Commmunications of the American Ceramic Society VoL. 90. No. 12 Sreceived 525°C igd boo ch for up o eoo h.T he behavi oad-deflection curves obtained after isothermal oxidation of Nicalon"/CAS at temperatures between 375 of the as-received material is also shown for reference expansion coefficient of the fiber due to oxidation.2. For the significant effect upon these interfacial properties. After heat case of Nicalon NLM-202, the fiber used in the present Cas treatment at 375C the debonding energy is significantly lower matrix composite, er, is determined to be 1.48.- Based on this, than for the as-received material, and essentially unchange an estimate for the sealing time can be determined from the ox- upon subsequent reloading, indicating that there is no rea idation kinetics of Nicalon NLM-202 fibers At the lowest interfacial chemical bond present after this heat treatment. temperature examined in that study, namely 700%C, sealing for a However, the frictional sliding stress is comparable with the in approximately 35 h for a 50-nm thick carbon layer, 5 h for from Fig 3 that only partial carbon removal occurs after 1.. glass-ceramic matrix composite can be predicted to occu as-received composite with a carbon interlayer. It can be seen a 20-nm thick layer, and I h 20 min for a 10-nm thick layer. The as weight loss continues essentially linearly up to 1000 h, which residual stress state of this composite is such that the matrix will infers that a thin carbon layer appears to be retained after 100 h. lamp down onto the fiber if the carbon layer is removed, thus The interfacial debonding energy and frictional sliding stress are aking the apparent fiber-matrix gap smaller and the sealing both increased by heat treatment at 450C (Table In). Given time shorter. Oxidation kinetics data are not readily available the absence of a lubricating carbon interlayer after this heat for lower temperatures, but extrapolation of the behavior ob- treatment, a dramatically increased frictional sliding stress can served at higher temperatures indicates that sealing will occur be anticipated if the matrix and fiber come into direct contact after approximately 20 h at 600C for a 20-nm thick fiber-m The increased debond energy is also believed to be due to the rix gap This estimation is in general accordance with the weigh arge frictional sliding stress that must be overcome before the gain response noted in Fig. 3, where weight gain ceases before fiber can be displaced, although the formation of isolated SiO2 0o h exposure. A more thorough determination of sealing be- bridges between the matrix and fiber cannot be completely dis- 5-600oC)would require counted (particularly given the observed weight gain after ex- continuous thermogravimetric analysis of both the fiber and tended heat-treatment at 450.C(Fig. 3). It was interesting to composite oxidation weight changes as a function of time, which note from this prior study that, upon reloading. the debond load is beyond the scope of the present study is essentially the same as the first loading cycle (i.e, 200 mN Prior work has demonstrated the effects of oxidation(for 10 which again indicates the lack of any interfacial chemical bond h)upon the interfacial micro-mechanical properties, namely the Examination of the surface of a polished sample, after sub- debonding energy (Gi) and frictional sliding stress([). Data sequent unstressed oxidation at 450.C for 100 h, demonstrates obtained in that prior study is presented in Table II for refer- that the Nicalon fibers protrude from the matrix by 1-2 um. ence. It is apparent that heat-treatment temperature has a This behavior arises from the mismatch in thermal expansion coefficient between the fibers, af. and the CAs matrix, am. Based upon published data, 31.32 a thermal expansion coefficient mis- 围 match,△x, can be determined, such that aa=am-ar≈1.9×10-6 K. Consequently, clamping of the fiber by the matrix can be nticipated when cooling from the initial processing temper ture. For example, for a 15-um diameter fiber the difference in shrinkage relative to the matrix sur 这0.04 17 nm, based on a AT of 1175.C (i.e, a processing temperature of 1200.C cooled to room temperature). At the same time. the 0.06 u>\able ll. Fiber Push Down Interfacial Properties of alon"/CAS in Both As-Received and Aged Conditions at treatment condition Debond energy, G,(/m) Sliding stress, t(MPa) 8.0±30 25+5 0.10 1.1+1.0 1200 40+14 Oxidation time(hrs) 450°C/100h 4.7+3.7 l44+63 525°C/100h 6.3+4.3 177+54 Fig 3. NIC loss after isothermal oxidation at 60 C for up to 1000 h 93±55 193+5expansion coefficient of the fiber due to oxidation.27,28 For the case of Nicalont NLM-202, the fiber used in the present CAS matrix composite, yf, is determined to be 1.48.28 Based on this, an estimate for the sealing time can be determined from the ox￾idation kinetics of Nicalont NLM-202 fibers.29 At the lowest temperature examined in that study, namely 7001C, sealing for a glass–ceramic matrix composite can be predicted to occur in approximately 35 h for a 50-nm thick carbon layer, 5 h for a 20-nm thick layer, and 1 h 20 min for a 10-nm thick layer. The residual stress state of this composite is such that the matrix will clamp down onto the fiber if the carbon layer is removed, thus making the apparent fiber–matrix gap smaller and the sealing time shorter. Oxidation kinetics data are not readily available for lower temperatures, but extrapolation of the behavior ob￾served at higher temperatures indicates that sealing will occur after approximately 20 h at 6001C for a 20-nm thick fiber–ma￾trix gap. This estimation is in general accordance with the weight gain response noted in Fig. 3, where weight gain ceases before 100 h exposure. A more thorough determination of sealing be￾havior at these lower temperatures (3751–6001C) would require continuous thermogravimetric analysis of both the fiber and composite oxidation weight changes as a function of time, which is beyond the scope of the present study. Prior work has demonstrated the effects of oxidation (for 100 h) upon the interfacial micro-mechanical properties, namely the debonding energy (Gi) and frictional sliding stress (t).30 Data obtained in that prior study is presented in Table II for refer￾ence. It is apparent that heat-treatment temperature has a significant effect upon these interfacial properties. After heat￾treatment at 3751C the debonding energy is significantly lower than for the as-received material, and essentially unchanged upon subsequent reloading, indicating that there is no real interfacial chemical bond present after this heat treatment.30 However, the frictional sliding stress is comparable with the as-received composite with a carbon interlayer. It can be seen from Fig. 3 that only partial carbon removal occurs after 100 h, as weight loss continues essentially linearly up to 1000 h, which infers that a thin carbon layer appears to be retained after 100 h. The interfacial debonding energy and frictional sliding stress are both increased by heat treatment at 4501C (Table II). Given the absence of a lubricating carbon interlayer after this heat treatment,27 a dramatically increased frictional sliding stress can be anticipated if the matrix and fiber come into direct contact. The increased debond energy is also believed to be due to the large frictional sliding stress that must be overcome before the fiber can be displaced, although the formation of isolated SiO2 bridges between the matrix and fiber cannot be completely dis￾counted (particularly given the observed weight gain after ex￾tended heat-treatment at 4501C (Fig. 3)). It was interesting to note from this prior study that, upon reloading, the debond load is essentially the same as the first loading cycle (i.e., B200 mN), which again indicates the lack of any interfacial chemical bond.30 Examination of the surface of a polished sample, after sub￾sequent unstressed oxidation at 4501C for 100 h, demonstrates that the Nicalont fibers protrude from the matrix by 1–2 mm. This behavior arises from the mismatch in thermal expansion coefficient between the fibers, af, and the CAS matrix, am. Based upon published data,31,32 a thermal expansion coefficient mis￾match, Da, can be determined, such that Da 5 am–af1.9
106 K1 . Consequently, clamping of the fiber by the matrix can be anticipated when cooling from the initial processing tempera￾ture. For example, for a 15-mm diameter fiber the difference in shrinkage relative to the matrix surrounding it is approximately 17 nm, based on a DT of 11751C (i.e., a processing temperature of 12001C cooled to room temperature). At the same time, the 375°C 450°C 525°C 600°C As-received 1 mm 100 N 1000 h 100 h 10 h Fig. 2. Schematic overview of the typical load-deflection curves obtained after isothermal oxidation of Nicalont/CAS at temperatures between 3751 and 6001C, for up to 1000 h. The behavior of the as-received material is also shown for reference. 0.00 0.02 0.04 0.06 0.08 0.10 0 200 400 600 1000 1200 375 450 525 600 Oxidation time (hrs.) 800 Mass loss (%) Fig. 3. Nicalont/CAS test bar weight loss after isothermal oxidation at temperatures between 3751 and 6001C for up to 1000 h. Table II. Fiber Push Down Interfacial Properties of Nicalont/CAS in Both As-Received and Aged Conditions30 Heat treatment condition Debond energy, Gi (J/m2 ) Sliding stress, t (MPa) As-received 8.073.0 2575 3751C/100 h 1.171.0 40714 4501C/100 h 4.773.7 144763 5251C/100 h 6.374.3 177754 6001C/100 h 9.375.5 193757 4052 Communications of the American Ceramic Society Vol. 90, No. 12