example Armor Products - Aircraft Armor Tiles (top), Vehicle Armor

Plate (middle), Personnel Armor Tiles (bottom).


SUMMARY

Two new reaction bonded ceramics were developed, one based on SiC and

one based on B4C. In both cases, the process conditions were selected to yield a

fine-grained structure relative to traditional liquid-Si infiltrated reaction bonded

ceramics. Both materials show excellent mechanical properties and high

hardness. In particular, the reaction bonded B4C was found to have an

unexpectedly high fracture toughness. A proposed mechanism for the high

toughness was presented based on fracture surface analysis and previous

observations in the literature. Finally, the ballistic performance of the new

ceramics was measured. Relative to the incumbent hot pressed B4C, the reaction

bonded SiC showed good performance versus a 7.62 mm ball round and the

reaction bonded B4C showed good performance versus a 7.62 mm AP round.

REFERENCES 1 K.M. Taylor, “Cold Molded Dense Silicon Carbide Articles and Methods of

Making the Same,” U.S. Pat. No. 3 205 043, Sept. 7, 1965. 2 P.P. Popper, “Production of Dense Bodies of Silicon Carbide,” U.S. Pat. No.

3 275 722, Sept. 27, 1966. 3 C.W. Forrest, “Manufacture of Dense Bodies of Silicon Carbide,” U.S.

Patent No. 3 495 939, Feb. 17, 1970. 4 R. Morrell, Handbook of Properties of Technical and Engineering Ceramics,

HMSO Publications, London, U.K., 1985. 5 K.M. Taylor and R.J. Palicka, “Dense Carbide Composite for Armor and

Abrasives,” U.S. Pat. No. 3 765 300, Oct. 16, 1973. 6 D.G. Munz, J.L. Shannon, and R.T. Bubsey, “Fracture Toughness

Calculation from Maximum Load in Four Point Bend Tests of Chevron Notch

Specimens,” Int. J. Fracture, 16 R137-41 (1980). 7

Engineered Materials Handbook, Vol. 4, Ceramics and Glasses, ASM

International, Metals Park, Ohio, 1991. 8 M.L. Wilkins, R.L. Landingham, and C.A. Honodel, “Fifth Progress Report

of Light Armor Program,” Report No. UCRL-50980, University of CA,

Livermore, Jan. 1971. 9 C. Hsieh, “Ceramic-Faced Aluminum Armor Panel Development Studies,”

Appendix 9 of Report No. JPL-D-2092, Jet Propulsion Laboratory, Feb. 1985. 10

J. Samueles, S.G. Roberts, and P.B. Hirsch, “The Brittle-to-Ductile

Transition in Silicon,” Materials Science and Engineering, A105/106 39-46

(1988).11

P.D. Warren, “The Brittle-Ductile Transition in Silicon: The Influence of

Pre-Existing Dislocation Arrangements,” Scripta Met., 23 637-42 (1989). 12

K. Sumino, “Dislocations and Mechanical Properties of Silicon,” Materials

Science and Engineering, B4 335-41 (1989).


13 P. Haasen, “Brittle-to-Ductile Transition in Silicon as a Model for

Intermetallics,” Materials Science and Engineering, A137 105-10 (1991). 14

Y.S. Touloukian [ed.], Thermophysical Properties of Matter, Plenum Press,

New York, 1970.


FLEXIBLE CERAMIC COATED FIBER FABRICS FOR LIGHT WEIGHT

PROTECTION SYSTEMS

Konstantin von Niessen and Rainer Gadow

University of Stuttgart

Institute for Manufacturing Technologies of Ceramic Components

and Composites (IMTCCC/IFKB)

Allmandring 7b

D-70569 Stuttgart, GERMANY

ABSTRACT

Based on thermal spray technologies a coating process for refractory oxide ce-

ramic layers even on temperature sensitive fiber substrates has been developed, so

that the coated fabrics retain their flexibility. High speed and high rate ceramic

coating is performed with simultaneous cooling so that refractory oxide ceramic

coatings can be applied on aramide and mullite fibers with potential for industrial

application. The penetration by bullets, knives and blades through such ceramic

coated multilayer fabrics is effectively prevented.

INTRODUCTION

For personnel protection as well as protection of aircrafts and cars, only light

and flexible materials can be used.1 Light and flexible fabrics made of aramide or

other high tenacity fibers meet some of these demands but their protection is not

sufficient. Sharp blades as well as high speed bullets can pierce these fabrics even

if several layers are used. This paper focuses on a new approach by coating fabrics

made of high tenacity fibers such as aramide and mullite fibers with a highly re-

fractory oxide ceramic by thermal spray technologies. By the combination of high

tenacity fiber woven fabrics and high performance ceramic coatings the penetra-

tion by bullets, knives and blades can be effectively prevented. The ceramic coat-

ing increases the fiber to fiber friction which prevents wave distortion and de-

lamination. The penetrating objects cannot change the fabric structure and push

the fibers aside. The hard oxide ceramic coating blunts sharp metal blades by

abrasion so they cannot trench the fabric, and the high friction between the ce-

ramic coating and the metal blade stops further penetration.



MATERIAL SCREENING

The material screening focusses on the use of high tenacity fiber fabrics and

highly refractory oxide ceramics. Two different commercially available fiber fab-

rics have been selected, the standard aramide fabric used for ballistic protection

Twaron CT 710 (Twaron Products, Wuppertal, Germany) and the mullite fiber

fabric Nextel 720 (3M, Minneapolis, MN, USA) consisting of 85% Al2O3 and

15% SiO2. The material properties of these fibers are summarized in table I.

Table I. Properties of fiber fabrics2

Fiber

fabricDensity

[g/cm3]

Tenacity

[MPa]

Initial

modulus E

[GPa]

De-

comp.

temp.

TD [°C]

Specific

heat CP

[J/kgK]

Max. appl.

Tem. TM

[°C]

Twaron

CT 710

1.45 2,800 85 500 1420 200

Nextel

720

3.40 2,100 260 2,000 800 1,204

Due to their high hardness and wear resistance the oxide ceramics Al2O3 and

TiO2 have been chosen as coating materials for thermal spraying. To improve the

bonding strength of the ceramic coatings on the fabric, AlSi is used as additional

bond coat. The bulk material properties of the ceramic materials are shown in ta-

ble II.

Table II. Bulk material properties of Al2O3 and TiO23

Oxide

ceramic

Density

[g/cm3]

Vickers

hardness

HV [-]

Youngs

modulus E

[GPa]

Melting

temp. TM

[°C]

Specific heat

CP [J/kgK]

Al2O3 3.98 2,200 400 2047 1,047

TiO2 4.25 1,150 205 1,860 730

In order to apply these oxide ceramics by thermal spraying, they have to be

available as spray powders. After a sintering process the used powders are me-

chanically broken and milled to a grain size of 10 – 22 µm.


DEPOSITION OF OXIDE CERAMIC COATINGS ON LIGHTWEIGHT FIBER

FABRICS BY THERMAL SPRAYING

The thermal spray process allows the application of a broad variety of metal-

lurgical, cermetic and ceramic coatings on a variety of substrates. The Atmos-

pheric Plasma Spray (APS) process uses an electric arc discharge between a water

cooled copper anode and a tungsten cathode as an energy source. This electric arc

discharge dissociates and ionizes the working gas and builds up a plasma that ex-

pands into the atmosphere forming a plasma gas jet (see Fig.1).4

cooling

powder injection

plasma

cathode

(tungsten)

anode

plasmagas

��

coating

substrate

courtesy Linde AG

energy source: el. arc / plasma

plasma temp.: up to 20.000 K

plasma gas: argon, helium, hydrogen, nitrogen

spray material: oxide ceramics, metals, alloys,

polymers

raw material form: powder

particle velocity: up to 450 m/s

deposition rate: 4 - 8 kg/h (oxide ceramics)

Fig. 1 The Atmospheric Plasma Spray (APS) process5

The spray powder, suspended in a carrier gas, is injected into the heat source of

the torch. After being totally or partially molten and being accelerated, the powder

particles impact on the substrate`s surface, where they are quenched and solidified

within 10-5

to 10-7

seconds. During atmospheric plasma spraying process tempera-

tures up to 20,000 °C are obtained.

For the application of thermally sprayed coatings on fiber woven fabrics the

torch movement is performed by a 6-axis robot system and a metal frame is used

for inserting and tentering the samples. The meandering movement and the metal

frame are shown in Fig. 2.

Two piece

metal frame

Screw joint

Wire cloth

Metal frame to

support and

stabilize

the fabric

APS plasmatorch

X

Y

Coating track configuration

Fig. 2 Mounting support for the fabrics and coating track configuration

In order to limit the thermal load on the fabrics a simultaneous cooling with

compressed air is used. Air nozzles are attached on both sides of the spraying


torch. In addition, the process is supervised by an infrared camera (Varioscan In-

fraTec ID, Dresden, Germany) and in that way the temperature of the coated sam-

ples can be controlled in real time. Fig. 3 shows a typical IR- picture during the

coating process.

Fig. 3 IR- picture of the temp. distribution during the coating process

MECHANICAL CHARACTERIZATION

With regard to the use of the coatings for ballistic protection, the main focus of

the characterization is on the determination of puncture resistance, hardness and

wear resistance as well as on the evaluation of the coating`s bonding strength on

the first fiber layers. During the coating buildup of thermally sprayed layers, po-

rosity and microcracks cannot be avoided. For the coating of flexible fabrics the

formation of porosity and microcracks in the coating is desired because it leads to

a higher flexibility of the fabric. But if the porosity is too high, the hardness and

other mechanical properties of thermally sprayed coatings decrease. So a balance

between porosity and mechanical properties has to be found.

The thickness of the oxide ceramic coatings on the fabric is in the range of 50 –

100 µm. Fig 4 shows a schematic drawing of the intended structure of the coated

fabric.


50-100µm

Fiber woven

fabric

Fig. 4 Intended structure of the oxide ceramic coated fabric

In Fig. 5 a cross section of a Twaron fabric coated with an Al2O3 oxide ce-

ramic layer is shown. The lamellar structure and the good wetting behavior of the

ceramic coating on the first layers of the fabric are visible. The macro-structure

and micro-structure of the coated fabric`s surface is typical for thermally sprayed

coatings (see Fig. 6). The structure of the fabric is still visible in the macro-

structure. Even though the TiO2- and Al2O3- coatings have melting points above

1800° and 2000°C respectively, there is no significant polymer fiber damage.

Fig. 5 Cross section of a thermally sprayed Al2O3 coating on a Twaron fabric

Al2O3- Coating

Twaron Fabric


Fig. 6 SEM micrographs of a thermally sprayed Al2O3 coating on a Twaron

fabric

Macro structure Micro structure

In order to evaluate the coating quality metallographic examinations have been

performed. The coating porosity determined by an image analysis is expressed by

the relative pore volume content VP [%]. An automized universal hardness in-

denter equipment (Fischerscope TM HCU) with a load of 500 mN is used to de-

termine the coating hardness HV0,05. In order to measure the hardness of an indi-

vidual fiber, the load was reduced to 10 mN (HV 0,001). Table III and table IV

show the measured porosity and hardness characteristics of the thermally sprayed

coatings and of the fibers, respectively.

Table III. Measured coating porosity and hardness (HV 0,05)

Coating VP [%] HV 0,05

Al2O3 5.8 1,240 +/- 300

TiO2 3.2 1,100 +/- 110

Al2O3/TiO2 4.1 1,025 +/- 180

AlSi 1.44 138 +/- 10

Table IV Microhardness of individual fibers (HV 0,001)

Fiber HV 0,001

Twaron CT 710 51.52 +/- 7

Nextel 720 1,610 +/- 405

The investigation of the coating`s adhesion on the fabric is performed on a

Zwick Z100 universal mechanical testing machine by pull testing. The coated fab-

ric samples are glued to a metal plate and a steel tension rod is glued to the coat-

ing surface by using an adhesive. After mounting the samples into the testing ma-

chine the tensile load is continuously increased. As soon as a delamination of the

coating occurs, the tensile load is measured and the bonding strength is deter-


mined. As the bonding strength of the coatings is limited by the maximum shear

strength of the first fiber layers which are in contact with the coating, the fabrics

are also tested without any coating. In this case the tensile rod is glued directly on

top of the fabric. Fig. 7 shows the measured bonding strengths for the used fabrics

with or without AlSi bond coat.

without coatin

Al 2O 3

TiO

2

Al 2O

3/T

iO2

AlS

i

Al 2O

3 - A

lSi

TiO

2 - A

lSi

Al 2O

3/T

iO2 - A

lSi

Nextel 720

Twaron

0

1

2

3

4

5

6

7

Fig. 7 Bonding strengths of the thermally sprayed coatings on fiber fabrics

Bo

nd

ing

str

en

gth

[N

/mm

2]

The results of the experiments with non–coated fabrics show the maximum

possible bonding strength a coating could reach on the fabrics. Because of its low

shear strength, Nextel already reaches its limit at a bonding strength of 3 N/mm2.

The TiO2 coatings reach this values with and without a AlSi bond coat. The bond-

ing strength of the Al2O3 and Al2O3/TiO2 coatings is rather low, however it can be

increased by using the additional AlSi bond coat. For the TiO2 coated Nextel

fabrics delamination occurs within the fabric itself, whereas the other coatings

with lower bonding strength delaminate at the fiber–coating interface. Due to a

higher shear strength the maximum bonding strength of Twaron is about 7

N/mm2. None of the oxide ceramic coatings reach this limit, but by the use of a

bond coat, the bonding strength on Twaron is increased. Especially the Al2O3-

AlSi coating shows a high bonding strength and the highest microhardness. All

coatings deposited on the Twaron fabrics delaminated at the fiber–coating inter-

face. The differences in the mechanical properties of the coated fabrics are obvi-


ous. This might be due to the differences in the process temperatures since the

fabrics have different thermophysical properties, which influence the wetting and

bonding behavior of the applied coatings.

Because of the good results obtained with the Al2O3-AlSi coating on Twaron

fabrics, comparative stab resistance tests on these coatings and uncoated Twaron

fabrics are performed. German standard engineered test blades K1 (A. Eickhorn

GmbH, Solingen, Germany) for stab resistance tests are mounted into a Zwick

Z100 universal mechanical testing machine. The fabrics are fixed in a specific de-

vice by hydraulic pressure to obtain a defined prestress. The puncture resistance

performance is measured in work [N mm] per penetration depth [mm]. In one

experimental run, 6 stabs are carried out on different samples of the same fabric,

using one test blade to evaluate the blunting of the blades. The test velocity of the

blade is varied from 50 to 1500 mm/min, but no influence on the results was ob-

served. The typical run of the curves show an increase of the puncture resistance

for every new stab. This increase, which is caused by the blunting of the blade, is

for the Al2O3-AlSi coated Twaron fabrics much higher than for the uncoated fab-

rics. The penetration work of the coated fabrics is 5 times higher in comparison to

the uncoated fabrics. Fig. 8 shows the measured puncture resistance of Al2O3-AlSi

coated Twaron fabrics and uncoated Twaron fabrics.

0 20 40 60

Fig. 8 Stab resistence of Al2O3-AlSi coated and uncoated Twaron fabric

80

0

500

1000

1500

2000

Penetration depht [mm]

Pen

etra

tion w

ork

[Nm

m]

Stab resistance in N*mm of Twaron

fabric coated with Al3O3-AlSi multi-

layer coating

Stab resistance in N*mm of uncoated

Twaron fabric

Twaron CT 710

1 layer

plain weave style

Increase of the penetration work for

every new stab, which is caused by

the blunting of the metal blade


CONCLUSIONS

The approach to combine highly refractory oxide ceramic coatings with high

modulus lightweight fiber fabrics has been successfully demonstrated. Atmos-

pheric plasma spraying with well defined parameter sets and simultaneous cooling

is a suitable process for the coating of oxide ceramics on top of fiber woven fab-

rics for ballistic protection. Even though the TiO2- and Al2O3- coatings have melt-

ing points above 1800° and. 2000°C respectively, no significant polymer fiber

damage has been seen. The adherent coatings remain flexible and reach a hardness

up to 1240 HV 0,05. The bonding strength is sufficient and mainly limited by the

maximum shear strength of the fibers. The adhesion of the coatings and the high

cycle flexibility can be improved by using metallurgical bond coats. So far the

best results have been reached with an Al2O3- coating on a Twaron fabric with a

AlSi bond coat. It has the highest microhardness and the highest bonding strength.

Stab resistance tests were carried out on Al2O3-AlSi multilayer coated Twaron

fabrics and the penetration work was increased by a factor of five compared to the

uncoated Twaron fabric. Further efforts will focus on the optimization of the in-

terface between oxide ceramic coating and fiber fabrics by tailoring the cooling

process during thermal spraying as well as by deposition of metallurgical thin

films as bond coats.

ACKNOWLEDGMENT

The authors would like to thank Mrs Katrin Keck (metallography) and Mr

Chuanfei Li (plasma spraying) for their support and Mr Scherer for the helpful

discussions.

REFERENCES 1 J-P. Charles, D. Guedra- Degeorges, “Impact Damage Tolerance of Helicop-

ter Sandwich Structures,” Aerospatiale, France (1999) 2“Product data sheet Twaron”, Twaro Products, D- 42097 Wuppertal, Kasi-

nostraße 19-21, (1995) 3 C. Friedrich, G. Berg, E. Broszeit, C. Berger: “Datensammlung zu Hartstof-

feigenschaften,” Materialwissenschaft und Werkstofftechnik, Vol. 28, No. 2,

(1997)4 L. Pawlowski, “The science and engineering of thermal spray coatings,” John

Wiley and sons, Chichester (1995)5“Das Verfahrensspektrum beim thermischen Spritzen,“ Linde AG, Werks-

gruppe technische Gase, Höffriegelskreuth (1990)


IMPROVED PERFORMANCE OF ALUMINA CERAMICS WITH CARBON

NANOTUBE REINFORCEMENT

Michael Sennett

U.S. Army SBCCOM

Natick Soldier Center

Natick, MA 01776-5020

Sekyung Chang, Robert H. Doremus, Richard W. Siegel, Pulickel M. Ajayan and

Linda S. Schadler

Materials Science and Engineering Department and

Rensselaer Nanotechnology Center

Rensselaer Polytechnic Institute

Troy, NY 12180-3590

ABSTRACT

Nanoscale alumina powder and carbon nanotubes were mixed and hot-pressed

to form dense ceramic-matrix composites. The strength and fracture toughness of

hot-pressed alpha-alumina was much greater than that of conventional grain size

polycrystalline alumina. The addition of carbon nanotubes to the alumina resulted in

composites with even greater strength and fracture toughness. Hot pressing in a

vacuum improved both of these properties over hot pressing in argon. These results

suggest that lightweight composites of high strength and fracture toughness can be

made from composites of nanophase alumina, or other ceramics, and carbon

nanotubes.

INTRODUCTION

Carbon nanotubes have high modulus and aspect ratio1,2

, and thus may be

excellent reinforcing fillers for ceramics. The mechanical properties of such

composites will depend strongly on the processing method and surface treatment of

the carbon nanotubes. Sintered alumina has high strength, hardness, and fracture

toughness. Improving these properties by incorporating carbon nanotubes in an

alumina-matrix composite is an exciting possibility as well as a processing challenge.

Here we report on the processing and mechanical properties of composites

made from nanoscale alumina particles to form the matrix and multi-wall carbon



nanotubes (MWNT) as the reinforcing material. We give special emphasis to

improved methods of dispersing the MWNT in the alumina powder before pressing

and sintering, and to purification and oxidation of the MWNT.

EXPERIMENTAL PROCEDURE

Gamma-phase alumina powder consisting of particles with a mean diameter

of 23 nm (Nanophase Technologies Corporation, Romeoville, IL) and MWNT,

synthesized by the arc-discharge method, were used to make the composites. The

gamma-phase alumina powder was transformed to alpha-alumina before sintering by

heating at 1300oC for 7 min. The mean particle size of the alpha powder determined

from X-ray line broadening was about 62 nm.

Alumina matrix composites with 5-20 vol.% MWNT were fabricated. The

MWNT were lightly oxidized by heating them at 640oC in air for various lengths of

time up to 150 min. This treatment removes some of the carbonaceous material and

makes it easier to disperse the nanotubes. The alpha-alumina powder and MWNT

were dispersed in dichloromethane (methylene chloride, CH2Cl2) with an ultrasonic

probe for about 4 min. The mixture of alumina and MWNT was held in the

ultrasonic bath until most of the CH2Cl2 evaporated, and then the mixture was dried

at 75oC for 24 hr. The weakly agglomerated mixture was ground and remixed in an

agate mortar and pestle and then dried at 130oC for 12 hr. Finally, the alumina-

MWNT mixtures were sintered by hot pressing in a graphite die at 1300oC, and a

pressure of 60 MPa, for 1 hr in an Ar atmosphere or in a vacuum hot press.

Alumina with un-oxidized MWNT composites, marked as “as received” were

prepared in the same way described in Reference 4

The density of the composites was measured by the Archimedes method. X-

ray analysis was performed on the composites to determine if the presence of the

MWNT causes the formation of any new phases.

To measure the hardness and fracture toughness, the surface of the

composites was polished with 1 µm diamond paste and then 0.3 µm alumina powder.

The hardness of the composites was measured with a micro-Vickers hardness

indenter (Model M-400, Leco Co.); a 1 kg load was applied on the surface for 10 sec.

To measure the fracture toughness of the composites, a Vickers hardness tester

(Vickers Limited) with a load of 5 kg was used, and the fracture toughness was

calculated from the lengths of cracks emanating from the indenter corners by the

“Evans & Charles” equation (Kc = 0.00824*(P/C1,5

), where P is equal to the applied

load in Newtons and C is equal to the crack length in meters.

The strength of the composite samples was measured with diametral tests of

sintered discs. In these tests a compressive load P is applied across the diameter d of

a disc sample. The result is a line of tensile stress along the sample surface and

through its volume to the other surface5,6

. The maximum stress S, which occurs on


the diametral plane between loading points, is S = 2P/dL, in which d is the diameter

of the disc (16 mm in these tests) and L its thickness (4 mm). A pad of soft material

(copper) is inserted between the hard loading plates and the specimen.

Vardar and Finnie5 compared the strength of “granodiorite” (presumably

grandidierite, (Mg, Fe) Al3BSiO9) and limestone measured in bending (tensile) and

diametral tests over a wide range of strengths. They found quite similar Weibull

distributions and mean strengths for both of these minerals in the two tests.

Grandidierite has a hardness of 7.5 and limestone is of course soft, so these results

demonstrate the validity of the diametral test as compared to bending tests over a

wide range of hardness and strength.

Polished surfaces of composites and hardness indents were examined with

optical microscopy. Fracture surfaces were coated with gold and examined in a

scanning electron microscope (SEM, JEOL-A 40).

RESULTS

The samples described in this section were prepared by hot pressing in Ar at

1300oC and 60 MPa for 1 h unless otherwise noted. X-ray diffraction patterns

showed that the composites consisted of alpha-alumina and graphitic carbon only.

Broadening of the graphite diffraction lines showed that the average diameter of the

MWNT was about 12 nm (see also Ref. 4). The structures of the MWNT were the

same before and after processing. The density of the sintered composites was above

97% of theoretical density.

The diametral strengths of alumina-MWNT composites with different

MWNT content are shown in Fig. 1. For each MWNT content three as-received

samples were tested and the mean taken; one sample of each specimen with MWNT

dispersed in CH2Cl2 was fractured. The bulk alumina made from nanoparticles alone

had a strength of 654 MPa, which is greater than the typical strength of 200 to 350

MPa for sintered alumina7

and was even comparable with strengths reported for

single-crystal alumina (sapphire) of from 350 to 1000 MPa. The composites

containing as-received MWNT had somewhat lower strengths than that of the bulk

alumina. When carbon nanotubes dispersed in CH2Cl2 were added to alumina to

form composites, the strength first increased at 5 and 10 vol.% MWNT and then

decreased to the strength of composites with as-received MWNT at 20 vol.%

MWNT.

The fracture toughness of bulk alumina and alumina-MWNT composites is

shown in Fig. 2. The average fracture toughness of bulk alumina and alumina-5

vol.% MWNT composites, hot-pressed in a vacuum, increased to about 4.9 MPa m

and 5.1 MPa m, respectively. These toughness values are higher than those reported

for single-crystal alumina (sapphire) and polycrystalline alumina.8


0 5 10 15 20 250

200

400

600

800

0 5 10 15 20 250

200

400

600

800

Content of MWNT ( vol. % )

MWNT ( as received )

MWNT ( purified in CH2Cl

2 )

Str

en

gth

( M

N/m

2 )

Fig. 1. Diametral strengths of alumina-matrix composites hot-pressed in Ar at

1300oC and 60 MPa for 1 h containing different amounts of MWNT.

Fig. 2. Fracture toughness of sintered nanophase alumina and alumina-MWNT

composites: black triangles, hot-pressed in argon; open triangles, vacuum hot

pressed.

0 5 10 15 20 250

1

2

3

4

5

6

Content of MW NT( vol. % )

MW NT ( Purified in CH2Cl

2 )

KC (

MP

a.m

0.5 )

MW NT ( as received )


In previous work,4

the Vickers hardness of alumina composites containing as-

received MWNT decreased linearly from the bulk pure alumina value of 18.4 GPa to

13.5 GPa at 20 vol.% MWNT. In the present work, the hardness of an alumina -10

vol.% MWNT composite increased from 16.2 GPa with no oxidation of the MWNT

to a maximum hardness of 20.4 GPa after 90 min. of heating the MWNT in air at

640 C.

SEMs from the fracture surfaces of alumina-MWNT composites are shown in

Fig. 3. They show that the MWNT purified in dichloromethane are more evenly

dispersed than in composites made from as-received MWNT.

Fig. 3. Scanning electron micrographs of fracture surfaces of nanophase alumina-5

vol.% MWNT composites (a) and (b), as-received MWNT, (c) and (d) MWNT

purified in dichloromethane, all hot-pressed in argon.

(b)

(c) (d)

(a)

1 m3 m

3 m3 m

(b)

(c) (d)

(a)

1 m3 m

3 m3 m

DISCUSSION

The strength and fracture toughness of bulk alumina hot-pressed from


nanophase alumina powder in Ar was much higher than typical strength and

toughness of the conventional polycrystalline alumina. The addition of 5 vol.%

MWNT to nanophase alumina to form a composite increased both the diametral

strength and fracture toughness even more. Purification of the MWNT in

dichloromethane improved the dispersion of the MWNT in the final composites. This

purification step removes excess carbon from the MWNT samples, leaving purer

MWNT that disperse better in the dichloromethane solvent. Vacuum hot pressing

removes entrapped gases in the composite powder mixture, preventing formation of

residual stresses and reduction of strength. These processing improvements show the

great promise of nanophase alumina-MWNT composites for lightweight, high-

strength materials. Further improvements in processing should lead to composites

with substantial content of MWNT and consequent low density, and having high

strength and fracture toughness.

ACKNOWLEDGEMENTS

This work was supported by the U.S. Army SBCCOM, Natick Soldier Center. We

thank Nanophase Technologies Corporation for supplying the nanophase alumina.

REFERENCES 1S. Iijima, “Helical Microtubules of Graphitic Carbon,” Nature, 35 [7] Nov.

56-58 (1991). 2O. Lourie and H. D. Wagner, “Evaluation of Young’s Modulus of Carbon

Nanotubes by Micro-Raman Spectroscopy,” J. Mater. Res., 13[9] 2418-2422 (1988). 3P. M. Ajayan and T. W. Ebbesen, “Nanometre-size Tubes of Carbon,” Rep.

Prog. Phys., 60 1025-1062 (1999). 4S. Chang, R. H. Doremus, P. M. Ajayan and R. W. Siegel, “Processing and

Mechanical Properties of C-Nanotube Reinforced Alumina Composites, Ceramic

Engineering and Science Proceedings, 21[3] 653-658 (2000). 5O. Vardar and I. Finnie, “An Analysis of the Brazilian Disk Fracture Test

Using the Weibull Probabistic Treatment of Brittle Strength,” Int. J. Fracture, 11 [3]

495-508 (1975). 6M. B. Thomas, R. H. Doremus, M. Jarcho and R. L. Salsbury, “Dense

Hydroxylapatite: Fatigue and Fracture Strength after Various Treatments, from

Diametral Tests,” J. Mat. Sci., 15 891-896 (1980). 7W. D. Kingery, H. K. Bowen and D. R. Uhlmann, Introduction to Ceramics,

John Wiley and Co., New York, 1976, p. 791. 8Y.-M. Chiang, D. P. Birnie, and W. D. Kingery, Physical Ceramics, John

Wiley and Co., New York, 1997, p. 484.


RECENT PROGRESS ON THE INFLUENCE OF MICROSTRUCTURE AND

MECHANICAL PROPERTIES ON BALLISTIC PERFORMANCE

J.C. LaSalvia



ABSTRACT

Recent work on a terminal ballistic phenomenon known as dwell has led to

the identification of the important ceramic characteristics that govern this

phenomenon. In the ballistics community, dwell is used to describe the non-

penetration phase (complete or partial) of a long-rod projectile impacting on a

target. Because of the typical densities and velocities of long-rod projectiles,

dwell is typically observed in targets containing ceramics with high hardness

values. Recovery of ceramics from experiments in which complete dwell

occurred has led to the observation and basic understanding of the damage

mechanisms. Most notable is the importance of shear with respect to these

mechanisms. Consequently, a model for the transition from dwell-to-penetration

(a ballistic performance measure) was formulated by combining a

micromechanics-based compressive failure model with Hertz’s theory for

frictionless contact between axisymmetric linear-elastic bodies. The resulting

model indicates the relative importance of a ceramic’s grain size, short-crack

fracture toughness, yield strength, Poisson’s ratio, coefficient of friction, and

critical crack-length on the dwell/penetration transition. The motivation,

derivation, and predictions of the model are presented.

INTRODUCTION

Despite over 30 years of research and development of ceramic-based armor

technologies1,2

, a coherent and comprehensive understanding of the effect of a

ceramic’s physical and mechanical characteristics on its ballistic performance

does not exist. While there have been a few notable attempts at identifying and

correlating the important physical and mechanical attributes of a ceramic with

performance3,4

, these remain qualitative and do not allow performance to be

predicted. Another major problem has been the reliance of ballistic performance

measure methodologies (e.g. depth-of-penetration, V50) that do not adequately



(a) (b)

Figure 1. (a) Early confinement scheme used by Hauver et al.9. (b) X-ray flash

radiograph of a impacting long-rod dwelling on the surface of the ceramic

(normal impact).

distinguish the effect of a ceramic’s characteristics from the total system

performance nor yield sufficient insight into the fundamental processes that lead

to ceramic failure (i.e. penetration)5,6

.

Recently, tremendous insight into the fundamental projectile/target interaction

was gained through the work by Hauver et al.7-9

, Lundberg et al.10,11

, and others12-

16. Hauver et al.

7-9 discovered that through proper target design, the projectile

could be completely defeated without penetrating the ceramic. As a result of this

discovery, two new terms were added to the terminal ballistics vocabulary,

“dwell” and “interface-defeat”. Dwell is used to describe the state of the

projectile/target interaction event where the projectile does not penetrate the

ceramic and therefore has a zero penetration velocity. Interface defeat is used to

describe the condition when there is no significant penetration of the ceramic by

projectile during the entire ballistic event.

Lundberg et al.10,11

followed Hauver’s work with several fundamental studies

on not only the effect of projectile velocity on the penetration velocity through the

ceramic, but also on the projectile velocity where the onset of penetration

occurred (i.e. below this velocity, complete dwell occurred). This work coupled

with observations made on recovered ceramics from successful interface defeat

experiments and application of the results from a micromechanics-based

compressive failure model by Shih17

led LaSalvia et al.18

to develop a physically-

based theory that provided a rationale explanation for both the localized damage

and dwell/penetration transition velocity observations.


5 mm5 mm5 mm5 mm5 mm5 mm5 mm5 mm5 mm5 mm5 mm

Figure 2. Cross-section of a titanium diboride tile recovered following a

successful interface defeat experiment17

.

The purpose of this paper is to provide a brief review of not only the theory

developed by LaSalvia et al.18

, but also the foundational work that led to it’s

development. A figure-of-merit is derived and its implications for connecting

microstructure and mechanical properties with the dwell/penetration transition

velocity are also presented.

BACKGROUND

Dwell and Interface Defeat

Dwell and interface defeat was first reported by Hauver and Melani7 in small-

scale reverse-ballistic experiments with heavily confined ceramic targets.

Subsequently, Hauver et al.9 conducted larger scale forward-ballistic experiments

in which subscale long-rod projectiles (L/D = 10 and 20, D = 5 - 6 mm) were

launched into ceramic targets at velocities up to 2000 m/s. The ceramics were

nominally 75 mm in diameter and 25 mm thick. Figure 1(a) is a schematic

illustration of an early confinement scheme that was used. In addition to the

ceramic being heavily confined on all sides, a shock-wave attenuator and a

tailored ceramic/front steel cover plate interface were incorporated in these larger

scale experiments. An X-ray radiograph taken during a dwell experiment is

shown in Figure 1(b)9. As can be seen, the long-rod projectile is dwelling on the

ceramic. Using the confinement scheme shown in Figure 1(a), Hauver et al.9

were able to achieve interface defeat against long-rod projectiles impacting at

1600 m/s for silicon carbide, titanium diboride, titanium carbide, and tungsten

carbide.

An important aspect of the larger-scale experiments conducted by Hauver et

al.9 was that the ceramics could be recovered and examined after ceramographic

preparation. Figure 2 show a cross-section from a recovered titanium diboride tile

that was impacted at 1600 m/s. In addition to the numerous lateral and cone


cracks, a region of severe localized damage just beneath where the long-rod

impacted is clearly evident in the titanium diboride tile. As can be seen, this

localized damage region (often referred to as the “comminuted” region) does not

extend to the top surface, but is apparently fully confined by “undamaged”

material. The shape of the comminuted region corresponds well with calculated

and/or observed deviatoric stress distributions in quasi-static and dynamic contact

problems19-21

. This indicates the importance of shear with respect to the damage

mechanisms. With the exception of tungsten carbide, those ceramics that were

recovered after successful interface defeat experiments exhibited this localized

damage region.

Dwell/Penetration Transition Velocity

Although Hauver and Melani7 had discovered the dwell phenomenon and

interface defeat, much of our fundamental understanding is due to the

experimental work of Lundberg et al.10,11

. Using the reverse-ballistic testing

methodology, Lundberg et al.10,11

systematically investigated both the

dwell/penetration transition and the penetration rate for a number of ceramics.

Highly confined (see Figure 3(a)) boron carbide, silicon carbide, Syndie+, and

titanium diboride specimens were launched into either tungsten- and

molybdenum-based subscale long-rods (L/D = 40, D = 2mm) at velocities up to

2500 m/s. The dwell/penetration transition velocities for these ceramics against

tungsten-based rods are plotted in Figure 3(b) as a function of their estimated

compressive yield strengths ( Y). The yield strengths for the ceramics were

calculated from hardness measurements22-24

. As can be seen, the data for the

ceramics evaluated fall nominally on two curves; hereafter referred to as the upper

and lower bound curves. Silicon carbide and titanium diboride fall on the upper

bound curve, while boron carbide and Syndie fall on the lower bound curve.

Lundberg et al.11

offered the following explanations for these two curves. The

upper bound curve corresponds to the critical pressure required to form an indent

on a rigid, perfectly-plastic body using a rigid punch. The lower bound curve

corresponds to the critical pressure required to initiate yielding beneath the area

loaded. At first thought, both explanations appear to be highly questionable,

given the fact that ceramics do not typically have a sufficient number of

independent slip systems to support any appreciable amount of bulk plasticity in

response to loads. However, with sufficient confinement, bulk plasticity is

possible as is known in the compressive failure of geological materials25-30

. If a

ceramic is physically confined as shown in Figures 1(a) and 3(a), is ductile failure

of the ceramic possible? The data in Figure 3(b) supports the possibility for

ductile failure (i.e. the upper curve). However, the hypothesis of a completely

+De Beers, Inc. synthetic polycrystalline diamond.


��

��

��

��

��

��

��

Rear

Plug

Front

Plug

Locking

Rings

Tube

12

20

8

Ceramic

20

28

��

��

��

��

��

��

��

��

��

��

��

��

��

��

Rear

Plug

Front

Plug

Locking

Rings

Tube

12

20

8

Ceramic

20

28

(a) (b)

Figure 3. (a) Schematic illustration of the confined ceramic design used by

Lundberg et al.11

. (b) Dwell/penetration transition velocity data for several

ceramics plotted as a function of the ceramic’s yield strength11

.

ductile response (i.e. explanation for both the upper and lower curves) is not

consistent with the damage observations from ceramics recovered by Hauver et

al.9. In order to reconcile the data shown in Figure 3(b) with the observed damage

in the recovered ceramics, the compressive failure of brittle solids must first be

considered.

Compressive Failure of Brittle Solids

The general features of compressive failure of geological materials has been

the subject of a large number of investigations given its’ importance in the proper

laying of foundations for concrete structures25-30

. The effect of confining pressure

on the observed failure mechanism (e.g. axial splitting, faulting, and plastic flow)

was identified. This led to a number of compression failure models that are based

upon growth or suppression of microcracks with the wing-crack or z-crack flaw

geometry28-30

. Wing-cracks are mixed-mode cracks that extend out from the

plane of the pre-existing flaw, nominally in the direction of the maximum

principal stress. A schematic illustration of a wing-crack is shown in Figure 4(b).

In general, the initiation and initial growth of the wing-cracks are governed by the

local shear failure (mode II) of the pre-existing flaw and the ability of the

surrounding material to accommodate this failure by plastic deformation. As

shown in Figure 4, this shear failure can be accommodated either by plastic

deformation, wing-crack formation, or a combination of both. The ability to

accommodate the shear failure by microplasticity is an important material


2

1

2

1

2

1

2

1

2

1

2

1

2

1

2

1

2

1

2

1

2

1

(a) (b) (c)

Figure 4. (a) Pre-existing flaw subjected to localized normal and shear stresses

due to far-field principal stresses 1 and 2. (b) Wing-crack initiation and growth

due to shear failure of pre-existing flaw. (c) Accommodation of shear failure of

the pre-existing flaw by dislocation generation.

characteristic since it leads to a suppression of wing-crack formation, and hence a

decreased potential for brittle failure (macro).

According to the compressive failure model proposed by Horii and Nemat-

Nasser29

, the propensity of a material to suppress the initiation and growth of

wing-cracks is indicated, though not exclusively, by its ductility parameter. The

ductility parameter is defined as29

:

2

cK

Y

IC* (1)

where KIC is the mode I fracture toughness, Y is the uniaxial compressive yield

strength for ductile failure, and 2c is the pre-existing flaw size. The ductility

parameter is dimensionless and represents the ratio of a brittle failure strength

measure to a ductile failure strength measure. A low ductility parameter would

indicate a stronger tendency towards wing-crack initiation and growth, while a

high ductility parameter would indicate a stronger tendency towards wing-crack

suppression.

The compressive failure model that was developed by Horii and Nemat-

Nasser29

is shown in Figure 5. A pre-existing flaw of length 2c is subjected to

far-field principal stresses 1 and 2. The pre-existing flaw makes an angle as

measured from the maximum principal stress direction. A plastic zone of length

and wing-crack of length are possible as a result of the sliding motion

(shear) of the pre-existing flaw surfaces. In the model, the plastic zones are

represented as collinear arrays of edge dislocations, while the wing-cracks extend

out at an angle . A frictional force, due to the combined effects of the resolved

pλ tλ


2

1

(2c)Pre-Existing Flawpλ

tλ

2

1


tλ

2

1

(2c)Pre-Existing Flaw

2

1


2

1



tλ

2

1


tλ

2

1


2

1


2

1


(2c)Pre-Existing Flawpλpλ

tλtλ

2

1


2

1


2

1



tλtλ

2

1


2

1



2

1



2

1



tλtλtλ

Figure 5. Wing-crack with plastic relaxation model proposed by Horii and Nemat-

Nasser29

for compressive failure.

normal stress and sliding friction resists this sliding motion. The critical shear

stress for wing-crack initiation and growth (assuming = 45crito and = 45

o):

1212critt1212

12*

Y

crit

11c112

1

λ (2)

Equation (2) can be used to predict the location and severity of compressive

damage in a solid by simply comparing it with the measured or estimated

maximum shear stress. Because failure is pressure dependent, the confining stress

must also be measured or estimated. For the problem under consideration in this

paper, it is assumed that the shear stress and confining stress generated during

dwell are given by the results of Hertz’s theory for frictionless contact between

axisymmetric linear-elastic bodies.

DWELL/PENETRATION TRANSITION MODEL DEVELOPMENT

Classic Hertzian Stress Distribution

The principal stresses and maximum shear stress in the elastic solid along the

center-line axis of contact (z-axis) arising due to the frictionless contact between

axisymmetric bodies with similar elastic moduli are given by31

:

2o

H1

az1

1

p (3a)

2

1

o

H3

o

H2

az12

1

z

atan

a

z11

pp (3b)


0.5 1 1.5 2

0

0.5

1

1.5

2

2.5

z/a

Damage

* = 0.05

No Damage

No Damage

Hertz/

crit

* = 0.1

Figure 6. Plot of the center-line distribution and severity of damage for several

values of *.

2

1

o

H

az12

3

z

atan

a

z11

2

1

p (3c)

where po is the maximum interface normal stress, 2a is the contact diameter, is

Poisson’s ratio for the elastic solid, and superscript “H” signifies Hertz’s solution.

The general solution for stresses is reported by Lawn19

.

Predicted Spatial Distribution of Damage and Severity

Substitution of Equations 3(a) and 3(b) into Equation 2 yields an expression

for the critical shear stress as a function of normalized depth and material

parameters. Dividing Equation 3(c) by this expression yields Figure 6 where the

distribution of damage and its severity along the center-line axis is plotted for

several values of the ductility parameter* with po = Y, = 0.2, = 0, and

critt cλ = 0.1. Damage would only be expected whereH > crit. Thus, as can

be seen, damage would not be expected near the surface or at depths significantly

greater than the diameter of the contact. Near the surface, the stresses are high,

but the stress-state is more hydrostatic. It can also be seen from this figure, that

the expected severity of damage rises quite rapidly, reaching a maximum less than

one-half contact diameters below the surface. From this maximum, the severity

of damage gradually tapers off. It can also be noted that the distribution and

severity of damage is strongly effected by the ductility parameter. These


predictions are entirely consistent with the localized damage shown in Figure 2,

as well as with previous observations18

.

Reconsidering the dwell/penetration transition velocity data shown in Figure

3(b) in light of these observations suggests the following possible explanations for

the upper and lower bound curves. Assuming the absence of shock-induced

damage, for ceramics such as boron carbide or Syndie, the comminuted region

forms and extends to the top surface. As a result, the damaged material within the

comminuted region becomes unconfined and is therefore easily displaced,

allowing penetration to occur. In the case of silicon carbide or titanium diboride,

this comminuted region forms, but does not extend to the top surface (i.e. it is

confined by the surrounding undamaged material). For penetration to occur, the

“undamaged” layer of ceramic separating the rod from the comminuted region

must fail. Therefore, for the silicon carbides and titanium diboride shown in

Figure 3(b), penetration is governed by the ductile failure of this relatively

undamaged layer. Consequently, the critical pressure for the dwell/penetration

transition would correspond approximately with 2.85 Y, that required to fully

indent a rigid, perfectly-plastic solid.

Critical Impact Pressure for the Dwell/Penetration Transition

Consideration of Equations 2 and 3 with the conditions that H = crit and z/a =

0, the critical mean pressure pm required to expand the damaged region to the top

surface is given by (for a Hertzian pressure distribution, the maximum po is equal

to 3/2 times the mean pressure pm31

):

critt

*

Y

m

c212321

93.0

85.2

p

λ (4)

Equation 4 relates the critical impact pressure pm to material properties and

characteristics. As such, if these material properties and characteristics are

known, Equation 4 can be used as a figure-of-merit and also used to predict the

dwell/penetration transition velocity. According to Lundberg et al.11

, the critical

impact pressure pm and the projectile dwell/penetration transition velocity Vp are

related by the following expression:

p

pYm

p

pp

K

27.3p211

K2V (5)

where Kp, p, and Yp are the bulk modulus, density, and yield strength of the

projectile, respectively. The predicted effect of the ductility parameter on the

dwell/penetration transition velocity is shown in Figure 7. The values assumed


500

1000

1500

2000

2500

3000

5 10 15 20 25 3

Vp

(m/s)

Y (GPa)

* = 0.05

* = 0.1

* = 0.15

pm

= 2.85Y

0

Figure 7. Predicted effect of *

on the dwell/penetration transition velocity based

upon Equations 4 and 5.

for the ceramic were = 0.2, = 0, and critt cλ = 0.1. The values for the long-

rod projectile were p = 17.6 x 103 kg/m

3, Kp = 285 GPa, and Yp = 1.2 GPa

11.

According to this plot, a ceramic that possessed a dwell/penetration transition

velocity given by point A could be improved if its ductility parameter is

increased. This could be done by increasing the fracture toughness KIC,

decreasing the governing pre-existing flaw size 2c, decreasing the yield strength

(or hardness), or a combination of all three of these parameters.

Considering the Equation 4 as a figure-of-merit and in terms of the

dwell/penetration transition velocity, the following conditions would be possible:

If 185.p Ym 2 , the dwell/penetration transition velocity would

be given by the upper-bound curve in Figure 7.

If 185.2p Ym , the dwell/penetration transition velocity would

be less than that given by the upper-bound curve in Figure 7.

If a ceramic’s dwell/penetration transition velocity was below that predicted by

the upper-bound curve shown in Figure 7, the predicted change in this velocity

Vp for a change in mean impact pressure pm required to expand the damaged

region to the top surface is given by:

Am

m

p

2App

2App

Am

Ap

p

p

p

K2V1

Vp

V

V (6a)


(6b) A

mi

mm ppp

criti

tiiii

*i

iY

im

c212321

93.0

85.2

p

λ (6c)

The maximum pm is given by:

(7) A

miYmaxm p85.2p

If the only difference between ceramic A and i is in their fracture toughness

values, then Equation 6(a) can be written as:

AIC

IC

p

2App

2App

Am

Ap

p

K

K

K2V1

Vp

V

V (8)

where KIC equals KiIC – K

AIC. Considering Equations 7 and 8, the maximum

change in fracture toughness is given by:

1p

85.2

K

K

Am

AY

AIC

maxIC (9)

For example, consider the boron carbide shown in Figure 3(b). The yield strength

is 15.8 GPa11

, while the mean impact pressure at 1450 m/s is approximately 23

GPa. Substitution into Equation 9 yields a maximum required change in fracture

toughness equal to 2.7 MPa*m1/2

assuming a base fracture toughness of 2.8

MPa*m1/2

(i.e. the KiIC = 5.5 MPa*m

1/2). The assumption that the only difference

between ceramics i and A are in their fracture toughness values was used in

deriving Equations 8 and 9. While it is acknowledged that actually producing a

ceramic with this quality would not necessarily be easy, the purpose was to

demonstrate the potential utility of the model for guiding ceramic developers.

SUMMARY

A hypothesis has been proposed to explain the upper bound and lower

bound(s) in the dwell/penetration transition velocity data of Lundberg et al.11

.

This hypothesis is based upon observations of the localized damage region in

ceramic tiles recovered from successful interface defeat experiments conducted

by Hauver et al.9. These observations led to the consideration of

micromechanical-descriptions for compressive failure of brittle solids. In

particular, the wing-crack model has been successfully used to explain the effect


of hydrostatic stress on the failure mode of geological materials. Combining the

wing-crack model developed by Horii and Nemat-Nasser29

with the stress

distribution developed by Hertz19,31

for the normal, frictionless contact between

axisymmetric linear-elastic bodies led to the development of a model which

captured the essential features of both the distribution and severity of damage

within the localized damage region. It also provides a physically-based rationale

for the dwell/penetration transition data of Lundberg et al.11

.

The model indicates that for ceramic armor applications where part of the

defeat mechanism is dwell, the first consideration for the ceramic should be its

“hardness”. The second consideration should be its ductility parameter, or in

particular, its grain size and fracture toughness. Assuming constant physical and

mechanical properties, with the exception of fracture toughness, an equation was

derived that relates the change in fracture toughness to the change in

dwell/penetration transition velocity. Using boron carbide as an example, the

utility of the model for ceramic developers was shown.

While it was not discussed, the fracture toughness value that one should

consider is for cracks whose length-scale is less than or equal to the grain size.

For cracks of this size, the effect of residual stress will also be important32

.

Residual stress was not accounted for in this model. Finally, it must be mentioned

that this model is applicable to ceramics that do not have “soft” grain-boundary

phases. The shear strength of the “soft” grain-boundary phase could provide an

even lower critical shear stress than that given by Equation 2.

ACKNOWLEDGEMENTS

The author would like to thank Mr. William Bruchey, Mr. William Gooch,

Mr. George Hauver, Mr. Edward Horwath, Dr. Michael Normandia, Mr. Edward

Rapacki, Dr. Bryn James, and Dr. Patrick Lundberg for sharing their knowledge

on the phenomenon of dwell and interface defeat. The author would also like to

thank Professor Marc Meyers for sharing his knowledge and insight into the

physical mechanisms that govern the dynamic behavior of ceramics.

REFERENCES1W.E. Snowden, “High Performance Ceramics for Armor Applications: A

Historical Perspective,” in Proceedings of the Symposium on Ceramic Armor

Materials by Design, PAC RIM 4, Wailea, Maui, HI, November 4-8, 2001.2B. Matchen, “Applications of Ceramics in Armor Products,” Key Eng. Mat.,

122-124 333-42 (1996). 3Z. Rozenberg and Y. Yeshurun, “The Relation Between Ballistic Efficiency

and Compressive Strength of Ceramic Tiles,” Int. J. Impact Eng., 7 [3] 357-62

(1988).


4J. Sternberg, “Material Properties Determining the Resistance of Ceramics to

High Velocity Penetration,” J. Appl. Phys., 65 [9] 3417-424 (1989). 5M.A. Adams, “Theory and Experimental Test Methods for Evaluating

Ceramic Armor Components,” in Proceedings of the Symposium on Ceramic

Armor Materials by Design, PAC RIM 4, Wailea, Maui, HI, November 4-8, 2001. 6M.J. Normandia and W.A. Gooch, “An Overview of Ballistic Testing

Methods of Ceramic Materials,” in Proceedings of the Symposium on Ceramic

Armor Materials by Design, PAC RIM 4, Wailea, Maui, HI, November 4-8, 2001. 7G.E. Hauver and A. Melani, “Behavior During Penetration of Long Rods

(U)”; pp. 149-160 in Proceedings of the Second BRL Topical Symposium:

Experimental Research and Modeling Support, Ballistic Research Laboratory,

Aberdeen Proving Ground, MD, May 24, 1988.8G.E. Hauver, P.H. Netherwood, R.F. Benck, and L.J. Kecskes, “Ballistic

Performance of Ceramic Targets”; pp. 23-34 in Proceedings of the 13th

Army

Symposium on Solid Mechanics, Plymouth, MA, August 17-19, 1993. 9G.E. Hauver, P.H. Netherwood, R.F. Benck, and E.J. Rapacki, “Interface

Defeat of Long-Rod Projectiles by Ceramic Armor,” ARL Technical Report, in

progress.10

P. Lundberg, L. Holmgren, and B. Janzon, Ballistics ’98, in Proceedings of

the Seventeenth International Symposium on Ballistics, Midrand, South Africa,

March 1998, 3 251- (1998).11

P. Lundberg, R. Renstrom, and B. Lundberg, “Impact of Metallic Projectiles

on Ceramic Targets: Transition Between Interface Defeat and Penetration,” Int. J.

Impact Eng., 24 259-75 (2000).12

J.E. Field, “High-Speed Photography,” Contemp. Phys., 24 [5] 439-59

(1983).13

D.A. Shockey, A.H. Marchand, S.R. Skagg, G.E. Cort, M.W. Burkett, and

R. Parker, “Failure Phenomenology of Confined Ceramic Targets and Impacting

Rods,” Int. J. Impact Eng., 9 [3] 263-75 (1990). 14

Y. Tanabe, T. Saitoh, O. Wada, H. Tamura, and A.B. Sawaoka, “An

Overview of Impact Damages in Ceramic Materials – For Impact Velocity Below

2 km/s,” Report of the Research Laboratory of Engineering Materials, Tokyo

Institute of Technology, 19, 1994. 15

E. Strassburger, H. Senf, C. Denoual, P. Riou, and C. Cottenot, J. Phys. IV

France, 7 [c3] 909-14 (1997). 16

D. Sherman, “Impact Failure Mechanisms in Alumina Tiles on Finite

Thickness Support and the Effect of Confinement,” Int. J. Impact Eng., 24 312-28

(2000).17

C.J. Shih, Dynamic Deformation of Silicon Carbide, Ph’d Dissertation,

UMI, Dissertation Information Service, 1998, 331 pp.


18J.C. LaSalvia, E.J. Horwath, E.J. Rapacki, C.J. Shih, and M.A. Meyers,

“Microstructural and Micromechanical Aspects of Ceramic/Long-Rod Projectile

Interactions: Dwell/Penetration Transitions”; pp.437-46 in Fundamental Issues

and Applications of Shock-Wave and High-Strain-Rate Phenomena, ed. K.P.

Staudhammer, L.E. Murr, and M.A. Meyers, Elsevier Science, New York, 2001. 19

B.R. Lawn, “Indentation of Ceramics with Spheres: A Century after Hertz,”

J. Am. Ceram. Soc., 81 [8] 1977-94 (1998). 20

D.A. Shockey, D.J. Rowcliffe, K.C. Dao, and L. Seaman, “Particle Imapct

Damage in Silicon Nitride,” J. Am. Ceram. Soc., 73 [6] 1613-19 (1990). 21

D.K. Kim, C-S. Lee, C.W. Kim, and S.N. Chang, “Indentation Damage

Behavior of Armor Ceramics,” in Proceedings of the Symposium on Ceramic

Armor Materials by Design, PAC RIM 4, Wailea, Maui, HI, November 4-8, 2001. 22

K. Zeng, E. Soderlund, A.E. Giannakopoulos, and D.J. Rowcliffe,

“Controlled Indentation: A General Approach to Determine Mechanical

Properties of Brittle Materials,” Acta Mat., 44 [3] 1127-41 (1996). 23

J. Alcala, A.E. Gainnakopoulos, and S. Suresh, “Continuous Measurements

of Load-Penetration Curves with Spherical Microindenters and the Estimation of

Mechanical Properties,” J. Mat. Res., 13 [5] 1390-1400 (1998). 24

Yu. V. Milman and S.I. Chugunova, “Mechanical Properties, Indentation

and Dynamic Yield Stress of Ceramic Targets,” Int. J. Impact Eng., 23 629-38

(1999).25

Fracture in Compression of Brittle Solids, National Materials Advisory

Board, Report 404, National Academy Press, August 1983, 70 pp. 26

B. Cotterell, “Brittle Fracture in Compression,” Int. J. Fracture, 8 [2] 195-

208 (1972). 27

E.Z. Lajtai, “Brittle Fracture in Compression,” Int. J. Fracture, 10 [4] 525-

36 (1974). 28

M.F. Ashby and S.D. Hallam, “The Failure of Brittle Solids Containing

Small Cracks Under Compressive Stress States,” Acta Mat., 34 [3] 497-510

(1986).29

H. Horii and S. Nemat-Nasser, “Brittle Failure in Compression: Splitting,

Faulting, and Brittle-Ductile Transition,” Phil. Trans. R. Soc. London A, 319 337-

374 (1986). 30

M.F. Ashby and C.G. Sammis, “The Damage Mechanics of Brittle Solids in

Compression,” PAGEOPH, 133 [3] 489-521 (1990). 31

K.L. Johnson, Contact Mechanics, Cambridge University Press, 1985,

452pp.32

S.J. Bennison and B.R. Lawn, “Role of Interfacial Grain-Bridging Sliding

Friction in the Crack Resistance and Strength Properties of Nontransforming

Ceramics,” Acta Mat., 37 7659-71 (1989).


Transparent Armor

TRANSPARENT ARMOR MATERIALS: NEEDS AND REQUIREMENTS

Parimal J. Patel and Gary A. Gilde

Army Research Laboratory,

Weapons and Materials Research Directorate

Attn: AMSRL-WM-MC

Aberdeen Proving Ground, MD 21005

ABSTRACT

There has been interest in improving transparent armor for use in Army

vehicles. Future combat and non-combat environments will require lightweight,

threat adjustable, multifunctional, and affordable armor. Significant

improvements can be achieved through insertion of new materials. However, an

emphasis must be placed on user needs and requirements in addition to

improvements in ballistic performance. Current glass/polycarbonate technologies

are not expected to meet the increased requirements for transparent protection.

Results over the past few years indicate that the use of transparent crystalline

ceramics and advanced polymers greatly improve the performance of a system.

An overview of user requirements, applications, and current efforts in transparent

armor will be discussed.

Keywords: transparent, ceramic, armor, aluminum oxynitride, spinel,

sapphire, polycarbonate, polyurethane

INTRODUCTION

Transparent Armor Requirements

Transparent armor is a system constructed of different materials that are

designed to defeat a particular threat or range of threats. The threats targeted are

dependent on the envisioned combat or non-combat scenarios. There are also

threat requirements for "operations other than war" where ballistic protection is

required. Though a system is designed for a particular threat, there are general

requirements common to most transparent armor systems. The paramount

requirement for a transparent armor system is the defeat of a designated threat.

The system must also provide a multi-hit capability with minimized distortion of



surrounding areas of the first hit. They must be transparent in the wavelengths of

interest, ranging in the UV, visible, and infrared frequencies. Other requirements

for transparent armor windows are that they are night vision compatible, and they

are affordable based on cost-performance models.1

For future land and air platforms, weight is a critical parameter that must be

minimized. Space efficiency can also be quite important for certain applications.

The system must be large enough to be useful. For example, a 6 inch square

armor plate may be useful as a face shield but would not be very useful as a truck

window. The size must be large enough for the user to perform their duties

appropriately. Baseline transparent armor systems generally rely on plastics,

plastic-plastic laminates, and glass/plastic laminates. The systems work and offer

protection for the threats they are designed for. As the defined threat become

more lethal, these systems no longer perform adequately. A simple solution that

increases the ballistic performance of a window is increasing the thickness of the

window. The material and design costs are thus, increased incrementally. For

many applications, very thick armor systems are not practical solutions, even if

they defeat the threat. Thick windows may be impractical for a few reasons. One

reason is due to the increased weight associated with thicker materials. Another

reason is the space limitations in many vehicles. Finally, thick sections of

transparent armor have greater optical distortion than thinner sections, reducing

the transparency. Therefore, new materials that are thinner, lightweight, and offer

better ballistic performance are sought. Affordability is a critical metric for

evaluating all armor systems and can be the limiting factor for given applications.

There are many methods to measure the ballistic performance of a system.

Several experimental techniques have been developed to aid in comparative

studies of armor systems. One of these tests, a V50 test2, was used to measure

ballistic performance for the systems mentioned in this paper.

A transparent armor system is comprised of many layers, joined by polymeric

interlayers. The front face is usually a hard face material that is designed to break

up or deform the projectile upon impact. The sequential plys are added to provide

additional resistance to penetration. These materials can be the same as or

different than the front ply material. An interlayer to join the two plates separates

the plys and provides a transition between two materials that may have thermal

expansion mismatches such as a glass and a polymer. The purpose of this

interlayer is to mitigate the stresses from thermal expansion mismatches, as well

as to stop crack propagation from ceramic to polymer during processing.

The armor system can be engineered to provide different levels of protection.

In addition to defeating the threat with multi-hit capability, the mass and space

efficiency should be optimized for a given application. The variables that can be

changed are plate material, thickness of plys, interlayer hardness, interlayer

thickness, number of plys and the order of constituent materials.


MATERIALS USED FOR TRANSPARENT ARMOR

Polymeric Materials

The most common plastic used for transparent armor applications is

polycarbonate. Polycarbonate offers excellent ballistic protection against small

fragments. Polycarbonate is an inexpensive material that is easily formed or

molded. Polycarbonate is used in applications such as the sun, wind, and dust

(SWD) goggles, spectacles, visors, face shields and laser protection goggles.

Polycarbonate is also used as a backing material for advanced threats.

Polycarbonate is more effective in the thin dimensions required for individual

protection than in the thicker sections required for vehicle protection. Though the

material is adequate for many applications, the search for lighter weight materials

has led to investigations into other polymeric materials such as transparent nylons,

polyurethane, and acrylics.3,4

The limiting factor for use of other transparent

polymeric materials is their durability and their optical properties. Improvement

in these properties would warrant an investigation into the ballistic properties of

the material.

There have been efforts to improve the properties of polyurethane. Simula

Technologies Inc. has recently introduced a new family of polyurethanes with

improved optical properties. These materials are marketed and sold by Simula

Polymer Systems Inc. Sim 2003 and Sim 1802 are both thermoset plastics that are

produced by casting or liquid injection molding. Sim 1802 is harder and more

brittle than Sim 2003. Due to their physical properties, Sim 2003 is a viable

candidate to replace polycarbonate as a riot visor or as a backing material. Sim

1802 is a better candidate for front or hard-face material. These improvements in

polyurethane have led to an investigation into these materials for face-shield

applications as will be discussed in the “Applications” section.

Glasses and Glass-Ceramics

There are several glasses that are utilized in transparent armor. Normal plate

glass (soda-lime-silica) is the most common glass used due to its low cost, but

greater requirements for optical properties and ballistic performance generate the

need for new materials. There are many different glasses including borosilicate

glasses and fused silica that can be used. Glasses can be strengthened using

chemical or thermal treatments. Controlled crystallization of certain glass systems

can also produce transparent glass-ceramics. TransArm, a lithium disilicate based

glass-ceramic is produced by Alstom++

for use in transparent armor systems.5

Simula Technologies, 10016 South 51st Street, Phoenix, AZ, 85044 ++

Alstom UK Ltd., Research & Technology Centre Stafford, Staffordshire, ST174LN, England.


Glasses and glass-ceramics have the overall advantage of having lower cost than

most other ceramics materials, and the ability to be produced in curved shapes and

large sheets.

Transparent Crystalline Ceramics

For advanced threats, transparent crystalline ceramics are used to defeat the

projectiles. However, there are not many candidate ceramic materials, however,

that are transparent. The three major candidates are aluminum oxynitride (AlON),

magnesium aluminate spinel (spinel), and single crystal aluminum oxide

(sapphire). There are advantages and disadvantages to each material.

Aluminum Oxynitride Spinel (Al23O27N5): One of the leading candidates for

transparent armor is aluminum oxynitride (AlON). It is produced by Raytheon

Corporation+

and marketed under the trade name Raytran. The incorporation of

nitrogen into an aluminum oxide stabilizes a spinel phase. Due to its cubic crystal

structure, AlON is an isotropic material that can be produced transparent as a

polycrystalline material. A polycrystalline material can also be produced in

complex geometries using conventional ceramic forming techniques such as

pressing and slip casting. The green body is processed to transparency and

polished. Some properties of AlON are listed on Table I. The limitations of AlON

are its high cost and the sizes that are currently available. Raytheon is currently

investigating the scale-up and costreduction of aluminum oxynitride.

Raytheon Corp has produced an 11in. x 11in. curved AlON window (Figure

1A). The Air Force Research Laboratory (AFRL) is currently funding Raytheon

to investigate cost reduction of AlON to produce larger windows. This will allow

Raytheon to scale-up AlON so that it can be produced in large sizes at reasonable

costs. Additionally, funding is sought to address the equipment issues to produce

very large size plates.

Concurrently, the U. S. Army Research Laboratory (USARL) is investigating

transient liquid phase sintering of aluminum oxynitride to reduce processing

costs.6

A reaction sintering technique with the aid of a reactive liquid is the focus

of the research. Small samples (Figure 1B) with a transmission of 85% and a haze

of 14% have been produced. The reduction of the haze and size scale-up are the

immediate objectives of the program. ARL also has a Small Business Innovative

Research (SBIR) program for processes that can produce affordable aluminum

oxynitride powders using scalable methods to reduce the cost of the raw

materials.

+ Raytheon Electronic Systems, Lexington Laboratory, 131 Spring Street, Lexington, MA 02421


Figure 1: Photographs of aluminum oxynitride produced by Raytheon and ARL

Magnesium Aluminate Spinel (MgAl2O4): Spinel is a transparent ceramic that

has a cubic crystal structure and can be transparent in its polycrystalline form.

Spinel produced by sinter/HIP, hot pressing, and hot-press/Hot Isostatic Pressing

(HIP) has yielded transparent materials.7,8

The use of a hot isostatic press has

been shown to improve the optical and physical properties of spinel.8

Table I

shows the properties of spinel. Spinel offers some processing advantages over

AlON. Spinel powder is available from commercial powder manufacturers while

AlON powders are proprietary to Raytheon. Spinel is also processed at much

lower temperatures that AlON. The optical properties are better than AlON, with

its IR cut-off at 6 microns compared to 5.5 microns for AlON and 6 microns for

sapphire, respectively.9

Though spinel shows promise for many applications, it is

not available in bulk form from any manufacturer, but there are efforts to

commercialize spinel.

Table I: Selected mechanical properties of AlON and spinel

AlON Spinel

Density g/cm3

3.67 3.58

Elastic Modulus GPa 315 277

Mean Flexure Strength MPa 228 241

Weibull Modulus 8.7 19.5

Fracture Toughness MPa m 2.4±0.11 1.72±0.06

Knoop Hardness (HK2) GPa 13.8±0.3 12.1±0.2

Ceramic Composite Inc.+

is currently investigating hot pressing of magnesium

aluminate spinel under a Phase II SBIR sponsored by the Army Research

Laboratory. Previous investigations have studied sinter-hot isostatic pressing

(HIP) techniques. Hot pressing was chosen for this program as the processing


technique based on comparative analysis of the several processing techniques for

producing spinel.8

The research has focused on hot pressing with an additive and

hot-press/hot isostatic pressing (HIP). Hot pressing has been shown to be a

successful technique to produce transparent parts. Figure 2 is a four-inch

diameter, 0.44-inch thick spinel plate that has been produced using this technique.

The plate has an 83 percent transmission with 9.32 percent haze. Scale-up to

twenty inch parts is underway using the hot-press technique. Subsequent HIPing

has been shown to improve the optical properties and mechanical properties of

spinel.7

Hipping is generally not cost-effective and its use should be minimized.

However, the improvement in the mechanical and optical properties may deem

HIPing necessary for given applications.

Figure 2: A hot pressed 4-inch diameter, 0.44" thick spinel plate produced at ARL

Single Crystal Aluminum Oxide (Sapphire - Al2O3): Polycrystalline aluminum

oxide is an armor ceramic material that is used in opaque armor systems.

Aluminum oxide is transparent when produced in single crystal form. The

material is grown using single crystal growth techniques such as HEM10

by

Crystal Systems Inc.+

or edge-defined film-fed growth (EFG)11

by Saphikon.++

The crystal structure of sapphire is rhombohedral and its properties are

anisotropic and vary with crystallographic orientation. Sapphire is currently the

most mature transparent ceramic and is available from several manufacturers. The

cost is high due to the processing temperature involved and machining costs to cut

parts out of single crystal boules. Sapphire is a very high strength material, but the

strength is very dependent on the surface finish.12

There are current programs to

scale-up sapphire grown by the HEM and EFG processes.

+ Crystal Systems, Inc., 27 Congress St., Salem, MA 01970

++ Saphikon, 33 Powers St., Milford, NH 03055


Another manufacturer of sapphire is Saphikon, Inc., which produces

transparent sapphire using an edge, defined growth technique. The process size

limitation is currently at 0.25 in. thick, in 10 in. x 10 in. sheets. The Army

Research Laboratory is currently investigating use of this material for transparent

armor systems using synergistic approaches in laminate design and construction.

The current objective is to determine a baseline of glass/plastic and

ceramic/plastic against the specified threat. Once the baseline is completed,

sapphire will be tested in different constructions and compared to the baseline.

Manufacturing: Scale-up to larger size poses several problems. The large

sizes generally cost more to produce due to the difficulty in scale-up. Also, larges

plates are more difficult to polish than smaller plates. Materials Systems Inc.+

is

investigating bonding sapphire plates using proprietary glass and glass-ceramic

bonding materials.13

To date, bonds have been produced that are 70 percent of the

strength of unbonded material.13

This innovative technique offers the ability to

make very large windows that may not be achievable in monolithic parts due to

lack of capital equipment. A 12.4" x 18.9" window bonded is shown in Figure 3.

The limitation of this process is the presence of bond lines that are presently

visible. There are efforts to remove or minimize these visual effects.

Figure 3: Ground sapphire bonded together by MSI to form a 12.4" by 18.9"

window

Machining and Polishing: Regardless of the ceramic material utilized,

machining and polishing costs are significant. The high hardness of AlON, spinel,

and sapphire require diamond grinding and polishing media. The finishing

+ Materials Systems Inc., 521 Great Road, Littleton, MA, 01460


process times are also quite long. Finishing costs can be as much as 50 percent of

the final cost of the materials. These costs are greater for curved windows.

There are some programs to reduce the costs of machining and polishing. The

Center for Optics Manufacturing+

is investigating advanced grinding and

polishing techniques for optics. Their processing has been shown to remove

AlON, sapphire, and SiC at removal rates of 3 um/min, 1.5 um/min, and

0.5um/min, respectively.14

Figure 4: Low cost alternative to polishing developed by MSI. The left

photograph is with no window while the photo on the right is the view looking

through coated ground sapphire plate.

The Army Research Laboratory is also is looking for low cost solutions to

polishing. The USARL has recently sponsored a Phase II SBIR to address low

cost alternatives for polishing. Materials Science Inc. of Littleton, MA, is

investigating various treatments on ground sapphire to make it transparent without

polishing. The initial results have been very successful, as can be seen in Figure 4.

The technique eliminates the final polishing step thus saving significant amounts

of time and cost in producing the transparent ceramic.

APPLICATIONS AND REQUIREMENTS

Common military applications for transparent armor are ground vehicle

protection, air vehicle protection, personnel protection, and equipment (sensor)

protection. There are also commercial applications such as riot gear, face shield,

security glass, armored cars and armored vehicles.

Personnel Protection

There are several applications of advanced transparent armor systems for

personnel protection. Personnel protection utilizes transparent armor against small

arms threats and fragments, such as high velocity, rocks and bottles. Goggles are

+ Center for Optics Manufacturing, 240 East River Road, Rochester, New York, 14623


required for protections against the sun, wind and dust and in some cases, lasers.

Increased use of military forces for "operations other than war" highlights the

need to protect forces involved in these peacekeeping missions. For these

operations, protective equipment such as riot gear is needed. Laser threats are also

significant, and protective materials and coatings are sought for these

applications. Once again, improved ballistic protection and lighter weight are the

major objectives and cost is a significant factor.

Face shields: Personnel protection for facial protection is one Army

application that requires transparent armor. The Army Research Laboratory has

completed a program to improve the current visor design.15

The two end items

identified for improvement were the riot visor and an explosive ordnance (EOD)

visor. The goal for the riot visor was to improve the ballistic performance by 30

percent without increasing the weight of the system. The overall goal for the EOD

visor was to reduce the weight of the visor by 30 percent while providing equal

protection.

Riot Visor: The riot visor is made from injection-molded polycarbonate that

has an areal density of 1.55 lb/ft2. The visor is designed to protect against large,

low-velocity projectiles such as rocks and bottles, as well as, from small, high

velocity fragments. Since the goal of this program was to improve the ballistic

performance without increasing the weight, an all-polymer solution was sought.

Previous investigations16,17

in the 1970's had shown the promise of polyurethane

as an armor material, but the optical properties were not sufficient for a

transparent armor material. Improvements in the optical properties of the

polyurethane by Simula warranted a ballistic evaluation.

Ballistic testing was conducted for the riot visor against a 0.22 cal fragment

simulating projectile (FSP).1

A helium gas gun was used for velocities below 2000

ft/sec and a 22 inch-long, 0.223 barrel with a 1:12 twist was used for velocities

above 2000 ft/sec. The results of the testing showed that the polyurethane (SIM

2003) behaves better than either polycarbonate or acrylic (PMMA). Overall, the

polyurethane performed 30-35 percent better than polycarbonate on an equal

weight basis. The conclusion was that with the improved optical properties of the

SIM 2003, this material would be an excellent replacement for polycarbonate to

reduce the weight of the system.

Explosive Ordnance Visor (EOD:) The other objective for the ARL program was

to reduce the weight of EOD visors. The goal is to reduce the areal density of the

current system using different materials and constructions. Several constructions

were investigated, including plastic/plastic laminates, glass/plastic laminates, and


glass-ceramic/plastic laminates.15

The plastic hard-face did not deform the FSP,

while the glass and glass-ceramics were able to deform the FSP. Many of the

constructions were better in weight than the current system weight of 4.27 lb/ft2.

The use of a polyurethane (Sim 2003) increased the performance of the system.

The optimum constructions used fused silica, Vycor, or TransArm, a transparent

glass-ceramic produced by GEC Alsthom.

The ballistic data obtained in this investigation can be used for comparative

purposes in designing a visor for use against the FSP threat for the range of areal

densities tested. Other considerations are cost, availability, and manufacturability,

for which there are trade-offs. For example, in visor applications, TransArm,

Vycor, and fused silica performed well. TransArm is currently more expensive

than fused silica. However, TransArm can be easily produced in curved shapes.

Currently, it is difficult to obtain fused silica in a curved shape of a visor. Thus,

while fused silica would be a lower cost solution that performs better (optically

and ballistically) it may not be used for visor applications until the manufacturing

problem of producing fused silica in curved shapes is overcome.

Ground Vehicles

Ground vehicle protection is required for equipment that is used on the

battlefield, such as HMMWVs, tanks, trucks, and resupply vehicles. Transparent

armor is necessary for the windshield and side windows. There are several general

requirements for these applications.18

One critical requirement is the ability to

withstand multiple hits since most threat weapons are automatic or

semiautomatic. The windows must also be full size so that the vehicle can be

operated in the manner in which it was designed. A small window on a truck can

increase ballistic survivability but can reduce operational safety if the driver does

not have an appropriate field of view. The windows also need to be durable and

withstand normal wear in non-combat situations and from user damage.

The fielded systems fulfill these requirements with varying degrees of

success. There are some requirements that future transparent armor systems need

to address.18

There is an overall requirement for future Army systems to be

lighter. The weight of a transparent armor system is a parasitic weight for a

vehicle. The added weight of a transparent armor appliqué can be significant,

often requiring a beefed up suspension and drive train to maintain the vehicles

performance capability. These upgrades also add weight to the system. Any

weight savings improves the ability to bring the vehicle into theater. Reduction in

weight increases the payload capacity for tactical vehicles and thus increases

operational capabilities. Thinner armor systems are also required for similar

reasons. Thinner windows can increase the cabin volume. Future systems also

need to be compatible with night vision goggle equipment while offering laser

protection.


Due to their size and shape, windows are constructed of glass and plastic. The

major drive for new windows for these applications is lower weight and improved

ballistic protection. Due to the number of these vehicles in service, the sizes of the

windshields needed and the costs, improved glasses, glass ceramics and polymers

are the materials of choice for these applications. Glass compositional variations,

chemical strengthening, or controlled crystallization can improve the ballistic

properties. Glasses can also be produced in large sizes and curved shapes. Most

importantly, glasses can be produced to provide incremental ballistic performance

and incremental cost.

For advanced threats, the weight of glass/plastic becomes prohibitively heavy

and thick. The use of a transparent ceramic as a front-ply has been shown to

improve the ballistic performance and reduce the weight of the system. The use of

a ceramic front ply can reduce the areal density by as much as 65 percent. This is

a significant weight savings over the state-of-the art. The ballistic performance of

these transparent ceramics offers great potential for weight savings on future

vehicles. Currently there are some challenges that must be overcome for these

materials to be utilized. The major limitations are cost, sizes available, and

curvature of the plates. There are several programs addressing these issues at

USARL and elsewhere as was described in the "Materials" section.

Aircraft

Helicopters and other aircraft used in combat or in support roles require

transparent armor. Applications include windshields, blast shields, lookdown

windows, and sensor protection. The general requirements for these systems are

similar to those for ground vehicles, though the importance of the requirements

varies. Weight is a critical factor for these applications. The current transparent

armor weight is the limiting factor for increasing ballistic protection. Heavier

vehicles use more fuel, are more difficult to move into theater, and reduce

maneuverability. The shields for aircraft applications need to be full size and

curved.

Electromagnetic windows

Many of the ceramic materials that are of interest for transparent armor

solutions are also applicable to electromagnetic (EM) windows. However, there

are many EM window applications where visible transparency is not critical. EM

window applications include radomes, IR domes, sensor protection, laser

windows,19

and multi-spectral windows. The requirements for these windows vary

greatly. There are some required properties mutual to many of the applications.

The optical properties are extremely important for window applications. The

transmission window and related cut-offs (uv, IR) control the electromagnetic

regime where the window is operational. Other properties of interest are abrasion


resistance, strength, and the thermal properties. The thermal stability of the

materials properties are also critical if the material will be heated as in the case of

missile windows during flight.

Commercial Applications

Many of these systems utilized for military applications would also have use

in commercial systems such as law enforcement protection visors, riot gear, and

windows in commercial car, trucks, and busses, as well as architectural

requirements in certain buildings. The desire for armored automobiles for

personal use is also growing.

CONCLUSIONS

There is a general push to reduce the weight of military systems. Increased

weight reduces maneuverability, transportability, and increases operation costs.

One approach to reduce weight is to reduce the weight of armor systems. In

addition to reduction of weight, new systems are required to defeat more

advanced threats and to perform in combat and non-combat scenarios.

The history of transparent armor has shown significant advances as new

materials are introduced into the marketplace. The current thrust into lighter

systems is also based on advances in materials technology. Advances in

polymeric materials utilized for transparent armor systems have led to a renewed

interest in these materials to reduce the overall weight of armor systems.

Polyurethane has been shown to improve the performance as compared to

polycarbonate backing. Transparent ceramics have been shown to offer significant

ballistic protection with reduced weights over conventional glass/plastic systems.

Advances in the processes of these ceramics and scale-up have lead to increased

interest in using these materials for transparent armor applications.

There are several programs that are investigating the cost reduction and scale-

up of these materials. Successful outcomes from these programs should initiate

their use for armor applications and fulfill the requirements to reduce weight on

Army systems.

REFERENCES

1. P.J. Patel.; G.A.Gilde.; P.G. Dehmer, J.W. McCauley; "Transparent ceramics

for armor and EM window applications," PROC. SPIE Vol. 4102, p1-14,

Inorganic Optical Materials II, Alexander J. Marker; Eugene G. Arthurs; Eds.,

10/2000.

2. U.S. Department of Defense, "V50 Ballistic Test for Armor", MIL-STD662,

18 December 1997.


3. A. L Lastnik,., M B Cleavly, J. B Brown,., "Development and fabrication of

polycarbonate eyeshield for the U.S. Army's Flyer's Helmet, TR-71-3-CE,

U.S. Army Natick Laboratories, Natick, MA, June 1970.

4. F. P Meyer, R Sacher, "Solarization effects on the materials employed in the

ballistic/laser eye protection spectacle system (B/LEPS), Interim Letter

Report, U.S. Army Materials Technology Laboratory, Watertown, MA, May,

1991.

5. A. R Hyde, J. G Darrant, "TRANSARM-Improved transparent armour,"

Proceedings of DARPA/ARL/ARO Transparent Armor Materials Workshop,

November 16-17, 1998, Annapolis, MD.

6. P.J Patel, G. A Gilde, J. W McCauley,., "Transient liquid phase sintering of

aluminum oxynitride (AlON), Army Research Laboratory Patent Disclosure

6-00, May 2000.

7. D.W Roy, J. L Hastert,L. E Coubrough, K. E Green, A Trujillo, "Method for

producing transparent polycrystalline body with high ultraviolet

transmittance," U.S. Patent # 5244849, September 14, 1993.

8. G.A. Gilde,P.J. Patel, M.Patterson, "A comparison of hot-pressing, rate

controlled sintering, and microwave sintering of magnesium aluminate spinel

for optical applications," Proceedings of SPIE Conference on Window and

Dome Technologies and Materials VI, Randal Tustison, SPIE Vol.3705, 94-

104, SPIE, Washington, April 1999.

9. D.C. Harris, Infrared window and dome materials, SPIE, Washington, pg.

32,1992.

10. Schmid, F., Viechnicki, D., J., Growth of Sapphire Disks from the Melt by a

Gradient Furnace Technique, J. Am. Ceram. Soc., 53, 528-29 1970.

11. H.E. Labelle, EFG, The Invention and Application to Sapphire Growth," J.

Cryst. Growth, 50, 8-17, 1980.

12. P.J. Patel, J.J. Swab,G.A. Gilde, Fracture properties and behavior of

transparent ceramics, PROC. SPIE Vol. 4102, p1-14, Inorganic Optical

Materials II, Alexander J. Marker; Eugene G. Arthurs; Eds., 10/2000.

13. P. McGuire, R. Gentilman, B. Pazol, J, Askinazi, J. Locher, "Mulitpane large

area and doubly-curved sapphire windows," Proceedings of the 8th DOD

Electromagnetic Windows Symposium, 27 April 2000.

14. H Policove,., "State of the Art in optical finishing," Proceedings of

DARPA/ARL/ARO Transparent Armor Materials Workshop, November 16-

17, 1998, Annapolis, MD.

15. P.J. Dehmer,., M. Klusewitz, "Proceedings of 8th DoD Electromagnetic

Windows Symposium at the USAF Academy, 24-27 April 2000"

16. R.W Lewis, and G.R Parsons, Ballistic Performance of Transparent Materials

for Eye Protection, AMMRC-TR-72-36, U.S. Army Material and Mechanics

Research Center, Watertown, MA, November, 1972.


17. M.E Roylance,., and Lewis, R.W., Development of Transparent polymers for

Armor, AMMRC-TR-72-23, U.S. Army Material and Mechanics Research

Center, Watertown, MA, July, 1972.

18. R Gonzalez, G.J Wolfe, Ballistic Transparencies for Ground Vehicles,

Proceedings of DARPA/ARL/ARO Transparent Armor Materials Workshop,

November 16-17, 1998, Annapolis, MD.

19. R. A Beyer, H Kerwien, "Evaluation of AlON for cannon window

application," Proceedings of SPIE Conference on Window and Dome

Technologies and Materials VI, Randal Tustison, SPIE Vol.3705, 113-118,

SPIE, Washington, April 1999.


MICROWAVE REACTIVE SINTERING TO FULLY TRANSPARENT

ALUMINUM OXYNITRIDE (ALON) CERAMICS

Dinesh Agrawal, Jiping Cheng, and Rustum Roy

Materials Research Institute

The Pennsylvania State University

University Park, PA 16802, USA

ABSTRACT

Fully transparent aluminum oxynitride (ALON) ceramic has been developed

by a single-step microwave sintering method. Starting with -alumina and

aluminum nitride powder mixture, the compacted pellets were microwave sintered

under an ambient pressure of pure nitrogen. It was found that single ALON phase

formed at 1650 C in 60 minutes by microwave processing, and the fully dense

and highly transparent ALON samples were made at 1800 C with residence time

of 60 minutes.

INTRDUCTION

Aluminum oxynitride (ALON) has an approximate composition of

Al23O27N5 (9Al2O3 5AlN). ALON can be sintered to fully transparent ceramic

material having mechanical and optical properties similar to those of sapphire

with the advantages of an isotropic cubic crystal structure. The transmission

range of ALON can extend from 0.2 m in the UV through the visible to 6.0 m

in the infrared, which makes it a very useful material for many electromagnetic

window applications. Combined with the high strength and high hardness, ALON

is an ideal material for transparent armor product [1,2].

The conventional fabrication of transparent ALON ceramics involves using

pre-synthesized ALON powder to form a green body, followed by sintering in a

nitrogen atmosphere at high temperatures (>1850 C) for extended period (20-100

hours) and often requires hot pressing [3]. A single-step preparation method was

also tried to make transparent ALON ceramics, by mixing Al2O3 and AlN

powders and subsequent reactive sintering at 1850 C for 1 hour at 3 bar nitrogen

atmosphere. But the sintered body in this case was translucent [4,5].

Microwave sintering is a novel sintering process that is fundamentally

different from the conventional sintering process. In conventional sintering, the



sintering driven force, temperature, is generated by external heating elements (in

resistance heating) and then is transferred to the samples via radiation, conduction

and convection. In microwave process, the processing materials themselves

absorb microwave power and then convert microwave energy in to heat within the

sample volume itself, and hence the heating is very rapid and uniform. The

microwave processing of materials has major advantages of higher energy

efficiency, enhanced reaction and sintering rate, cycle time and cost savings [6].

In the last four years, in this laboratory we have successfully sintered various

ceramics, composites, and even powdered metals to full density using microwave

processing [7,8]. Some highly transparent ceramic samples, such as alumina,

spinel, and aluminum nitride, have been successfully prepared by microwave

sintering process in our lab. Compared to the conventional sintering process, the

microwave sintering to highly transparent ceramic samples can be conducted at

lower sintering temperatures and much shorter sintering times [9].

EXPERIMENTAL

The ALON green samples in this work were prepared by mixing high purity

-Al2O3 powder (SM8, Baikalox, Baikoski International, NC, USA) and AlN

powder (Grade C, H.C. Starck, Laufenburg, Germany). The properties of the

starting powders are shown in Table 1. It was found that the addition of a small

amount of Y2O3 increased the densification and improved the transparency of the

sintered bodies during microwave sintering. Therefore the starting mixture

contained 67.5 mole percent of Al2O3, 33.5 mole percent of AlN, to which 0.5%

(by weight) Y2O3 in form of Y(NO3)3 6H2O was added. The powders with 3

wt.% of binder (Acryloid) were ball-milled in acetone for 24 hours. After drying,

the mixture was compacted uniaxially into pellets of diameter 12.7 mm and height

3 mm at a pressure of 30 MPa. Finally, the pellets were cold isostatically pressed

at 250 MPa for 5 minutes.

Table 1. The physical properties of the starting powders.

Manufacturer Purity Particle size Main phase

AlN powder H.C Starck (Grade C) >98% 2.41 m AlN

Al2O3 Powder Baikowski (SM8) 99.99% 0.15 m -Al2O3

Microwave sintering was carried out by using a 1.5 kW, 2.45 GHz single

mode microwave applicator in flowing pressure nitrogen at ambient pressure.

The heating rate was kept around 100 C/min by controlling the incident

microwave power. The phase composition of the samples was determined by x-

ray diffractometry (XRD). The densities of the sintered ALON samples were

measured by the Archimedes method. The optical microscope (Olympus) was


used to study the microstructures, and the Varian spectrophotometer (CARY

2300) was used to measure the transmittance of the sintered samples.


Figure 1 shows the XRD patterns of the starting mixture and microwave

processed samples under different synthesis conditions. The phase composition of

the starting material is pure -Al2O3 and AlN. The ALON phase appeared when

the sample was microwave heated at 1650ºC for only 10 minutes. The content of

ALON phase increased with the firing time at that temperature. A single phase

ALON was found after microwave firing at 1650ºC for 60 minutes.

Figure 1. The X-ray diffraction patterns of the starting mixture and

microwave synthesized ALON samples.

(a) Starting Material;

(b) Microwave synthesized at 1650°C for 10 minute;

(c) Microwave synthesized at 1650°C for 60 minute;

Figure 2 shows the densification behaviors of the ALON samples during

microwave sintering process. All microwave sintered samples exhibited only a

pure ALON phase composition which was confirmed by XRD. The theoretical

density (T.D.) of ALON is around 3.67 g/cm3. It was found that the samples

sintered at 1700 C for 1 hour with the density of 3.60 g/cm3 (~98.1% T.D.) were

still opaque or very slightly transparent. The samples sintered at 1750 C for 1

hour with the density of 3.67 g/cm3 (~100% T.D.) were quite translucent.


However, by raising the sintering temperature to 1800 C and keeping the dwell

time unchanged, the grain size increased, and the transparency of the samples was

greatly improved. We tried to microwave sinter ALON at 1850 C, but the

sintering process was unstable because sometimes a discharge occurred which

resulted into partial melting of the samples. At the sintering temperature of

1800 C, the density of the samples increased with the increasing dwell time and

the transparency improved as well.

10 20 30 40 50 60

3.55

3.60

3.65

3.70

Sintering temperature = 1800o

C

Den

sity, g

/cm

3

Sintering time, min.1700 1750 1800

3.55

3.60

3.65

3.70

Sintering time = 60 min.

Density, g/c

m3

Sintering temperature,o

C

(a) (b)

Figure 2. Densification behavior of the ALON samples during microwave

sintering with (a) temperature, and (b) time.

The microstructural developments of the ALON samples during microwave

sintering are shown in Figure 3. As mentioned above, the samples sintered at

1750 C and 1800 C both for 1 hour showed the same density. But the grain size

of the 1800 C sintered sample was around 40-50 m, much higher than that of the

1750 C sintered samples (around 10-20 m), and also the grain boundaries

became narrower and cleaner, total grain boundary volume also reduced

dramatically. This obviously resulted in a transparency improvement. It was very

difficult to find pores in the sample sintered at 1700 C for 1 hour, that means the

sample had had a good densification, but the grain size was very small (less than

1-2 m), with very high grain boundary volume.


(a) (b) (c)

Figure 3. Microstructures of microwave sintered ALON samples at (a)

1700 C, (b) 1750 C and (c) 1800 C for 1 hour.

Compared to single crystals, sintered polycrystalline ceramics (such as

ALON) have much more complicated microstructures including grains, grain

boundaries, second phases and pores. A light incident to a sintered body

experiences a diffuse reflection at the surface, and is subsequently absorbed and

scattered by the inhomogeneities inside the sintered body. To increase the

transmissivity of a sintered polycrystalline ceramic body, it is very important to

reduce porosity and the grain boundary phases since they strongly scatter light.

The ALON sample sintered at 1700 C for 1 hour had a high density up to 98%

T.D., but the grain structure was not developed well enough, and the grain

boundary volume was too large to cause considerable scattering of light. This

made the sample opaque.

0.5 1.0 1.5 2.0 2.5

0

10

20

30

40

50

60

70

80

90

100

Tra

nsm

itta

nce

, %

Wavelength, m

Figure 4. Transmittance of the ALON sample made by microwave sintering

at 1800 C for 1 hour (sample thickness = 0.6 mm).


Figure 5. Appearance of the ALON sample made by microwave sintering at

1800 C for 1 hour.

Figure 4 shows the transmittance data of the microwave sintered ALON at

1800C for 1 hour. The total transmission of 60% was achieved for the polished

sample with a thickness of 0.6 mm. The sample shown in Figure 5 was optically

transparent. The results shown in this work have demonstrated that the

microwave sintering process can offer lower sintering temperature and much

shorter sintering times, in comparison with the conventional sintering process, to

make fully transparent ALON ceramics.

CONCLUSION

Fully transparent aluminum oxynitride (ALON) ceramic was successfully

prepared by pressureless microwave sintering processing. It was found that single

ALON phase formed at 1650 C in 60 minutes during microwave processing, and

the fully dense and highly transparent ALON samples were made at 1800 C with

residence time of 60 minutes.

ACKNOWLEDGMENTS

This work is partially funded by by DARPA /ONR under Grant No.

N00014-98-1-0752.

REFERENCES

1. T.M. Hartnett, S.D. Bernsein, E.A. Maguire, and R.W. Tustison, Optical

Properties of ALON (aluminum oxynitride), in Window and Dome

Technologies and Materials V, Proceedings of SPIE, edited by R.W.

Tustison, Vol. 3060 (1997)

2. N.D. Corbin, Aluminum Oxynitride Spinel: A Review, J. Euro. Ceram.

Soc., 5, 143-154 (1989)

3. R.L. Gentilman, E.A. Maguire, and L.E. Dolhert, Transparent Aluminum

Oxynitride and Method of Manufacture, US patent, 4720362 (1988)


4. H.X. Willems, M.M.R.M. Hendrix, G. de With, and R. Metsalaar,

Production of Translucent –Aluminum Oxynitride, in Euro-Ceramics II,

edited by G.Ziegler and H. Hausner, Vol.3, 2443-2447 (1991)

5. J.W. McCauley, and N.D. Corbin, Phase Relations and Reaction Sintering

of Transparent Cubic Aluminum Oxynitride Spinel (AlON), J. Amer. Cer.

Soc. 62, 476-479 (1979)

6. W. H. Suttoon, in Microwave Processing of Materials III (R. L. Beatty,

W. H. Sutton, and M. F. Iskander, eds), Proceedings of the Materials

Research Society, Vol. 269, pp. 3-20 (1992)

7. R. Roy, D. Agrawal, J. Cheng, and M. Mathis, in Microwave: Theory

and Application in Materials Processing IV, Ceramic Trans., Vol. 80,

3-26 (1997)

8. R. Roy, D. Agrawal, and J. Cheng, Microwave Electromagnetic

Processing Invades New Materials, presented at the 2nd

World Congress

on Microwave & Radio Frequency Processing, Orlando, FL, USA, April

2-6, (2000)

9. J. Cheng, D. Agrawal, Y. Zhang, B. Drawl, and R. Roy, Fabrication of

Transparent Ceramics by Microwave Sintering, American Ceramic Society

Bulletin, Vol. 79, No. 9, 71-74, Sept. (2000)


AN INVESTIGATION OF THE TRANSMISSION PROPERTIES AND

BALLISTIC PERFORMANCE OF HOT PRESSED SPINEL

Mark C.L.Patterson

Technology Assessment & Transfer Inc.,

133, Defense Highway

Annapolis, MD 21401

Don W. Roy

21210 North 132 Drive,

Sun City West, AZ 85375

Gary Gilde

US Army Research Laboratory

AMSRL-WM-MC Building 4600

Aberdeen Proving Grounds

Aberdeen, MD 21005

ABSTRACT

The fabrication of transparent polycrystalline spinel (MgAl2O4) is being

pursued by Technology Assessment and Transfer Inc. (TA&T), with the goal of

becoming a commercial producer of transparent armor as well as optical windows

and domes. The process is based on hot pressing followed by hot isostatic

pressing to further improve the optical properties. This approach promises to

produce spinel at a cost significantly less than sapphire or ALON and at a scale up

to 22” diameter in the near term and possibly up to 36” in diameter. The larger

plates should be possible if the hot isostatic pressing step can be eliminated. This

paper discusses the effort underway to improve the optical properties of spinel

during hot pressing alone thereby establishing a low cost approach for transparent

armor.

The ballistic performance of spinel has been evaluated against ALON and

sapphire and the key properties of spinel are discussed with reference to its use in

infra-red windows and domes. High transmission in the mid infra-red is driving

a renewed interest in spinel for many optical systems. This paper provides an

overview of a joint effort between the Army and Technology Assessment and

Transfer Inc. to establish a capability for large spinel plate fabrication, and of

efforts to improve the optical transmission for multimode window and dome

applications.



INTRODUCTION

Magnesium aluminate spinel (MgAl2O4), a cubic oxide ceramic, has been

successfully sintered from selected reactive powders to transparency in the 0.3 m

to 5.5 m range. Transparency was first demonstrated in 1961 by the General

Electric Company and since that time there has been an intermittent effort to

develop optical quality spinel for a range of IR window, dome and armor

applications. A good summary of the early development efforts can be found in

the literature1 and in earlier work by the present authors2.

There are two main approaches to the fabrication of transparent spinel; the

first is by hot-pressing (HP) to transparency followed by hot isostatic pressing

(HIP) and the second is by pressureless sintering to produce an opaque product

which can be HIPed to transparency. Using HP/HIP processing, excellent optical

performance was achieved previously with spinel in small sizes and thin wall

thickness by Coors Ceramics and Alpha Optical Systems. Spinel domes were

qualified for at least two IR guided missiles and for the stinger missile launch tube

window, prior to the shutdown of production when military budgets were reduced

following the Gulf War.

An effort to fabricate transparent spinel by RCS Technologies Inc., using “rate

controlled pressureless sintering” followed by HIP processing in the early 90’s

showed considerable promise, but could not be sustained because of the lack of

financial support3. Consequently there has been no commercial spinel production

since 1993.

There is a need within the military to reduce the weight and increase the size

capability of transparent armor systems while simultaneously increasing ballistic

protection capabilities. Additionally, there is growing need for window and

dome materials which extend further into the IR and can be used for multimode

weapons systems that are exposed to very demanding environments.

Polycrystalline MgAl2O4 spinel has been recognized for many years as a material

with great potential for transparent armor and for UV, visible and mid IR optical

component applications. Based on these needs, TA&T is scaling up to produce

spinel commercially for both optical and armor applications

This work is being driven not only by the need for spinel based on its unique

properties but based on current as well as prior manufacturing information. It is

expected that the cost of large spinel plates will be significantly less than the

competitive ALON and sapphire materials.

Properties of Spinel

Spinel crystal structure is cubic and optically isotropic; thus polycystalline

shapes may be fabricated without severe scattering problems inherent in


polycrystalline non-cubic materials. In the microwave region the isotropy of

spinel prevents localized absorption and heating that occurs in non-cubic

materials because of differing grain boundary orientation and anisotropic

dielectric loss index. Spinel undergoes no polymorphic transformations, so it is

free of problems due to thermally induced phase changes. Extensive programs

were carried on in the 1980’s to measure the properties of spinel as well as other

candidate window materials, including sapphire, ALON and yttria at Johns

Hopkins University Applied Physics Laboratory4 and Honeywell Systems

Research Center5. The typical physical properties for polycrystalline spinel are

listed in Table I.

Table I Typical Physical Properties of Polycrystalline MgAl2O4 Spinel

Density 3.58 gm/cc

Hardness, Knoop [100gm] 1398kg/mm2

Minimum Strength @ 25oC

4-pt bending 15x 103psi [103x10

6Pa]

Biaxial 25x 103psi [172x10

6 Pa]

Tension 16x103psi [110x10

6 Pa]

Compression

390x103psi[2.69x10

9Pa]

Elastic Mod. 39x106psi 273x10

9Pa]

Bulk Mod. 27.9x106psi 192x10

9Pa]

Shear Modulus 15.9x106psi[110x10

9Pa]

Thermal Coefficient of Expansion

25 - 200oC 5.6x10

-6/oC

25 - 5000C 7.3x10

-6/C

25 - 1000oC 7.9x10

-6/C

Specific Heat, cal/gm/oC

20oC 0.21

Poisson’s Ratio 0 .26

Dielectric Strength, kV/mm

.05’’[1.27mm] thick 490

.01’’[.25mm] thick 580

Melting Point 21350C [ 3875

oF]

Volume Resistivity, ohm-cm

25oC >10

14

300oC 5x10

14

500oC 2x10

11

700oC 4x10

8

Thermal Conductivity,

gm-cal/cm2/sec/

oC [W/m-

oK]

25oC 0.060 [24.7]

100oC 0.0357 [14.8]

1200oC 0.0130 [5.4]

Dielectric Constant & Loss index

1KHz 8.2 0.00025

1MHz 8.2 0.0002

9.3GHz 8.3 0.0001

The refractive index of spinel has been measured to vary between approximately

1.74 and 1.66 over the range of its transparency as shown in Table II.

Table II. Reflective index of spinel at different wavelengths.

( m) 0.49 0.59 0.66 1.0 2.0 3.0 4.0 5.0

Ref.Ind 1.736 1.727 1.724 1.704 1.702 1.698 1.685 1.659

Spinel has distinct optical property advantages over both sapphire and ALON.

In contrast to cubic spinel, single crystal Al2O3 [sapphire] is anisotropic and

birefringent, causing optical design problems11

. ALON has a shorter transmission

cut-off in the 4.5 to 5.5 micron spectral region, resulting in a significantly higher


coefficient of absorption in that critical band. This is shown dramatically in Table

III where the relative transmission properties at RT, 250 C and 500 C are shown

for spinel, sapphire and ALON. At 4.8 microns and 250 C spinel offers a 4% and

14% improvement in transmission over sapphire and ALON respectively.

Table III. Transmission properties for spinel, sapphire and ALON vs. wavelength

at temperatures up to 500 C6.

Percent Transmission at Wavelength in ( m)

Sample & Temp.( C) 3.0 4.0 4.5 5.0 5.5 6.0

25 88 87 77 59 11 -

ALON 250 87 84 71 46 7 -

500 87 81 62 33 2 -

25 87 87 84 71 49 -

Sapphire 250 86 86 79 61 32 -

500 84 82 72 50 20 -

25 84 87 84 76 54 22

Spinel 250 84 87 82 67 39 11

500 82 83 76 55 23 4

Spinel Applications

As an optical material spinel is similar in nature to both ALON and sapphire

in that it has a high hardness, erosion resistance, and transmits from

approximately 0.25 m to 6.0 m. It is isotropic and does not therefore exhibit the

birefringence seen in sapphire and as discussed above exhibits a lower absorption

coefficient than either sapphire or ALON in the mid infrared, particularly at

elevated temperatures.

Based on these properties and the hope that spinel can be fabricated at a

considerably reduced cost over either ALON or sapphire, spinel is being

developed for use as erosion resistant multimode windows and domes for a wide

range of defense applications. It is also being investigated for optical lenses as

well as armor against hard, armored piercing projectiles. For armor applications,

the possibility of producing large panels, possibly up to 36” in diameter through a

low cost hot-pressing process is driving the continued interest at present.

Ballistic Evaluation of Spinel

The Army has been interested in spinel for transparent armor since the late

sixties7. When all factors including transparency, hardness, impact resistance,

strength, modulus, ease of fabrication, and crystal size capability are taken into

consideration, spinel appears particularly well suited for armor applications.

Because of the U.S. Army's continued interest in spinel for transparent armor and


electromagnetic windows, ARL and TA&T signed a cooperative Research and

Development Agreement (CRADA) in 1998 for the development and dual use

assessment of transparent spinel, using hot-press/HIP processing. Good

transparency has been achieved in flat plates up to 5 inches square and 0.5 inches

thick. While the main focus of recent work has been for the fabrication of thick

section (>0.4 inches) spinel plate, efforts have also included the evaluation of

multiple pieces of thinner sectioned plates.

A recent evaluation of transparent spinel and ALON carried out by the ARL8

demonstrated that both materials dramatically improve the performance of

transparent armor systems over the traditional glass/plastic systems currently in

use, based on areal density and velocity requirements9. The results of this and

other ceramic based systems compiled between 1969 and 1996 are shown in

Figure 1. For this particular threat, the data shows that spinel backed with

polycarbonate performs approximately 4% better than ALON and 10% better than

sapphire, both backed with polycarbonate. The spinel/polycarbonate was

approximately 1.5 lbs/ft2 heavier than the spinel/polycarbonate system. With an

areal density of 12 lbs/ft2 it exhibited a V50 between 2900 ft/sec and 2950 ft/sec.

ALON backed with both glass and polycarbonate exhibited a V50 of 3000 ft/sec.

The weight of the armor system, however, was 33% higher at 16 lbs/ft2.

- Spinel

- ALON

- Sapphire

- Ball. glass

V50

Inverse areal density

Desired ballistic

performance

Figure 1. Relative ballistic data for spinel, sapphire and AlON against an

unspecified threat.

There is presently a growing need for lightweight transparent armor concepts

against armored piercing (AP) 12.7mm projectiles, which can fulfill the mass

and/or thickness (scale) requirements for air and light armored vehicle programs.

Spinel in thin layers has in the past been laminated with glass and polycarbonate

backing to defeat 7.62mm AP projectiles. Recently, thicker sections of spinel (up


to 20mm) have been evaluated for this application10 and its performance

compared with Al2O3 and SiC against 12.7mm projectiles at two different

velocities. The spinel tested for this application was fabricated in tiles up to 20cm

square by the French company, Ceramiques & Composites. The ceramic front

face was laterally confined and bonded to an aluminum honeycomb back surface

(no ballistic influence). Ballistic evaluation was carried out at 2880 ft/sec and

1800 ft/sec. In their study they determined that spinel outperformed alumina at

both projectile velocities but was inferior to SiC. (The alumina was not sapphire

but was a 92% Al2O3 ballistic grade). They also determined that ballistic

protection against AP 12.7mm at 2880 ft/sec could be obtained from a sapphire

front surface backed with polycarbonate at an areal density of approximately 21.5

PSF or by a glass/polycarbonate laminate with an areal density of approximately

41.0 PSF. They estimated that the same protection with spinel could be achieved

with an areal density of 20.5 PSF.

Recent evaluation of spinel tiles tested at the Army Research Laboratory in

Aberdeen showed comparable performance for both spinel and ALON, and a

significant weight reduction over a glass system. The tiles were 4” square by

0.375” thick and shot with a steel core, small caliber projectile. The actual data is

not available but is shown normalized in Table IV together with a baseline

glass/plastic system.

Table IV. Normalized ballistic performance (V50 and areal density) for spinel

ALON and a glass/plastic baseline against a small caliber, steel cored projectile.

Glass/plastic Spinel ALON

Areal Density 1.0 0.43 0.44

V50 1.0 0.88 0.89

Although ALON and sapphire are very promising transparent armor materials,

spinel may be able to offer the best balance of both performance and affordability.

PROCESS DEVELOPMENT

Technology Assessment and Transfer Inc., is pursuing a HP process using LiF

as a sintering aid. At present the process requires subsequent HIP to produce

good optical transmission with low haze, but the goal of future development will

be to investigate if the final HIP step can be eliminated, thereby significantly

reducing the cost and allowing the fabrication of parts up to 36” in diameter.

The theory behind the present pressing approach is to ensure that volatile

contaminants such as LiF are allowed to escape from the sintering body before

closed porosity is attained. If the grains are allowed to sinter at too high a

temperature bridges are formed, which upon application of further load cannot be


broken, resulting in opacity. It is important therefore to balance application of the

load with out gassing of the sintering body and temperature to ensure continuous,

yet gradual microstructural development. To date, satisfactory pressing

procedures have been established to produce good transmission (>80%) in 0.40”

thick sections of spinel up to approximately 5” in diameter. The present focus of

work is threefold:

understanding the effect of processing environment on microstructural

development and properties

optimize pressing procedures for shaped configurations including domes

increase the size capability to 22” in diameter by year end 2002

Processing, Microstructural Development and Properties

It has been shown previously that the transmission properties of spinel can be

improved if the temperature is increased, annealed11, or if it is HIPed12. Figure 2

shows the increase in transmission that has been observed when samples are

annealed for longer periods of time at a temperature below the sintering

temperature.

Figure 2. Percent transmission up to 3 microns following HP at 1650 C for 3 hours

(right) and following an additional anneal at 1550 C for 12 hours (left).

Initially it was proposed that the increase in transmission following HIPing

was attributed to a growth in the grain size and a reduction in the number of grain

boundaries. A complimentary increase in the strength following HIPing, which

was also observed previously by Don Roy, was thought to be due to grain

boundary development (a reduction in the impurity levels or realignment of

adjacent grains). A brief study was undertaken to investigate the microstructural

development that takes place during HP, annealing and HIP. Some of theses

results are described below.


Orientation Imaging Microscopy: Microstructural analysis was performed on

three samples following HP at 1650 C, HP at 1650 C followed by annealing at

1550 C for 12 hours, and following HP at 1650 C and HIP at 1700 C. Once

polished, each of these samples showed transmission values summarized in Table

V showing the significant increase in transmission that can be achieved in the

visible and near UV regions. In the near IR regions the increase in transmission is

less pronounced. It was hoped that the cause for this increase in the optical

transmission could be seen in the microstructure and so orientation imaging

microscopy was performed on these samples using a FEI XL-30 FEG SEM with a

beam current of 1.5nA. Data was collected from an area 1,200 m by 2,400 m in

size using a step size of 5 m.

Table V. Transmission and microstructural data for spinel following HP -

1650 C, HP & annealing at 1550 C, and following HP & HIP - 1700 C.

Attribute HP only HP/anneal HP/HIP

Wavelength 0.5 m 2.0 m 0.5 m 2.0 m 0.5 m 2.0 m

Transmission 45% 70% 65% 78% 78% 79%

Av grain size 38.3 m 39.5 m 45.1 m

Aspect ratioa

0.55 0.58 0.65

Av.misorientationb

~ 45 ~ 45 ~ 45

Textc. all grains 2.678 2.927 3.329

Textc. small 2.061 2.111 2.365

Textc. medium 2.556 2.328 2.836

Textc. large 6.864 6.000 7.405

In an effort to establish whether or not grain texture was contributing to the

overall transmission properties of the spinel, density pole figure plots were

generated for each of the samples for the 100, 101 and 111 axis. These were

generated for all the grains as well as for individual groups of grains as described

in Table V above. The maximum values measured for any of the reference

orientations are shown plotted as a times random (1.0) value in Table V, showing

that there is slightly more texture seen in the HIP spinel and that texture is driven

primarily by the large grains. The average grain size is similar for the HP and HP

annealed sample but is larger in the HP/HIP sample, as shown in Table V. The

a grain aspect ratio compared with 1.0 being equiaxed.

b av. misorientation angle distribution for all grains is estimated as being the same

c maximum times random values (1.0) from summary of pole figure plots. Small refers to grain

less than 50 m, medium 50-120 m and large above 120 m in diameter.


grain aspect ratio increased for the annealed and the HIPed samples. However, the

misorientation angle distribution was centered around 45 for all the samples.

Figure 3. Image quality maps showing the grain size distributions for the samples

described in Table V. The shaded regions in the distributions refer to bands of

50 m to 100 m and 300 m to 500 m respectively.

Figure 3 shows the grain boundary structure for each of the three samples and

highlights very low angle grain boundaries (1 -5 ), low-angle grain boundaries

(5 -15 ), and high angle grain boundaries (>15 ).


Although it may not be clear in these micrographs, it was interesting to note

that the very low angle grain boundaries (as indicated by the arrows) seemed

concentrated within certain grains and were not evenly distributed throughout the

microstructure in any of the samples. No orientation correlation was apparent

with any of the selected grains.

Results of the Orientation Imaging Microscopy: This initial analysis of the spinel

microstructures following different treatments revealed little insight as to the

significantly higher transmission values which were observed following annealing

of the samples. Following HP/HIP it can be seen that there is a larger average

grain size, a shift in the grain size distribution towards a higher fraction of large

grains, a higher fraction of near equiaxed grains, and a higher fraction of

directional alignment. These differences are not observed between the HP and

HP/annealed spinel samples. It is interesting to note that no correlation could be

seen in the distribution of very low angle grain boundaries, which appeared

localized within specific grains and not evenly distributed throughout the

microstructure of all three spinel types. Additionally, no correlation could be seen

between these grains and their orientation.

The increased transmission values were most pronounced in the near UV as

shown in Figure 2, and it is expected that the influences may occur at a smaller

scale than were observed and evaluated in this study.

PRODUCTION APPROACH

A facility is presently being made ready through the installation of processing

and quality assessment equipment. The first of these to be installed into the new

facility is a 600 ton Birdsboro press with 72” of daylight. A heating package and

vacuum enclosure will be installed separately, thereby allowing the fabrication of

plates up to 22” in diameter (and possibly up to 36” in diameter in the future).

The 600 ton Birdsboro press is shown partially constructed in Figure 4. Each

of the four posts (not shown in this picture) are 12” in diameter and the main ram

is 16” in diameter. It is expected that through further changes to the hydraulics

system it will be possible to increase the load capability to 1000 tons, thereby

allowing even larger spinel plates to be fabricated. In addition to fabrication of

single components with large diameters, this press will also be used to process

multiple parts in the form of domes, lenses or smaller windows.


Figure 4. 600 ton press during construction following alterations

The spinel fabrication process that has been selected by TA&T is hot-

pressing, followed by hot isostatic pressing. The initial hot-pressing process uses

LiF as a sintering aid and results in a transparent product which can readily be

inspected for internal flaws, inclusions and discoloration. The second hot

isostatic pressing step further improves the optical properties of the spinel and

reduces variation in other mechanical properties such as strength. A process flow

chart for the individual operations is shown in Figure 5.

The hot-pressing process is a forgiving one in which good optical properties

can be obtained within a single processing step. At this time the optical properties

required for window and dome applications can only be achieved through both

hot-pressing and hot isostatic pressing. It is expected that with future

development it maybe possible to achieve the required optical properties from a

single hot-pressing step, thereby leading to a significant reduction in the

processing costs. This approach will be investigated over the next 2 years. As yet,

a two step process is still required. To date the hot-pressing process has been


capable of producing flat plates of spinel up to a thickness of approximately 0.5”

and has also shown the ability to fabricate near hemispherical domes

approximately 6” in diameter. Additionally, the hot-pressing approach has

historically provided a high yield in excess of 60% as compared with alternative

approaches such as pressureless sintering13.

RejectReject

Final finishing

processes

Hot, isostatic

pressing of partQuality

controlQuality

control

Hot-press

spinel part

Binder

burnout

Formation of

“green” shape

Powder blending

and mixing

Figure 5. Process flow chart showing the individual operations required for

optical spinel fabrication

DISCUSSION

Technology Assessment and Transfer Inc. is establishing a facility to become

a commercial supplier of transparent spinel parts for armor, IR window and dome

applications. The focus of the present work is to understand the effect of

processing environment on microstructural development and properties of the

spinel. The pressing procedures are being optimized for flat plates and shaped

configurations including domes, and the size capability is being increased up to

22” in diameter by year end 2002.

An improvement in the optical transmission is seen following HP if the spinel

is annealed or HIPed. Following HP/HIP it can be seen that there is a larger

average grain size, a shift in the grain size distribution towards a higher fraction

of large grains, a higher fraction of near equiaxed grains, and a higher fraction of

directional alignment. However, following annealing, these differences were not

observed and the cause of the improved optical performance over the HP spinel is

as yet unknown.

The HP processing of spinel promises to be a low cost approach to producing

transparent armor with similar performance to both ALON and sapphire but with

the capability of producing larger panels than are possible with either ALON or

sapphire. In the near term 22” diameter plates will be made and this will possibly

be increased to 36” in the future. Microstructural development of spinel is

presently being investigated with the purpose of improving the optical properties


of spinel without the HIP requirement, thereby reducing significantly the overall

fabrication costs. The HP process produces high yields and is capable of

fabricating near net shape parts, such as full hemispherical domes. Greater than

80% transmission has presently been achieved in 0.4” thick parts 5” in diameter

and a number of parts formed from near net-shape HP/HIP are presently being

evaluated for optical application.

ACKNOWLEDGEMENTS

The authors would like to thank Matt Nowell at TexSEM Laboratories for

performing orientation imaging analysis on the spinel samples. This work was

funded in part by the Army Research Laboratory Aberdeen under an SBIR

contract # DAAD17-00-C-0080.

REFERENCES

1 W.H.Rhodes, “Phase Chemistry in the Development of Transparent

Polycrystalline Oxides”, GTE Laboratories, TR-0209-07-92-082, 1992. 2 D.W.Roy, M.C.L.Patterson, J.E.Caiazza and G.Gilde, “Progress in the

Development of Large Transparent Spinel Plates”, 8th

DoD Electromagnetic

Windows Symposium Proceedings, ASAFA Colorado Springs, CO 24th

-27th

April

2000.3 M.L.Huckabee, “ Near net shape spinel optics for broadband windows,

lenses and domes” Final report contract DAAH04-95-C-0010, RCS Technologies

Inc. 1995 4 M.E.Thomas, R.L.Joseph and W.J.Tropf, “Infrared Properties of Sapphire,

Spinel and Yttria as a Function of Temperature”, SPIE vol. 683, 1986. 5 J.A.Cox, D.Greenlaw, G.Terry, K.McHenry and L.Fielder., “Infrared and

Optical Transmitting Materials”, SPIE vol. 683, 1986. 6 SWIR/LWIR “Optical Sensor Window Development Program”. Final

Report DASG60-85-C-0018. 7 A. Gatti and J. Noone, Feasibility Study for Producing Transparent Spinel,

General Electric Company, Space Sciences Laboratory , Space Division, King of

Prussia, PA Final Report for Contract DAAG46-69-C-0096 8 M.C.L.Patterson, J.E.Caiazza, and D.W.Roy, “Transparent Spinel

Development”, Inorganic Optical Materials II, Alexander J.Marker III, Eugene G.

Arthurs Editors, Proc, of SPIE Vol. 4102 pp.59-68. 2000. 9 J.Conners., “Magnesium Aluminate Spinel, Material and Prototype

Development,” ARL Internal Communication, 17th June 1997. (This article was

not seen by the author but has been quoted directly from a paper by J.J.Swab,

J.C.LaSalvia, G.A.Gilde, P.J.Patel and M.J.Motyka, “Transparent Armor

Ceramics:ALON and Spinel”.


10 C.E. Cottenot, “Transparent Ceramic for Lightweight Armors,” Lightweight

Armor Systems Symposium ‘95 Cranfield, England, 28-30th June, 1995. 11

M.C.L.Patterson, G.Gilde and D.W.Roy, “Fabrication of Thick Panels of

Transparent Spinel” Inter. Symp. Proc. Optical Science & Technology. SPIE 46th

Annual Meeting San Diego, CA. 29th

July – 3rd

August 2001. 12

G.Gilde, P.Patel and M.C.L.Patterson, “A comparison of hot-pressing, rate

controlled sintering and microwave sintering of magnesium aluminate spinel for

optical applications”, SPIE Conf on Window and Dome Technologies and

Materials VI, Orlando FL. April 1999. Vol. 3705. pp. 94-104. 13

Private communication, Don Roy August 2001.


Microstructure and Macrostructure Effects

THE EFFECT OF MICROSTRUCTURE ON THE DYNAMIC BEHAVIOR OF

COMPOSITE ALUMINA/TITANIUM DIBORIDE

Kathryn V. Logan, Ph.D., P.E.

School of Materials Science and Engineering

Georgia Institute of Technology

Atlanta, Georgia 30332-0245

ABSTRACT

Past work has shown that the dynamic behavior of a dense, hot pressed

ceramic-ceramic composite that is composed of a nominal 70wt% alumina/

30wt% titanium diboride formed from powders that were either manually mixed,

or synthesized using self-propagating high temperature synthesis (SHS) is

significantly affected by the microstructural bias, including phase distribution and

grain morphology, formed during synthesis and processing. A method has been

developed to bias the composite microstructure such that the titanium diboride

grains are caused either to be dispersed amongst, or to surround the alumina

grains. A review of past work on the significance of processing/forming

consistency, as well as results to date on efforts towards quantitative

characterization of the microstructure are presented.

INTRODUCTION AND REVIEW OF PAST WORK

A number of ceramic materials having potential application as high strain rate

armor materials have exhibited superior ballistic mass efficiencies comparable to

steel, but the results have not been consistent and the material properties that induce

a resistance to high strain rate penetration have not been completely determined,

especially for the effects of various microstructures.

In order to elucidate some of the mechanisms in ceramics that optimize armor

performance, a systematic study on controlled microstructure alumina/titanium

diboride ceramic composite materials has been carried out over the last several

years. Long rod penetrator (LRP), depth of penetration (DOP) tests conducted at

Aberdeen Proving Ground (APG) on these materials exhibited superior performance

with ballistic mass efficiencies up to four.1 The multi-phase material was a dense

ceramic particulate composite with a preferentially biased microstructure of hard

TiB2 grains congregating around grains of less hard Al2O3.



Previous work by Logan2 demonstrated the ability to influence the

microstructural bias of a hot pressed alumina/titanium diboride composite, both in

starting powders that were either produced using Self-Propagating High

Temperature Synthesis (SHS) or commercially available powders that were

manually mixed (MM). The microstructure of the composites can be preferentially

biased such that the morphology and distribution of the component phase grains can

be partially controlled.3,4

The microstructure designated as “A” (figure 1)

represents SHS powders hot pressed to >98% theoretical density ( t) forming a

microstructure that is biased towards 1-10 m titanium diboride grains (white

areas) comprising an average 7.4 m phase size surrounding 20-40 m alumina

(gray areas) grains. The microstructure designated as “B” (figure 2) represents

SHS powders hot pressed to >95% t forming a microstructure that is biased

towards 1-5 m titanium diboride grains (white areas) comprising an average 6.2

m titanium diboride phase size that is uniformly distributed amongst 10-20 m

alumina (gray areas) grains. The microstructure designated as “C” (figure 3)

represents MM powders hot pressed to >98% t forming a microstructure that is

biased towards 1-10 m titanium diboride grains (white areas) comprising an

average 8.7 m titanium diboride phase surrounding an alumina phase (gray

areas) comprised of grains up to 100 m in diameter. The microstructure

designated as “D” (figure 4) represents MM powders hot pressed to >98% t

forming a microstructure that is biased towards 1-10 m titanium diboride grains

(white areas) uniformly distributed in an alumina phase (gray areas) comprised of

grains averaging 12.3 m.

Composites having the biased microstructures have exhibited quasi-static and

dynamic behaviors that indicate a tendency to vary because of the microstructural

bias. Keller and Zhou5 have found dynamic compressive strengths of the four

biased microstructures described above to range from 4.4 to 5.3 GPa; values

which are 27% higher than the quasi-static values. Also, the measured

compressive strength directly correlates with the fraction of titanium diboride rich

areas on the fracture surfaces. The failure associated with the alumina phase is

transgranular; while the failure associated with the titanium diboride phase is both

transgranular and intergranular. Table I is a summary of dynamic properties of

four representative biased microstructures A, B, C and D. Figures 5-10 are

fracture surfaces of SHS and MM composites after MOR bar breaks. Figure 5

(boxed area) shows cleavage in the relatively large alumina phases (gray areas)

with the titanium diboride (white areas) tending to be localized around large

(gray) areas of alumina (figures 5 and 7). Figures 6 and 8 show evidence of

titanium diboride grains being more homogeneously distributed in the alumina,

crack pinning by titanium diboride, and a high concentration of microcracks.

Figure 9 shows relatively large TiB2 grains (light gray) localized at the alumina


grain boundaries (darker gray). Figure 10 shows the relatively large TiB2 grains

(white areas) homogenously distributed amongst the alumina grains (gray areas).

Figure 1. Sample A, (SHS T@A) Figure 2. Sample B, (SHS TinA)

[_____] = 100 microns

Figure 3. Sample C, (MM T@A) Figure 4. Sample D, (MM TinA)


Table I. Summary of dynamic properties

Compressive

Strength

Spall

Strength Wave Speed

Hugoniot

Elastic Limit

Tensile

Yield

Strength

Sample (GPa @103 s-1) (GPa) (KM/S (GPa) (GPa)

A 5.2 0.32 8.24 +/- .83 6.2 +/- 3.1 4.2 +/- 2.1

B 4.6 N/A 9.67 +/- 1.0 4.4 +/- 1.2 3.11 +/- 0.84

C 4.4 0.311 9.08 +/- .74 5.5 +/- 2.3 4.02 +/- 1.7

D 5.3 0.222 8.31 +/- .78 8.5 +/- 4.5 6.23 +/- 3.3

TiB2 4.2 0.33 9-18

Al2O3 4.0 0.45 ~6.7

Figure 5. MOR bar fracture surface Figure 6. MOR bar fracture surface

Sample A, SHS (T@A) Sample B, SHS (TinA)

Kennedy, et. al.6 have determined that while the Hugoniot Elastic Limit and the

compressive strengths of the four biased microstructures are dependent on the

average polycrystalline grain (phase) size, the tensile (spall) strength scales with

the titanium diboride-phase connectivity. This result implies that the

interconnected microstructure provides a higher resistance to failure in tension

compared with a microstructure having homogeneously dispersed particles. Table

I is a summary of the observed dynamic properties. Table II is a summary of the

phase sizes.



m m

Sample A, SHS (T@A) Sample B, SHS (TinA)


Sample C, MM (T@A) Sample D, MM (TinA)

Ferranti7 has found that processing parameters influence the development of

microstructural bias and composite properties in that the interconnected TiB2

polycrystalline phase forms directly during the SHS reaction. Mixing the B2O3


with Al prior to combining with TiO2 promotes TiB2 phase connection following

the SHS reaction. Ball milling of the resultant SHS product reduces the inherent

phase connectivity with the phase size decreasing as milling periods are longer. A

bimodal particle size distribution of synthesized powders produced high-density

parts; hot-press parameters did not appear to affect TiB2 phase connectivity.

Table II. Summary of phase sizes.

Sample Average Integral TiB2 Phase Al2O3 Phase

Curvature ( m-1) Size ( m) Size ( m)

A -0.316 +/- 0.022 7.0 10.4

B -0.476 +/- 0.046 6.2 9.1

C -0.074 +/- 0.028 8.7 25.1

D -0.375 +/- 0.031 7.9 12.3

The properties of composite ceramics usually follow the Rule of Mixing and

are influenced by the properties of the continuous phase.8 However, the

processing mechanisms that allow control and reproducibility of a specific

microstructure in multi-phase ceramic materials is not completely understood; and

accordingly, how the processing factors would directly affect and optimize

properties in high performance structural applications. The potential high

performance properties of a material are defined by intrinsic properties such as

crystal structure, bond strength and composition. However, the actual material

performance is significantly influenced by extrinsic properties such as grain size,

porosity and phase distribution. Therefore, the ability to form materials having

superior performance properties requires a detailed knowledge and control of the

processing routes that influence microstructural development.

A number of processing parameters have been shown to influence the

resulting microstructure of the hot pressed SHS and MM alumina/titanium

diboride composites: the initial state of the hot pressed powders (stoichiometry,

particle morphology and proximity), and densification variables (pressure,

temperature, and time at temperature).

The compositions to date have been based on a (nominal) stoichiometry of

30wt% TiB2 and 70wt% Al2O3 according to equation (1)

3 TiO2 + 3 B2O3 + 10Al => 3TiB2 + 5Al2O3 (1)

(Note: The SHS reaction produced a product that was (adiabatically) 29% TiB2

and 71% Al2O3; while the MM powders were mixed to be 30% TiB2 and 70%

Al2O3.) It is probable that the compositions used were close to a eutectic or


eutectoid composition since characteristic eutectic-like microstructures have been

observed in the SHS composites (figure 11).

Figure 11. Eutectic-like microstructure.

TiB2 (white areas) and Al2O3 (gray areas)

The crystal structure of titanium diboride is simple hexagonal close packed

(HCP) with a c/a lattice ratio of 1.07. The unit cell lattice parameters have been

reported as a = 3.03A, and c = 3.23A with the c/a lattice ratio remaining 1.07 from

25 C to 1200 C.9. The c/a lattice ratio of an ideal HCP unit cell is 1.63. Alpha-

alumina also has a hexagonal close packed crystal structure. The lattice parameters

are a = 4.76A, and c = 12.99A with a c/a ratio of 2.73. Grain morphology of a

polycrystalline, HCP crystalline ceramic material can vary in shape from equiaxed,

to hexagonal, to high aspect ratio lamellae. The number of faces and edges on a

grain, as well as the bond strength between the grains, will influence the ultimate

mechanical strength and behavior.

Prior observations have also shown that the microstructure of the SHS

composite titanium diboride is influenced by the rate of application of pressure to

the red hot, plastic product immediately after the reaction has occurred. A

comparison was made between the microstructure that was formed in-situ just

after the SHS reaction had occurred and the microstructures formed as pressure

was applied immediately after the reaction occurred when the product was still

red-hot and plastic. The applied pressures were 12.42MPa (1800psi), 17.25MPa

(2500psi), 20.70MPa (3000psi) and an explosively applied pressure. Figure 12 is

a micrograph of the microstructure formed in-situ after the SHS reaction has

occurred showing a heterogeneous localization of TiB2 (white area) in Al2O3

(gray area). The effect of application of pressure is shown in Figures 13-15: as

the rate of applied pressure is increased, the aspect ratio of titanium diboride

decreases.10


Ceramic materials generally show some increase in compressive strength with

an increase in strain rate loading. The compressive strength of TiB2 increases with

application of pressure.11

and increases significantly with strain rate loading by

exhibiting a reported HEL of 160 kbar at a shock stress of 240 kbar.12

A marked

strengthening of alumina is shown with increasing strain rate.13

The strain rate

during chip formation in metal cutting is determined to be 104 s

-1 so the process

would be expected to approach adiabatic conditions.14

m

m

Figure 12. In-situ foam Figure 13. 12.42 MPa applied pressure

ELECTRICAL RESISTIVITY CHARACTERIZATION OF

MICROSTRUCTURES

Since the properties of alumina/titanium diboride materials are microstructure

dependent, it would therefore be advantageous if a simple non-destructive

measurement could be used to determine the degree of microstructural bias.

Titanium diboride (TiB2) is an intermetallic compound that has very impressive

performance characteristics:15,16,17

it acts very much like a metal with an electrical

resistance comparable to that of copper at 1000oC. The conductivity of TiB2 is

approximately 10-55 micro-ohm-cm at 300-1200K. As in a metal, the resistivity

of TiB2 decreases with increasing temperature.18

Therefore, the connectivity and

percolation path of the titanium diboride will govern the electrical properties. 19

Test bars were cut to dimensions specified in MIL-SPEC 1942B from each

hot pressed, three-inch OD disk of the four biased microstructures A, B, C and D.

The bars sampled from the discs were used to obtain mechanical and electrical

property data. The quasi-static and dynamic mechanical property results have

been previously reported,20

so only electrical property results21

will be reported

here. Because the surface conductivity of the composite was poor, silver paint

was used to create a good connection to the Fluke multi-meter leads. The


m m

Figure 14. 20.70 MPa applied pressure Figure 15. Explosive pressure

resistance for each bar was recorded at two different times. Two readings were

taken to check the stability of the measurement and the consistency of the

measurement techniques.

Figure 16 is a summary of the electrical resistivity measurements. In general,

the average resistivity of the MOR bars that were cut from the hot pressed

manually mixed powders (0.213-4.137 ohm-cm) was lower than the average

resistivity of the MOR bars that were cut from the hot pressed SHS powders

(0.234-53.866 ohm-cm). The lower resistivity in the MM samples was probably

due to the relatively large titanium diboride grains as compared with the smaller

grains in the SHS sample. No significant affect on the resistivity could be

discerned between the composites with TiB2 segregated at the alumina grain

boundaries (0.213-0.336 ohm-cm) and the composites with TiB2 uniformly

distributed in alumina (0.234-0.266 ohm-cm). It was also found that a four hour

hot pressing hold time at temperature reduced the resistivity of both powder types:

MM (0.21-0.27 ohm-cm), SHS (0.23-0.40 ohm-cm). After the four-hour hold at

temperature, the SHS composite resistivity was comparable to that of the MM

composites. It was apparent that the TiB2 provided a path allowing reduction of

the overall resistivity.

SUMMARY

A. Processing variables significantly affect resulting microstructure and thus

performance of hot pressed SHS and MM powders.

B. Although Samples A, B, C and D showed tendencies towards a preferentially

biased microstructure allowing one to discern trends in microstructural effects on


performance, further research is necessary to determine the specific processing

parameters to produce a totally biased microstructure.

C. The dynamic behavior of composite alumina/titanium diboride is significantly

affected by the microstructural bias, including phase distribution and morphology,

formed during synthesis and processing.

D. In general, the average resistivity of the MOR bars that were cut from the hot

pressed manually mixed powders was lower than the average resistivity of the

MOR bars that were cut from the hot pressed SHS powders.

E. No significant affect on the resistivity could be discerned between the

composites with TiB2 segregated at the alumina grain boundaries (0.213-0.336

ohm-cm) and the composites with TiB2 uniformly distributed amongst the

alumina grains (0.234-0.266 ohm-cm).

F. After the four-hour hold at temperature, the SHS composite resistivity was

comparable to that of the MM composites.

G. Electrical resistivity measurements have the potential of being a non-

destructive means of screening candidate armor materials.

ACKNOWLEDGEMENTS

The author gratefully acknowledges support from the U. S. Army Research

Office Contract No. DAAG55-98-1-0454; Mr. Matthew Burkins at the U. S.

Army Research Laboratory for the ballistic test results; and the U. S. Army

TACOM, Warren MI, Contract No.DAAE07-95-C-R040.

REFERENCES

1 K. V. Logan, “Composite Ceramics,” Final Technical Report, USATACOM,

Warren, MI, Contract #DAAEO7-95-C-R040, November 1996. 2 Ibid 1.

3 K. V. Logan, “Process for Controlling the Microstructural Bias of Multi-Phase

Composites,” U. S. Pat. No 6,090,321, July 18, 2000. 4 K. V. Logan, “Process for Controlling the Microstructural Bias of Multi-Phase

Composites,” U. S. Pat. Notice of Allowability, Application No. 09/549,648,

March 12, 2002. 5 A. R. Keller and M. Zhou, “Effect of Microstructure on Dynamic Failure

Resistance of Titanium Diboride/Alumina Ceramics,” Journal of the American

Ceramic Society, to be published in 2002. 6 G. Kennedy, L. Ferranti, R. Russell, M. Zhou and N. Thadhani, “Dynamic High

Strain-Rate Mechanical Behavior of Microstructurally-Based Two-Phase

TiB2+Al2O3,” Journal of the American Ceramic Society, to be published in 2002. 7 Louis Ferranti Jr., “Processing and Characterization of Microstructurally Biased

Two-Phase Titanium Diboride/Alumina Ceramic (TiB2+Al2O3),” Masters


Thesis, School of Materials Science and Engineering, Georgia Institute of

Technology, Atlanta, Georgia, December 2001. 8 L.H. Van Vlack, pg 493 in Elements of Materials Science and Engineering,

Addison-Wesley Publishing Company, Reading, MA, 1985. 9 E. C. Skaar and W. J. Croft, "Thermal Expansion of TiB2," Journal of the

American Ceramic Society, 56 pg 45 [1] (1973). 10

K. V. Logan, G. R. Villalobos, and J. T. Sparrow, "Synthesis/Densification

Using SHS of Composite TiB2/Al2O3," presented at The First International

Ceramic Science & Technology Congress, Anaheim, California, 31 October - 3

November, 1989. 11

Z. Rosenberg, S. N. Brar, et al., "Shear Strength of Titanium Diboride Under

Shock Loading Measured By Transverse Manganin Gauges," presented at The APS

1991 Topical Conference on Shock Compression of Condensed Matter,

Williamsburg, VA, June 17-20, 1991, Elsevier. 12

D. P. Dandekar, "Effect of Shock Reshock on Spallation of Titanium Diboride,"

presented at the APS Topical Conference on Shock Compression of Condensed

Matter, Williamsburg, VA, June 17-20, Elsevier. 13

J. Lankford, "Compressive Strength and Microplasticity in Polycrystalline

Alumina," Journal of Materials Science, 12 791-796 (1977). 14

M. G. Stevenson and P. L. B. Oxley, "Experimental Investigation of the Influence

of Speed and Scale on the Strain-Rate in a Zone of Intense Plastic Deformation,"

Proc. Inst. Mech. Engr., 184, [31] 561-74 (1969-70). 15

W. P. Holbrook, ed., "Technical Data," Ceramic Source, 7, 269-369, (1991-

1992).16

D. Viechnicki, W. Blumenthal, et al., "Armor Ceramics - 1987," Proceedings of

the Third TACOM Coordinating Conference, Monterey, CA (1987). 17

D. P. Dandekar and P. J. Gaeta, "Extent of Damage Induced in Titanium

Diboride Under Shock Loading," pp.1059-1068 in Shock Waves and High Strain-

Rate Phenomena in Materials, Marcel Dekker, NY (1992). 18

K. P. Ananthapadmanbhan, P.V. Sreekumar, “Electrical, Resistivity of Plasma-

Sprayed Titanium Diboride Coatings,” Journal of Materials Science 28, [6],

1665-1658 (March 1993) 19

A. J. Moulson and J. M. Herbert, Electroceramics: Materials, Properties,

Applications. Chapman and Hall, 1990 20

Ibid 1,3,4 21

J. K. Phillips, K. V. Logan and R. Gerhardt, “Effects of Hot Press Parameters

on Microstructure and the Effects of Microstructure on Electronic Properties of a

70% Al2O3/30% TiB2 Composite,” Independent Research Report, Mate 4951-2-3,

Georgia Institute of Technology, 1996.


ELECTRICAL RESISTIVITY

0

10

20

30

40

50

60

70

80

M M

500

30

1

SHS

500

30

2

M M

500

150

3

SHS

500

150

4

M M

3375

30

5

SHS

3375

30

6

M M

3375

90

7

SHS

3375

90

8

M M

3375

150

9

SHS

3375

150

10

M M

5/5*

30

11

SHS

5/5*

30

12

M M

5/5*

150

13

SHS

5/5*

150

14

M M

5000

30

15

SHS

5000

30

16

M M

5000

150

17

SHS

5000

150

18

M M

5000

240

21

SHS

5000

240

22

M M

5/5*

240

24

SHS

5/5*

240

23

PRESSURE

HOLD TIM E

SAM PLE NO.

Resistivity (ohm-

0

20

40

60

80

100

120

Density (% theo

RESIST. 1 stat. avg. RESIST. 2 stat. avg. BAR DEN. % theo.

5/5*=500/5000

i

Figure 16. Summary of electrical resistivity measurements


PHASE EQUILIBRIUM STUDIES IN AL2O3-TIB2

Isabel K. Lloyd Kevin J. Doherty and Gary A. Gilde

Materials and Nuclear Engineering U.S. Army Materials Research Laboratory

University of Maryland Aberdeen Proving Ground

College Park, MD 20742-2115 Aberdeen, MD 21005

ABSTRACT

In this study, high temperature anneals were preformed on Al2O3-TiB2

mixtures containing 10, 20 and 40 mole % TiB2 to determine if the eutectic

reaction suggested by the microstructure of self-propagating high temperature

synthesis powders and thermodynamic calculations occurred. Energy dispersive

spectroscopy and X-ray diffraction of the annealed mixtures suggested that there

was a eutectic near 1925°C at a composition near 80 mole % Al2O3 and 20 mole

% TiB2. The melting behavior of the mixtures and the microstructures of the

annealed powder mixtures supported this conclusion.

INTRODUCTION

Al2O3-TiB2 has received some attention as a potential ceramic armor material

since it was expected to retain some of the hardness and stiffness of TiB2 while

being easier to process than monolithic TiB2. Interest in the system was piqued by

initial ballistic tests that suggested it exhibited significant resistance to high strain

rate penetration and mechanical properties tests that indicated its static mechanical

properties were similar to TiB2 [1]. Al2O3-TiB2 bodies can be made by hot-

pressing either mechanically mixed Al2O3 and TiB2 powders or composite

powders made by SHS, self-propagating high temperature synthesis, around

1600°C. The microstructure [2,3] of the SHS powders suggests that there may be

a eutectic between Al2O3 and TiB2. This conclusion is supported by the

microstructure of hot-pressed samples that experienced significant temperature

overshoots. This study explored the existence of a eutectic since an understanding

of the phase equilibrium in a system can aid in the development of processing

routes that produce tailored microstructures or that are amenable to large scale

manufacturing.

Before any experiments were done, the potential eutectic composition and

temperature were estimated from freezing point depression and liquidus surface



calculations. It was assumed that the liquid in the Al2O3-TiB2 system was an ideal

solution. Then, mechanically mixed powders were annealed in W foil packets in

Ar at temperatures between 1850 and 2070°C. After annealing the powder

mixtures were examined visually, optically, in the SEM with EDS (energy

dispersive spectroscopy) and back scattered electrons, and by XRD (X-ray

diffraction).

THERMODYNAMIC CALCULATIONS

Freezing point depression was used to estimate the eutectic composition

and temperature. First the freezing point depression for alumina as a function of

TiB2 addition was calculated assuming an ideal solution using equation 1 [4]:

ln XA=-( Hf/R) [(Tm-T)/( Tm * T)] (1)

where XA is the mole fraction TiB2, Tm is the melting point of alumina (2327 K

[5]), Hf is the enthalpy of fusion (-1675.7 kJ/mol [5]) and R is the Universal Gas

Constant. Then, the freezing point depression of TiB2 was similarly calculated.

Both liquidus curves were plotted as shown in Figure 1 and the temperature and

composition where the two liquidus curves crossed was taken as the estimate of

the eutectic composition and temperature.

The intersection of the liquidus surfaces for both Al2O3 and TiB2 was used

to confirm the estimate from freezing point depression. To calculate the liquidus

surfaces, the change in free energy for ideal solution as a function of temperature

from 2200 to 3100 K was first calculated using equation 2 [4]:

G = RT [X lnX + (1-X) ln (1-X)] (2)

where R is the gas constant, T is the temperature, and X is the mole fraction of

solute. Next, the difference in free energy between the solid and liquid was

calculated using equation 3 [4]:

Gs – Gl = - Hf ln (Tm /T) (3)

where Gs is the free energy of the solid, Gl is the free energy of the liquid and Hf

is the enthalpy of fusion for either Al2O3 or TiB2. A tangent to the G curve

through the Gs – Gl value was then drawn to estimate the liquidus composition at

that temperature. These values are also shown in Figure 1. The intersection of

both sets of curves occurs at about 80 mole % alumina and about 2250 K

(1977°C).

.


1800

2000

2200

2400

2600

2800

3000

3200

3400

0 0.2 0.4 0.6 0.8 1

X alumina

T (

K)

Al2O3 (FP dpress)

TiB2 (FP dpress)

Al2O3 (liquidus)

TiB2 (liquidus)

Figure 1: Estimate of eutectic temperature and composition

EXPERIMENTAL PROCEDURE

Mixtures of commercial Al2O3(Alcoa, A16) and TiB2 (Stark, Grade D)

powders with the compositions given in Table I were ball milled in ethanol for 20

hours and then dried under a heat lamp. After drying, large agglomerates were

crushed with a spatula and 2-5 g of mixed powder was placed in a loosely sealed

W foil packet. Then, a packet of each composition was placed in its own covered

graphite crucible and all three compositions were annealed in a graphite furnace

under the conditions in Table II.

After annealing the samples were examined visually and in a optical

microscope for signs of melting and reaction with the W foil. Then they were

examined using XRD with Cu K radiation to determine phase composition. The

strong alumina peaks (2 = 35.2, 25.6, 43.4, 66.5 and 68.3°) in the as-milled,

unannealed powders were used as standards. Quantitative comparison with these

peaks was used to determine the relative amounts of Al2O3 and TiB2 in the

annealed powders. The annealed powder was lightly coated with gold to prevent

charging before it was examined in the scanning electron microscope (SEM) using

secondary and back-scattered electrons and EDS.


Table I: Compositions used for eutectic studies

Al2O3 TiB2 Predicted Melting Behavior

mole% wt.% mole% wt/%

80 (85.4) 20 (14.6) predicted eutectic composition

90 (93) 10 (7) 50 m% eutectic L, 50 m% Al2O3

60 (70) 40 (30) 50 m% eutectic L, 50 m% TiB2

Table II: Heat Treatments

Run Temperature (°C) Time Atmosphere

1 1850 4 hours vacuum

2 1900 4 hours Ar

3 2070 15 min. Ar

4 1950 4 hours Ar

5 1925 4 hours Ar

6 1925 4 hours Ar


Evidence of melting was observed visually, under the optical microscope,

and in the SEM for samples annealed at 1925°C and above. No melting was

observed below 1925°C. More melting was observed in the 80 mole % alumina

mixtures. The microstructures were not uniform in the samples annealed

1925°C. In these samples, EDS indicated that the top of the powder bed was

highly deficient in alumina and that the bottom of the powder bed was less

alumina deficient. The 80 mole % alumina samples were more alumina deficient

than the other samples. These observations were supported by the quantitative

XRD results shown in Figure 2. Since alumina would be more volatile in the

liquid state than the solid state, this was taken as evidence that there was more

alumina in the liquid state in the 80 mole % mixtures. XRD indicated that Al2O3

and TiB2 were the major phases in all samples. Later experiments used an Ar

atmosphere since a number of minor phases were formed under vacuum in the

first experiment. There was no evidence of reaction with the W foil in the XRD

results.

The microstructures observed for both polished and as-annealed powders

were consistent with the proposed eutectic. Below 1925°C, the TiB2 grains were

angular and separated by an alumina matrix. Above 1925°C, as shown in Figure

3, the TiB2 grains were rounded and they tended to be more interconnected which

indicates melting and suggests the possibility of a eutectic.


0

10

20

30

40

50

60

70

80

90

100

1700 1800 1900 2000 2100

Anne aling T (°C, as re c 'd =1800)

80mole%alumina

(peak at 68.3°)80mole%alumina




(peak at 66.5°)

Figure 2: Semi-quantitative XRD results for 80 mole% alumina. Ratios were

calculated using different alumina peaks in the as-milled powder as standards.

Figure 3: Secondary electron image of an 80 mole % alumina powder

mixture annealed at 1950°C. The rounded dark grains are TiB2.


CONCLUSIONS

Thermodynamic calculations and annealing experiments in the Al2O3-TiB2

system from 60 to 90 mole % Al2O3 indicate a liquidus minimum above 1925°C

and a possible eutectic at about 80 mole % Al2O3. Additional experiments will be

needed to confirm these tentative conclusions. Given the high liquidus

temperature of the proposed eutectic, it is unlikely that eutectic processing would

offer any advantages for Al2O3-TiB2 composites compared to conventional hot-

pressing [1].

REFERENCES

1. K. V. Logan, “Composite Ceramics,” Final Technical Report A002, Army

Materials and Mechanics Research Center Contract DAAE07-95, Nov.

1996.

2. K.V. Logan and J.D. Walton, “Ti Formation Using Thermite Ignition,”

Ceram. Eng. Proc. 5 [7] 712-38 (1985).

3. L.J. Kecskes, A. Niiler, T. Kottke, K.V. Logan, and G.R. Villalobos,

“Dynamic Consolidation of Combustion Synthesized Alumina-Titanium

Diboride Composite Ceramics,” J. Am. Ceram. Soc. 79 [10] 2687-95

(1996).

4. C.F. Bergeron and S.H. Risbud, Phase Equilibrium Studies in Ceramics,

pp. 52-62, Am. Ceram. Soc. 1984.

5. JANAF Tables Al2O3, TiB2, J. Phys. Chem. Ref. Data, Monograph 9


MICROSTRUCTURE DEVELOPMENT OF ALUMINUM OXIDE/TITANIUM

DIBORIDE COMPOSITES FOR PENETRATION RESISTANCE

J.W. Adams, G.A. Gilde and M. Burkins



L. Prokurat Franks

U.S. Army Tank-Automotive and Armaments Command

Warren, MI 48397

ABSTRACT

Early research on aluminum oxide/titanium diboride (Al2O3/TiB2) composites

focused on exploiting their potential as a low cost armor ceramic. Limited

ballistic data demonstrated that the microstructure has a dramatic effect on

ballistic performance. With the "preferred" microstructure, the penetration

resistance of Al2O3/TiB2 approached that of monolithic TiB2 ceramics. Challenges

were encountered both in quantifying the microstructural detail and fabricating

the desired microstructure.

Our research focused on microstructure control during fabrication and

correlation of microstructure with mechanical properties and penetration

resistance of the composite. Composites were made from mixed Al2O3 and TiB2

powders, as well as a composite Al2O3/TiB2 powder prepared via a self-

propagating high-temperature synthesis (SHS) reaction. A summary of depth of

penetration ballistic analyses for several projectiles is given. Our results show that

although the penetration resistance of Al2O3/TiB2 composites is good, the results

fall within the expected experimental scatter shown by commercial state-of-the-

art armor ceramics.

BACKGROUND

A brief history of the interest in Al2O3/TiB2 composite materials shows that in

1982 the Army became aware of Soviet technology for Self-Propagating High

Temperature Synthesis (SHS) to produce TiB2. Within ten years fully dense SHS

Al2O3/TiB2 composites were produced at Georgia Tech.1 Ballistic evaluations of

those materials performed by the University of Dayton Research Institute showed

that there could be composite a possible correlation between microstructure and



ballistic performance. The composite structure that had TiB2 localized at the grain

boundaries of the aluminum oxide exhibited a high mass efficiency. In 1997

TARDEC contracted with the Army Research Laboratory for further study based

on the following factors:

•Promising new materials processing technology

•Ballistic test results showed potential

•Comparable performance to state-of-the-art armor ceramics, with

possible cost savings

•Potential Future Combat System (FCS) applications

In particular, the early Al2O3/TiB2 composites that were processed using

powders derived by self-propagating synthesis (SHS) and evaluated by long-rod

penetrators in Depth of Penetration (DOP) tests gave intriguing results. The

ballistic mass efficiencies were greater than expected from the rule of mixtures,

and were high enough to generate interest in these materials as potential armor

(see Figure 1.) 2,3

The reason for interest in the composite is twofold: 1) initial

screening indicated that the material may perform as well as, or equivalent to

titanium diboride (TiB2) armor ceramics at substantially lower cost, and


Figure 1. Early DOP ballistic data for L/D=13 rod at 1550 m s-1

against alumina-

titanium diboride ceramics.

2) Al2O3/TiB2 has higher space efficiency than can be achieved with silicon

carbide, as well as a mass efficiency that is almost equal to that of silicon carbide

against medium caliber threats.

The purpose of our work had several aspects. We wanted to explore other

powder processing and sintering routes to systematically determine and quantify

differences in microstructures, to evaluate the composites against small and

medium caliber penetrators and assess consistency with previous ballistic results,

and lastly, to correlate the microstructure to the ballistic properties.

EXPERIMENTAL

We fabricated composites according to several processes, characterized them

and performed DOP ballistic tests using three different penetrators in the course

of investigating this system. Details of the processing matrix, mechanical and

DOP evaluation methods and test results are given in Gilde et al.4 Processes used

in this study were:

• SHS Al2O3/TiB2 powders + ball milling + hot pressing (HP)

• Al2O3 and TiB2 powders + ball milling + HP

• Co-extrusion Al2O3 and TiB2 powders + HP

• Colloidal powders + sintering

• Al2O3 and TiB2 powders shaken to mix electrostatically (ESD) + HP

Our processing study maintained the nominal 75/25% composite ratio weight

ratio of Al2O3/TiB2 using various green powder processing routes to achieve

different microstructural textures. Density, hardness, 4-point flexure strength,

fracture toughness and fracture analyses were performed on all composites prior

to ballistic evaluation. DOP ballistic testing was conducted using 7.62 mm AP M2

(armor piercing) projectiles and L/D=10 tungsten alloy rods. However, the larger

L/D=13 tungsten alloy rods that had been used in the 1990s were not a part of this

study.

RESULTS

Figure 2 summarizes the results of the ballistic testing against the 7.62mm AP

M2 projectile and compares it to the ballistic performance of hot pressed silicon

carbide, boron carbide and a sintered aluminum oxide tested against the same

penetrator.5 As can be seen from the graph, the aluminum oxide/titanium diboride

composites performed slightly better than the sintered aluminum oxide, but were

less effective than silicon carbide and boron carbide. The composites made from

the SHS powders performed the best against this small projectile.


Penetr

ation

resis

tan

ce x

Are

aldensity, kg m

-2

0

20

40

60

80

100

120

140

0 5 10 15 20 25

Areal density, kg m-2

Al2

O3

: 123.12 + 13.590(1-e (0.11659)C

tC

) R2 = 0.9864

B4

C: 123.12 + 5.1494(1-e (0.3086)C

tC

) R2 = 0.98066

SiC: 123.12 + 25.113(1-e (0.13106)

C

t

C

) R2 = 0.97985

FM CP

SHSMM CP

Al2O3/TiB2 composites

Figure 2. Residual penetration areal density vs. ceramic areal density against the

7.62 AP M2 projectile at 841 m s-1

.

Ballistic properties of Al2O3/TiB2 composites impacted with a 131W tungsten

alloy rod at 1550 m/s as compared to other armor ceramics are presented in Table

2. Several preferred Al2O3/TiB2 microstructures were evaluated. Despite the

overlap in the ballistic data for aluminum oxide, titanium diboride and silicon

carbide, the average em follow the expected trend that silicon carbide is better than

titanium diboride, and titanium diboride performs better than aluminum oxide. It

Table 2. Ballistic properties of armor ceramics impacted with a L/D=13 rod at

1550 m s-1.

Sample em es q2

Commercial AD995 3.2 + 0.5 1.6 + 0.3 5.0 + 1.9

Commercial SiC 4.4 + 0.5 1.8 + 0.3 8.0 + 2.3

Commercial TiB2 3.9 + 0.3 2.2 + 0.2 8.7 + 1.6

SHS TiB2 around Al2O3 3.3 + 0.2 1.7 + 0.1 5.6 + 0.6

MM TiB2 around Al2O3 4.1 + 0.4 2.1 + 0.3 8.7 + 2.1

SHS TiB2 within Al2O3 2.5 + 0.1 1.3 + 0.1 3.4 + 0.2

MM TiB2 within Al2O3 3.3 + 0.6 1.7 + 0.4 5.5 + 2.7


can be seen that the manually mixed (MM) Al2O3/TiB2 composites have higher em

values than AD 995 aluminum oxide. The average em is higher than that of

titanium diboride and slightly less than hot-pressed silicon carbide. When both

space and weight are critical to the armor design, the es and q2 values indicate that

the Al2O3/TiB2 composite could have an advantage over titanium diboride armor

ceramics for an armor package designed against medium cal threats.

Ballistic properties of Al2O3/TiB2 composites impacted with the L/D=10

tungsten alloy rod at 1500 m s-1

as compared to other armor ceramics are

presented in Figure 2 and Table 3.

x

0

5

10

15

20

25

30

35

40

45

40 60 80 100 120 140 160 180

Ceramic areal density, kg m-2

Pe

ne

tra

tio

n,

mm

HP SiC 25mm

HP SiC 30mm

TARDEC GTRI

TARDEC ARL ESD

1992 GTRI UDRI 39MM

X HP TiB2

SiC

Al2O3/TiB2

xx

x

Figure 2. Penetration into RHA backing vs. ceramic areal density (25 mm

thickness) against the L/D=10 rod at 1500 m s-1.

Table 3. Ballistic properties of armor ceramics (25 mm thickness) impacted with

a L/D=10 rod at 1500 m s-1.

Material Density, gcm-3

em

Commercial AD995 3.6 2.4

Commercial SiC 3.2 4.2

Commercial TiB2 4.5 3.2

Al2O3/TiB2 (ESD) 4.1 3.1

All Al2O3/TiB2 composites 4.1 3.0


SUMMARY

Our investigation to assess Al2O3/TiB2 composites’ potential as an armor

ceramic demonstrated that distinctive microstructural textures can be developed

and controlled by a variety of processing methods. A systematic ballistic

evaluation was completed for small and medium caliber projectiles at velocities

ranging from ~850 m/s to 1500 m s-1

. All TiB2/ Al2O3 composite structures were

effective at defeating the projectile in all cases. For the 7.62 AP round, the

composites made from the SHS powder performed slightly better. In the case of

medium caliber long rod penetrators, SHS-derived composites did not offer any

advantage. Composites made from mixed Al2O3 and TiB2 powders performed

better. The process of mixing dry powders to electrostatically disperse the TiB2

around the Al2O3 grains resulted in composite structures that were as effective as

those that were ball milled for hours.

In order for a ceramic to offer attractive potential as armor, the material must

offer effective protection, and be manufacturable and affordableThe promise

based on early ballistic data and probable cost savings for Al2O3/TiB2 composites

has not been borne out in this study. Serious manufacturability issues, including

the lack of commercial SHS powder suppliers and little market pull for products

beyond armor for the titanium diboride plus alumina system, override the

estimates for favorable raw material cost/processing savings. Early anecdotal high

ballistic penetration resistance results were shown to be within the range of

expected DOP test variability.

REFERENCES 1 K.V. Logan, “Elastic-plastic Behavior of Hot-pressed Composite Titanium

Diboride/Alumina Powders Produced Using Self-propagating High-temperature

Synthesis,” PhD Thesis, Georgia Institute of Technology, 1992. 2 G. Abfalter, N.S Brar. and D. Jurick, “Determination of the Dynamic

Unload/Reload Characteristics of Ceramics,” University of Dayton Research

Institute, Dayton OH, June 1992, Contract No. DAAL03-88-K-0203. 3 P. Woolsey, D. Kokidko and S. Mariano, “An Alternative Test Methodology

for Ballistic Performance Ranking of Armor Ceramics,” MTL TR 89-43, U.S.

Army Materials Technology Laboratory, Watertown, MA, 1989. 4G.A. Gilde, J.W. Adams, M. Burkins, M. Motyka, P.J. Patel, E. Chin, L.

Prokurat Franks, M.P. Sutaria and M. Rigali, "Processing of Aluminum

Oxide/Titanium Diboride Composites for Penetration Resistance," Cer. Eng. Sci.

Proc., 22 (2001) 331-342.5 T. J. Moynihan, S. Chou, and A.L. Mihalcin, “Application of the Depth-of-

Penetration Test Methodology to Characterize Ceramics for Personnel Protection”

ARL-TR-2219, April 2000, Army Research Laboratory, Aberdeen Proving

Ground, MD 21005-5066.


THE EFFECT OF METAL-CERAMIC BONDING ON BALLISTIC IMPACT

Kevin J. Doherty

US Army Research Laboratory

Weapons Materials Research Directorate

AMSRL-WM-MC


ABSTRACT

Lightweight armor systems are crucial to the survivability of future Army

vehicles. The combination of ceramics and lightweight metals is a key element in

modern armor packages. The interface created when joining metals and ceramics

can have a significant influence on the behavior of the entire system. In this

study, the joining of SiC and Ti-6Al-4V plates was demonstrated using an active

solder, Sn-4Ag, containing ~4 wt% Ti. This configuration was compared with

plates joined using an epoxy. Preliminary ballistic evaluation and microstructural

analysis of the joints in the different armor systems will be discussed.

INTRODUCTION

The desire for smaller, lighter Army vehicles has motivated the need for

lightweight metal and ceramic armor systems. The process of fabricating an

armor package from lightweight metals and ceramics is complicated by the need

to bond very dissimilar materials both together as well as attaching these armor

packages to the vehicle structure. A typical joining method for ceramics-metals is

adhesive bonding. Joining with adhesives, such as epoxy, is convenient because it

is performed near room temperature, in air and is compatible with most materials.

The drawbacks to adhesive bonding are the resulting low bonding strength and the

low modulus. The combination of low modulus and low density creates a

substantial elastic impedance mismatch with the ceramic and metal substrates.

Other bonding options such as brazing and soldering typically have higher moduli

and higher densities that decrease the elastic impedance mismatch with the

ceramic and metal substrates in comparison with adhesives.

The desire for stronger bonding in metal-ceramic systems has led to the

examination of joining techniques that involve beneficial chemical reactions at the



metal-ceramic interfaces. During the early ‘80s, Mizuhara and coworkers [1,2]

adapted this idea from the ‘50s by putting an “active” component, such as

titanium (Ti), directly into a brazing alloy, typically a silver-copper eutectic, to

significantly improve the wetting of both metal and ceramic substrates. This

initiates a one-step vacuum brazing process that wets most materials (including

ceramics, Ti alloys and stainless steels) and forms strong, metallurgical bonds.

The major disadvantage in using “active” brazing for metals and ceramics is the

high processing temperature required that results in large strain (stress) build-up

from the inherent differences in coefficient of thermal expansion (CTE) between

metals and ceramics during cooling. There are some techniques available to

alleviate the strains on the ceramic, such as using an interlayer, which either has

an intermediate (between the metal and ceramic) value of CTE and/or is “soft”

(compliant). However, it is still extremely challenging to actively braze specimens

that are larger than 5 cm in diameter when there is a considerable CTE gradient.

Active solder joining is an emerging technology that incorporates many of

the ideas from active brazing. A reactive element (typically Ti) is added to a

solder alloy to enable direct wetting and bonding. Currently, two lead-free

systems are being investigated: Sn-Ag-Ti and Zn-Ag-Ti [3]. No chemical fluxes

are used, so mechanical agitation (such as brushing or ultrasonic vibration) is used

to disrupt the oxide naturally on the solder to promote wetting. The use of lead-

free solders, without chemical fluxes, creates an “environmentally friendly”

process that offers additional cost benefits by eliminating extra cleaning steps

associated with the fluxes. The active soldering process offers a compromise of

lower joining temperatures (<450ºC), reasonable elastic impedance, but little

improvement in strength over epoxy. The flexibility of this process allows the

joining of larger specimens with a significant mismatch in CTE, which offers the

chance to test the effect of elastic impedance and adherence of the bonding layer

on ballistic samples. Thus, this paper will examine the effect of active soldering

and epoxy joining for SiC and Ti-6Al-4V substrates and it will present some

preliminary ballistic results for the different bonding conditions.

EXPERIMENTAL

Square ballistic targets were assembled by joining hot pressed silicon

carbide (10.2 cm x 10.2 cm) to annealed Ti-6Al-4V (MIL-DTL-46077F, 15.2 cm

x 15.2 cm). The elastic and bulk moduli of the SiC plates were measured by the

pulsed excitation method (ASTM C1259) to verify the homogeneity within the

lot. The density of each plate was also determined to eliminate any anomalous

plates from the study. The SiC plates were used in an as-ground condition and the

Ti-6Al-4V pieces were machined to achieve flat and parallel surfaces. Four

different bonding conditions were investigated to join the SiC to the Ti-6Al-4V.


The first set of targets was bonded using a two-part epoxy (Epon resin 815 and

Versamid 125). The epoxy was applied to the mating surfaces and the targets

were cured for 72 hours at room temperature. The other three sets of targets were

prepared and bonded using the proprietary active solder S-Bond™ Alloy 220 from

Materials Resources International in Lansdale, PA. The S-Bond™ Alloy 220 is a

Sn-4Ag based alloy with 4 weight percent Ti. The first set of SiC and Ti-6Al-4V

plates were grit blasted prior to joining in air with the active solder at 250ºC. The

next set of SiC plates was pre-treated with the active solder in a vacuum furnace

at 850ºC for one-half hour and furnace cooled. The Ti-6Al-4V was grit blasted

and then bonded to the pre-treated SiC in air at 250ºC. The final set of SiC and

Ti-6Al-4V plates were all pre-treated in a vacuum furnace at 850ºC and furnace

cooled prior to final joining in air at 250ºC. An example of the ballistic targets is

shown in Figure 1.

Figure 1. Sample ballistic targets

Projectiles were fired at various velocities at the targets to establish a

modified V50 protection ballistic limit (MIL-STD-662E, V50 Ballistic Test For

Armor). Modifications from the standard procedure included using x-ray analysis

instead of a witness plate for the determination of partial penetration (PP) versus

complete penetration (CP). The targets bonded with epoxy were tested first to

establish the baseline V50 for comparison with the various soldered targets. While

the epoxy targets provided an initial testing velocity, there was still an insufficient

number of pre-treated solder targets to achieve a statistically significant V50. A

photograph of the ballistic setup is shown in Figure 2. The V50 ballistic limit may

be defined as the velocity at which 50% of the attacking projectiles may be

statistically expected to completely penetrate the target.

Optical metallography was performed on the impacted Ti-6Al-4V plates

and SiC fragments. In addition, scanning electron microscopy (with a Robinson

backscatter detector) was utilized to investigate the impacted interface structure.

Hardness measurements were taken from segments of the Ti-6Al-4V plates


following the different bonding procedures to determine any consequence from

the heating and cooling cycles.

Figure 2. Sample ballistic setup

RESULTS

The ballistic testing produced comparable V50 results for all of the

different bonding conditions. In all cases, the ballistic event produced extremely

fragmented SiC pieces detached from the Ti-6Al-4V plates. The Ti-6Al-4V plates

can be separated into two categories of ballistic damage. The first set of plates

has undergone PP (Figure 3) where the projectile did not penetrate the target. A

plastically deformed bulge in the center that is embedded with SiC rubble

characterizes these plates. In addition, the plates contain several cracks, but are

still intact. The second set was CP (Figure 4), characterized by slight bulge in the

center with a crack proceeding entirely through the center of the plates, from top

to bottom. All of the Ti-6Al-4V plates contain scratches and an adhered powdery

residue emanating from the center out to the edges.

SiC fragments were collected and characterized after the ballistic tests.

Typically, the fragments examined came from the corners and edges of the

ballistic targets. Optical microscopy was utilized to examine the SiC/solder

interface in the grit blasted and soldered condition. Only a small fraction of the

solder is still adhered to the SiC following the ballistic event (indicated by the

lighter regions in Figure 5). Optical images of the SiC fragments, joined by

soldering with a pre-vacuum treatment, are presented in Figure 6. For this joining

condition the solder is still adhered to the surface of the SiC. The raised regions

(islands) in Figure 6b constitute failure at or near the Ti-6Al-4V/solder interface,

while the recessed regions constitute failure within the solder. This behavior is


further demonstrated in a backscatter SEM image (Figure 7) of another SiC

fragment from the same ballistic event as Figure 6. The bonding of the solder to

the SiC is fully intact while the failure is both within the solder and at or near the

Ti-6Al-4V/solder interface.

Figure 3. Ti-6Al-4V plates after PP. Bonded with (a) epoxy, (b) solder, and (c)

solder with vacuum treatment.

Figure 4. Ti-6Al-4V plates after CP. Bonded with (a) epoxy, (b) solder, and (c)

solder with vacuum treatment.

Figure 5. Optical images of post-ballistic SiC, grit blasted and bonded with solder.


Hardness measurements taken from the Ti-6Al-4V plates after the epoxy

or grit blast/solder bonding produced results consistent with the standard MIL-

DTL-46077F (33.0-33.5 HRC). With the addition of a vacuum treatment to

promote improved bonding, the softening of the Ti-6Al-4V to 32 HRC was

measured and attributed to the slow furnace cool.

Figure 6. Optical images of post-ballistic SiC, soldered with vacuum treatment.

Figure 7. SEM backscatter image, post-ballistic SiC soldered, vacuum treatment.

DISCUSSION

Using active soldering offers a new alternative for joining large samples of

materials where there is a significant CTE mismatch, such as SiC and Ti-6Al-4V.

The obvious contrasts between epoxy and the Sn-Ag-Ti alloy offers an excellent

opportunity to explore both material property differences (Table I) and bonding

differences as they relate to ballistics. The increased elastic modulus and density

of the active solder results in a tenfold increase in the elastic impedance over the

epoxy. This increase brings the elastic impedance more in line with the SiC,

decreasing the impedance mismatch, and causing less reflection of the stress wave

back into the SiC during a ballistic event. The different processing routes used for

active soldering of the SiC and Ti-6Al-4V also allows for a variation in bond

strength. Active soldering without a thermal treatment provided the lowest bond


strength of all the techniques. In the epoxy and the grit blast/solder cases, the

bond strength comes exclusively from van der Waals forces, which is verified in

the post-impacted images. The thermal treatment improves adherence of the

solder to the SiC resulting from chemical reactions between the titanium in the

solder and the SiC. Thermal treatment of the substrates prior to active soldering

can improve the bond strength by 40% [4]. However, even with the variations in

elastic impedance and bond strength among the different joining techniques, the

ballistic results show no difference in performance in any of the configurations.

Table I. Typical properties for joining materials and substrates***^^^"

MaterialDensity

(g/cm3)

Elastic

Modulus

(GPa)

Long.

Velocity"

(km/s)

Elastic

Impedance"

(kg/m2s)*10

7

CTE

(m/m°C)

[RT-x°C]

Tensile

Strength

(MPa)

Bonding

Temp.

Epoxy* Epon 815 1.1 3-4 1.7 0.2 73 <100°C

Solder**S-Bond™

Alloy 2207.4 50-56 2.7 2

19*10-6

(250°C)53 250°C

Ceramic^Hot Pressed

SiC3.2 455 11.9 3.8

4.5*10-6

(1000°C)

Metal^^Annealed

Ti-6Al-4V4.4 114 5.1 2.3

9.7*10-6

(650°C)

Grit blasting of the SiC surface is the initial attempt to improve the

adherence of the solder to the SiC. Typically, grit blasting adds surface roughness

that increases bond area and introduces mechanical interlocking. However, in

post-ballistic analysis the adherence is limited and is generally confined to the

depressions in the SiC surface (Figure 5). The thermal treatment of the SiC prior

to active soldering leads to a substantial improvement of the adherence of the

solder to the SiC (Figure 6) which relates to an increase in the shear strength of

the bond. Such an improvement can bring the shear strength of the bond above

the tensile strength of the solder. This is evident in Figures 6 and 7 where the

failure of the bond occurs within the solder and near the solder/Ti-6Al-4V

interface instead of at the SiC/solder interface.

The thermal treatment did not have the same remarkable effect with the

Ti-6Al-4V as it did on the SiC surface. Active soldering of grit blasted Ti-6Al-4V

* Epoxy information from Resolution Performance product data sheets.

** Solder information from Materials Resources International product data sheets.

^ SiC information from Cercom, Inc. product data sheets.

^^ Ti-6Al-4V information from TIMET product data sheets.

" Longitudinal velocity and elastic impedance values calculated from the above data.


is still a van der Waals type bond. The thermal treatment of the Ti-6Al-4V did

not induce the level of reaction between the active titanium in the solder with the

Ti-6Al-4V surface that was observed in the SiC. This behavior occurred because

the active titanium most likely reacted more with the protective oxide on the Ti-

6Al-4V surface and reacted less with the base metal.

Even though there was not a correlation between ballistic results and better

elastic impedance or improved bond strength, some basic questions are still

relevant. Were the changes in joint properties too insignificant to effect the

ballistics? Would more substantial changes in strength and/or elastic impedance

correlate with better ballistic performance? The use of higher temperature active

solders and brazes should allow experiments to be performed to answer these

questions. However, some basic material changes may be required in order to

diminish the higher strains (stresses) associated with the higher temperature

joining. In addition, quantification of joint shear strength for all of the different

bonding techniques is required to directly relate the bonding and the ballistics.

SUMMARY

• Active soldering is a viable option for joining large pieces of ceramic and

metal where there is a substantial coefficient of thermal expansion mismatch.

• No relationship was observed between ballistic performance and either

bond adherence or elastic impedance in this set of experiments.

• A vacuum thermal treatment prior to joining improves the solder

adherence to the SiC significantly, but has a small effect on the adherence of the

solder to Ti-6Al-4V.

• Higher temperature active solders and brazes allow a larger variation in

bond strength and elastic impedance to further test the relationship between

bonding and ballistics.

ACKNOWLEDGEMENTS

The author would like to thank Dr. Ernest Chin and Dr. Joseph Wells for

reviewing the manuscript and discussing the research.

REFERENCES 1H. Mizuhara and K. Mally, “Ceramic-to-Metal Joining with Active

Brazing Filler Metal,” Welding Journal, 64 [10] 27-32 (1985). 2H. Mizuhara and E. Heubel, “Joining Ceramic to Metal with Ductile

Active Filler Metal,” Welding Journal, 65 [10] 43-51 (1985). 3R.W. Smith, “Active Solder Joining of Metals, Ceramics and

Composites,” Welding Journal, 80 [10] 30-35 (2001). 4R.W. Smith, Personal Communication, October, 2001.


ASPECTS OF GEOMETRY AFFECTING THE BALLISTIC PERFORMANCE

OF CERAMIC TARGETS

I M Pickup, A K Barker, R Chenari and B J James


Chobham Lane, Chertsey

Surrey, KT16 0EE, UK.

V Hohler, K Weber and R Tham

Faunhofer-institut fur Kurzzeitdynamik (EMI)

Eckerstrasse 4,

79104 Freiburg, Germany

ABSTRACT

Some ceramic armour configurations have the ability to erode kinetic energy

long rod penetrators on the impact surface either totally or partially before

subsequent penetration. This phenomenon (often called dwell) can result in very

high ballistic efficiency. The occurrence of dwell is very sensitive to subtle

changes in experimental conditions leading to extreme variation in performance.

The effect of some geometrical parameters such as obliquity, target thickness and

impact surface configuration on ballistic performance of silicon carbide is

assessed at impact velocities ranging from 1400 to 1800 ms-1

. Significant benefits

in ballistic performance may be realised by addressing impact surface and

ceramic back surface configurations to maintain reproducibly high ballistic

performance.

INTRODUCTION

Ceramic armour is capable of exhibiting very high ballistic efficiencies

against kinetic energy (KE) long rod penetrators. Some non-oxide ceramics such

as boron carbide and silicon carbide have a high potential to cause the KE rod to

dwell at the impact surface, i.e. the rod may be eroded at the impact surface

without penetration. This is due to the relatively low density and the particularly

high initial deviatoric strength of ceramics under hydrostatic pressure and in some

cases their relatively slow damage kinetics [1,2]. According to the exact nature of

the experimental targets, KE rods have been totally eroded at the impact surface at



velocities in excess of 1600 ms-1

[3]. The ballistic efficiency of ceramic armours

can be very significantly increased if such dwell is harnessed reproducibly.

The dwell phenomenon is difficult to quantify. However, there has been some

good experimental measurement of dwell and subsequent KE penetration rates of

slender KE rods into various ceramics by Lundberg et al [4-6] and Orphal and

Franzen [7,8]. This was achieved using multiple flash X-ray photography of

reverse ballistic shots in which the target is fired at the rod. This allowed the

assessment of dwell period as a function of impact velocity for this very specific

experimental situation.

To allow design of improved armour which utilises dwell, it is necessary to

understand what promotes and terminates dwell in a wide range of geometrical

configurations. This precludes the use of high intensity flash X-ray on the grounds

of cost. The work presented here is the initial part of such a study. At this stage

the experimental techniques for determining dwell parameters directly are not yet

mature. Consequently other measurands are employed to evaluate the armour

configuration effects. These are based on a comparison of residual depth of

penetration (DOP) into armour steel back blocks with an estimated penetration in

the absence of dwell.

The ceramic used in this programme is silicon carbide, SiC PAD-B,

manufactured by Cercom Inc., USA. UK specification RHA steel back blocks

were used for DOP measurement. The parameters examined here are:

i) Obliquity. Normal impact and 60 impact angles are compared for

equivalent line of sight (LOS) ceramic thickness.

ii) Velocity. KE rod velocities ranging from 1200 ms-1

to >1800 ms-1

were

employed.

iii) Ceramic thickness. Two LOS thicknesses were used, 30 and 40 mm.

iv) Impact surface configuration. Experiments were conducted with and

without a front coverplate.

EXPERIMENTS

Flat ended tungsten alloy rods (Plansee, Densimet FNC, density = 17600

kgm-3

) with an aluminium flare were fired from a 40 mm smooth bored gun using

a base pushed launch assembly with a three part sabot. The rods were 5 mm

diameter and 100 mm in length. A gun muzzle to target distance of 10 m was used

with 2 pairs of flash X-ray heads positioned 0.1 m and 0.5 m from the target to

monitor rod velocity, pitch and yaw.

All targets were laterally confined using steel adjustable clamps. Annealed

brass inserts were used as an interface between the ceramic and the steel

confinement frame. This was to ensure excellent mechanical contact and

consequently to improve the acoustic impedance match between the confinement

and the ceramic.


The normal impact targets and the 60 obliquity targets had lateral dimensions

of 100 x 100 mm and 100 x 200 mm respectively. Front and back surfaces were

ground flat and parallel to 0.01 mm, as were the steel DOP backblocks. The depth

of penetration was assessed by machining the backblocks to determine the

maximum penetration. The cover plate system, where used, was based on a

system used by Hauver et al [3] to accommodate dwell by allowing the possible

lateral spread of rod material as it dwells on the ceramic surface. It consisted of 5

mm of RHA, 2 mm of copper and 1.5 mm of graphite.

RESULTS

Tabulated results are presented in Table I. These identify the experimental

configuration, impact velocity, resolved yaw and measured depth of penetration

for all the shots. The system ballistic mass efficiency, Em, calculated as in

Equation 1, where AD is areal density, is also tabulated.

ceramicDOPRHA

ArefernceRHm

ADAD

ADE (1)

The effect of obliquity on DOP over a range of velocities is presented for 40

and 30 mm thick ceramic targets in Figures I and II respectively. Figures III and

IV show the effect of target thickness for normal and oblique rod impact

respectively. The effect of cover plate addition for both thickness targets is

presented in Figures V and VI.

DISCUSSION

Two methods are employed to provide a reference DOP vs. velocity curve

which represents the DOP after penetration through the ceramic if no penetrator

surface dwell occurred. No direct methods are available so the following

approximations have been made. The first was calculated using steady state

penetration data based on flash X-ray photography [4] from impacts of the same

rod and silicon carbide material and the same velocity regime as in the

experimental programme presented here. Assumptions were made that the same

penetration rates would be applicable to the 5 mm rod, 20:1 aspect ratio (the rods

from reference 4 were 2 mm diameter and 40:1 aspect ratio), and that the tail

velocity was reduced by an amount equal to twice the particle velocity in the rod,

each time an elastic rebound reached the free surface of the tail. Using these

assumptions, tip and tail velocities were calculated as a function of penetration

into the ceramic and the residual length of the rod and its velocity upon entering

the RHA DOP block were estimated. The residual penetration into RHA from this

starting point was calculated by numerical simulation using the Lagrangian code

ELFEN. Johnson-Cook models were used for the rod and for the RHA backblock.

This reference curve is identified in the Figures as ‘no-dwell calculation’.


TABLE I. Experimental results

Shot ID Target Type Obliquity Yaw Impact Residual Em

Velocity DOP

( ) ( ) (m/s) (mm)

3094 0 2.1 1593 83.0 0.97

3095 0 2.0 1756 97.0 1.01

3096 0 1.8 1463 69.0 0.97

3098 0 0.6 1200 37.0 1.06

3103 0 2.0 1833 104.0 1.02

3199

RHA

reference shots

0 1.6 1416 61.9 1.00

3204 0 1.6 1630 31.1 1.54

3205 0 1.7 1560 0.0 3.25

3206 0 2.1 1770 0.0 4.19

3315 0 0.2 1708 34.7 1.59

3207 0 - 1717 9.6 2.81

3265 0 2.5 1691 17.1 2.23

3251

40mm SiCB

with coverplate +

75mm RHA

backblock

0 1.5 1802 16.0 2.58

3203 0 0.7 1355 0.0 2.83

3317 0 0.3 1582 22.5 1.89

3319 0 2.5 1703 29.3 1.88

3318 0 0.8 1814 17.9 2.77

3389 0 0.6 1780 24.0 2.30

3387

30mm SiCB

with coverplate +

75mm RHA

backblock

0 0.4 1464 0.0 3.42

3320 0 1.38 1350 17.00 1.88

3316 0 0.24 1467 14.70 2.49

3314 0 0.65 1574 19.20 2.49

3262 0 1.0 1588 28.4 1.97

3258

30mm SiCB

No coverplate +

75mm RHA

backblock0 2.0 1702 36.1 1.90

3211 0 1.9 1709 16.5 2.82

3376 0 0.9 1590 28.8 1.77

3377 0 0.2 1668 37.76 1.63

3381

40mm SiCB

No coverplate +

75mm RHA

backblock 0 3.6 1788 29.5 2.20

3209 60 1.1 1631 16.6 1.87

3210 60 1.7 1784 25.8 1.85

3244 60 1.5 1689 15.7 2.04

3245

20mm SiCB

No coverplate +

50mm RHA

backblock 60 3.6 1549 0 2.65

3266 60 4.1 1563 20.8 1.71

3253 60 3.2 1672 24.7 1.80

3261 60 5.4 1506 20.1 1.60

3388 60 0.1 1783 35.8 1.67

3386

15mm SiCB

No coverplate +

50mm RHA

backblock60 0.1 1396 0.0 2.43


Figure I. 40mm SiC, normal and

oblique impact

0

20

40

60

80

100

120

1000 1200 1400 1600 1800 2000


Re

sid

ual D

OP

into

RH

A(m

m)

RHA reference

CPS/40SiC-B, 0 deg

CPS/40SiC-B, 60 deg

Em=1.3

No-dwell calc.

Figure II. 30mm SiC, normal and

oblique impact

0

20

40

60

80

100

120

1000 1200 1400 1600 1800 2000


Re

sid

ual D

OP

into

RH

A(m

m)

RHA Reference

CPS/30 SiC-B, 0 deg

CPS/30 SiC-B, 60 deg

Em=1.3

No-dwell calc.

Figure III. 30mm & 40mm SiC,

normal impact

0

20

40

60

80

100

120

1000 1200 1400 1600 1800 2000


Re

sid

ual D

OP

into

RH

A(m

m)

RHA Reference

CPS1/30SiCB, 0 deg.

CPS/40SiC-B, 0 deg

EM=1.3 (30mm)

EM=1.3 (40mm)

Figure IV. 30mm & 40mm SiC,

oblique impact

0

20

40

60

80

100

120

1000 1200 1400 1600 1800 2000


Resid

ualD

OP

in

to R

HA

(mm

)

RHA Reference

CPS/30SiC-B, 60 deg

CPS/40SiC-B, 60 deg

EM=1.3 (30mm)

EM=1.3 (40mm)

Figure V. 40mm SiC, normal impact.

Effect of coverplate

0

20

40

60

80

100

120

1000 1200 1400 1600 1800 2000


Re

sid

ual D

OP

into

RH

A(m

m)

RHA Reference

CPS/40SiC-B, 0 deg

NCP/40SiC-B, 0 deg

Em=1.3

Figure VI. 30mm SiC, normal impact.

Effect of coverplate

0

20

40

60

80

100

120

1000 1200 1400 1600 1800 2000


Re

sid

ual D

OP

into

RH

A(m

m)

RHA Reference

CPS1/30 SiC-B, 0 deg

NCP/30 SiC-B, 0 deg

Em=1.3

CPS = Cover plate system

NCP = No cover plate


The second reference curve which represents minimal or no rod dwell on

silicon carbide is based on ballistic penetration shots on a second silicon carbide

SiC-100. The shots were performed as an integral part of this current programme

using exactly the same experimental configuration. The targets had no cover

plate. The DOP results from this material yielded a reasonably constant Em of 1.3

across the velocity regime. Previous shock studies on this material [1,2] have

indicated that, on impact, the initial deviatoric strength of this material is

significantly lower than that for SiC B. The quasi-static strengths of the two

materials are almost identical and the density is very similar; 3163 and 3217

kgm-3

for SiC-100 and SiC B respectively.

It is believed that the difference in DOP between SiC100 and SiC B impacted

under identical conditions (compare Em =1.3(30mm) line and the SiC B with no

coverplate data, Figure 6.) is due to the degree of dwell, with SiC-100 exhibiting

little or no dwell. This reference DOP is marked on Figures I-VI as Em =1.3 and

the resulting DOP is calculated for either 30mm thick tiles or 40mm thick tiles.

From Figures I and II it is apparent that the two estimates of zero dwell

penetration estimates do not coincide. It is interesting to note that the ‘no-dwell’

calculation indicates a velocity at which zero penetration terminates is coincident

with the experimental data for both 30mm and 40 mm thick targets. However, the

curve has a very steep slope compared to the Em =1.3 estimate and the

experimental data sets. This may be due to the fact that the penetration rate

measured in reference 4 was measured post-dwell and this could be substantially

different to the penetration rate where no dwell occurred. It was felt that this

estimate is not accurate and the Em=1.3 reference from SiC100 data was adopted

for further comparison.

The residual DOP’s for normal impact and 60 impact angle are compared in

Figure I for a ceramic line-of-sight thickness of 40mm. The reference penetration

of the rod into RHA is shown as a thick solid line. The normal impact data

exhibited a large degree of scatter, with some extremely high Em results (4.4), for

which zero DOP values were measured at high velocities (1770 ms-1

) and some

low Em results (1.6), giving 31 mm DOP at an impact velocity of 1630 ms-1

.

The scatter is much reduced for the oblique targets with Em’s ranging from

1.87 to 2.64. In addition a linear relationship with velocity (correlation coefficient

0.94) is observed for the oblique impact, whilst little correlation is seen for the

normal impact targets. The scatter for the normally impacted targets is broadly

distributed around the oblique impact data suggesting a similar underlying

relationship. When the thickness of the ceramic is reduced to 30 mm both normal

and oblique results follow a very similar trend. Excluding zero penetration values

the Em’s for the normal targets were 2.27±0.44 and for the oblique were 1.69 ±

0.09. At the highest impact velocities the normal 30mm targets deviated from the

oblique producing lower DOP values. It would appear that obliquity does not

offer improved ballistic resistance but tends to reduce the extreme results at both


low and high efficiency. Similarly when the target thickness was reduced from 40

mm to 30 mm extreme behaviour was reduced.

It should be noted that for an equivalent LOS thickness at 60 the tile

thickness is halved. One reason for this study was to examine the effect of stress

wave release paths on ballistic performance. Ceramics have high deviatoric

strength under high hydrostatic pressure. When this pressure is released ballistic

performance is reduced. When an oblique target is struck the compressive stress

pulse travels radially from the impact sight but the release will travel back

normally from the back surface of the tile (assuming poor transmission into the

RHA). This means that the release path for the oblique target is half that of the

normal target. The similarity of the normal and oblique results for the thinnest

target (30 mm) would seem to indicate this effect is not dominating the results.

The effect of thickness is plotted in Figures III and IV. Even though there is a

wide distribution in results for the normal impact targets of 40 mm thickness there

is a clear difference between the 30 and 40 mm results. The nominal ‘zero-dwell’

reference lines (Em =1.3, 30 and 40 mm) indicate the difference in DOP that

would result from a purely 10 mm path difference for a constant Em. The 40 mm,

normal impact targets can offer significantly increased performance over the

30mm, exceeding that of purely path difference effects. The possible

improvement is not so great for the oblique targets, even so, there is still an

improvement over and above the path length difference, particularly at lower

velocities. The difference in ballistic performance between 40 and 30 mm normal

impact targets suggests that there are ceramic back surface effects which can

reduce the chances of attaining the extremely high Em’s that SiC B is capable of.

It is difficult to attribute quantitative differences in ballistic performance to

front or back surface effects. The effect of using an impact surface cover plate on

normal incidence targets was investigated for the two thicknesses of ceramic,

Figures V and VI. The cover plate used had a graphite layer adjacent to the

ceramic, used to allow radially spreading rods that were undergoing dwell to

continue easily to spread. For both 40 and 30 mm ceramic targets there appears to

be an increase in ballistic efficiency using the cover plate. For the 40 mm normal

incidence targets there was an apparent improvement in ballistic efficiency at

higher impact velocities. For the 30 mm targets there were improvements in

performance at both high and low velocities. At this stage it is not clear how the

configuration of the cover plate affects the dwell characteristics. In part it may be

due to the reduction of shock impact effects on the surface. It is also possible that

the system promotes dynamic axial confinement of the impacted ceramic surface

by channelling the eroding rod material. Further experiments are continuing with

different configurations.


CONCLUSIONS

The factors which promote dwell in silicon carbide are very sensitive to slight

changes in the experimental conditions, resulting in a large degree of statistical

scatter in the determination of ballistic mass efficiency. The effect of obliquity,

thickness and impact surface configuration have been investigated for long rod

impact velocities ranging from 1450 to 1850 ms-1

. Significant benefits in ballistic

performance may be realised by addressing impact surface and ceramic back

surface configurations to maintain reproducibly high ballistic performance.

ACKNOWLEDGEMENT

The work reported in this paper was funded jointly by the UK Government

Corporate Research Programme and by the German Government and was

performed under the auspices of a UK-German collaborative research project.

REFERENCES

1. I. M. Pickup and A. K. Barker, ‘Damage Kinetics in silicon carbide’, in Shock

Compression of Condensed Matter-1997, edited by S.C. Schmidt et al , AIP

press, pp513-516, 1998.

2. I. M. Pickup and A. K. Barker, ‘The deviatoric strength of silicon carbide

subject to shock’, in Shock Compression of Condensed Matter-1999, edited

by M.D. Furnish et al , AIP press, pp573-576, 2000.

3. G. E. Hauver, P. H. Netherwood, R.F. Benck and L. J. Kecskes, ‘Ballistic

performance of ceramic targets,’ Army Symposium on Solid Mechanics,

Plymouth, Massachusetts US, 1993

4. P. Lundberg, R. Renstrom and B Lundberg, ‘Impact of metallic projectiles on

ceramic targets: transition between interface defeat and penetration, Int. J.

Impact Engng., 24, pp259-275, 2000.

5. L. Westerling, P. Lundberg and B Lundberg, , Int. J. Impact Engng. (in

Press),2001.

6. P. Lundberg, R. Renstrom and L Holmberg, ‘An experimental investigation of

interface defeat at extended interaction time’, 19th

International Symposium of

Ballistics, Interlaken, Switzerland, pp1463-1469, 2001.

7. D. L. Orphal and R.R. Franzen, A. J. Piekutowski and M.J. Forrestal,

‘Penetration of confined aluminium nitride targets by tungsten long rods at at

impact velocities from 1.5 to 4.5 kms-1

, Int. J. Impact Engng.,18, pp355-368,

1996.

8. D. L. Orphal and R.R. Franzen, ‘Penetration of confined silicon carbide

targets by tungsten long rods at impact velocities from1.5 to 4.5 kms-1, Int. J.

Impact Engng., 19, pp355-368, 1997.



3-D finite element analysis,317

Acoustic impedance, 33Adams, J. W., 629Adams, Marc A., 139Aghajanian, M.K., 527Agrawal, Dinesh, 587Ajayan, Pulickel M., 551Alumina, 63, 83, 91, 103,

185, 233, 269, 441, 463,511, 551, 611, 623, 629

Alumina-zirconia, 91Aluminum nitride, 151Aluminum oxynitride, 573,

587Anderson, Charles E., Jr.,

485Applications, 3

Ballistic performance map(BPM), 139

Bar impact, 225Barker, A.K., 643Bless, Stephan J., 197, 225Bonding, 635Boride, 73Boron carbide, 73, 151, 269Burkett, M.W., 385Burkins, Matthew S., 53, 629

Carbon nanotube, 551Ceramic to metal bonding,

635Chang, Sekyung, 551Chang, Soon Nam, 261, 429Chen, W., 217Chen, Z., 329Chenari, R., 643Cheng, Jiping, 587Chhabildas, Lalit C., 233,

269Cimpoeru, S.J., 361Coated fabric, 541Coating, 541Composite, 73, 185, 551,

611, 623, 629Compressibility, 249Compressive fracture, 197

Compressive layers, 499Computational modeling,

299, 309Cort, G.E., 385Corundum, 463Cost reduction, 451

Damage assessment, 441Damage mechanisms, 557Damage models, 281Dandekar, D.P., 249, 269Danforth, S.C., 473Depth of penetration (DOP),

83, 165, 361Design, 3, 33, 473, 511Doherty, Kevin J., 623, 635Doremus, Robert H., 551Dwell, 113, 173, 309, 557Dynamic fracture, 185Dynamic indentation, 261

Erim, Zeki, 103Ernst, Hans-Jürgen, 23Espinosa, Horacio D., 349

Fabrics, coated, 541Failure mechanism, 103Failure model, 371Fiber, fabric 541Fine grained alumina, 463Finite element analysis, 337,

349Flexible ceramic, 541Forrestal, M.J., 217Fracture mechanics, 185Fragmentation behavior, 103Franks, L. Prokurat, 629Frew, D.J., 217Fused deposition of ceramics,

473Future Combat System, 3Future direction, 421

Gadow, Rainer, 541Galanov, B.A., 73Geometry, 643Gilde, Gary A., 573, 595,

623, 629Glass, plates, 329

Gooch, William A., Jr., 3, 53,113

Grady, Dennis E., 233Grain level analysis, 349Green, William H., 441Grigoriev, O.N., 73Grove, David J., 299, 371

Hbaieb, K., 499High-density ceramic, 45, 53Historical developments, 421Hohler, V., 643Holmquist, Timothy J., 299,

309

Impact surface configuration,643

Impact testing, 113Impact, high-velocity, 23Indentation damage, 429Infra-red windows, 595Interface defeat, 173, 309Isaacs, Jon B., 511Ivanov, S.M., 73

James, Bryn, 33, 165, 643Johnson, Gordon R., 309Joining, 635

Kanel, G.I., 197, 329Kartuzov, V.V., 73Kim, Chang Wook, 261, 429Kim, Do Kyung, 261, 429Kim, Young-Gu, 261Kobayashi, Albert S., 185Kolsky bar technique, 217,

261Konduk, B.A., 103Krell, Andreas, 83, 463

Laminar ceramics, 499Lange, F.F., 499Lanz, W., 63LaSalvia, J.C., 557Layered manufacturing, 473Leavy, Brian, 299Lee, Chul-Seung, 261, 429Lexow, B., 83Lightweight armor, 485

KEYWORD AND AUTHOR INDEX


Lischer, David W., 511Lloyd, Isabel K., 623Logan, Kathryn V., 611Long rod penetration, 151,

385Long rod penetrator, 23Lundberg, Patrik, 173

Manufacturing, 91, 451, 473 Marchand, A.H., 385Mashimo, Tsutomu, 233Matthewson, M.J., 473McCuiston, R.C., 473McMeeking, R.M., 499Mears, J., 527Medvedovski, Eugene, 91Membranes, 511Metal-ceramic bonding, 635Meyer, Hubert W., Jr., 299Microcracking diffusion, 329Micro-cracks, 403Micro-mechanisms, 403Microstructure, 349, 557,

611, 629Microwave sintering, 587Modeling, 317, 329, 337,

349, 361, 371, 557Models, comparison of, 299Models, damage, historical

perspective of, 281Molinari, Jean-Francois, 317Morgan, B.N., 527

Nanopowder, alumina, 551Nanotube, carbon, 551Nemat-Nasser, Sia, 403, 511Niesz, D.E., 473Nitride, 73Nondestructive testing, 441Normandia, Michael, 113

Obliquity, 643Orphal, D.L., 151Overview, 3

Palicka, Richard, 53Parker, R., 385Patel, Parimal, J., 573Patterson, Mark C.L., 595Penetration mechanism, 385Penetration model, 337Peron, Pierre-François, 45

Phase equilibrium, 623Pickup, I.M., 643Plane shock wave loading,

249Plastic deformation, 197Polycarbonate, 573Polyurethane, 573Porous silicon nitride, 63Protection areal density

(PAD), 139

Radome, 595Rajendran, A.M., 281, 371Rajendran-Grove model, 371Rao, M.P., 499Rapacki, E.J., 249Razorenov, S.V., 329Reaction bonded silicon car-

bide, 527Reinforcement, 551Reinhart, William D., 233,

269Renström, René, 173Roy, Don W., 595Roy, Rustum, 587Rupert, Nevin L., 441

Sapphire, 233, 573Sarva, Sai, 403, 511Schadler, Linda S., 551Sennett, Michael, 551Shear strength, 249Shear, 557Shen, L., 329Shock compression, 233Shock wave loading, 197Shockey, Donald A., 385Siegel, Richard W., 551Silicon carbide, 63, 73, 151,

269, 309, 441, 527, 635,643

Silicon nitride, 63, 185Singh, J.R., 527Skaggs, S.R., 385Solid freeform fabrication,

473Song, B., 217Spinel, 573, 595Split Hopkinson pressure bar

(SHPB), 217, 269Stassburger, Elmar, 463Stepp, D.M., 421

Strassburger, E., 83Stress propagation, 33Structural ceramics, manu-

facturing, 451Submicron alumina, 83Submicron powders, 463

Target thickness, 643Templeton, Douglas W., 299Test method, 113, 139, 165,

173Tham, R., 643Theory, 139Thermal spray coating, 541Threshold strength, 499Tiles, 33, 103Titanium carbide, 441Titanium diboride, 249, 441,

611, 623, 629Transparent armor, 573, 587,

595Tressler, Richard E., 451Tungsten carbide, 45, 53

Ucisik, A.H., 103Ultra-lightweight armor, 482

von Niessen, Konstantin, 541Vural, Murat, 103

Walker, James D., 337Weber, K., 643Wells, Joseph M., 441Westerling, Lars, 173Wiesner, Volker, 23Wolf, Thomas, 23Wolffe, R.A. 527Woodward, R.L., 361

X-ray computed tomography,441

Zavattieri, Pablo D., 349Zhou, Fenghua, 317Zirconia, partially stabilized

(PSZ), 185

ceramic armor materials by design

Documents

ceramic armor materials

theamerican ceramic

blank ceramic armormaterials

view of ceramic element

copyright law

copyright clearance

advanced ceramics

photocopying items