A

Oco(lNt©

K

1

mreattaahZtta

ZaR

b

0h

Available online at www.sciencedirect.com

Journal of the European Ceramic Society 33 (2013) 1591–1598

Oxidation of zirconium diboride with niobium additions

Maryam Kazemzadeh Dehdashti ∗, William G. Fahrenholtz, Greg E. HilmasDepartment of Materials Science and Engineering, Missouri University of Science and Technology, Rolla, MO 65401, United States

Received 20 January 2013; accepted 26 January 2013Available online 17 February 2013

bstract

xidation of ZrB2 ceramics containing Nb additions at 1500 ◦C resulted in the formation of a two-layer oxide scale. The outer surface was partiallyovered by a glassy layer containing B2O3 with smaller amounts of Nb and Zr oxides dissolved into it. With increasing exposure time, evaporationf B2O3 from the outer layer resulted in precipitation of oxide particles in the receding glassy phase. Between the outer layer and the unoxidizedZr,Nb)B2 was a porous layer that consisted of particles containing Zr, Nb, and O. The formation of Nb2Zr6O17 was observed in the porous oxideayer. Since this compound is solid at the oxidation temperature, liquid phase sintering of the ZrO2 scale was not possible. However, dissolution of

b into B2O3 increased the stability of the liquid/glassy layer, which acted as a barrier to the transport of oxygen at higher temperatures compared

o the scale formed on nominally pure ZrB2. 2013 Elsevier Ltd. All rights reserved.

eywords: ZrB2; Nb; Oxidation; ZrO2; Composites

at

Z

∼lficdSioig

. Introduction

Ultra high temperature ceramics (UHTCs) are a group ofaterials that includes ZrB2, ZrC, HfB2 and HfC. These mate-

ials are candidates for applications that require exposure toxtreme thermal and chemical environments. The performancedvantages of the diboride-based UHTCs come not only fromheir high-temperature stability but also from the capability toransfer and redistribute heat at elevated temperatures. This char-cteristic is attractive for sharp leading edges for hypersonicerospace vehicles, which must transfer heat away from theottest areas and redistribute it to cooler areas.1 Among UHTCs,rB2 has the lowest theoretical density combined with reported

hermal conductivity values as high as ∼100 W/m K at roomemperature, which is an advantage over other candidates forerospace applications.2

Oxidation behavior is a restriction to the development of

rB2-based ceramics for rocket propulsion and hypersonic flightpplications. Assuming stoichiometric oxidation according toeaction (1), exposure of ZrB2 to air at temperatures of 800 ◦C

∗ Corresponding author. Tel.: +1 573 578 2853; fax: +1 573 341 6934.E-mail addresses: mk7y8@mst.edu (M. Kazemzadeh Dehdashti),

illf@mst.edu (W.G. Fahrenholtz), ghilmas@mst.edu (G.E. Hilmas).

taalsffu

955-2219/$ – see front matter © 2013 Elsevier Ltd. All rights reserved.ttp://dx.doi.org/10.1016/j.jeurceramsoc.2013.01.033

nd above results in formation of B2O3 and ZrO2, which leadso measurable mass gain.3

rB2(cr) + 5/2 O2(g) → ZrO2(cr) + B2O3(l) (1)

Evaporation of B2O3 is considerable at temperatures above1200 ◦C. The loss of B2O3(l) leaves behind a porous ZrO2

ayer with a columnar microstructure, which offers channelsor rapid oxygen transport to the reaction interface and resultsn significant mass gain at temperatures above 1200 ◦C.4 Theonventional approach to improving the oxidation resistance ofiboride ceramics is to add Si-containing compounds such asiC1,5–13, MoSi214–16, or TaSi2.17,18 The formation of a borosil-

cate layer on the surface of the diborides provides improvedxidation resistance in air compared to the borate glass on nom-nally pure ZrB2 due to the increased stability of the borosilicatelass compared to the borate material.3,19 At elevated tempera-ures, SiO(g) forms beneath the borosilicate glass as a result ofctive oxidation of SiC. When the pressure of SiO(g) exceedsmbient, the resulting pressure can rupture the protective glassyayer, which can result in a cyclic protective/non-protective

20

cale-forming sequence. Further, some authors have noted theormation of the SiC-depleted layer in ZrB2-SiC samples, whichacilitates the transport of oxygen through the oxide scale to thenoxidized matrix.21,22 Hence, SiC may not be the best choice

1 e Eur

fh

acidroanWZgstrocNtd1sttotii

2

Nttttbwsae2Cpcwpmem

Tal

wdwfintuBaBwt

hThwosae1(rtaf

sIouNtcoH

3

3

tamacvtfwh

592 M. Kazemzadeh Dehdashti et al. / Journal of th

or improving the oxidation resistance of ZrB2 ceramics at ultraigh temperatures.

Several studies showed that additions of Cr-, Ti-, Nb-, V-,nd Ta-borides improved the oxidation resistance of ZrB2–SiComposites.22–26 Hence, additions of transition metals offermproved oxidation resistance to ZrB2 ceramics without theeleterious effects of silica formers. Similarly, Zhang et al.eported that WC additions improved the oxidation resistancef ZrB2 ceramics. The addition of WC led to the formation of

two-layer scale structure, which consisted of a porous zirco-ia outer layer and a dense inner layer containing ZrO2 andO3, in contrast with the single, highly porous and columnar

rO2 layer formed on nominally pure ZrB2.27,28 It was sug-ested that during oxidation, the presence of WO3 in the oxidecale resulted in liquid phase sintering of ZrO2, which increasedhe relative density of the scale, resulting in improved oxidationesistance.28 Other transition metal additives, such as Nb, mayffer similar beneficial effects on the oxidation behavior of zir-onium diboride. Unlike the presence of WO3, the formation ofb2O5 during oxidation is not expected to result in the forma-

ion of a liquid phase. Examination of the Nb2O5–ZrO2 phaseiagram shows that at 1500 ◦C, the presence of small (less than0 mol%) concentrations of Nb should lead to the formation ofolid compounds such as Nb-doped ZrO2 or Nb2Zr6O17 ratherhan a liquid phase, such as the WO3–ZrO2 solution predicted forhe presence of small amounts of W with ZrO2.29,30 The purposef this paper was to examine the effect of niobium additions onhe oxidation behavior of ZrB2 at elevated temperatures to gainnsight into the mechanism by which transition metal additionsmprove oxidation resistance.

. Experimental procedure

High purity (>99%) ZrB2 powder (Grade B, H.C. Starck,ewton, MA) with an average particle size of ∼2 �m was used

o prepare the specimens for this study. To enhance densifica-ion, 2 wt% B4C (∼0.8 �m, Grade HS, H.C. Starck) was addedo all batches to react with and remove oxide impurities fromhe powder particle surfaces. For some batches, 6 mol% nio-ium was added in form of Nb powder (Johnson Matthey, MA),hich had an average particle size of ∼1 �m. To reduce particle

ize and promote intimate mixing, the as-received ZrB2, B4C,nd Nb (for Nb containing batches) were dispersed in methylthyl ketone (MEK) by ball milling with zirconia media for4 h. An organic dispersant (DISPERBYK-110, BYK-Chemieo., Wesel, Germany) was added at a level of 0.54 mg of dis-ersant per m2 of ZrB2 surface area. The amount of zirconiaontamination added to the batches as a result of ball millingas determined to be less than 1 wt% based on the mass of ZrB2owder by measuring the mass of the media before and afterilling. After mixing, the solvent was removed using rotary

vaporation, and then the powder was ground and sieved to −80esh.

Powders were densified by hot pressing (Model HP-3060,

hermal Technology, Santa Rosa, CA) at 2100 ◦C for 45 min at pressure of 32 MPa. Powders were loaded into a graphite dieined with graphite foil that was coated with BN spray. Billets

lpmT

opean Ceramic Society 33 (2013) 1591–1598

ith a diameter of ∼25 mm and a thickness of ∼5 mm were pro-uced. Specimens with dimensions of 10 mm by 4 mm by 4 mmere diced from the billets and polished on all sides to a 15 �mnish for testing and characterization. Images obtained by scan-ing electron microscopy (SEM; S-4700, Hitachi, Japan) fromhe polished surfaces of as processed (Zr,Nb)B2 and ZrB2 weresed to study the microstructure of the specimens. The amount of4C remaining after densification was calculated using imagenalysis software (ImageJ, U.S. National Institutes of Health,ethesda, MD). The bulk densities of the hot pressed billetsere measured using the Archimedes technique with water as

he immersing medium.Oxidation studies were performed in a MoSi2 resistance-

eated horizontal tube furnace (Model 0000543 Rapidemperature Furnace, CM Inc., Bloomfield, NJ) equipped with aigh-purity alumina tube with a diameter of 6.35 cm. Specimensere cleaned in acetone in an ultrasonic bath and then placedn a zirconia foam setter that was on an alumina D-tube. Thepecimen assembly was inserted into the center of the furnacend leveled. The ends of the tube were sealed using gas-tightnd caps. Specimens were heated at ∼5 ◦C/min to 1500 ◦C or600 ◦C and held for up to 3 h in air with a flow rate of 0.2 cm/slinear flow rate was calculated according to the volumetric flowate and the size of the tube). To minimize changes such as fur-her oxidation that may occur during cooling, specimens wereir quenched to room temperature by removing them from theurnace after the desired oxidation time.

The thicknesses of the resulting oxidation layers were mea-ured from fracture surfaces that were observed in SEM.n addition, the microstructures of the oxide scales werebserved using SEM and chemical compositions were analyzedsing energy dispersive spectroscopy (EDS; EDAX, Mahwah,J). X-ray diffraction (XRD; Philips X-Pert Pro diffractome-

er, Westborough, MA) analysis was used to identify majorrystalline phases present in both the pre-oxidized and the post-xidized composites and the data were analyzed using X‘Pertigh Score software.

. Results and discussion

.1. Densification, microstructure, and phase analysis

A microstructure typical of the (Zr,Nb)B2 specimens used inhis investigation is presented in Fig. 1. The darker phase is B4Cnd it appears to be uniformly dispersed in the lighter (Zr,Nb)B2atrix. Based on image analysis, the amount of B4C remaining

fter densification was 1.4 wt%. The average bulk density of barsut from hot-pressed (Zr,Nb)B2 billets was 6.03 g/cm3. Using aolumetric rule of mixtures calculation, and assuming true densi-ies of 6.09 g/cm3 for ZrB2, 2.52 g/cm3 for B4C, and 8.57 g/cm3

or Nb, the theoretical density of ZrB2 containing 6 mol% Nbas calculated to be 6.05 g/cm3. Using this true density, theot-pressed bars had relative densities of >99%. The calcu-

ated relative density is consistent with the minimal amount oforosity revealed by SEM analysis. Also, Archimedes’ measure-ents showed the amount of open porosity to be insignificant.hus, porosity was not considered to have a significant effect on

M. Kazemzadeh Dehdashti et al. / Journal of the European Ceramic Society 33 (2013) 1591–1598 1593

FZ

tSta

3

isstempdadiZs1iNootNAom

(

qmp

Fig. 2. Surface of (Zr,Nb)B2 oxidized at 1500 ◦C for (a) 0 h, (b) 3 h, showingtp

w(ats

tHpsbAg1pc

ig. 1. SEM image of a polished cross section showing the microstructure ofrB2 containing 1.5 vol% B4C and 6 mol% Nb.

he oxidation behavior. Microstructure and phase analysis usingEM and XRD were consistent with the dissolution of Nb into

he matrix to form a (Zr,Nb)B2 solution, which was identifieds hexagonal by indexing to ZrB2 (PDF card number 34-0423).

.2. Surface morphology and composition

Assuming that the oxidation of (Zr1−xNbx)B2 proceeds sto-chiometrically, reaction at temperatures of 800 ◦C and abovehould produce molten B2O3 (melting temperature ∼450 ◦C),olid ZrO2, and solid Nb2O5 in the molar ratios shown in Reac-ion (2). In this case, the addition of 6 mol% Nb to ZrB2 isquivalent to x = 0.06, which produces an oxide scale with aolar ratio of Nb2O5 to ZrO2 of 1 to 33 or 93.5 wt% ZrO2

lus 6.5 wt% Nb2O5. According to the ZrO2–Nb2O5 phaseiagram29,30, the primary crystalline phases that should formre a monoclinic solid solution based on ZrO2 that containsissolved Nb and an orthorhombic compound, Nb2Zr6O17. Byncreasing the temperature, the solubility limit of Nb2O5 intorO2 increases and the ratio of Nb2Zr6O17 to the ZrO2 solidolution decreases. The melting temperature of Nb2Zr6O17 is670 ◦C and the solubility limit of Nb2O5 in ZrO2 at 1500 ◦Cs about 7 mol% (14 mol% Nb). Due to the formation of solidb2Zr6O17, no liquid phase is predicted for the compositionf the oxide scale, which should be ZrO2 containing 6 mol%f dissolved Nb at 1500 ◦C. Upon cooling to room tempera-ure, the solubility limit of Nb2O5 in ZrO2 decreases and someb2Zr6O17 should precipitate from the ZrO2 solid solution.t room temperature, Nb2Zr6O17 comprises about 18 mol%f the oxide phase along with an amorphous phase that isainly B2O3.

Zr1−xNbx)B2(cr) + (5/2 + x/4) O2(g) → (1−x) ZrO2(cr)

+ x/2 Nb2O5(cr) + B2O3(l) (2)

◦

The surface of (Zr,Nb)B2 oxidized at 1500 C for 0 h (i.e.,uenched as soon as it reached 1500 ◦C) is shown in Fig. 2a. Aajority of the surface of the specimen was covered by a dark

hase that had a glassy appearance (∼90% of the surface area)

pcoB

he presence of a glassy oxide (dark phase) and crystalline oxide particles (lighthase).

ith a small fraction of the surface composed of oxide particles∼10% of the surface area). After 3 h at 1500 ◦C (Fig. 2b), therea fraction of the glassy phase had decreased to ∼60% andhe glassy phase was concentrated in several pools that wereurrounded by oxide particles.

Due to low sensitivity of EDS to light elements, quantifica-ion of the boron content in the glassy phase was not possible.owever, EDS results indicated that the matrix of the glassyhase contained O and Nb along with a small amount of Zr, pre-umably all dissolved in B2O3. Small particles that containedoth Zr and Nb (Fig. 3a) were also observed in the glassy phase.ccording to the ZrO2–B2O3

31 and Nb2O5–B2O332 phase dia-

rams, approximately 12 mol% ZrO2 can dissolve into B2O3 at500 ◦C while both Nb2O5 and B2O3 are liquids at that tem-erature. Hence, the particles observed at room temperatureould be ZrO2, Nb2O5, or an oxide containing Zr and Nb. Thearticles could have formed either during oxidation or during

ooling. They could form during oxidation due to evaporationf B2O3 that would result in supersaturation of the remaining2O3 with ZrO2, which could result in precipitation. Conversely,

1594 M. Kazemzadeh Dehdashti et al. / Journal of the European Ceramic Society 33 (2013) 1591–1598

Fig. 3. SEM images of the surface of a (Zr,Nb)B2 specimen oxidized at 1500 ◦Cfor 3 h that were (a) close to the edge of the liquid pool, and (b) in the middle ofthe liquid pool. The images show a homogeneous distribution of oxide particlesc

tstAcfaNtfu(mbft

f

Fig. 4. SEM images of a (Zr,Nb)B2 specimen oxidized at 1500 ◦C for 3 h (a) atthe edge of the liquid pool, (b) in the crystalline oxide region close to the liquidpool, showing the growth of the oxide particles by joining the small precipitatedp

Noetog(l

1awpbremained, particularly in the middle of the glassy pools, the pre-

ontaining Zr and Nb and formation of elongated particles.

he particles could precipitate from the glassy phase when thepecimen was cooled from the processing temperature due tohe change in solubility of ZrO2 in B2O3 with temperature.ccording to the ZrO2–Nb2O5 phase diagram29, two different

rystalline phases could form when ZrO2 and Nb2O5 precipitaterom the B2O3 melt. For Nb2O5 contents less than about 5 mol%,

(Zr1−xNbx)O2+0.5x solid solution is the stable phase. If theb2O5 concentration in the glassy phase is higher, then the crys-

alline phase Nb2Zr6O17 (also designated 6ZrO2·Nb2O5) couldorm in addition to the ZrO2 solid solution. X-ray diffraction wassed to characterize the phases present on the surface of oxidizedZr,Nb)B2. The major phases detected were triclinic H3BO3,onoclinic ZrO2 and orthorhombic Nb2Zr6O17 (PDF card num-

ers 30-0199, 83-0944 and 72-1745, respectively). Boric acidormed after cooling the specimen to room temperature due to

he instability of B2O3 in the humid ambient air.

With increasing exposure time, more B2O3 should evaporaterom the surface, which would increase precipitation of Zr- and

cFt

articles.

b-rich oxide particles from the liquid phase. Near the edgesf the pools, spherical particles precipitated as the B2O3 liquidvaporated. Fig. 3a shows that the particles were uniformly dis-ributed in the glassy phase and that they had an average diameterf about 0.5 �m. Closer to the centers of the glassy pools, elon-ated precipitates appeared to grow from the spherical particlesFig. 3b). The elongated precipitates were typically about 3 �mong.

As the glassy phase receded during extended exposure at500 ◦C, the underlying oxide particles were revealed (Fig. 4and b). As the liquid receded, the small spherical particles thatere observed in the glassy phase appeared to attach to the largerarticles that were exposed. Some smaller particles were visi-le between the larger ones. In areas where the glassy phase

ipitated particles became more concentrated. As can be seen inig. 5, clusters of elongated particles formed in the areas where

he last of the glassy pools were present.

M. Kazemzadeh Dehdashti et al. / Journal of the European Ceramic Society 33 (2013) 1591–1598 1595

Fig. 5. SEM image of the (Zr,Nb)B2 specimen oxidized at 1500 ◦C for 3 hst

3

rpt

snldfttll

fworasc

F

F ◦a

etwB

g

howing the clustering of the equiaxed and elongated particles on the surface ofhe crystalline oxide particles.

.3. Oxide scale morphology and thickness

Initial attempts to polish cross sections of oxidized samplesevealed that the oxide scales were damaged by the preparationrocess. To produce cross sections that were representative ofhe oxide scales, fracture surfaces were examined.

Fig. 6 shows a low magnification SEM image of the fractureurface of a (Zr,Nb)B2 sample oxidized at 1500 ◦C for 3 h. Sig-ificant differences were observed in the thickness of the glassyayer between the middle and edge of the glassy pool. After oxi-ation at 1500 ◦C for 3 h the thickness of the glassy layer rangedrom a maximum of about 0 �m to ∼40 �m. In addition, thehickness of the porous oxide layer ranged from about 45 �mo 60 �m. Areas of thick glassy oxide had thinner porous oxideayers while areas with thinner glassy oxide had thicker porousayers.

Fig. 7 shows cross sectional SEM images of the oxide scalesrom regions with maximum glassy layer thickness. The scalesere formed on the surfaces of the (Zr,Nb)B2 samples afterxidation at 1500 ◦C for 0, 1.5 and 3 h. The oxide scales in theseegions consisted of two layers: (1) a dense outer glassy layer,

nd (2) an inner layer that appeared to be porous. The two-layercale is believed to be formed due to volume expansion upononversion of ZrB2 to ZrO2 and B2O3, which is ∼300% volume

ig. 6. Fracture surface of a (Zr,Nb)B2 specimen oxidized at 1500 ◦C for 3 h.

trioena

ftnfeirW

ig. 7. Fracture surfaces of (Zr,Nb)B2 oxidized at 1500 C for: (a) 0 h, (b) 1.5 h,nd (c) 3 h.

xpansion based on density calculations. Oxidation produceswo phases because of the immiscibility of the two materialshile the large volume expansion associated with formation of2O3 causes it to be forced to the surface of the specimen.3

Several small elongated particles can be observed in thelassy layer in Fig. 7a and b. The particles were uniformly dis-ributed through the thickness of the glassy layer. Also, someesidual glassy phase was observed between the oxide particlesn the porous layer. The oxide particles in the porous layer ofxidized (Zr,Nb)B2 were less than 10 �m in diameter and hadquiaxed shapes. For comparison, the scale formed on nomi-ally pure ZrB2 was composed of larger ZrO2 particles that hadn elongated morphology.33

Fig. 8 shows the results of the scale thickness measurementsor nominally pure ZrB2 compared to (Zr,Nb)B2 after oxida-ion at 1500 ◦C for 0, 1.5 and 3 h. While the scales formed onominally pure ZrB2 were mostly uniform, considerable dif-erences in the thickness of the scales between the middle anddge of the glassy pools were observed in the (Zr,Nb)B2 spec-

mens. Therefore, the measurements were performed on theegions with maximum glassy layer thickness for (Zr,Nb)B2.

hen nominally pure ZrB2 and (Zr,Nb)B2 reached 1500 ◦C,

1596 M. Kazemzadeh Dehdashti et al. / Journal of the Eur

F ◦n

tt(otgtgwoZ3o1s

deov1oo

iDaeta

ptbflwwtttntkNt

3

tapuddtwteqBigvious studies have shown that the addition of W to ZrB2 leads to

ig. 8. Scale thickness as a function of oxidation time at 1500 C comparingominally pure ZrB2 to areas with the maximum glassy thickness for (Zr,Nb)B2.

he glassy layers for both materials were ∼4 �m thick, whilehe porous oxide scale was ∼26 �m for ZrB2 and ∼18 �m forZr,Nb)B2. After 1.5 h at 1500 ◦C, the glassy and porous layersn the nominally pure ZrB2 were ∼2 and ∼59 �m thick, respec-ively. For the same oxidation time, (Zr,Nb)B2 had a maximumlassy layer thickness of ∼18 �m. The porous layer beneathhe glassy layer with the maximum thickness was ∼35 �m. Nolassy layer was observed on pure ZrB2 after 3 h at 1500 ◦C,hile the glassy layer on (Zr,Nb)B2 had a maximum thicknessf ∼37 �m. The thickness of the porous layer on nominally purerB2 was ∼73 �m compared to ∼45 �m for (Zr,Nb)B2 after

h at 1500 ◦C. Although the surface area of the glassy poolsn (Zr,Nb)B2 decreased with increasing the exposure time at500 ◦C, the thickness of the glassy layer increased with expo-ure time.

For nominally pure ZrB2, the thickness of the glassy layerecreased with increasing oxidation time at 1500 ◦C due tovaporation of B2O3. Previous studies have reported that evap-ration of B2O3 is substantial above 1100 ◦C due to its highapor pressure.3 In contrast, increasing the oxidation time at

500 ◦C for (Zr,Nb)B2 resulted in an increase in the thicknessf the glassy layer. The presence of glassy phase after oxidationf (Zr,Nb)B2 at 1500 ◦C for 3 h indicated that the addition of Nb

ltt

Fig. 9. Model of the evolution of the

opean Ceramic Society 33 (2013) 1591–1598

ncreased the stability of the liquid phase at high temperatures.issolution of Nb2O5 into the liquid phase should decrease the

ctivity of B2O3 and, consequently, reduce its vapor pressure andvaporation rate. The presence of a glassy layer should improvehe oxidation resistance of the ceramic since borate glasses acts a barrier to oxygen transport.34

After oxidation at 1500 ◦C for 3 h, the thickness of theorous layer was ∼75 �m for nominally pure ZrB2 comparedo ∼45 �m for (Zr,Nb)B2. Based on this observation, the sta-ility of the glassy layer resulted in a decreased oxidation rateor (Zr,Nb)B2. The addition of Nb increased the stability of theiquid phase, providing better protection at high temperatures,hich decreased the oxidation rate of the (Zr,Nb)B2 comparedith nominally pure ZrB2. Not only was the initial thickness of

he porous oxide layer thinner for (Zr,Nb)B2 (∼18 �m comparedo ∼26 �m for nominally pure ZrB2), the porous oxide layer washinner after 3 h at 1500 ◦C (∼45 �m compared to ∼75 �m forominally pure ZrB2). Based on the calculations performed onhe thickness of the porous layers, the oxidation exhibited linearinetic behavior (R2 > 0.94) at 1500 ◦C. Further, the addition ofb to ZrB2 resulted in a lower oxidation rate (9 �m/h) compared

o nominally pure ZrB2 (16 �m/h).

.4. Evolution of structure

Fig. 9 is a schematic description of the evolution of the struc-ure of the oxide scale on (Zr,Nb)B2. At the early stages, Zr, Nb,nd B oxides form as (Zr,Nb)B2 is oxidized. The Zr–Nb oxidearticles form on the (Zr,Nb)B2 surface, but are covered by a liq-id phase composed of mainly B2O3 with smaller amounts ofissolved Nb and Zr oxides. During oxidation, Nb and Zr oxidesissolve into the liquid borate phase. At higher temperatures,he evaporation rate of B2O3 from the outer surface increases,hich results in precipitation of particles in the liquid. The par-

icles contain Zr, Nb, and O and join together to form either thequiaxed or elongated grains visible in the glassy phase of theuenched specimens (e.g., Fig. 3). As exposure time increases,2O3 evaporation continues and the liquid phase concentrates

n pools that are separated by regions of crystalline oxide. As thelass recedes, it exposes the underlying porous oxide scale. Pre-

iquid phase sintering of the porous ZrO2 scale, which improvedhe oxidation resistance by decreasing oxygen transport throughhe scale. In contrast, Nb2O5 and ZrO2 form a solid Zr–Nb–O

oxide structure of (Zr,Nb)B2.

e Eur

cntofgemoop

4

Al(aocgcetwttilfBupc

A

S1(pa

R

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

M. Kazemzadeh Dehdashti et al. / Journal of th

ompound at the oxidation temperature, namely Nb2Zr6O17, soo liquid phase sintering of ZrO2 occurred. However, dissolu-ion of Nb2O5 into the B2O3 liquid phase increased the stabilityf the liquid phase compared to the nominally pure B2O3 thatorms when ZrB2 is oxidized. The improved stability of thelassy layer leads to improved oxidation behavior because thexternal layer and glassy phase trapped among the particles thatake up the porous oxide layer acts as a barrier to the transport

f oxygen. Therefore, the addition of Nb to ZrB2 increased thexidation resistance, but only as long as the Nb-containing B2O3hase was present.

. Conclusion

The oxidation behavior of (Zr,Nb)B2 ceramics was studied.t 1500 ◦C, exposure to air resulted in the formation of a two-

ayer oxide scale structure on (Zr,Nb)B2. The two layers were:1) an outer layer of a glassy phase containing B2O3 with Nbnd Zr dissolved in it; and (2) a porous oxide layer composedf oxide particles containing Zr and Nb. Small spherical parti-les, presumably a ZrO2 containing dissolved Nb, grew in thelassy phase with increasing exposure time. Some of the parti-les were spherical while others were elongated. As the B2O3vaporated, the particles became concentrated and were even-ually incorporated into the newly exposed porous oxide layer,hich contained both ZrO2 and Nb2Zr6O7. As the glass receded,

he small precipitated particles joined the porous oxide layerhat was present under the glassy layer. Because of high melt-ng point of Nb2Zr6O7 that was formed in the porous oxideayer, liquid phase sintering was not active as has been reportedor W-containing ZrB2. However, dissolution of Nb into the2O3 liquid phase increased the stability of the protective liq-id layer by reducing the volatility of B2O3 from the liquidhase. Hence (Zr,Nb)B2 showed improved oxidation resistanceompared with pure ZrB2.

cknowledgments

This work was supported as part of the National Hypersoniccience Center for Materials and Structures (Grant FA9550-09--0477) with Dr. Ali Sayir (AFOSR) and Dr. Anthony CalominoNASA) as program managers. The authors wish to thank projectrincipal investigator Dr. David Marshall of Teledyne Scientificnd Imaging for his support and guidance.

eferences

1. Monteverde F, Savino R, Fumo MDS, Di Maso A. Plasma wind tunneltesting of ultra-high temperature ZrB2–SiC composites under hypersonicre-entry conditions. J Eur Ceram Soc 2010;30(11):2313–21.

2. Guo SQ. Densification of ZrB2-based composites and their mechanical andphysical properties: a review. J Eur Ceram Soc 2009;29(6):995–1011.

3. Rezaie A, Fahrenholtz WG, Hilmas GE. Evolution of structure during theoxidation of zirconium diboride-silicon carbide in air up to 1500 ◦C. J Eur

Ceram Soc 2007;27(6):2495–501.

4. Parthasarathy TA, Rapp RA, Opeka M, Kerans RJ. Effects of phase changeand oxygen permeability in oxide scales on oxidation kinetics of ZrB2 andHfB2. J Am Ceram Soc 2009;92(5):1079–86.

2

3

opean Ceramic Society 33 (2013) 1591–1598 1597

5. Rezaie A, Fahrenholtz WG, Hilmas GE. Oxidation of zirconium diboride-silicon carbide at 1500 ◦C in a low partial pressure of oxygen. J Am CeramSoc 2006;89(10):3240–5.

6. Carney CM, Mogilvesky P, Parthasarathy TA. Oxidation behavior of zir-conium diboride silicon carbide produced by the spark plasma sinteringmethod. J Am Ceram Soc 2009;92(9):2046–52.

7. Guo WM, Zhang GJ. Oxidation resistance and strength retention ofZrB2–SiC ceramics. J Eur Ceram Soc 2010;30(11):2387–95.

8. Monteverde F, Savino R, Fumo MDS. Dynamic oxidation of ultra-hightemperature ZrB2-SiC under high enthalpy supersonic flows. Corros Sci2011;53(3):922–9.

9. Zhang X, Xu L, Du S, Han W, Han J. Preoxidation and crack-healing behav-ior of ZrB2–SiC ceramic composite. J Am Ceram Soc 2008;91(12):4068–73.

0. Karlsdottir SN, Halloran JW, Henderson CE. Convection patterns in liquidoxide films on ZrB2–SiC composites oxidized at a high temperature. J AmCeram Soc 2007;90(9):2863–7.

1. Akin I, Cinar Sahin F, Yucel O, Goller G. Oxidation behavior of zirconiumdiboride-silicon carbide composites. In: 34th International Conference onAdvanced Ceramics and Composites, 2010. 2010. p. 105–11.

2. Karlsdottir SN, Halloran JW. Oxidation of ZrB2–SiC: influence of SiCcontent on solid and liquid oxide phase formation. J Am Ceram Soc2009;92(2):481–6.

3. Mallik M, Ray KK, Mitra R. Oxidation behavior of hot pressed ZrB2-SiCand HfB2-SiC composites. J Eur Ceram Soc 2011;31(1-2):199–215.

4. Guo S, Mizuguchi T, Ikegami M, Kagawa Y. Oxidation behav-ior of ZrB2–MoSi2–SiC composites in air at 1500 ◦C. Ceram Int2011;37(2):585–91.

5. Chamberlain AL, Fahrenholtz WG, Hilmas GE, Ellerby DT. Characteriza-tion of zirconium diboride for thermal protection systems. In: 8th Conferenceand Exhibition of the European Ceramic Society. 2004. p. 493–6.

6. Sciti D, Brach M, Bellosi A. Oxidation behavior of a pressureless sinteredZrB2–MoSi2 ceramic composite. J Mater Res 2005;20(4):922–30.

7. Talmy IG, Zaykoski JA, Opeka MM. High-temperature chemistry and oxi-dation of ZrB2 ceramics containing SiC, Si3N4, Ta5Si3, and TaSi2. J AmCeram Soc 2008;91(7):2250–7.

8. Sciti D, Medri V, Silvestroni L. Oxidation behaviour of HfB2-15 vol.% TaSi2at low, intermediate and high temperatures. Scr Mater 2010;63(6):601–4.

9. Opila EJ, Halbig MC. Oxidation of ZrB2–SiC. In: 25th Annual Conferenceon Composites, Advanced Ceramics, Materials and Structures. 2001. p.221–8.

0. Opeka MM, Talmy IG, Zaykoski JA. Oxidation-based materials selection for2000 ◦C + hypersonic aerosurfaces: theoretical considerations and historicalexperience. J Mater Sci 2004;39(19):5887–904.

1. Opila EJ, Halbig MC. Oxidation of ZrB2-based ultra-high temperatureceramics. Ceram Eng Sci Proc 2001;22(3):221–8.

2. Levine SR, Opila EJ. Tantalum addition to zirconium diboride for improvedoxidation resistance: NASA/TM-2003-212483; 2003.

3. Peng F, Berta Y, Speyer RF. Effect of SiC, TaB2 and TaSi2 additives on theisothermal oxidation resistance of fully dense zirconium diboride. J MaterSci 2009;24(5):1855–67.

4. Talmy IG, Zaykoski JA, Opeka MM, Dallek S. Oxidation of ZrB2 ceramicsmodified with SiC and group IV–VI transition metal diborides. ElectrochemSoc Proc 2001;12:144–58.

5. Hu P, Zhang XH, Han JC, Luo XG, Du SY. Effect of various additives onthe oxidation behavior of ZrB2-based ultra-high-temperature ceramics at1800 ◦C. J Am Ceram Soc 2010;93(2):345–9.

6. Zhang XH, Hu P, Han JC, Xu L, Meng SH. The addition of lanthanumhexaboride to zirconium diboride for improved oxidation resistance. ScrMater 2007;57(11):1036–9.

7. Zhang SC, Hilmas GE, Fahrenholtz WG. Improved oxidation resistanceof zirconium diboride by tungsten carbide additions. J Am Ceram Soc2008;91(11):3530–5.

8. Zhang SC, Hilmas GE, Fahrenholtz WG. Oxidation of zirconium diboridewith tungsten carbide additions. J Am Ceram Soc 2011;94(4):1198–205.

9. Wang F, Northwood DO. Oxides formed between ZrO2 and Nb2O5. J MaterSci 1995;30(16):4003–8.

0. Roth RS, Coughanour LW. Phase equilibrium relations in the systemstitania–niobia and zirconia–niobia. J Res Nat Bur Stand 1955;55:209–13.

1 e Eur

3

3

3

598 M. Kazemzadeh Dehdashti et al. / Journal of th

1. Karlsdottir SN, Halloran JW, Grundy AN. Zirconia transport by liquid con-

vection during oxidation of zirconium diboride–silicon carbide. J Am CeramSoc 2008;91(1):272–7.

2. Levin EM. Phase equilibria in the system niobium pentoxide-boric acid. JRes Natl Bur Stand, Sect A 1966;70A(1):11–6.

3

opean Ceramic Society 33 (2013) 1591–1598

3. Fahrenholtz WG. The ZrB2 volatility diagram. J Am Ceram Soc

2005;88(12):3509–12.

4. Kuriakose AK, Margrave JL. The oxidation kinetics of zirconiumdiboride and zirconium carbide at high temperatures. J Electrochem Soc1964;111(7):827–31.