Influence of Oxidation Healing for Cracks on the Strengthof Hot-Pressed ZrB2–SiC–AlN Ceramics

Jun Liang,* Yu Wang, Guodong Fang, and Guangfei Wang

Key Laboratory of Science and Technology for National Defence, Center for Composite Materials andStructure, Harbin Institute of Technology, Harbin 150080, People’s Republic of China

Precracks were made using a Vickers indenter on hot-pressed ZrB2–20 vol% SiC–10 vol% AlN ceramics (sample ZSA).The residual strength for the preindented samples oxidized at 10001C for different durations of time was examined using athree-point bending test. According to the analysis of SEM and EDS on the surface and cross-section of the ZSA composite,the evolution of the oxidizing layer was obtained for the precracked samples. The influence of oxidation on the crack healingand residual strength was also investigated. The residual strength of the samples after oxidation increased with an increase ofthe oxidation time. The oxidation layer versus oxidation time and the crack-healing mechanism were also studied. When theoxidation time approached 60 min, the strength reached the maximum value. With the continuous increase of the oxidationtime, the residual strength began to decrease.

Introduction

ZrB2-based ceramics have been regarded as prom-ising candidate materials for high-temperature structuralapplications, especially for reusable thermal protectionsystems of hypersonic vehicles1–4 exposed to a high-temperature oxidation atmosphere. Therefore, whetheroxidation has an influence on the service performance or

not for ZrB2-based ceramics with a microcrack or a mi-crodefect should be investigated. Considerable attentionhas been focused on the influence of oxidationfor ZrB2–SiC ceramics. Chamberlain et al.5 studiedthe oxidation behavior of ZrB2–SiC ceramics underdifferent oxidation conditions, which revealed thepresence of a SiC depletion region between the protec-tive oxide and unaltered ZrB2–SiC when the ceramicswere oxidized at 15001C in a furnace. But the oxidizingstructures, which were composed of a layer of ZrO2, apartially oxidized layer, and a SiC depletion layer,

Ceramic Product Development and Commercialization

*liangj@hit.edu.cn

Int. J. Appl. Ceram. Technol., 9 [2] 441–446 (2012)DOI:10.1111/j.1744-7402.2011.02660.x

r 2011 The American Ceramic Society

were observed on the surface of the specimens underreentry conditions simulated by an arc jet test. Theresearches of Liang6 and Zhang7 on the influence ofoxidation on the characterization of surface heat ex-change indicated that preoxidation could improve thethermal shock resistance of ceramics considerably, pri-marily due to the fact that coefficients of surface heattransfer for oxides were much lower than the corre-sponding diborides and carbonides. For ZrB2–SiCceramics, oxidation products of ZrB2 and SiC wereB2O3 and SiO2, which took on a liquid state and a glassstate, respectively. And they were able to combine toform an adhesive borosilicate, which could inflood intothe surface cracks at a high temperature and heal thecracks. In early years, some researchers8–12 found thatAl2O3/SiC ceramics had the preferable capability of selfcrack healing.

In this paper, ZrB2–20 vol% SiC–10 vol% AlNceramics with indented cracks were oxidized at 10001Cfor different durations of time. And the residual strengthafter oxidation was measured to evaluate the influence ofoxidation on strength and crack healing. Then, theoxidizing layer structures and the mechanism of crackhealing versus time were analyzed.

Experimental Procedures

Commercially available zirconium diboride (ZrB2),silicon carbide (SiC), and aluminum nitride (AlN) pow-ders were selected as the raw materials. The averageparticle sizes of ZrB2, SiC, and AlN powders were 2 mm,1 mm, and 100 nm, and had purities of 99.5%, 98.7%,and 98.1%, respectively. The as-processed powder mix-tures of ZrB2–20 vol% SiC–10 vol% AlN (sample ZSA)were hot pressed at 18501C for 60 min in an argonatmosphere, with an applied uniaxial pressure of30 MPa using a BN-coated graphite die.

The dimensions of samples for oxidation were3 mm� 4 mm� 36 mm. One indentation was madewith a Vickers diamond indenter (model HvS-5) usingan indentation load of 49 N on the polished bottomsurface of each specimen. The diagonal of indentationwas vertical to the edge of the sample, and the holdingtime of each indentation was 10 s. Each indentationgave rise to four cracks as shown in Fig. 1. The oxida-tion was performed by heating the precracked samplesat 10001C in a furnace with air atmosphere for 10, 20,40, 60, and 90 min, respectively, and then cooling themto room temperature in the furnace. The radial cracklengths of each indentation were measured before andafter oxidation using an optical microscope (ZeissMC80DX, Jena, Germany). The configurations ofoxide layers for both surface and cross-section were an-alyzed by SEM microscopy equipped with an energy-dispersive spectrometer. Five specimens were tested foreach selected oxidation temperature. Then the residualstrength for the samples oxidized for different durationsof time was measured using a three-point bending test atroom temperature with the indentation at the tensionsurface. The loading span of the three-point bendingtest is 30 mm and the crosshead speed is 0.5 mm/min.

Results and Discussions

The residual flexural strength for the samples oxi-dized at 10001C for different durations of time is shownin Fig. 2. It can be found that the strength for the sam-ples oxidized for a random time increases evidently incomparison with the unoxidized indented samples. Andthe strength increases with the increase of the oxidationtime. When the sample is oxidized at 10001C for60 min, the residual strength reaches the maximumand even exceeds the strength of the polished sample,which indicates that oxidation can make a good

Fig. 1. Surface micrograph of indentation and its dimension.

442 International Journal of Applied Ceramic Technology—Liang, et al. Vol. 9, No. 2, 2012

contribution toward the healing of the indented cracksand the microflaws that might exist on the surface. Butthe residual strength decreases with an increase of theoxidation time when the oxidation time exceeds 60 min.The research of Nickel12 and Opila and Halbig13 on theoxidation behavior for ceramic materials indicated thatthe variation of thickness and weight of the oxidelayer with the oxidation time were all in accordancewith the parabolic law. Thus, it can be supposed thatthe magnitude of crack healing of oxide can be charac-terized by a parabola function in the present study.Therefore, the variation of residual strength withthe oxidation time can also be expressed by a parabolalaw. The experimental results of residual strength versustime are fitted by a parabola in Fig. 2, which can bewritten as

s ¼ 229:915þ 16:48t � 0:122t2 ð1Þ

It can be found that the residual strength at theinitial oxidation stage increases gradually with an in-crease of the oxidation time, and the given parabola maybe able to fit the experimental results well. After oxida-tion for 60 min, the residual strength decreases gentlywith the oxidation time, and the given parabola can alsofit it preferably, which may be accidental and needs tobe studied further.

Because of the components of ZrB2–SiC–AlNcomposites oxidized at 10001C, the main reactions in-volved in the oxidation process are given as follows:

ZrB2þ2:5O2ðgÞ ! ZrO2þB2O3ðlÞ ð2Þ

SiCþ1:5O2ðgÞ ! SiO2ðlÞþCOðgÞ ð3Þ

AlNþ 0:75O2ðgÞ ! 0:5Al2O3ðsÞþ0:5N2ðgÞ ð4Þ

B2O3ðlÞ ! B2O3ðgÞ ð5Þ

The SEM micrograph of the surface after oxidationat 10001C for 10 min is shown in Fig. 3a. A dark grayoxide layer is produced on the surface. Analysis by EDS(not shown here) indicates that they are mainly B2O3

and dispersive ZrO2 particles. The indentation becomesindistinct and the precracks cannot be identified. Al-though the oxide layer is very thin, the residual strengthwas enhanced considerably as a result of the oxidation,which can make the crack tip blunt. After the sample isoxidized for 40 min, the surface still remains incompactand takes on ‘‘stream-like’’ and ‘‘ridge-like’’ structures asshown in Fig. 3b and d. Examination by EDS indicatesthat the ‘‘ridge’’ is mainly unvolatilized B2O3 and glassysilicate, and the ‘‘valley floor’’ is an uncovered surface ofsilicate. Meanwhile, white bare ZrO2 can also be seen inboth the ‘‘ridge’’ and the ‘‘valley floor.’’ Owing to thehealing and bluntness of oxidation to precracks, the re-sidual strength of the samples after oxidation enhancesfurther. Actually, the generation and volatilization ofB2O3 and the generation of SiO2 glass occur simulta-neously. When the oxidation time approaches 60 min,due to the huge production of SiO2, the surface is cov-ered by a black compact glassy material as shown in Fig.3c. Analysis by EDS suggests that the black glass mainlyincludes Si, O, B, and Zr. The white block is ZrO2 andthe white acicular particles are Al2O3. The compactglass had healed both the indented crack and the surfacedefects completely. Therefore, the residual strength canreach or exceed the flexural strength of polished un-cracked samples.

With the increase of the oxidation time, the surfacebecomes more and more incompact due to the volatil-ization of B2O3. Figure 4 shows the surface micrographof a ZSA composite after oxidation at 10001C for90 min. The surface is incompact, with many voids.Examination of EDS indicates that the surface layer asshown in Fig. 4b contains primarily Si, Al, and O, pos-sibly SiO2 and Al2O3. The formation of this structure ismost possible due to the volatilization of B2O3, whichleft voids, and the blowout of SiO2 and Al2O3 by highvapor pressure species such as B2O3, B2O2, and SiO.The oxide layer in Fig. 4c includes mainly Zr, Si, O, andAl. The gray matrix, the white block, and the white

0 10 20 30 40 50 60 70 80 90

200

300

400

500

600

700

800

Polished samples without pre-crackPolished and pre-cracked samplesParabolic fitting

Fle

xura

l str

engt

h,σ

/ M

Pa

Oxidation time, t / min

Fig. 2. Variation of the flexural strength of ZSA ceramics versusthe oxidation time.

www.ceramics.org/ACT Oxidation Healing for Cracks of ZrB2—SiC—AlN Ceramics 443

acicular particle are the borosilicate, ZrO2, and Al2O3,respectively.

In order to make clear the crack-healing mechanismof oxidation and its influence on the residual strength,the cross-section of the oxidized sample was observed by

SEM, as shown in Fig. 5. After oxidation for 40 min,the thickness of the oxide layer was about 2.5 mm andthe surface was scraggly and incompact, which is ingood agreement with the results shown in Fig. 3d.When the oxidation time approached 60 min, the

Fig. 4. SEM images of the ZSA composite after oxidation at 10001C for 90 min. (a) The entire surface microstructure; (b) and (c) themagnification of the regions in (a).

Fig. 3. SEM micrograph of ZSA ceramics after oxidation for different durations of time. (a) 10 min; (b) 40 min; (c) 60 min;(d) magnification of the box in (b).

444 International Journal of Applied Ceramic Technology—Liang, et al. Vol. 9, No. 2, 2012

thickness of the oxide layer increased to about 8 mm andthe oxidizing layer was very smooth and compact, whichcan heal the cracks and defects perfectly.

However, when the sample is oxidized for 90 min,the cross-section can be divided into three layers obvi-ously as shown in Fig. 6a: surface oxidizing layer, tran-sition layer, and matrix. The superficial oxidizing layerwith about 10 mm thickness is incompact and has many

pores in it. Analysis by EDS as shown in Fig. 6b suggeststhat the main components of this layer are SiO2 andAl2O3. Most of the B2O3 was volatilized. The transitionlayer is a SiC depletion layer. This structure is moreincompact and its components are mainly ZrO2 andlittle SiO2, and Al2O3 as given in Fig. 6c. Because of thegeneration of this incompact layer, the strength of thesamples after oxidation begins to decrease.

Fig. 5. SEM images of the cross-section of the ZSA composite at various oxidation times. (a) 40 min; (b) 60 min.

Fig. 6. SEM images of the cross-section and the EDS spectrum of the ZSA composite after oxidizing for 90 min. (a) SEM image of thecross-section; (b) EDS spectrum of area A in (a); (c) EDS spectrum of area B in (a).

www.ceramics.org/ACT Oxidation Healing for Cracks of ZrB2—SiC—AlN Ceramics 445

Conclusions

According to the analysis of SEM and EDS on thesurface and the cross-section of a ZSA composite, anoxidizing layer is obtained for precracked samplesoxidized at 10001C for different durations of time.Oxidation is extremely effective in improving the resid-ual strength as a result of crack-tip blunting and crackhealing. The residual strength of the samples after ox-idation increases with an increase of the oxidation time.When the oxidation time approaches 60 min, the oxidelayer is the most compact, which can heal the cracks anddefects perfectly; hence, the residual strength reaches itsmaximum value. With the continuous increase of theoxidation time, an incompact transition layer emerges;therefore, the residual strength begins to decrease. Be-sides, the preoxidation technique is an applicable andeffective method to improve the strength of brittle ma-terials with microcracks.

Acknowledgements

This work was supported by Program for NewCentury Excellent Talents in University (NCET-05-0346), the National Natural Science Foundation ofChina (90916027), the National Natural Science Fundsfor Distinguished Young Scholar (No. 10725207), andthe Foundation for Innovative Research Groups ofthe National Natural Science Foundation of China(10821201).

References

1. W. G. Fahrenholtz, G. E. Hilmas, I. G. Talmy, and J. A. Zaykoski, ‘‘Re-fractory Diborides of Zirconium and Hafnium,’’ J. Am. Ceram. Soc., 90 [5]1347–1364 (2007).

2. J. D. Bull and D. J. Rasky ‘‘Stability Characterization of Diboride Compos-ites under High Velocity Atmospheric Flight Conditions,’’ the 24th Interna-tional SAMPE Technical Conference, Toronto, Canada, October 20–22, 1992.

3. D. J. Rasky, J. D. Bull, and H. K. Tran, ‘‘Ablation Response of AdvanceRefractory Composites,’’ Advanced Ceramics Association 15th Annual Con-ference on Composites, Materials & Structures, Restricted Session, CocoaBeach, FL, January 1991.

4. M. M. Opeka, I. G. Talmy, E. J. Wuchina, J. A. Zaykoski, and S. J. Causey,‘‘Mechanical, Thermal, and Oxidation Properties of Refractory Hafnium andzirconium Compounds,’’ J. Eur. Ceram. Soc., 19 [13–14] 2405–2414 (1999).

5. A. Chamberlain, W. Fahrenholtz, G. Hilmas, and D. Ellerby, ‘‘Oxidation ofZrB2–SiC Ceramics under Atmospheric and Reentry Conditions,’’ Refract.Appl. Trans., 1 [2] 1–8 (2005).

6. J. Liang, C. Wang, Y. Wang, L. Jing, and X. Luan, ‘‘The Influence of SurfaceHeat Transfer Conditions on Thermal Shock Behavior of ZrB2–SiC–AlNCeramic Composites,’’ Scr. Mater., 61 [6] 656–659 (2009).

7. X. H. Zhang, L. Xu, S. Y. Du, W. B. Han, and J. C. Han, ‘‘Preoxidation andCrack-Healing Behavior of ZrB2–SiC Ceramic Composite,’’ J. Am. Ceram.Soc., 91 [12] 4068–4073 (2008).

8. H. S. Kim, M. K. Kim, S. B. Kang, S. H. Ahn, and K. W. Nam, ‘‘BendingStrength and Crack-Healing Behavior of Al2O3/SiC Composites Ceramics,’’Mater. Sci. Eng. A, 483–484 672–675 (2008).

9. W. Nakao, M. Ono, S. K. Lee, K. Takahashi, and K. Ando, ‘‘Critical Crack-Healing Condition for SiC Whisker Reinforced Alumina Under Stress,’’J. Eur. Ceram. Soc., 25 [16] 3649–3655 (2005).

10. I. A. Chou, H. M. Chan, and M. P. Harmer, ‘‘Effect of Annealing Envi-ronment on the Crack Healing and Mechanical Behavior of Silicon Carbide-reinforced Alumina Nanocomposites,’’ J. Am. Ceram. Soc., 81 [5] 1203–1208(1998).

11. S. K. Lee, W. Ishida, S. Y. Lee, K. W. Nam, and K. Ando, ‘‘Crack-HealingBehavior and Resultant Strength Properties of Silicon Carbide Ceramic,’’J. Eur. Ceram. Soc., 25 [5] 569–576 (2005).

12. K. G. Nickel, ‘‘Multiple Law Modeling For the Oxidation of Advanced Ce-ramics and a Model-independent Figure of Merit,’’ Corrosion of AdvancedCeramics/Measurements and Modelling. ed. K. G. Nickel. Kluwer AcademicPublishers, Dordrecht, the Netherlands, 59–71, 1994.

13. E. J. Opila and M. C. Halbig, ‘‘Oxidation of ZrB2–SiC,’’ Ceram. Eng. Sci.Proc., 22 [3] 221–228 (2010).

446 International Journal of Applied Ceramic Technology—Liang, et al. Vol. 9, No. 2, 2012