L Zou et al. /Journal of the European Ceramic Society 23(2003)1987-1996 1995 lenser, and no eminent big pores existed(Fig. 8d), cor- of tending to be a constant. Obviously, these were con- respondingly, the interfacail toughness was too big to tradictory with the initial assumption for the three-point allow crack deflection and propagaton in the interphase. bending fracture model, and they were not correct However, in the case of the Bn interphase modified by results. To solve this problem, we concentrated all the Al_O3, the densification rate was not so fast as that of deviations from the effecting factors upon the Young Si3 N4 modified interphase. With the increase of volume Modulus in Eqs. (5)and(10). In this way, The E value percentage of AlO3, the Bn interphase was densified was successfully adjusted to offset the earlier-mentioned gradually and slowly, and the interfacial toughness also influence The adjustment was not finished until the inter increased. Even if the percentage approaches 63%, the facial toughness value almost tended to be a constant after interphase still seemed to be loosen and porous(Fig. 9c), the occurrence of crack deflection and propagation within this is the reason why crack deflection and propagation interphase. To obtain a preliminary e value that is near still could happen in this interphase. As the interphase to the prospective one, sometimes the critical condition. was pure Al2O3, the interfacial toughness was bi i.e. a=0 or a=L employed for the purpose enough to hinder the occurrence of crack deflection and according to Eq. (5). As a=0, it was correspondi propagation From the microstructure analysis, it is also the initiation point of interfacial crack propagation, clear that the strengthening effect of Si3N4 is much big- there was a sudden load descent in the load-displace- ger than that of Al2O3, at the same volume percentage ment curve For a=L, it was only applicable to the case of modifier, the interfacial toughness value of Si3N4 in which the interfacial crack reached the ends of the doped bn interphase is much higher that of Al_O3 loading span. Based on the obtained preliminary E doped one value, the prospective E value was determined easily to The earlier-mentioned results indicate that the make the interfacial toughness nearly tend to be a con- method we used to measure and characterize interfacial stant. Comparing with the four-point bending test to toughness is viable. Compared with the method Phillips measure the interfacial toughness, the three-point bend et al. used for measuring the interfacial toughness of a ing test measurement results would be less accurate laminated SiC/C composite, the present method allowed us to measure the interfacial toughness using smaller sized samples, which were much easier to obtain. Thus 4. Conclusions the method of presetting a crack directly connecting to the interphase is viable, making the experimental testing simpler and more flexible 1. By considerably reducing sandwiching sample For the method presented here, the most important size, and presetting a crack source connecting step is to determine the crack propagation length by directly to a BN interphase in the middle of a compliance method (5). According to the equation, the single half-Si3 N4 matrix, a new method was crack length is also determined by the moment of inertia presented for measuring interfacial toughness of of half Si3N4 matrix (Is. Because the matrix thickness the Si3N4/BN composite (h2) was obtained by measuring its thickness in different 2. The interfacial toughness of the Si3N4/BN com- positions under common optical microscopy and then posite with different interphases strengthened by taking an average value, this introduced some errors Si3N4 or Al,O3 were measured, the method turned due to irregularities of the matrix surface contacted to out to be viable, and is simpler and more flexible the interphase. Because the inside surface of Si3N4 compared with other method in literatures matrix layer was not absolutely flat and its thickness 3. The interfacial toughness for Si3 N4/BN compo- had some fluctuations after the specimen was fabri sites with bN, BN+ 15vol% Si3N4, and cated. The crack propagation length was considerably BN+25vol Si3N4 interphases was 35.42, affected by the fuctuation of Is and finally the inter- 96.45, and 107.67 J/m, respectively. The com facial toughness was influenced indirectly. Interfacial posite with BN+ 16vol Al2O3, BN+ 36vo- toughness was also influenced by the Is [see Eq(10)1 1 %Al,O3, and Bn+ 63 vol% Al,O3 interphases so there existed a double influencing factors in three was 43.44J /m, 49.97J/m- and 68.27J/m- respec- point bending method, this was the cause of giving rise tively. For the both interphase systems, as the to deviation of the result. Another important factor amount of the modifier Si3 N4 or AlO3 increased that affected the measurement results was the Youngs the interphase was over-strengthened, the crack modulus of the matrix. For several samples, the mea deflection and propagation sured E value could not stand for that of the real Si3N4 and resulted in brittle fracture matrix in the sample for measuring interfacial tough 4. The method was affected by the fluctuation of the ness, leading to a result that was not reasonable. Some- moment of inertia of half Si3 N, matrix (Is)due to times the crack propagation length was negative, or the irregularities of the matrix surface contacted to interfacial toughness value had a big fluctuation instead the interphase. Also it was influenced by thedenser,and no eminent big pores existed (Fig. 8d),cor￾respondingly,the interfacail toughness was too big to allow crack deflection and propagaton in the interphase. However,in the case of the BN interphase modified by Al2O3,the densification rate was not so fast as that of Si3N4 modified interphase. With the increase of volume percentage of Al2O3,the BN interphase was densified gradually and slowly,and the interfacial toughness also increased. Even if the percentage approaches 63%,the interphase still seemed to be loosen and porous (Fig. 9c), this is the reason why crack deflection and propagation still could happen in this interphase. As the interphase was pure Al2O3,the interfacial toughness was big enough to hinder the occurrence of crack deflection and propagation. From the microstructure analysis,it is also clear that the strengthening effect of Si3N4 is much big￾ger than that of Al2O3,at the same volume percentage of modifier,the interfacial toughness value of Si3N4 doped BN interphase is much higher that of Al2O3 doped one. The earlier-mentioned results indicate that the method we used to measure and characterize interfacial toughness is viable. Compared with the method Phillips et al.11 used for measuring the interfacial toughness of a laminated SiC/C composite,the present method allowed us to measure the interfacial toughness using smaller￾sized samples,which were much easier to obtain. Thus, the method of presetting a crack directly connecting to the interphase is viable,making the experimental testing simpler and more flexible. For the method presented here,the most important step is to determine the crack propagation length by compliance method (5). According to the equation,the crack length is also determined by the moment of inertia of half Si3N4 matrix (Is). Because the matrix thickness (h2) was obtained by measuring its thickness in different positions under common optical microscopy and then taking an average value,this introduced some errors due to irregularities of the matrix surface contacted to the interphase. Because the inside surface of Si3N4 matrix layer was not absolutely flat and its thickness had some fluctuations after the specimen was fabri￾cated. The crack propagation length was considerably affected by the fluctuation of Is and finally the inter￾facial toughness was influenced indirectly. Interfacial toughness was also influenced by the Is [see Eq. (10)], so there existed a double influencing factors in three￾point bending method,this was the cause of giving rise to deviation of the result. Another important factor that affected the measurement results was the Young’s modulus of the matrix. For several samples,the mea￾sured E value could not stand for that of the real Si3N4 matrix in the sample for measuring interfacial tough￾ness,leading to a result that was not reasonable. Some￾times the crack propagation length was negative,or the interfacial toughness value had a big fluctuation instead of tending to be a constant. Obviously,these were con￾tradictory with the initial assumption for the three-point bending fracture model,and they were not correct results. To solve this problem,we concentrated all the deviations from the effecting factors upon the Young’s Modulus in Eqs. (5) and (10). In this way,The E value was successfully adjusted to offset the earlier-mentioned influence. The adjustment was not finished until the inter￾facial toughness value almost tended to be a constant after the occurrence of crack deflection and propagation within interphase. To obtain a preliminary E value that is near to the prospective one,sometimes the critical condition, i.e. a=0 or a=L,was employed for the purpose according to Eq. (5). As a=0,it was corresponding to the initiation point of interfacial crack propagation, there was a sudden load descent in the load–displace￾ment curve. For a=L,it was only applicable to the case in which the interfacial crack reached the ends of the loading span. Based on the obtained preliminary E value,the prospective E value was determined easily to make the interfacial toughness nearly tend to be a con￾stant. Comparing with the four-point bending test to measure the interfacial toughness,the three-point bend￾ing test measurement results would be less accurate. 4. Conclusions 1. By considerably reducing sandwiching sample size,and presetting a crack source connecting directly to a BN interphase in the middle of a single half-Si3N4 matrix,a new method was presented for measuring interfacial toughness of the Si3N4/BN composite. 2. The interfacial toughness of the Si3N4/BN com￾posite with different interphases strengthened by Si3N4 or Al2O3 were measured,the method turned out to be viable,and is simpler and more flexible compared with other method in literatures. 3. The interfacial toughness for Si3N4/BN compo￾sites with BN,BN+ 15vol.% Si3N4,and BN+25vol.% Si3N4 interphases was 35.42, 96.45,and 107.67 J/m2 ,respectively. The com￾posite with BN+ 16vol.% Al2O3,BN+36vo￾l.%Al2O3,and BN+ 63Vol% Al2O3 interphases was 43.44J/m2 ,49.97J/m2 and 68.27J/m2 respec￾tively. For the both interphase systems,as the amount of the modifier Si3N4 or Al2O3 increased, the interphase was over-strengthened,the crack deflection and propagation no longer happened, and resulted in brittle fracture. 4. The method was affected by the fluctuation of the moment of inertia of half Si3N4 matrix (Is) due to irregularities of the matrix surface contacted to the interphase. Also it was influenced by the L. Zou et al. / Journal of the European Ceramic Society 23 (2003) 1987–1996 1995