Preparation and Mechanical Properties of Alumina Zirconia Composites with Agglomerated Structures Using Pre-Sintered Powder

Materials Transactions, Vol. 44, No. 8 (2003) pp. 1571 to 1576 #2003 The Japan Institute of Metals Preparation and Mechanical Properties of Alumina Zirconia Composites with Agglomerated Structures Using Pre-Sintered Powder Kensuke Kageyama, Youhei Harada* and Hiroshi Kato Department of Mechanical Engineering, Faculty of Engineering, Saitama University, Saitama 338-8570, Japan It is known that thermal residual stress in particulate ceramics results from the mismatch of thermal expansion coefficients of particulates and matrix and contributes to toughening of ceramic composites. In this study, alumina-zirconia composites with agglomerated structures were prepared using alumina or alumina-zirconia powder to obtain large-sized compressive zones in particulate ceramics without degrading flexural strength. Agglomerated powder was obtained by pre-sintering. Then several samples used different fraction and size of agglomerated powder were prepared by pressureless sintering. Microstructure and crack paths of prepared samples were examined by scanning electron microscopy (SEM); flexural strength and fracture toughness of samples were evaluated by four-point flexural test and controlled surface flow method, respectively. Alumina-rich agglomerated structures and a zirconia-rich matrix were formed in samples that were produced using pre-sintered powder. Addition of zirconia to pre-sintered alumina powder prevented coarsening of alumina grains in agglomerated structures. Grain coarsening and cracking caused the decrease in flexural strength of samples with agglomerated structures. Agglomerated structures enhanced fracture toughness. In particular, a specimen using 21.1 vol% of pre-sintered alumina-rich powder of 32 to 150 mm exhibited increase in fracture toughness by approximately 30% without sacrificing average flexural strength. A SEM observation of crack paths showed that grain bridging did not occur in samples. Thereby, we inferred that the compressive residual stress zone in agglomerated structures played an important role in raising fracture toughness. (Received May 8, 2003; Accepted June 26, 2003) Keywords: ceramics, composites, agglomerated structure, thermal residual stress, fracture toughness, alumina, zirconia 1. Introduction Zirconia-toughened alumina (ZTA) has been used in cutting and implant applications because it has higher fracture toughness than monolithic alumina and shows chemical stability as well as very good resistance to wear. 1 4) Further improvement of fracture toughness of ZTA, however, has been required to offer increased component life and performance. Toughening by thermal residual stress is one toughening mechanism in ceramic composites including ZTA. 5 8) Thermal residual stress in particulate ceramics results from the mismatch between coefficients of thermal expansion (CTE) of the ceramic matrix and particulates. In principle, thermal residual stress in a particulate composite can be postulated as a periodic tensile-compressive stress field. Evans et al. 9) proposed that such a periodic tensilecompressive stress field enhances fracture toughness. According to the model by Cutler and Vikar, 10) toughening by periodic residual stress of particulate ceramic composites is provided as rffiffiffiffiffiffiffi 2D K 1 ¼ 2q ð1þ where K 1 is the increase in fracture toughness, q is the local residual compressive stress, and D represents the length of the compressive stress zone. Equation (1) supposes that CTE of particulates is smaller than that of a matrix so that the compressive thermal residual stress in the particulates is generated and D equals the average diameter of particulates. In this case, local residual compressive stress q is equivalent to the average stress in the particle hi p. According to the model of Taya et al., 8) the average stresses in the particle hi p *Graduate Student, Saitama University. Present address: DaiNippon Printing Co. Ltd., Tokyo 162-8001, Japan. and in the matrix hi m are given as hi p E m hi m E m ¼ 2ð1 f p f v ÞT A ¼ 2f pt A A ¼ð1 f p f v Þð þ 2Þð1 þ m Þþ3f p ð1 m Þ ð4þ ¼ ð1 þ mþ E p ð5þ ð1 2 p Þ E m where f p and f v are volume fractions of particles and voids (pores), respectively. Young s modulus and Poisson s ratio of the ith phase are E i and v i, where i ¼ m and p represent the matrix and particle, respectively. The magnitude of the CTE mismatch is, whereas T is the difference between sintering temperature and room temperature. Cutler and Vikar 10) pffiffiffiffi observed the D dependence of KIC as shown in eq. (1) experimentally. On the other hand, Taya et al. 8) compared this model with TiB 2 particulate SiC matrix composites where the compressive thermal residual stress in the matrix was generated. They supposed that the length of the compressive stress zone, D, was equivalent of d, where is average interparticulate spacing and d is the average diameter of TiB 2 particles. They concluded that predicted fracture toughness from eq. (1) agreed well with experimental results. Equation (1) suggests that the increase in q and D, i.e. the increase in mismatch of CTE between the matrix and particle and grain coarsening, causes toughening of particulate composites. However, excessive residual stress causes cracking in the tensile stress zone during cooling; coarse grains then inhibit densification. Ultimately, the fracture strength is degraded as a result. In this study, formation of agglomerated structures was used instead of grain coarsening ð2þ ð3þ

1572 K. Kageyama, Y. Harada and H. Kato A B Agglomerated Structure Table 1 Composition of polycrystalline powder. Powder PP1 PP2 PP3 PP4 Size (mm) 32 32 32 32 150 Al 2 O 3 (vol%) 100 99.5 95 95 ZrO 2 (vol%) 0.50 5.0 5.0 Table 2 Composition of samples. Fig. 1 Schematic of particulate composites where B particulates are uniformly distributed and where B particulates are agglomerated. to enhance fracture toughness of composites. If a polycrystalline powder of different composition from the virgin powder (single-crystal powder) is mixed with the virgin powder, the sintered body after pressing and sintering the mixed powder would have agglomerated structures formed from the polycrystalline powder as shown in Fig. 1. When both of polycrystalline powder and single-crystal powder consist of two materials (A and B in Fig. 1), but the composition of the polycrystalline powder differs from that of the single-crystal powder, the average CTE of agglomerated structures can be controlled by arrangement of composition of both powders while keeping the total amount of A and B constant. Size of agglomerated structures also can be controlled by the size of polycrystalline powder. Hence if average CTE of agglomerated structures is set to be lower than that of the matrix, average residual stress in the agglomerated structures would be compressive and higher fracture toughness would be expected as larger D, i.e. average diameter of agglomerated structures. Furthermore, the densification problem could be solved by using polycrystalline powder instead of coarse grains because the polycrystalline powder, consisting of fine grains, has sufficient grain boundaries within for pores to diffuse to the surface during sintering. Alumina and zirconia are suitable to fabricate particulate composites with agglomerated structures, as described above, because: CTE of zirconia is higher than that of alumina; zirconia has no solubility with alumina; and they are already familiar to the structural ceramics community. To obtain large-sized compressive stress zones in this study, alumina-zirconia composites with agglomerated structures were prepared using alumina or alumina-zirconia polycrystalline powder that was prepared by pre-sintering. Flexural strength, fracture toughness, and crack paths of alumina-zirconia composites with agglomerated structures were investigated. 2. Experimental Procedure The stabilized zirconia powder (TZ-3Y; Tosoh, Inc.) and alumina powder (AES11; Sumitomo Chemical) were selected as raw materials. Virgin alumina powder was put in crucible without pressing and sintered for 10 min at 1773 K with a rate of 50 K/min. The sintered powder was put through a sieve to obtain polycrystalline alumina powder SC1 SP1 SP2 SP3 SP4 SP5 Powder PP1 PP2 PP3 PP4 PP4 Size Polycrystalline Powder (mm) 32 32 32 32 150 32 150 (vol%) Al 2 O 3 80 80 80 80 20 ZrO 2 0.40 4.2 4.2 1.1 Virgin Al 2 O 3 80 60 Powder (vol%) ZrO 2 20 20 19.6 15.8 15.8 18.9 with <32 mm diameter. The virgin alumina powder and zirconia powder were mixed by ball milling and dried by a rotary evaporator. Granular alumina-zirconia was sintered and sieved as mentioned above to obtain polycrystalline alumina-zirconia powder under 32 mm or 32 150 mm of a diameter. Table 1 shows the polycrystalline powder composition. The virgin powder and the polycrystalline powder were mixed by ball-milling at the composition shown in Table 2, dried by a rotary evaporator, pressed into a green body under pressure of 200 MPa by cold isostatic pressing, and sintered in air at for 2 h at 1873 K with a heating rate of 5 K/min. Although all samples had the same fraction of total volume of alumina and that of zirconia (80 vol% of alumina), composition or diameter of polycrystalline powder varied according to samples. Specimens of 3 4 40 mm for measuring flexural strength and fracture toughness were cut from sintered bodies. The microstructures of polished and thermal etched surface of specimens were examined by SEM (Hitachi S-2150) at 20 kv accelerating voltage. Four-point flexural strength was measured with an upper span of 10 mm, a lower span of 30 mm, and a displacement rate of 0.5 mm/ min. Specimens were also served for fracture toughness measurement using the controlled surface flaws method. 11) A Knoop indenter was loaded on the specimen surface under 294 N for 30 s to induce a crack perpendicular to the long axis of a specimen. Indented specimens were annealed for 2 h at 1523 K because Cooks et al. 12) reported that annealing removed residual stress associated with the indentation and gave reliable values of fracture toughness of ceramics. The four-point flexural test was applied to the indented specimens, which had loaded tensile stress on the indented surface when fracture strength was measured. Length and depth of the indented crack were measured from the fracture surface by optical microscope; fracture toughness was calculated using Newman s equation. 13) Three specimens of each sample were examined for flexural testing and fracture toughness testing, respectively. Other indented specimens suffered thermal etching and paths of indented cracks were examined by SEM.

Preparation and Mechanical Properties of Alumina Zirconia Composites with Agglomerated Structures Using Pre-Sintered Powder 1573 3. Results and Discussion 3.1 SEM observation of microstructure Figure 2 shows that sample SC1 has the microstructure of uniform distributed fine alumina-zirconia grains because it was prepared solely from virgin alumina and zirconia powder (dark area is alumina and light area is zirconia). Figures 2 (d) show that samples SP1 to SP3 have agglomerated alumina-rich structures in the zirconia-rich matrix because of addition of the polycrystalline alumina-rich powders PP1 to PP3. There should be compressive residual stress in the agglomerated structures because the average CTE of agglomerated alumina-rich structure is lower than that of the zirconia-rich matrix. The sizes of agglomerated structures of samples SP1 to SP3 were <32 mm because the diameter of the used polycrystalline powder was <32 mm. Figure 2 also shows that there are more pores remaining in sample SP1 to SP4 than in sample SC1. The constituent polycrystalline powder is considered to decrease the density of the green body. Eventually, as a higher ratio of polycrystalline powder to virgin powder is mixed, more pores remains in the sintered body because decrease in density of the green body results in an increase in pores remaining in the sintered body. Figure 2(e) shows that sample SP4 has larger agglomerated structures than samples SP1 to SP3 because of the use of larger polycrystalline powder PP4 (32 150 mm diameter). SP4 was prepared to obtain larger compressive stress zone in the agglomerated structures, but there was much (c) (d) 50µm (e) (f) Fig. 2 SEM micrographs (secondary-electron images) of the etched surface of SC1, SP1, (c) SP2, (d) SP3, (e) SP4, and (f) SP5.

1574 K. Kageyama, Y. Harada and H. Kato 10µm (c) (d) Fig. 3 SEM micrographs (secondary-electron images) of the etched surface of SP1, SP2, (c) SP4 (within an agglomerated structure), and (d) SP5 (within an agglomerated structure). cracking at the interface of the agglomerated structures and the matrix. These cracks resulted from excess thermal residual stress; therefore, the CTE or volume ratio of agglomerated structures compared to the matrix must be decreased to remove these cracks. Figure 2(f) shows the microstructure of sample SP5 prepared from the polycrystalline alumina-rich powder PP4 and alumina-zirconia virgin powder as shown in Table 2. The volume fraction of agglomerated structures fell to 21.1 vol%. The average CTE of agglomerated structures is closer to the matrix than that of sample SP1 to SP4 because of addition of alumina to the matrix. Eventually, there was no cracking caused by thermal residual stress on sample SP5 even though the size of agglomerated structures was much larger than that of samples SP1 to SP3. Figure 3 shows that grain coarsening apparently occurred in agglomerated structures of sample SP1. It resulted from lack of grain-boundary pinning by zirconia particles. Figure 3 shows that addition of 0.5 vol% zirconia was not efficient in preventing grain coarsening. s SP3, SP4 and SP5 have 5 vol% zirconia in agglomerated structures; in those samples, grain coarsening was prevent by zirconia pinning as shown in Figs. 3(c) and (d). 3.2 Mechanical Properties s containing agglomerated structures were prepared from a mixture of virgin powder and the polycrystalline powder, which resulted in pores remaining, as mentioned Relative Density (%) 100 99.5 99 98.5 98 97.5 97 96.5 96 Fig. 4 SC1 SP1 SP2 SP3 SP4 SP5 Relative density of fabricated samples. above. Relative density of samples SP1 to SP5 was therefore lower than that of the conventional sample SC1, as shown in Fig. 4. Figure 5 shows that samples SP1 and SP2 had slightly lower flexural strength than conventional sample SC1 because of grain coarsening, as shown in Figs. 3 and. Figure 5 also shows that sample SP4 had much lower flexural strength than sample SC1 because there was cracking caused by an excess of thermal residual stress as shown in Fig. 2(e). Grain coarsening was inhibited in samples SP3 and SP5; there was no cracking caused by thermal residual stress.

Preparation and Mechanical Properties of Alumina Zirconia Composites with Agglomerated Structures Using Pre-Sintered Powder 1575 Flexural Strength, σ f / MPa 650 600 550 500 450 400 350 300 SC1 SP1 SP2 SP3 SP4 SP5 Fig. 5 Flexural strength of fabricated samples. 2.50 Experimental data 2.00 Predicted data 1.50 1.00 0.50 0.00 SP1 SP2 SP3 SP4 SP5 Fig. 7 Comparison of K IC of experimental data (average values) and predicted data. K IC / MPa.m 1/2 Fracture toughness, K IC / MPa.m 1/2 5 4.8 4.6 4.4 4.2 4 3.8 3.6 3.4 3.2 3 Fig. 6 SC1 SP1 SP2 SP3 SP4 SP5 Fracture toughness of fabricated samples. Hence the average flexural strength of samples SP3 and SP5 fairly approximated that of sample SC1. The scatter of flexural strength of sample SP5 was, however, still larger than sample SC1. If remaining pores are improved by hot pressing or other sintering techniques, decreased scatter of flexural strength will be expected. 14) Figure 6 shows that agglomerated samples SP1 to SP5 had higher fracture toughness than the conventional sample SC1. Especially, sample SP5 showed noticeable toughening by agglomerated structures without sacrificing average flexural strength close to the conventional sample. Figure 7 shows measured and predicted values of the increase in fracture toughness of samples SP1 to SP5. Predicted values were calculated from eqs. (1) to (3) assuming that an average Young s modulus, Poisson s ratio, and CTE of the matrix and those of agglomerated structures were subject to a simple law of mixture. The prediction also assumed that D is half the value of mesh size of a sieve (32 mm) for samples SP1 to SP3 or an intermediate value of mesh size of two sieves (32 and 150 mm) for samples SP4 and SP5, i.e. D ¼ 16 mm for samples SP1 to SP3 or D ¼ 91 mm for samples SP4 and SP5. Predicted values show the same magnitude relationship of measured values between samples, but are higher than measured values. As shown in eq. (1), the predicted value is proportional to the square root of length of the compressive zone, D, i.e. the average diameter of agglomerated structures. Total area and number of agglomerated structures of samples SP4 and SP5 were measured from low magnification (80) SEM micrograph; the calculated average diameter of agglomerated structures of sample SP4 is 101 mm and that of sample SP5 is 39 mm. Although polymer balls were used when the polycrystalline powder and virgin powder were blended by ball-milling, crushing of the polycrystalline powder during ball-milling could not be avoided. On the other hand, higher volume fraction of polycrystalline powder and more coalescence between polycrystalline powders occurred during sintering. In the case of sample SP5, crushing the polycrystalline powder during ball-milling influenced the decrease in the average diameter of agglomerated structures because of its low volume fraction of polycrystalline powder (21.1 vol%). For D ¼ 39 mm, the predicted value of K 1 is 1.5 MPam 1=2 and still higher than the experimental value (1.1 MPam 1=2 ). Predicted values are based on the assumption that tensile residual stress generated in the matrix has only a negligible influence on toughening as estimated by eq. (1). The tensile residual stress in the matrix, however, caused cracking in sample SP4, as shown in Fig. 2(e), so that it could result in decrease in fracture toughness of the samples in this study. Furthermore, fracture toughness of alumina is much lower than that of zirconia (3.4 MPam 1=2 for alumina and 7.1 MPam 1=2 for zirconia used in this study). Such heterogeneity of fracture toughness is not considered in eq. (1). Another problem is the pores remaining in samples. Tanaka et al. 15) measured residual stress in alumina-zirconia particulate composites using the X-ray diffraction method. They reported that the theoretical prediction based on eqs. (2) and (3) agreed well with observed residual stress. However, samples used in their study were obtained by hot isostatic pressing. Therefore, the porosity was negligible. s in the present study had porosity of 1.3 to 2.8%, eq. (2) assumes that pores are distributed in the matrix. It predicts that hi p of sample SP5 is 152 MPa. Figure 3, however, shows there are also pores in the agglomerated structure as well as the matrix; thus magnitude of hi p of samples SP1 to SP5 should be lower than the predicted value by eq. (2). Such a decrease in hi p leads to degrading K 1. If remaining pores are

1576 K. Kageyama, Y. Harada and H. Kato 20µm Fig. 8 SEM observation (secondary-electron images) of crack paths on the surface of samples SP2 and SP5. improved, it is expected that the value of experimental K 1 will move close to that of the prediction. 3.3 Observation of surface crack Swanson et al. 16) proposed that grain bridging was an effective means of toughening ceramics because grains became pinned behind an advancing crack tip. Chantikul et al. 17) observed the role of grain size in fracture toughness of alumina. Figure 8 shows crack paths of indented surfaces of samples SP2 and SP5. The cracks penetrated agglomerated structures in all samples. Hence grain bridging was not the reason for the increase in fracture toughness observed in samples with agglomerated structures. Consequently, the most possible toughening mechanism by agglomerated structures is that cracks go through the local residual compressive zone without going around them. Thereby, the stress intensity factor of crack tip is reduced in the compressive zone, i.e. agglomerated structures. 4. Conclusions Several samples of alumina-zirconia composites with alumina-rich agglomerated structures were prepared using pre-sintered powder. Microstructure and crack paths of prepared samples were examined by SEM. Then flexural strength and fracture toughness of samples were evaluated by four-point flexural test and CSF method, respectively. (1) Alumina-rich agglomerated structures and a zirconiarich matrix were formed in samples using pre-sintered powder of which diameter was <32 mm or 32 to 150 mm. Addition of 5 vol% of zirconia to pre-sintered alumina powder prevented coarsening of alumina grains in agglomerated structures. Use of 84.2 vol% of presintered alumina-rich powder of 32 to 150 mm diameter caused cracking at the interface between agglomerated structure and matrix because of excess of thermal residual stress. (2) Grain coarsening and cracking caused a decrease in flexural strength of samples with agglomerated structures. Agglomerated structures enhanced fracture toughness. In particular, use of 21.1 vol% of presintered alumina-rich powder of 32 to 150 mm diameter caused an increase in fracture toughness by approximately 30% without sacrificing average flexural strength compared to conventional sample. (3) SEM observation of crack paths showed that grain bridging did not occur in samples of this study. Therefore the compressive residual stress zone in agglomerated structures played an important role in enhancing fracture toughness. REFERENCES 1) B. Mondal, A. B. Chattopadhyay, A. Virkar and A. Paul: Wear 156 (1992) 365 383. 2) T. Sornakumar, M. V. Gopalakrishnan, R. Krishnamurthy and C. V. Gokularathnam: Inter. J. of Refractory Metals and Hard Materials, 13 (1995) 375 378. 3) D. Galusek and J. Majling: Ceramics International 21 (1995) 101 107. 4) S. Ishihara, K. Akashiro, T. Tanizawa, N. Furushiro, Y. Umakoshi and S. Hori: Mater. Trans., JIM 41 (2000) 376 382. 5) G. C. Wei and P. F. Becher: J. Am. Ceram. Soc. 67 (1984) 571 574. 6) C. H. McMurtry, W. D. G. Boecker, S. G. Seshadri, J. S. Zanghi and J. E. Garnier: Am. Ceram. Soc. Bull. 66 (1987) 325 329. 7) M. O. Nandy, T. Nose, M. Enoki and T. Kishi: Mater. Trans., JIM 37 (1996) 769 775. 8) M. Taya, S. Hayashi, A. S. Kobayashi and H. S. Yoon: J. Am. Ceram. Soc. 73 (1990) 1382 1391. 9) A. G. Evans, A. H. Heuer and D. L. Porter: Int. Conf. Fract. 4th I (1977) 529 556. 10) R. A. Cutler and A. V. Virkar: J. Mater. Sci. 20 (1985) 3557 3573. 11) P. Chantikul, G. R. Anstis, B. R. Lawn and D. B. Marshall: J. Am. Ceram. Soc. 64 (1981) 539 565. 12) S. G. Cook, J. E. King and J. A. Little: Mater. Sci. Technol. 12 (1996) 366 370. 13) J. R. Newman and I. S. Raju: Eng. Frac. Mech. 15 (1981) 185 192. 14) M. Naito, T. Hotta, H. Abe, N. Shinohara, Y. Cho, M. Okumiya and K. Uematsu: Mater. Trans. 42 (2001) 114 119. 15) K. Tanaka, M. Motoyasu, R. Shikata and T. Nishikawa: J. Soc. Mat. Sci. Japan 41 (1992) 593 599. 16) P. L. Swanson, C. J. Fairbanks, B. R. Lawn, Y. W. Mai and B. J. Hockey: J. Am. Ceram. Soc. 70 (1987) 279 289. 17) P. Chantikul, S. J. Bennison and B. R. Lawn: J. Am. Ceram. Soc. 73 (1990) 2419 2427.