ELEVATED TEMPERATURE COATINGS: SCIENCE AND

TECHNOLOGY IV

Edited by: Narendra B. Dahotre Janet M. Hampikian

John E. Morral

ELEVATED TEMPERATURE COATINGS: SCIENCE AND

TECHNOLOGY IV

Proceedings of a Symposium sponsored by the Surface Engineering Committee of

the Materials Processing & Manufacturing Division (MPMD) and the Corrosion and Environmental Effects Committee (Jt. with ASM/MSCTS) of

the Structural Materials Division (SMD) of TMS (The Minerals, Metals & Materials Society).

Held at the TMS 2001 Annual Meeting in New Orleans, Louisiana, USA

February 11-15,2001.

Edited by: Narendra B. Dahotre Janet M. Hampikian

John E. Morral

A Publication of

TIÜS

Partial funding for this publication was provided by the Seeley W. Mudd Fund

A Publication of The Minerals, Metals & Materials Society 184 Thorn Hill Road

Warrendale, Pennsylvania 15086-7528 (724) 776-9000

Visit the TMS web site at http://www.tms.org

The Minerals, Metals & Materials Society is not responsible for statements or opinions and is absolved of liability due to misuse of information contained in this publication.

Printed in the United States of America ISBN Number 0-87339-489-5

Authorization to photocopy items for internal or personal use. or the internal or personal use of specific clients, is granted by The Minerals, Metals & Materials Society for users registered with the Copyright Clearance Center (CCC) Transactional Reporting Service, provided that the base fee of $7.00 per copy is paid directly to Copy-right Clearance Center. 27 Congress Street, Salem. Mas-sachusetts 01970. For those organizations mat have been granted a photocopy license by Copyright Clearance Center, a separate system of payment has been arranged.

©2001

If you are interested in purchasing a copy of this book, or if you would like to receive the latest TMS publications catalog, please telephone 1-800-759-4867 (U.S. only) or 724-776-9000. EXT. 270.

TMS

TABLE OF CONTENTS

PREFACE vii

THERMAL BARRIER COATINGS

ACCELERATED DURABILITY TESTING OF COATINGS FOR GAS TURBINES 1 M.J. Stiger, R. R. Handoko, J.L. Beuth, P.S. PettitandG.H. Meier

SYNTHESIS OF ALPHA-ΑΙ,Ο, TEMPLATE ON Ni SUPERALLOY SURFACE BY CHEMICAL VAPOR DEPOSITION 15

Y.-F. Su, M. Torzilli, J.D. Meyer, W.Y. Lee, M.J. Lance, C.J. Rawn andS. Ruppi

CHARACTERIZATION OF COMMERCIAL EB-PVD TBC SYSTEMS WITH CVD (Ni,Pt)Al BOND COATINGS 29

J.A. Harnes, M.J. Lance, B.A. Pint and LG. Wright

NTERFACIAL MICROSTRUCTURE FOR As-DEPOSITED AND CYCLED-TO-FAILURE THERMAL BARRIER COATING 45

AltafH. Carim, Tab bet h A. Dobbins, Merrilea J. Mayo and Lucille A. Giannuzzi

ADVANCED THERMAL COATING SYSTEMS: RESEARCH AND DEVELOPMENT TRENDS 61

C. Leyens, U. Schulz and M. Peters

EFFECT OF Hf ADDITIONS TO Pt ALUMINIDE BOAND COATS ON EB-PVD TBC LIFE 77

James Nesbitt, Ben Nagaraj and Jeffrey Williams

DAMAGE INDUCED BY THERMAL CYCLING OF THERMAL BARRIER COATINGS 93

Vladamir K. Tolpygo and David R. Clarke

MODELING THERMAL STRESSES AND MEASURING THIN FILM CTE IN MoSi, AND MoSL+SiC COMPOSITE COATINGS ON MOLYBDENUM ". 109

Earl C. Hixson, C. Suryanarayana, Graham G. W. Mustoe and John J. Moore

v

INTERDIFFUSION OF COATINGS

INTERDIFFUSION BEHAVIOR IN AN ALUMINIDE COATED NICKEL-BASE ALLOY AT 1150°C 119

B. Gleeson, £. BasukiandÂ. CrosÁy

PREDICTING INTERDIFFUSION MICROSTRUCTURE USING THE PHASE FIELD APPROACH 133

Kaisheng Wu, Yunzhi Wang and John E. Morral

SYNTHESIS OF Hf-DOPED CVD ß-NiAl COATING BY CONTINUOUS DOPING PROCEDURE 143

G. Y Kim, J.D. Meyer, L.M. He, W.Y. Lee and J.A. Haynes

A NEW ANALYSIS FOR THE DETERMINATION OF TERNARY INTERDIFFUSION COEFFICIENTS FOR Ni-Cr-Al AND Fe-Ni-Al ALLOYS 159

Y.H. Sohn and M. A. Dayananda

IN-SITU PROCESSING OF NICKEL ALUMINIDE COATINGS ON STEEL SUBSTRATES 171

Rajesh Ranganathan, Olga Vayena, Teüchi Ando, Charalabos C. Doumanidis and Craig A. Blue

METALLIC/INTERMETALLIC COATINGS AND OXIDATION

DEVELOPMENT OF PROTECTIVE COATINGS FOR HIGH-TEMPERATURE METALLIC MATERIALS 181

R. Keith Bird, Teriyl A. Wallace and San/cara N. Sankaran

RARE EARTH OXIDE COATINGS FOR LIFE EXTENSION OF CHROMIA FORMING ALLOYS 197

Stela M.C. Fernandes and Lalgudi V. Ramanathan

HIGH TEMPERATURE SURFACE OXIDATION CHEMISTRY OF IN-738LC 209 Sudipta Seal, Ley da A. Bracho, Vimal Desai and Kirk Sea m mon

OXIDATION KINETICS AND MORPHOLOGY OF LASER SURFACE ENGINEERED HARD COATING ON ALUMINUM 219

Narendra B. Dahotre and Lalitha R. Katipelli

THE INFLUENCE OF METALLIC COATINGS ON THE STRUCTURE, WETTING, AND MECHANICAL STRENGTH OF CERAMIC/METAL INTERFACES 233

Natalia Sobczak and Rajiv Asthana

VI

CERAMIC COATINGS

FUNCTIONALLY GRADED MATERIALS PRODUCED BY LASER CLADDING 247 Je/fT. De Hosson and Yuta o Pe i

ELECTROPHORETIC AND ELECTROLYTIC DEPOSITION OF CERAMIC FILMS 263 / Zhitomirsk}·

YTTRIA STABILIZED ZIRCONIA/ALUMINA COATINGS DEPOSITED BY COMBUSTION CHEMICAL VAPOR DEPOSITION 277

D. W. Stollberg, J.M. Hampikian, M. Mclntosh and W.B. Carter

HIGH TEMPERATURE OXIDATION OF VC COATED H13 STEEL 291 Swapnil Shah and Narendra B. Dahotre

NEAR NET SHAPE FORMING OF HAFNIUM-BASED CERAMIC COMPONENTS: SYNTHESIS AND CHARACTERIZATION 301

Arvind Agarvial, Tim McKeechnie, StuartS ta rett and Mark M. Opeka

PROTECTIVE CVD MULLITE COATINGS WITH CONTROLLED COMPOSITION AND MICROSTRUCTURE 317

S.M. Zemskova, J.A. Haynes and K.M. Cöoley

THE SPECTRAL EMITTANCE AND STABILITY OF COATINGS AND TEXTURED SURFACES FOR THERMOPHOTO VOLTAIC (TPV) RADIATOR APPLICATIONS 327

B. V. Cockeram andJ.L. Hollenbeck

TEXTURED DIAMOND FILMS ON Si AND Cu SUBSTRATES BY HFCVD TECHNIQUE 343

Ashok. Kumar, A.K. Sikder, J. Mark Anthony and D.S. Misra

AUTHOR INDEX 357

SUBJECT INDEX 359

Vll

PREFACE

Elevated Temperature Coatings: Science and Technolog}' IV\% the fourth volume in a series of invited and contributed papers presented in the symposium: High Temperature Coatings IV. This sym-posium was organized by Narendra B. Dahotre, Janet M. Hampikian and John E. Morral and held in New Orleans, Louisiana, during the TMS annual meeting, February 11-15, 2001. This volume consists of invited and contributed papers from national and international researchers representing universities, federal laboratories and industries. Thus, it provides a rich diversity of material in the research area of High Temperature Coatings. The sponsorship of the TMS Surface Engineering Committee of Materials Design and Manufacturing Division, the Joint TMS/ASM Corrosion and Environmental Effects Com-mittee of Structural Materials Division, and the Materials Science Critical Technology Sector of ASM is gratefully acknowledged.

The topic of High Temperature Coatings is one that is motivated by the increasing need for improved surface characteristics from a wide range of materials without compromising bulk character-istics such as mechanical performance. An example of this is the current thrust toward achieving higher operating temperature in industrial gas turbine engine components through use of thermal barrier coat-ings in high temperature gradient areas such as on turbine nozzles and blades. The specific materials topics covered in this symposium include: Thermal Barrier Coatings, Interdiffusion of Coatings, Metal-lic/Intermetallic Coatings, and Oxidation and Ceramic Coatings. Thus this time in addition to being concerned with the adherence of thermal barrier coatings and the oxidation resistance of metallic coat-ings, one group of papers was concerned with interdiffusion and microstructural changes that occur in metallic coatings during service.

We are grateful for the institutional support provided by the University of Tennessee Space Institute, the School of Materials Science and Engineering at the Georgia Institute of Technology, and the Institute of Materials Science and the School of Engineering at the University of Connecticut. Fi-nally, we appreciate the continuing assistance from TMS for the Symposia on High Temperature Coat-ings.

Professor Narendra B. Dahotre Department of Materials Science and Engineering Laser Materials Processing Group The University of Tennessee Space Institute 10521 Research Drive, Suit 400 Knoxville, Tennessee 37932

Professor Janet M. Hampikian School of Materials Science and Engineering Georgia Institute of Technology Atlanta, Georgia 30332-0245

Professor John E. Morral Department of Metallurgy and Materials Engineering Institute of Materials Science University of Connecticut Storrs, CT 06268-3136

viii

ELEVATED TEMPERATURE COATINGS: SCIENCE AND

TECHNOLOGY IV Edited by:

Narendra B. Dahotre Janet M. Hampikian

John E. Morral

Session 1: Thermal Barrier Coatings

Accelerated Durability Testing of Coatings for Gas Turbines

M.J. Säger, R. Handoko, J.L. Beuth, F.S. Pettit and G.H. Meier

Pgs. 1-14

TIMS 184 Thorn Hill Road

Warrendale, PA 15086-7514 (724) 776-9000

ACCELERATED DURABILITY TESTING OF COATINGS FOR GAS TURBINES

M. J. Stiger1, R. Handoko2, J. L. Beuth2, F. S. Pettit1, and G. H. Meier'

1 Department of Materials Science and Engineering, University of Pittsburgh, Pittsburgh, PA 15261, USA

2 Department of Mechanical Engineering, Carnegie Mellon University, Pittsburgh, PA 15213, USA

Abstract

Oxidation resistance and thermal barrier coatings for components on the hot section of gas turbine engines are desired to have lifetimes on the order of tens of thousands of hours. This presents a problem in evaluating new coatings and modifications to existing coatings tests, which completely replicate the operating conditions, could take years to complete. Therefore, a reliable accelerated testing protocol is required. In this paper efforts directed toward developing a mechanism-based protocol for evaluating the life times of oxidation resistant coatings under thermal cyclic and hot corrosion conditions and thermal barrier coatings under thermal cyclic conditions is described. The cyclic lifetimes of oxidation resistant and thermal barrier coatings are determined by spalling behavior. Spallation is a function of oxide thickness and stress level, which control the elastic energy available to drive spallation, and the structures and morphologies of the various layers and interfaces in a given system, which control the fracture toughness at possible planes of weakness. Efforts to evaluate these quantities in relatively short duration tests are described. Specific techniques include acoustic emission studies, indentation techniques, and detailed metallographic observations. The extrapolation of results from high temperature tests, where failure can be achieved in relatively short times, to lower temperatures, which are characteristic of service conditions is also described. An approach to control these variables in a manner to produce accelerated failures under conditions, which allow estimation of lifetimes under typical operating conditions, are described and preliminary results are presented.

Introduction

The adherence and durability of oxide scales, particularly alumina scales, is critical to the long-term performance of most high temperature alloys. For example, the adherence of oxide scales to hardware in the hot section of gas turbines is essential for their long-term corrosion resistance. Oxides also play a key role in other types of coating systems. In particular, the resistance of ceramic thermal barrier coatings (TBCs) to spallation from nickel-based substrates is dependent on the adherence of the alumina scale which forms between the TBC and the substrate. The loss of adherence along the oxide/substrate interface maybe a fundamental barrier to more widespread use of TBCs on rotating components such as turbine blades and the significant increases in turbine efficiency that would result. The problem of in-service loss of adhesion in TBC systems is thus a major one in the aircraft and power generation gas turbine industries.

Despite the importance of issues associated with oxide adherence and durability, state-of the-art oxidation testing in industry relies on time-consuming and expensive tests (e.g. burner-

2

rig tests), exposing alloys to highly simulative cyclic thermal environments. Such tests are required because of a lack of understanding of fundamental chemical and mechanical processes controlling oxide degradation. Existing limits in the understanding of oxide degradation serve as a barrier to lower-cost accelerated testing methods and rapid development of new alloy and coating systems.

The problems involved in accelerated testing are not easy to solve. Harper and Gleeson (1) have shown that in the case of alloys that develop chromia scales, substantial changes in oxidation resistance can occur after as much as 180 days of oxidation in air at 982°C. Smialek et al. (2) have shown that the oxidation behavior of alloys is influenced by the cooling rate where the effects vary depending upon the alloy. The only way to predict the behavior of such alloys is to understand the degradation process.

Recent developments in two fields have the potential to have significant impact on accelerated testing. First, in the study of oxidation, an increased understanding of fundamental oxidation mechanisms has evolved. Meanwhile, fracture mechanics theory and modeling methods for the failure of thin films and coatings have advanced tremendously. The work described in this paper is an attempt to exploit these advances to gain an increased fundamental understanding of the degradation of oxide scales and thermal barrier coating systems. Figure 1 shows schematic cross-sections of the two systems of interest in this paper. State-of-the-art coatings for oxidation resistance (Fig. la.) form a protective alumina film during exposure. The length of time that this film can resist breakdown by chemical or mechanical means coupled with the capability to redevelop another alumina scale determines the life of the coating. Thermal barrier coatings (Fig. lb.) pose similar issues in that an alumina layer (often called a "thermally grown oxide", TGO) forms between the metallic bond coat and ceramic topcoat. This situation has the additional feature that strain energy accumulated in the ceramic layer can contribute to the mechanical degradation of the TGO.

a) b)

TBC

alumina ^M alumina (TGO)

coating bond coat

alloy aildy

Figure 1. Schematic diagram of cross-sections of the coatings considered in this paper, (a.) An alumina-forming coating for oxidation protection, deposited on a nickel-base superalloy. (b.) A thermal barrier coating on a nickel-base superalloy.

The lifetime of thermal barrier coatings illustrates some of the important issues associated with accelerated testing. Figure 2 presents a macroscopic photograph of an EB-PVD TBC on a Pt-aluminide bond coat, which has failed after 1287 cycles (45 min. at high temperature and 15 min. cooling) at 1100°C in dry air. The failure initiated at a specimen edge and propagated as an elongated buckle. Figure 3 indicates the effect of temperature on the time to failure. Here the inverse of the time at elevated temperature (0.75 times the number of

3

cycles) is plotted versus reciprocal temperature. The data follow a reasonable straight line with failure times decreasing from about 1000 h at 1100°C to 50 h at 1200°C. Importantly, if these data are extrapolated to the temperatures currently experienced by bond coats (« 1000°C) the failure times are predicted to be on the order of 50,000 hours. Experiments of this duration are generally not feasible and the effectiveness of changes made to bond coats must be evaluated on a much shorter time scale. This example clearly shows the need for a reliable accelerated method for determining failure times.

Figure 2. Macroscopic photograph of the failure of a thermal barrier coating consisting of an EB-PVD topcoat with a Pt modified aluminide bond coat on a disk of a nickel-base superalloy.

-10

-12

Temperature (°C) 1200 1150 1100 1050 1000

Q=390kJ/mole

♦ dry

50

■ 500 ~ 3 CD

■ 5000

■ 50000

6.6 6.8 7.0 7.2 7.4 7.6 7.8 8.0 1ΑΓ(χ10^)

Figure 3. The effect of exposure temperature on the time to failure of an EB-PVD TBC with a Pt modified aluminide bond coat

Philosophy of Accelerated Testing

A reliable accelerated testing protocol requires, as a prerequisite, a knowledge of the failure mechanism and the important parameters which control failure. The accelerated test must then modify the important parameters to produce failure in a reasonable time. However, it must be clear that the changes do not move the system into a regime where a different failure mechanism becomes operative. For example, increasing the exposure temperature is one common way of accelerating a test. However, there are a number of well-known cases where changing the temperature also changes the mechanism. One such case is the oxidation of the intermetallic compound M0SÍ2 (3). This compound oxidizes very rapidly at 500°C by growing a surfece layer of M0O3. Increasing the temperature to 600°C actually decreases the oxidation

4

rate because it results in the formation of a slowly-growing surface film of S1O2. The following is a description of work directed toward the accelerated testing of alumina-forming coatings and thermal barrier coatings.

Important Parameters

Oxide Growth and Aluminum Depletion The exposure of oxidation resistant coatings, such as that depicted in Fig. la., results in the formation of an alumina film, which depletes aluminum from the coating. The factor which controls life is the loss of Al in forming alumina on the surface and through spallation during thermal cycling, as well as interdiffusion with the substrate. This is illustrated for a NiCoCrAlY coating in Figure 4. The Al-depletion is manifested by dissolution of the Al-rich ß phase and, eventually the appearance of spinels in the oxide scale. After selecting the failure criterion the rate of Al loss is needed. Nesbitt and coworkers have analytically modeled this type of a process and predicted the lifetimes associated with Al losses from alumina-forming systems (4). This will not be addressed further in this paper, rather the factors associated with mechanical breakdown of alumina scales will be addressed i. e. loss of adherence. The adherence of protective oxide scales to substrates is governed by the stored elastic energy in the scale which drives delamination and the fracture toughness of the alloy/oxide interface which quantifies the resistance to fracture.

as-processed 209 cycles

~ "y^wfw"' "**»· TÍ.*

~ ' Í55J3L 100 urn

Figure 4. Cross-sections of a NiCoCrAlY coaling on a nickel-base superalloy showing the Al-depletion (indicated by the dissolution of the ß phase), which occurs during cyclic oxidation at 1100°C.

Elastic Energy Stored in the Scale The stored elastic energy is determined primarily by the scale thickness h and the stress level σ in the scale such that (Elastic a ha2). The scale thickness is determined by the rates of diffusion of metal and/or oxygen through the scale and is generally represented by the parabolic rate constant kP. The growth rates of continuous alumina scales are strongly dependent on oxidation temperature but are only influenced slightly by alloy composition. However, for systems which require extremely long lifetimes, a moderate change in kP can result in measurable changes in oxide thickness and, therefore, metal consumption and elastic strain energy in the scale.

5

The stress state in the scale is determined by stresses which arise during the oxide formation (growth stresses), stresses produced during temperature changes as the result of thermal expansion mismatch between the oxide and the alloy (thermal stresses), and any stress relaxation which may occur. Growth stresses can, in principal, be affected by the growth mechanism of the alumina and, therefore, by doping with elements such as yttrium which can alter alumina growth. Thermal stresses are primarily a function of the temperature is change (ΔΤ) but can be influenced by the rate of temperature change if stress relaxation processes occur. Stress relaxation processes arise primarily from creep of the oxide and/or substrate and are influenced by the relative thicknesses and creep strengths of the oxide and substrate.

Fracture Toughness of the Alloy/Oxide Interface The fracture energy of the interface is a function of the composition at the interface, the microstructure in the interfacial region, and the composition of the exposure environment. It is now well established that small additions of reactive elements, such as yttrium, hamium, and cerium, substantially improve the adherence of alumina films to alloy substrates (5). While the effects produced by the reactive elements are widely known, the mechanisms whereby they improve adherence are not completely understood. Over the last fifty years a number of mechanisms have been proposed.

The precious metals, such as Pt, have also been known for many years to have beneficial effects on the cyclic oxidation of alurnina-forming alloys and this effect is the basis for the platinum-aluminide coatings which are widely used to protect Ni-base superalloys. However, there is still much which is not understood with respect to the mechanisms by which Pt affects alumina adhesion (6).

A critical step in understanding mechanisms leading to oxide adhesion loss and in matching adhesion loss mechanisms between industry standard and accelerated oxidation tests is periodic measurement of interfacial fracture toughness. A reliable interfacial toughness test is needed, suitable for application to relatively small-sized specimens.

Response to Stresses Generally, alumina scales will be loaded in compression when cooled to room temperature. Possible responses to these stresses are shown in Figure 5. The mechanisms shown if Fig. 5(a), buckling, and Fig. 5(b), wedge cracking, are the most damaging in that they expose fresh metal to the oxidizing gas.

Gas

Oxide

Alloy

Gas Gas Gas

r$ägei^>- — > — < CAS L

Alloy Alloy

(a) (b) (c)

Figure 5. Schematic diagram of responses of an oxiue which is loaded in compression, (a) buckling of the oxide, (b) shear cracking of the oxide, and (c) plastic deformation of the oxide and alloy.

Thermal Barrier Coatings

6

There are a number of degradation modes which can limit the life of a TBC and these must be understood in order to make lifetime predictions for existing systems and to provide the basis for the development of improved TBC systems (7). The modes most important for the thermal cycling degradation of EB-PVD systems are:

(i) Cracking along the interface between the TBC and the bond coat for EBPVD coatings which results in spalling of the entire TBC (ii) Sintering of the TBC at the outer surface where the temperature is highest which can affect the strain energy stored in the TBC.

It is now generally accepted that oxidation of the bond coat is a critical factor controlling the lives of EBPVD TBCs. It is now well established that the ability of a bond coat to form an o> alumina layer with negligible transient oxidation and the adherence of the alumina to the bond coat are critical factors in controlling the durability of TBCs. Experience with aircraft engines has shown that bond coat oxidation and the ability to resist spalling of the TBC from the bond coat are critical factors determining coating life. The bond coats develop a thermally grown oxide (TGO) layer during fabrication. The TGO grows thicker during exposure of the TBC. Therefore, this problem is related to that for the alumina-forming coatings with the added effect of strain energy stored in the TBC also influencing the spallation behavior.

Evaluation of Important Variables

Long Term Cyclic Oxidation Testing

Figure 6 shows a comparison of the cyclic oxidation behavior of the various NiCrAl alloys at 1100°C. All of the reactive element doped alloys show substantially better cyclic oxidation resistance than the undoped alloy. There are, nevertheless, quantitative difíerences among the various alloys.

The concentrations and distributions of the reactive elements play significant roles in the cyclic oxidation resistance of coatings alloys. When concentrations are too high, preferential oxidation of the reactive elements can occur which adversely affects protective alumina formation. Low concentrations of these elements may not provide the desired effects on cyclic oxidation resistance. Moreover, control of reactive element concentrations is often difficult because of their high reactivity. It may be possible to predict cyclic oxidation resistance by definition of reactive element concentrations and distributions on alloys.

Figure 6. Cyclic oxidation kinetics for several Ni-Cr-Al alloys exposed at 1100°C.

7

XRD Stress Measurements

XRD stress measurements are based on the determination of the strain by measuring the d-spacing of specific lattice planes. The following relationship, which is subject to several experimentally verifiable assumptions, including a biaxial stress state (8), exists between strain and stress:

£¡¥ = °Ψ ~a0 = J. ^ ( / / j t / ) i r s i n2 ψ + 2 ( S ] ( M / ) f 7 ( 1 )

do

where εψ is the strain measured for a particular set of (hkl) planes in a laboratory coordinate system, άψ is the lattice spacing of (hkl) planes when the specimen is tilted by the angle ψ, do is the strain-free lattice spacing and σ is the stress. Si (hkl) and S2(hkl) are the X-ray elastic constants. In this sinV technique the stress is calculated from the slope of the d«, vs. sin2iy plot.

The sin2\y method allows the calculation of the stress without measuring do, which could yield great errors when using one point methods. In order to calculate the stress from the slope of d vs. sin2 ψ plots the knowledge of S2(hkl) is sufficient. For randomly oriented grains S2 can be calculated from single crystal data using the Reuss,Voigt and other approximations. In the following S2 is calculated within the Reuss approximation for hexagonal crystals

S2(hkl) = (2su - su - 5!3) - (5.9,, + S33 - 3 ^ - su - 5su)a23i

+(su + S33 - 544 - 2su)a¡3 + 6sl4a22a33(3a^ - a222) (2)

with

33 \{cla)(h2 +hk + k2) + l2 ' The Sjj are the single crystal elastic compliances and a^ is the direction cosine between the c-axis and the normal of the selected (hkl). It should be noted that for (hkO) planes a^ = 0 and equation (2) is reduced to its first term.

Standard techniques based on equation (1) are the Rocking and Tilting methods, illustrated schematically in Figures 7a. and 7b., respectively.. In the Tilting technique, the tilt axis is the intersection of the sample surface with the diffraction plane. The diffraction plane is fixed and by definition contains the incident and reflected X-ray beams. The angle ψ is zero when the normal to the sample surface is within the diffraction plane, ψ increases as the sample is tilted and its surface normal is rotated out of the diffraction plane. In the Rocking technique the sample normal is tilted within the diffraction plane and ψ is the angle between the surface normal and the normal [hkl] of the measured set of (hkl) planes. In both techniques only one selected set of (hkl) planes is used for the measurement. To achieve high accuracy during stress measurements, only reflections with high 2 ΘΒ (Bragg angle) can be used, since errors in the measurement of d decrease with increasing θβ.

High Temperature Stress Measurements by XRD Tilting out of the diffraction plane is generally not possible with a hot stage for high-temperature measurements, so the Tilting technique cannot be used. There are also concerns that during rocking of the hot-stage at high temperature the sample might shift or fall off. Therefore, a new method, the Fixed Incidence Multiplane technique (FIM) was developed (9). Instead of tilting the sample, the angle Ψ is varied by measuring different (hkl) reflections (Fig . 7c). The angle Ψ is defined as Ψ = ΘΒ - a where a is the angle between the incident beam and the specimen surface. Since this method uses different reflections, the cell dimension a calculated from the d-spacings is plotted vs. sin2lF. Furthermore, in order to calculate a stress from the slope of this line all (hkl) planes must have the same S2(hkl) value. For hexagonal crystals this is fulfilled for planes of the type

8

(hkO), for which a33 is zero, as can be seen in equation 2. One consequence of this method is that only a few (hkO) planes are available for hexagonal substances and high Bragg angles.

Figure 7. Geometry of stress measurement by XRD (a) using the Tilting Technique, (b) using the Rocking Technique, (c) using the FIM Technique.

In this study XRD measurements of the stresses formed in the alumina on a NiCrAlY alloy were performed in a Philips X'pert diffractometer with a CuKa line source. Parallel beam optics with a flat graphite monochromator and a proportional counter were used. The room temperature residual stresses were measured using the Tilting technique.

The high temperature stress measurements where conducted with a resistively heated Pt hot-stage (Anton Paar KG) in ambient air using the rocking technique. The specimens were fixed to the Pt-strip using a special bonding agent (Zapon-Lacquer). Temperature control was achieved with a thermocouple welded to the underside of the Pt stripe. This thermocouple was correlated with the desired sample surface temperature by use of a Laser-pyrometer.

Residual Stress (1.2.10)

0.998

0.996

£. 0.994 i

0.992 - - ^ ^ |

0.99 0.5 0.6 0.7 0.8 0.9 1

sine squared psi

Figure 8. Residual stress measurement (sinV plot) for alumina scale that formed on NiCrAlY exposed to 1000°C.

The residual stress in the alumina layer that grew on NiCrAlY during an isothermal exposure to 1000°C for 250h was measured by the tilting technique on the (1.2.10) plane. A compressive stress of 3.90GPa was determined from the sin2vy plot in Figure 8. This value agrees with the calculated thermal stress value of 4.10GPa from the CTE mismatch between the oxide and the substrate. In support of the above results, high temperature work on the same material exposed tol000°C revealed that there is no growth stress.

9

♦ 29h, 4.1GPa

■ 250h,3.9GPa

Acoustic Emission Measurements

Acoustic emission measurements can be used to detect cracking and spalling of oxide scales in-situ. The apparatus used in this study has been described in detail previously (10). A lmm diameter Pt rod was spot welded to the oxidation specimens. This rod served both to support the specimen in a vertical oxidation furnace and to provide a wave guide to transmit the acoustic signal from the specimen to a stainless steel cone, outside the furnace, which was connected to a transducer. The acoustic emission monitoring system used was the Dunegan/Endevco, 3000 series. The parameters measured in the present study were mainly acoustic emission counts, which scale with the amount of energy released by the event(s) producing the emission, and the peak amplitude of the acoustic emission event. The specimens had the same dimensions as those used for the oxidation studies

Figure 9 illustrates an important aspect of acoustic emission measurements. Here cumulative A.E counts were measured after a 24-hour exposure at 1100°C for various alloys. Comparison of these data with long term cyclic oxidation data in Figure 6 show there is a correlation between short term acoustic emission measurements and long term cyclic oxidation behavior i.e. less emission indicates better resistance. It is believed that acoustic emission measurements can provide an important component to an accelerated testing protocol.

8

o" ° c o o x O O

Έ 4

O ü LU < 2

1000 800 600 400 200 0 temperature (°C)

Figure 9. Acoustic emission count rate during cooling of three NiCrAl based specimens as a function of temperature after oxidation at 1100°C in air for 24 hours.

Fracture Toughness Measurements

A critical step in understanding mechanisms leading to oxide adhesion loss and in matching adhesion loss mechanisms between industry standard and accelerated oxidation tests is periodic measurement of interfacial fracture toughness. A reliable interfacial toughness test is needed, suitable for application to relatively small-sized specimens. Despite its importance, no tests have been developed for quantifying toughness loss in oxide scales or TBC/oxide systems. As a result, existing observations of oxide adherence loss have not been linked to measured changes in interfacial toughness.

NiCrAl

i ¿ B o i o i r ô » ¡CIA1-0.2Y

Indentation Test for Fracture Toughness The authors have developed and applied an indentation test for quantifying degradation of interfacial toughness in TBC (11,12). The principal advantages of this type of test are that it is easy to perform and that it can be performed on relatively small specimens. The dimensions of specimens used thus far by the authors are approximately 2.5 cm in diameter and 0.3 cm thick, and smaller specimens can be used.

The indent test is diagrammed in Figure 10. In the current form of the test, a specimen is placed in a Rockwell hardness tester using a brale C indenter. The coating (in this case the TBC and the alumina scale beneath it) is penetrated by the indenter and the plastic deformation of the underlying substrate induces compressive radial stresses in the substrate, away from the indent crater. This compressive radial stress is transferred to the coating and acts to drive the extension of an axisymmetric interface crack (shown in cross section in Fig. 10). Figure 11 shows optical micrographs of a typical debond produced in an Electron Beam Physical Vapor Deposited (EBPVD) TBC system, as viewed from the side (Fig. 11a) and from the top (Fig. 1 lb). The toughness of the interface where debonding occurs (in this case the interface between the alumina scale and the PtAl bond coat) can be determined from a mechanics analysis of the indentation problem and a measurement of the delamination radius.

Debonding TBC

Superalloy a = Contact Radius R = Debond

Figure 10. Schematic diagram of the indentation test for measuring interfacial fracture toughnesses.

The indent test has also been used to study oxide-only systems (with no TBC on top). For such systems, indentation is not accompanied by a single axisymmetric delamination. but instead induces either localized delaminations or, in the case of adherent scales, compressive failures of the oxide. By measuring the radial extent of such failures, interfacial toughnesses or compressive strengths of the oxide can be estimated. Regardless of the failure mechanism, the goal of this type of test is to induce an oxide scale failure via compressive applied stresses. In this way, it mimics in-service failure mechanisms caused by compressive stresses in the oxide, whatever they may be.

Figure 11. Optical photograph of a typical indent produced in a TBC after high temperature exposure.

11

Exposure-Induced Toughness Loss in TBC Systems Figure 12 shows a plot of "apparent" losses of interfacial toughness for EB-PVD TBC systems subjected to 1100°C, 1135°C, and 1200°C isothermal exposures in dry air, as determined by the indent test. They are designated as apparent toughness losses because calculations used to obtain toughness values from measured delamination radii do not include the effects of known changes in the TBC system that could affect adherence. For instance, oxide growth or increases in stress magnitudes in the debonded coatings could cause an apparent decrease in toughness even if the interface itself were not weakened or embrittled. Toughnesses (in the form of a critical stress intensity factor, Kc) in the as-processed state are in the range of 3.1 - 3.7 MPaVm or higher. The applied K, resulting from residual stresses alone (with no indentation) is approximately 1.0 MPaVm, so that times to failure are designated as occurring when Kc reaches this value.

W -Θ-

-èr -*-D

•

1200°C 1135°C 1100°C As-Processed TBC Fails

100 600 700

Figure 12.

200 300 400 500

Exposure Time (hrs) Plot of apparent toughness as a function of exposure time for

TBC systems at various temperatures.

The results plotted in Figure 12 are the first available data quantifying toughness degradation vs. time for TBC systems and they show that much of the loss in toughness occurs at times that are a fraction of the time needed for spontaneous failure (Fig. 3). This loss in toughness could be due to one or more of a number of mechamsms, including chemical or mechanical damage at the interface, oxide growth and sintering of the TBC at high temperature (which changes the overall stiffness and residual stress in the TBC). Each of these mechanisms is thermally activated (where mechanical damage is likely linked to thermally activated creep deformation). Of these mechanisms only chemical or mechanical damage at the interface results in a "true" loss of toughness at the interface.

Insights into Accelerated Testing Techniques Insight into accelerated testing methods for TBC systems can be obtained by plotting the times to reach a given interfacial toughness (taken from Figure 12) on an Arrhenius plot similar to that in Figure 3. Such a plot is shown in Figure 13. In Figure 13, the line at the bottom of the plot reproduces average failure times plotted in Figure 3, except these have now been related to an apparent interfacial toughness of 1.0 MPaVm. Times to reach higher values of apparent interfacial toughness are plotted as three additional lines, with all data in the temperature range of 1100°C to 1200°C.

Two types of accelerated testing methods for TBC systems have been considered in this research. They are high-temperature testing to failure and mechanical testing for interfacial toughness loss. The first method allows shorter testing times by shortening the time to failure.

12

The second method allows shorter testing times by probing toughness loss at early times, before failure occurs. The plot of Figure 13 and on-going research by the authors gives insight into the validity of these two methods. First, because the slopes of all four lines are similar, it is suggested that the mechanisms leading to apparent toughness loss are the same as those that lead to TBC system failure. As a result, measurement of interfacial toughness losses in TBC systems at early exposure times appears to be a valid accelerated testing method for understanding TBC failure.

0) £

T—

sz

0

-1 ;

-2;

-3:

-4!

-5;

-6

-7

^ o-

: -Δ-- Kc= 2.5 : Í -> · Kc= 2.0 !

* -« . ^ ! a K C = I . 5 -. , ' " -» . „_ i -Θ- Kc= 1.0

"""""'--.. ~ - - , ~^-~_ " ' ""■■- . ~ ~ o ~ . . . ~ ~ " \

~-~^ " " - · . "~""·- · . "~"--Δ

^ ^ ^ b ~ ^ ^ " ~"'D

- 1 0 -67 6.8 7.2 7.3 7.4 69 7 71

1/Temperature (K) x io"" Figure 13. Arrhenius plot showing TBC failure times as a function of

exposure temperature (similar to Fig. 3) and the time to reach a given level of fracture toughness as a function of temperature.

In order for high temperature testing to be valid, the failure times at operating temperatures (near 1000°C) need to fall on the bottom line of Figure 13, extrapolated to these lower temperatures. As previously noted, determining whether this occurs (requiring tests on the order of 50,000 hours) is not feasible. However, exposing a TBC sample at 1000°C to reduce the apparent interfacial toughness to 2.5 - 2.0 MPaVm could be accomplished in a much shorter time. If such data points were to lie on lines extrapolated from the top two lines of Figure 13, it could be presumed that the bottom line could also be accurately extrapolated to operating temperatures. The authors are currently carrying out low temperature toughness tests of this type.

In this way, the plot of Figure 13 can serve a roadmap for understanding these two types of accelerated testing methods for TBC systems. Although it gives feedback on whether mechanisms leading to failure are independent of the test method, it does not give insight into what mechanisms may or may not be dominant for individual TBC systems. Identifying the relative contribution of failure mechanisms is the goal of current research by the authors, which includes optical and scanning electron microscopy, stress measurements, acoustic emission monitoring and other techniques.

13

References

1. M. A. Harper and B. Gleeson "Long-Term, Cyclic Oxidaiton Behavior of Three Chromia-Forming Alloys Exposed to Air at 982°C," Cyclic Oxidation of High Temperature Materials, eds. M. Schütze and W. J. Quadakkers, (IOM Communications Ltd. 1999), 273.

2. J.L. Smialek, J. A. Nesbitt, C. A. Barrett, and C. E. Lowell, "Cyclic Oxidation Testing and Modelling: a NASA Lewis Perspective," Cyclic Oxidation of High Temperature Materials, eds. M. Schütze and W. J. Quadakkers, (IOM Communications Ltd. 1999), 148.

3. H. J Grabke. and G. H. Meier. "Accelerated Oxidation, Internal Oxidation, Intergranular Oxidation, and Pesting of Intermetallic Compounds," Oxid. Metals. 44 (1995), 147.

4. J. L. Smialek, J. A. Nesbitt, W. J. Brindley, M. P. Brady, J. Doychak, R. M. Dickerson, and D.R. Hull, Service Limitations for Oxidation Resistant Intermetallic Compounds," Mat. Res. Soc. Svmp. Proc. Vol. 364 (1995), p. 1273.

5. C. Sarioglu, M. J. Stiger, J. R. Blachere, R. Janakiraman, E. Schumann, A. Ashary, F. S. Pettit, and G. H. Meier, "The Adhesion of Alumina Films to metallic Alloys and Coatings," Materials and Corrosion. 51 (2000), 1-15.

6. 7. J. S. Schaeffer, G. M. Kim, G. H. Meier, and F. S. Pettit, "The Effects of Precious Metals on the Oxidation and Hot Corrosion of Coatings," The Role of Active Elements in the Oxidation Behavior of High Temperature Metals and Alloys, ed. E. Lang, (Elsevier 1989), 231.

7. M. J. Stiger, N. M. Yanar, M. G. Topping, F. S. Pettit, and G. H. Meier, "Thermal barrier Coatings for the 21st Century," Z. für Metallkunde. 90 (1999), 1069-1078.

8. I. C. Noyan and J. B. Cohen, Residual Stresses. (Springer-Verlag, 1987).

9. C. Sarioglu, J. R. Blachere, F. S. Pettit, and G. H. Meier, "Room Temperature and In-Situ High Temperature Strain (or Stress) Measurements by XRD Techniques," Microscopy of Oxidation 3, eds. S. B. Newcomb and J. A. Little, (The Institute of Materials. London, 1997), 41.

10. A. Ashary, G. H. Meier, and F. S. Pettit, "Acoustic Emission Study of Oxide Cracking During Alloy Oxidation," High Temperature Protective Coatings, ed. S. C. Singhal, (AIME 1983), 105.

H . A . Vasinonta, and J.L. Beuth, "Measurement of Interfacial Toughness in Thermal Barrier Coating Systems by Indentation," accepted to Engineering Fracture Mechanics.

12. R. A. Handoko, J.L. Beuth, G.H. Meier, F.S. Pettit, and M.J. Stiger, "Mechanisms for Interfacial Toughness Loss in Thermal Barrier Coating Systems," accepted for the Proceedings of the Materials Division Symposium on Durable Surfaces. 2000 ASME International Mechanical Engineering Congress and Exposition, Orlando, November, 2000.

KEYWORDS Alumina scales, thermal barrier coatings, adherence, durability.

14

ELEVATED TEMPERATURE COATINGS: SCIENCE AND

TECHNOLOGY IV Edited by:

Narendra B. Dahotre Janet M. Hampikian

John E. Morral

Session 1: Thermal Barrier Coatings

Synthesis of Alpha-A1203 Template on Ni Superalloy Surface by

Chemical Vapor Deposition Y.-F. Su, M. Torzilli, J.D. Meyer, W.Y. Lee,

M.J. Lance, C.J. Rawn and S. Ruppi

Pgs. 15-28

TIMS 184 Thorn Hill Road

Warrendale, PA 15086-7514 (724) 776-9000

Synthesis of alpha-Al203 Template on Ni Superalloy Surface by Chemical Vapor Deposition

Y.-F. Su, M. Torzilli, J. D. Meyer, and W. Y. Lee Department of Chemical, Biochemical, and Materials Engineering

Stevens Institute of Technology Hoboken, NJ 07030

M. J. Lance and C. J. Rawn Oak Ridge National Laboratory

Oak Ridge, TN 37831

S. Ruppi SECO Tools AB Fagersta, Sweden

Abstract

The life of thermal barrier coatings (TBCs) used in aircraft engine and power generation turbines is largely dictated by: (1) the ability of a metallic bond coating to form an adherent thermally grown oxide (TGO) at the metal-ceramic interface and (2) the rate at which the TGO grows upon oxidation. It is postulated that a thin 01-AI2O3 template, if appropriately synthesized on a Ni-based alloy, will guide the alloy surface to form a TGO that is more tenacious and slower growing than what is attainable with state-of-the-art bond coatings. The feasibility of preparing such a template was examined using a chemical vapor deposition (CVD) process previously developed for cutting tool applications. A coating directly deposited by this method on a single crystal Ni superalloy substrate consisted of ~1 μπι OC-AI2O3 crystals in a matrix of amorphous AI2O3. The coating, although not entirely OC-AI2O3, promoted the formation of a highly stressed and adherent TGO layer (-6.0 GPa) on the alloy surface upon subsequent oxidation. When the alloy surface was modified with an electroplated Pt layer, the coating was entirely OC-AI2O3, but with the presence of fine microcracks on the coating surface. In comparison to results observed for pure Pt and Ni substrates, it appeared that the role of the Pt interlayer was to promote the formation of K-AI2O3 nuclei, which subsequently transformed to OC-AI2O3 during the CVD growth step. These results suggested that the nucleation and growth of a fully CC-AI2O3 layer on the Ni superalloy surface, without forming microcracks, would be difficult through simple adaptation of the CVD process previously developed for the cutting tool industry.

Elevated Temperature Coatings: Science and Technology IV

Edited by N.B. Dahotre, J.M. Hampikian and J.E. Morral TMS (The Minerals, Metals & Materials Society), 2001

16

Introduction

TBCs are currently used, in conjunction with air cooling, to prolong the life of metallic "hot-section" turbine components in revenue generating aircraft engine services.1 State-of-the-art TBCs used for rotating airfoils in aircraft engines consist of a strain-tolerant Y203-stabilized Zr02 (YSZ) layer prepared by electron beam physical vapor deposition (EBPVD) and a metallic bond coating which provides high-temperature oxidation protection. It is recognized that, among various failure mechanisms observed, the principal failure mode of the EBPVD-TBCs is progressive fracture along the interface region between the metallic bond coating surface and its thermally grown oxide (TGO) upon oxidation and thermal cycling.1"4 Bond coatings commonly used with the EBPVD-YSZ layer are: (1) MCrAlY (where M is Ni, Co, or NiCo) prepared by vacuum plasma spray (VPS)1 and (2) Pt-aluminide prepared by Pt electroplating and subsequent aluminizing by pack cementation or chemical vapor deposition (CVD).5

There are some intriguing observations reported in the literature about enhancing the oxidative stability of the metal-ceramic interface by incorporating a thin-layer of high-quality AI2O3. As reported in his patent, Strangman6 observed that the presence of an AI2O3 interlayer (~1 μηι) prepared by CVD between NiCoCrAlY and EBPVD-YSZ increased the TBCs burner rig life by five-fold. The increased oxidation resistance was attributed to the dense morphological quality and high chemical purity of the CVD-AI2O3 layer, although the claims were made without substantial characterization results. Another independent study by Sun et al.7 showed that the presence of a CVD AI2O3 layer (4 μηι thick) between a plasma sprayed YSZ layer and a NiCoCrAlY layer substantially increased the cyclic oxidation life of the YSZ layer. The rate of bond coating oxidation was observed to be lower due to the presence of the artificial AI2O3 layer. It was also claimed that the formation of spinels (i.e., transient oxide phases that form during the initial stage of Ni alloy oxidation) was not observed at the YSZ-TGO interface.

Both Strangman and Sun et al. used a CVD process which utilizes AICI3, CO2, and H2 as precursors at a deposition temperature of ~1000°C. This chloride-based CVD process was previously developed, and is being widely used for the cutting tool industry.8 The non-line of sight, atomistic growth technique is attractive for manufacturing, since engineering components with intricate shapes and complex surface features can be readily coated. The CVD process is the only technique currently capable of commercially producing CC-AI2O3 in the form of coherent and dense coatings. PVD methods such as sputtering,9 reactive sputtering,10'11 reactive evaporation,12 ion-assisted deposition,13,14 and cathodic arc plasma deposition15 are generally known to produce metastable or amorphous AI2O3, unless post-deposition annealing above 1000°C is applied.

In the cutting tool industry, WC/Co substrates are deposited with Ti(C,N) and T12O3 interlayers prior to the OC-AI2O3 growth step. The Ti(C,N) interlayer is used as a diffusion barrier, since some of substrate elements and impurities (particularly Co) are found to cause the formation of metastable phases or undesired morphological features such as whiskers during the coating growth.8 It has been postulated that the role of the T12O3 layer is to promote the preferential nucleation of α-Αΐ2θ3.16 Another important processing feature is that air leaks into the CVD process must be tightly controlled in order to avoid the development of "cauliflower" growth morphology and powder formation in the gas phase. Despite these difficulties, remarkable engineering progress has been made with controlling the nucleation and growth behavior of AI2O3 on the cutting tool surface. For example, the selective growth of CC-AI2O3 or K-AI2O3 (even a-A^CVic-A^Ch multilayers) is possible by adding a small concentration (~ 0.1%) of T1CI4 or ZrCU as a dopant to the gas phase during the initial period of coating growth.8'16

17

In the present study, an industry standard CVD process, previously developed at SECO Tools for the cutting tool industry, was used to prepare CÍ-AI2O3 coating specimens on the (100) surface of a single crystal Ni alloy (René N5) and on the same alloy coated with an electroplated Pt interlayer. The Pt-plated René N5 was investigated for the following reasons: (i) Pt may mitigate adverse effects of some alloying elements of the Ni alloy (particularly Co and Cr), which are known to cause the growth of AI2O3 whiskers;8'17 (ii) beneficial effects of Pt on TGO adhesion are well known;18'19 and (iii) Pt electroplating is a routine part of turbine coating production and repairing. While René N5 and Pt-plated René N5 substrates were of engineering importance for this study, pure Ni and Pt substrates were also used to study the nucleation and growth behavior of CVD-AI2O3 in the absence of impurity effects. Impurities such as Co and Cr on the substrate surface and chlorides of Si, Fe, Ni, Cr, and Co from the gas phase have been shown to influence both the morphology and phase content of CVD-AI2O3 coatings prepared for the cutting tool industry.8,17'20'21

Experimental

CVD-AI2O3 coating specimens were prepared at SECO Tools, Sweden, using the procedures described elsewhere 16,22'23

s at a deposition temperature of ~1000°C and a reactor pressure of -10 kPa with AICI3, CO2 and H2 as precursors. A single crystal Ni alloy (René N5; René is a trademark of General Electric Company, Fairfield, CT) was cast as a cylindrical rod with the [100] seed direction. The alloy rod was sliced radially to produce disc specimens (0.2 cm thickness x 1.2 cm diameter) while preserving the (100) orientation on the disc surface. The nominal composition (in weight %) of the alloy is: 6.2 Al, 0.05 C, 7.5 Co, 7.0 Cr, 0.16 Hf, 1.5 Mo, 3.0 Re, 6.5 Ta, 0.02 Ti, 5.0 W, and Ni as the balance. This alloy was melt-desulfurized to below 1 ppm sulfur by PCC Airfoils for Oak Ridge National Laboratory (ORNL), and its oxidation characteristics have been extensively studied as described elsewhere. 4' The specimen surface was polished prior to coating experiments with 0.05 μπι ΑΙ2Ο3 suspension. General Electric Aircraft Engines provided René N5 discs (0.3 cm thickness x 2.5 cm diameter) electroplated with a layer of pure Pt. The thickness of the Pt layer was nominally -7 μιη. The Pt-coated alloy surface was not polished prior to the coating experiments. Pt and Ni substrates were cut to 10 x 10 mm squares from Pt and Ni foils (99.99+% for both foils, Goodfellow, Berwyn, PA). The Pt and Ni surfaces were not polished for the coating experiments.

Coating morphology and compositions were examined using a field emission scanning electron microscope (SEM, LEO 982, LEO Electron Microscopy Inc., Thornwood, NY) equipped with an energy dispersive spectrometer (EDS). Phase contents and microstructural orientation of coating specimens were studied by X-ray diffraction (XRD, Siemens Diffractometer D5000) with capability to perform grazing incidence (GIXRD) measurements. The GIXRD technique was used to analyze the coating layer while reducing contributions from substrate peaks, as low incidence angles provide shallow penetration depth into the sample surface.26 For selected coating specimens, high temperature x-ray diffraction (HTXRD) measurements were conducted at ORNL using a Scintag PAD X vertical Θ/2Θ goniometer equipped with a modified Buehler HDK-2 diffraction furnace and an MBRAUN Position Sensitive Detector (PSD-50M). All data were collected in vacuum or air. The temperature of the sample heating assembly was measured with a Pt/Pt-10%Rh thermocouple spot-welded to the Pt-30%Rh heater strip on which the sample was placed. The temperature of the surface of the samples was measured with an optical pyrometer (Pyrometer Instrument Company, Inc. Bergenfield, NJ. Model 95).

18

In addition to XRD, luminescence spectroscopy was used to determine the presence of CC-AI2O3 in the coating layer and to measure the average stress of the CÍ-AI2O3 phase. OC-AI2O3 has trigonal symmetry with rhombohedral Bravis centering and can possess trace amounts Cr3+

cation impurities in the alumina crystal lattice. When excited by a laser, the electrons of the Cr3+

cations emit photons.27 The two characteristic photon energies for the (X-AI2O3 phase are: 14400 (Rl) and 14430 cm"1 (R2). A stress in the OC-AI2O3 phase results in a change in the electronic

„3+ environment of the Cr cations and therefore a corresponding shift in the Rl and R2 lines. A Dilor XY 800 triple stage Raman microprobe and a coherent Innova 308C Argon ion laser operating at 514.5 nm with a 100 mW output power were used. This instrument had a spatial resolution of 2 or 10 μπι, depending on the objective used.

Results

SEM/EDS, XRD, and luminescence characterization data suggested that the CVD alumina coating directly deposited on René N5 was essentially made of ~1 μπι α-Α1203 crystals in a matrix of amorphous AI2O3. Figure la shows that the coating surface consisted of two distinct features: (i) faceted crystals of ~0.5-1.5 μιη and (ii) a glassy, web-like phase. The coating surface was not fully "dense" due to the lack of coalescence among the crystals as well as the presence of the glassy phase between the crystals. Figure lb shows that the coating was ~1 μηι thick, and contained Al and O only when examined by EDS. The interface at the coating layer and the alloy surface did not appear to be particularly sharp, although the alloy surface was polished to -0.05 μπι before the coating process.

René N5 (b)

Figure 1. Surface and cross-section images of CVD-AI2O3 on René N5: (a) and (b) as-deposited and (c) and (d) after annealing and oxidation.

The Θ-2Θ XRD pattern in Figure 2a suggested that the coating contained OC-AI2O3 crystals that were randomly orientated. The penetration of the XRD beam appeared to be relatively small as the substrate peak was barely visible from the Θ-2Θ XRD scan, but significant broadening of the

1Q

baseline was observed. The high background suggested the presence of some amorphous content within and/or underneath the coating layer, although a more quantitative measure could not be possible with the XRD analysis. The GIXRD scan in Figure 2b showed that the top region of the coating layer was mostly α-Α1?03.

As shown in Figure 3, relatively broad Rl and R2 fluorescence peaks were observed for the coating layer with a 10-μπι laser probe. A smaller 2-μηι probe revealed that the faceted crystals (-0.5-1.5 μιη) shown in the SEM micrograph (Figure la) were indeed OÍ.-AI2O3. The Rl and R2 peaks became more intense and shaper as the probe was focused on the crystals. On the other hand, the glassy phase was found to contain neither a- nor Θ-ΑΙ2Ο3, as evidenced by significant reductions in the intensity of the Rl and R2 peaks. Due to the coefficient of thermal expansion (CTE) mismatch between the α-Αΐ2θ3 coating and the René N5 substrate, compressive residual stress was expected to develop in the in-plane direction of the coating layer upon cooling from the CVD deposition temperature to room temperature. The average stress in the coating layer was estimated to be -3.2 GPa. However, the broadness of the Rl and R2 peaks suggested a considerable stress distribution in the coating layer.

(b ) a : a -A 120 3

S : S u b s t r a t e

20 3 0 40 50 60 70 2Θ

Figure 2. XRD patterns of CVD-AI2O3 on René N5: (a) Θ-2Θ scan and (b) grazing incidence angle of Io.

The coating deposited on René N5 substrate was annealed in the HTXRD chamber under vacuum as the temperature was gradually increased from 1020 to 1150°C over a period of 2.5 hours. After the sample was cooled to room temperature, it was examined by luminescence spectroscopy and XRD. Subsequently, the same sample was oxidized in the HTXRD chamber in air at 1200°C for 1 hour. Comparison of the SEM micrographs before and after the annealing

20