《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_SiC-SiC-6

Availableonlineatwww.sciencedirectcom ScienceDirect E噩≈RS ELSEVIER Journal of the European Ceramic Society 27(2007)377-388 www.elsevier.com/locate/jeurceramsoc Degradation mechanisms of a sic fiber reinforced self-sealing matrix composite in simulated combustor environments Ludovic Quemard, francis rebillat", Alain Guette Henri Tawil. Caroline louchet-Pouillerie Laboratoire des Composites Thenmostructuraux, UMR 5801(CNRS-SAFRAN-CEA-UB1), 3 Allee de la boetie, 33600 Pessac, France b Snecma Propulsion Solide, Les Cing Chemins, 33187 Le Haillan, france Received 23 November 2005: received in revised form 10 February 2006: accepted 17 February 2006 Available online 26 May 2006 Non-oxide ceramic matrix composites are potential candidates to replace the current nickel-based alloys for a variety of high temperature applications in the aerospace field. The durability of a SiC(/Py Co/Si, C, Blm) composite with a multi-layered self-sealing matrix and Hi-Nicalon fibers was investigated at 1200C for exposure durations up to 600 h. The specimens are aged in a variety of slow-flowing air/steam gas mixtures and total pressures, ranging from atmospheric pressure with a 10-50% water content to l MPa with 10-20% water content. The degradation of the composite was determined from the measurement of residual strength and strain to failure on post-exposure specimens and correlated with microstructural observation of the damaged tows. The most severe degradation of the composite occurred at 1 MPa in an air/steam(80/20)gas mixture. Correlation between this degradation and the dissolving of the Sic fibers in the generated boria-containing glass, is discussed. 2006 Elsevier ltd. all rights reserved. Keywords: Ceramic matrix composites; Corrosion; Mechanical properties; Lifetime; Engine components; SiC fibers 1. Introduction self-sealing approach are to consume part of the incoming oxy- gen and limit access of residual oxygen to the PyC interphase Non-oxide ceramic matrix composites such as SiCon/PyCa/ by sealing the matrix microcracks with a Sio2-B2O3 oxide SiC(m) consist of SiC matrix reinforced with SiC fibers and pyro- phase. However, previous studies showed that B2034-6. 8. 12. 13 carbon(PyC)interfacial coating. These composites exhibit a and Sio2 5, 7 can volatilize, respectively, at 600 and 1100oC low density associated with high thermomechanical properties under water vapor-containing environments. This phenomenon and are potential candidates to replace the current nickel-based can cause the self-sealing capability to degrade, thus reducing alloys for a variety of long-term applications in the aerospace the lifetime of the SiC(/PyCo/[Si, C, b](m) field. In these applications, SiCon/Py Co/SiC(m) components The matrix layers are SiC, B4C and a phase noted Si-B-C. can be subjected to service conditions that include mechani- The efficiency of the self-sealing process under environments cal loading under intermediate to high temperatures and high containing both oxygen and water vapor, results from the com- pressure complex environment containing oxygen and steam. petition between the oxidation of the matrix layers and the The oxidation of the PyC weak interphase can occur under volatilization of the generated oxide phase dry air at a temperature lower than 500C and leads to inter- Under dry air, B4C undergoes oxidation and volatiliza- facial degradations of SiC(o/PyCo/SiC(m). SiC(n/PyCo/Si, C, tion reactions below 600 and 900C, respectively, as shown B(m) composites with a sequenced ealing matrix have been below 4-6,, 13 developed,and investigated-Ito protect the PyC interphase against oxidation effects up to 1400C. The principles of the B4C(s)+4O2(g)=2B2O3+CO2(g) (1) (2) Corresponding author. Fax: +33 5 5684 12 25. Under water vapor-containing environments, B2O301 may E-mail address: rebillat@lcts. u-bordeauxl fr(F. Rebillat) react significantly at 600C to form hydroxydes by the following 0955-2219/S-see front matter o 2006 Elsevier Ltd. All rights reserved. doi: 10. 1016/j-jeurceramsoc. 2006.02.042

Journal of the European Ceramic Society 27 (2007) 377–388 Degradation mechanisms of a SiC fiber reinforced self-sealing matrix composite in simulated combustor environments Ludovic Quemard a, Francis Rebillat a,∗, Alain Guette a, Henri Tawil b, Caroline Louchet-Pouillerie b a Laboratoire des Composites Thermostructuraux, UMR 5801 (CNRS-SAFRAN-CEA-UB1), 3 All´ee de la Bo´etie, 33600 Pessac, France b Snecma Propulsion Solide, Les Cinq Chemins, 33187 Le Haillan, France Received 23 November 2005; received in revised form 10 February 2006; accepted 17 February 2006 Available online 26 May 2006 Abstract Non-oxide ceramic matrix composites are potential candidates to replace the current nickel-based alloys for a variety of high temperature applications in the aerospace field. The durability of a SiC(f)/PyC(i)/[Si, C, B](m) composite with a multi-layered self-sealing matrix and Hi-Nicalon fibers was investigated at 1200 ◦C for exposure durations up to 600 h. The specimens are aged in a variety of slow-flowing air/steam gas mixtures and total pressures, ranging from atmospheric pressure with a 10–50% water content to 1 MPa with 10–20% water content. The degradation of the composite was determined from the measurement of residual strength and strain to failure on post-exposure specimens and correlated with microstructural observation of the damaged tows. The most severe degradation of the composite occurred at 1 MPa in an air/steam (80/20) gas mixture. Correlation between this degradation and the dissolving of the SiC fibers in the generated boria-containing glass, is discussed. © 2006 Elsevier Ltd. All rights reserved. Keywords: Ceramic matrix composites; Corrosion; Mechanical properties; Lifetime; Engine components; SiC fibers 1. Introduction Non-oxide ceramic matrix composites such as SiC(f)/PyC(i)/ SiC(m) consist of SiC matrix reinforced with SiC fibers and pyro￾carbon (PyC) interfacial coating. These composites exhibit a low density associated with high thermomechanical properties and are potential candidates to replace the current nickel-based alloys for a variety of long-term applications in the aerospace field. In these applications, SiC(f)/PyC(i)/SiC(m) components can be subjected to service conditions that include mechani￾cal loading under intermediate to high temperatures and high pressure complex environment containing oxygen and steam. The oxidation of the PyC weak interphase can occur under dry air at a temperature lower than 500 ◦C and leads to inter￾facial degradations of SiC(f)/PyC(i)/SiC(m). SiC(f)/PyC(i)/[Si, C, B](m) composites with a sequenced self-sealing matrix have been developed1,2 and investigated3–11 to protect the PyC interphase against oxidation effects up to 1400 ◦C. The principles of the ∗ Corresponding author. Fax: +33 5 56 84 12 25. E-mail address: rebillat@lcts.u-bordeaux1.fr (F. Rebillat). self-sealing approach are to consume part of the incoming oxy￾gen and limit access of residual oxygen to the PyC interphase by sealing the matrix microcracks with a SiO2–B2O3 oxide phase. However, previous studies showed that B2O3 4–6,8,12,13 and SiO2 15,17 can volatilize, respectively, at 600 and 1100 ◦C under water vapor-containing environments. This phenomenon can cause the self-sealing capability to degrade, thus reducing the lifetime of the SiC(f)/PyC(i)/[Si, C, B](m). The matrix layers are SiC, B4C and a phase noted Si–B–C. The efficiency of the self-sealing process under environments containing both oxygen and water vapor, results from the com￾petition between the oxidation of the matrix layers and the volatilization of the generated oxide phase. Under dry air, B4C undergoes oxidation and volatiliza￾tion reactions below 600 and 900 ◦C, respectively, as shown below4–6,8,12,13: B4C(s) + 4O2(g) = 2B2O3(l) + CO2(g) (1) B2O3(l) = B2O3(g) (2) Under water vapor-containing environments, B2O3(l) may react significantly at 600 ◦C to form hydroxydes by the following 0955-2219/$ – see front matter © 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.jeurceramsoc.2006.02.042

78 L. Quemard et al. Journal of the European Ceramic Sociery 27 (2007)377-388 reactions 46,8,12,13 H3B3O Tranverse 3B2030+2H2O(g=H3 BO3e fiber tows Bulk porosity +5H2O HBO Multi-layered The competition between the oxidation()and volatilization reactions(2)(5)can lead to the recession of B4C In high temperature environments and under dry air, the active oxidation of Sic occurs above 1000C and below Po,=100 Pa Fig. 1. Polished cross-section of the as-received C410 material. according to(6). At a higher Po,, a paralinear oxidation of Sic 4 occurs to form a protective Sio2 scale according to(7) In environments containing O2 and H2O, the formation of Sioz different matrix layers are crystallised SiC, amorphous B4C and is dramatically enhanced by reaction(8)4-16 a SiC-B4C phase named Si-B-C which can be described as a SiC(s)+O2(g)=Siog)+CO( (6) mixture of Sic nanocrystals in a B, C amorphous phase.7The sequenced matrix is reinforced by Hi-Nicalon "SiC fibers and 2SiO(s)+2C0 (7) plane multi-layer reinforcement is used to eliminate delamina- tion sensitivity that is common for 2D ceramic matrix compos SiO2(s)+CO(g)+3H (8) ites Fiber volume fraction, material density and mainly closed Under steam-containing environments, the SiO2 scale may bulk porosity, as reported by the composite manufacturer, are, volatilize. by the main following reaction respectively, 34%0, 2.25+0.05 and 13+1%. The interphase is pyrocarbon SiO2(s)+2H2 O(g)= Si(oh)4(g) The test specimen geometry used in this study has a reduced gauge section(Fig. 2). It is 200 mm long, with a grip sec- The volatilization rate of SiO2 5, 7 is much lower than for tion width of 24 mm, a reduced gauge section width of 16mm B2034-6. 8, 12,13 However, as for boron matrix layers, the com- and a thickness of 4.4 mm. Two different material batches petition between the oxide formation(7)and (8)and the oxide manufactured in similar conditions, are used for these works phase volatilization(9)reactions can lead to the recession of Test samples are machined from composite plates using dia- mond grinding and then are seal-coated with CVI layers of The Si-B-C matrix layer can be described as a mixture Sic, B4C and Si-B-C. The sequenced seal-coat thickness, of Sic nanocrystals in an amorphous B4C phase. A previous with a SiC final layer, is 120 um on the composite surface Idy investigated the oxidation of Si-B-C coatings under oxy- and about 40 um on the machined edges. Corrosion tests gen and steam-containing environments by thermogravimetric are also performed on Si-B-C coating and SiC/SiC coupons analysis. 4.5 It has been shown that SiC nanocrystals can oxidize for comparison purposes. The Si-B-C coating, with a thick significantly at 600C and at atmospheric pressure, simultane- ness of 30+3 um, is deposited on Sic chips(diameter of ously to the B2 O3 formation, to form silica according to reaction 8 mm and thickness of 2 mm)via chemical vapor deposition Previous works 9 focused on the corrosion behavior of the (CVD) SiC(o/PyCo/Si, C, Blm)at high pressure and at 600C. The 3. Test procedures aim of this study is to evaluate, at a higher temperature, the effects of both oxygen and water vapor on the self-sealing 3. 1. Pre-damaging process of SiC(o/PyCo/Si, C, B](m) composites subjected to high-pressure environments. Corrosion tests are conducted for The dog-bone specimens are loaded in tension monotonically long periods of time on SiC(o/PyC(/Si,C, B](m)composites at at room temperature to a tensile stress of 150 MPa(the stress cor- 1200C in oxygen and steam-containing environments at atmo spheric pressure and high pressure. Post-exposure mechanical tests are performed at room temperature(rT)to investigate the effects of corrosion phenomena on the retained mechanical prop- erties 2. Materials and test specimens The material investigated is the CERASEPA41010, I (C410) manufactured by Snecma Propulsion Solide(France) via chemical vapor infiltration(CVD). It is a woven-SiC-fiber Fig 2. C410 specimen geometry used in this study. Dim re in millime. reinforced [Si, C, B] sequenced matrix composite(Fig. 1). The ters

378 L. Quemard et al. / Journal of the European Ceramic Society 27 (2007) 377–388 reactions4–6,8,12,13: 3 2B2O3(l) + 3 2H2O(g) = H3B3O6(g) (3) 3 2B2O3(l) + 1 2H2O(g) = H3BO3(g) (4) 1 2B2O3(l) + 1 2H2O(g) = HBO2(g) (5) The competition between the oxidation (1) and volatilization reactions (2)–(5) can lead to the recession of B4C. In high temperature environments and under dry air, the active oxidation of SiC occurs above 1000 ◦C and below PO2 = 100 Pa according to (6). At a higher PO2 , a paralinear oxidation of SiC14 occurs to form a protective SiO2 scale according to (7). In environments containing O2 and H2O, the formation of SiO2 is dramatically enhanced by reaction (8)14–16: SiC(s) + O2(g) = SiO(g) + CO(g) (6) 2SiC(s) + 3O2(g) = 2SiO2(s) + 2CO(g) (7) SiC(s) + 3H2O(g) = SiO2(s) + CO(g) + 3H2(g) (8) Under steam-containing environments, the SiO2 scale may volatilize, by the main following reaction17: SiO2(s) + 2H2O(g) = Si(OH)4(g) (9) The volatilization rate of SiO2 15,17 is much lower than for B2O3. 4–6,8,12,13 However, as for boron matrix layers, the com￾petition between the oxide formation (7) and (8) and the oxide phase volatilization (9) reactions can lead to the recession of SiC.18–20 The Si–B–C matrix layer can be described as a mixture of SiC nanocrystals in an amorphous B4C phase.7 A previous study investigated the oxidation of Si–B–C coatings under oxy￾gen and steam-containing environments by thermogravimetric analysis.4,5 It has been shown that SiC nanocrystals can oxidize significantly at 600 ◦C and at atmospheric pressure, simultane￾ously to the B2O3 formation, to form silica according to reaction (7). Previous works9 focused on the corrosion behavior of the SiC(f)/PyC(i)/[Si, C, B](m) at high pressure and at 600 ◦C. The aim of this study is to evaluate, at a higher temperature, the effects of both oxygen and water vapor on the self-sealing process of SiC(f)/PyC(i)/[Si, C, B](m) composites subjected to high-pressure environments. Corrosion tests are conducted for long periods of time on SiC(f)/PyC(i)/[Si, C, B](m) composites at 1200 ◦C in oxygen and steam-containing environments at atmo￾spheric pressure and high pressure. Post-exposure mechanical tests are performed at room temperature (RT) to investigate the effects of corrosion phenomena on the retained mechanical prop￾erties. 2. Materials and test specimens The material investigated is the CERASEP®A41010,11 (C410) manufactured by Snecma Propulsion Solide (France) via chemical vapor infiltration (CVI). It is a woven-SiC-fiber reinforced [Si, C, B] sequenced matrix composite (Fig. 1). The Fig. 1. Polished cross-section of the as-received C410 material. different matrix layers are crystallised SiC, amorphous B4C and a SiC–B4C phase named Si–B–C which can be described as a mixture of SiC nanocrystals in a B4C amorphous phase.7 The sequenced matrix is reinforced by Hi-Nicalon® SiC fibers and plane multi-layer reinforcement is used to eliminate delamina￾tion sensitivity that is common for 2D ceramic matrix compos￾ites. Fiber volume fraction, material density and mainly closed bulk porosity, as reported by the composite manufacturer, are, respectively, 34%, 2.25 ± 0.05 and 13 ± 1%. The interphase is pyrocarbon. The test specimen geometry used in this study has a reduced gauge section (Fig. 2). It is 200 mm long, with a grip sec￾tion width of 24 mm, a reduced gauge section width of 16 mm and a thickness of 4.4 mm. Two different material batches, manufactured in similar conditions, are used for these works. Test samples are machined from composite plates using dia￾mond grinding and then are seal-coated with CVI layers of SiC, B4C and Si–B–C. The sequenced seal-coat thickness, with a SiC final layer, is 120m on the composite surface and about 40 m on the machined edges. Corrosion tests are also performed on Si–B–C coating and SiC/SiC coupons for comparison purposes. The Si–B–C coating, with a thick￾ness of 30 ± 3m, is deposited on SiC chips (diameter of 8 mm and thickness of 2 mm) via chemical vapor deposition (CVD). 3. Test procedures 3.1. Pre-damaging The dog-bone specimens are loaded in tension monotonically at room temperature to a tensile stress of 150 MPa (the stress cor￾Fig. 2. C410 specimen geometry used in this study. Dimensions are in millime￾ters.

L Quemard et al. Journal of the European Ceramic Society 27(2007)377-388 Table 1 Summary of the test conditions for C410 composites exposed at 1200C in various environments PTot(MPa) Air/steam Po,(kPa) PH,o(kPa) PH,o/Po, v(cms) Furnace ABCDEF Furnace 1200±15 1220±25 0.56 HP furna 1225±30 1 HP furnace 8020 responding to twice their elastic limit) then unloaded before the 3.2. Corrosion tests corrosion exposures. The aim of this pre-damaging is to generate a controlled crack network in the matrix, which facilitates the The corrosion test conditions are reported in Table 1. Two ingress of the corrosive species. The residual strain is very low corrosion test equipments are used for these tests. High pressure (0.001%)and can be neglected for post-exposure mechanical corrosion tests are conducted in the high pressure-high tem- tests. The pre-damaging microcracks are mainly located in the perature furnace-(Fig 3). High pressure air is provided by a seal-coat of the gauge section of the specimens. At room tem- pressurized gas supply system then mixed with water in an evap- perature, their mean spacing distance is 230+ 30 um and their orator. The air and water flows are independently controlled by width is 0.5-3 um. In addition, few microcracks with a width mass flow meters and the air/H2o gas mixture is injected in the lower than 1 um at RT are present at the edge of the macrop- alumina test tube (i d: 34 mm, purity: 99.7%, OMG, France) porosities of the furnace. A system of pneumatically driven back pressure allv driven hack nn val HyOn mass now me Pneumatically driven PRESSURIZED GAS let tuhe Pneumatically driven Dry air mass nkm meter SIFAM GENERAT(N Fig 3. Schematic of the high temperature-high pressure corrosion test equipment(a)and view of the fumace(b)

L. Quemard et al. / Journal of the European Ceramic Society 27 (2007) 377–388 379 Table 1 Summary of the test conditions for C410 composites exposed at 1200 ◦C in various environments No. Exposure T ( ◦C) PTot (MPa) Air/steam PO2 (kPa) PH2O (kPa) PH2O/PO2 v (cm s−1) A Furnace 1210 ± 10 0.1 90/10 18 10 0.56 5 B Furnace 1210 ± 10 0.1 80/20 16 20 1.25 5 C Furnace 1200 ± 15 0.1 50/50 10 50 5 10 D HP furnace 1220 ± 25 0.45 90/10 81 45 0.56 8 E HP furnace 1225 ± 30 1 90/10 180 100 0.56 8 F HP furnace 1225 ± 30 1 80/20 160 200 1.25 8 responding to twice their elastic limit) then unloaded before the corrosion exposures. The aim of this pre-damaging is to generate a controlled crack network in the matrix, which facilitates the ingress of the corrosive species. The residual strain is very low (∼=0.001%) and can be neglected for post-exposure mechanical tests. The pre-damaging microcracks are mainly located in the seal-coat of the gauge section of the specimens. At room tem￾perature, their mean spacing distance is 230 ± 30m and their width is 0.5–3m. In addition, few microcracks with a width lower than 1 m at RT are present at the edge of the macrop￾orosities. 3.2. Corrosion tests The corrosion test conditions are reported in Table 1. Two corrosion test equipments are used for these tests. High pressure corrosion tests are conducted in the high pressure–high tem￾perature furnace21 (Fig. 3). High pressure air is provided by a pressurized gas supply system then mixed with water in an evap￾orator. The air and water flows are independently controlled by mass flow meters and the air/H2O gas mixture is injected in the alumina test tube (i.d.: 34 mm, purity: 99.7%, OMG, France) of the furnace. A system of pneumatically driven back pressure Fig. 3. Schematic of the high temperature–high pressure corrosion test equipment (a) and view of the furnace (b).

380 L. Quemard et al Journal of the European Ceramic Sociery 27 (2007)377-388 Bubbling exhaus presaturator Flow meter Fig. 4. Schematic of the high temperature furnace and the water vapor saturator at atmospheric pressure. reducers maintains a slight difference of pressure between the After exposures, the test specimens are cut perpendicularly tube interior and the metallic vessel(Ptube -Pvessel=-3 kPa). and parallel to the loading axis then polished for examina This permits to minimize stresses on the tube and increase its tion using an optical microscope. Moreover, the fractured sur- airtightness. The uniform heating zone of the furnace is approx faces are analyzed by scanning electron microscopy(SEM) imately 120 mm long which is longer than the gage length of the The chemical composition of the oxide phase formed during the corrosion tests is determined on polished cross-sections of A high temperature furnace associated with a water saturator the composites using electron probe micro analysis(EPMA is used to run the corrosion tests at atmospheric pressure( Fig 4). CAMECA SX100) The dry air flows through a heated water column in order to be saturated in steam before its introduction in the alumina tube 4. Results (i.d. 34 mm, purity: 99.7%, OMG, France)of the furnace. The temperature of the water in the column is slightly higher than the 4.1. Post-expo chanical results dewpoint corresponding to the desired water vapor partial pres sure. For example, an air/steam(90/10)gas mixture is obtained The post-exposure mechanical properties of the C410 pi for a column temperature of 48C(dewpoint for PH20= 10kPa damaged specimens are determined using cyclic tensile tests at is 46C). Water content in the gas stream is monitored by mea- RT. The results are shown in Fig. 5 and reported in Table 2. suring the condensate in the gas exhaust daily and the amount Three C410 specimens are used to determine the mechanical of water in the stock which supplies the heated column. The properties up to failure of the two batches of the as-re uniform heating zone of the furnace is approximately 220 mm material using cyclic tensile tests at RT. The failures of all the specimens occurred in the gauge In both corrosion test equipments, the C410 specimens are section. Small brittle rupture areas and wide non-brittle rup- oriented parallel to the gas flow and placed on alumina sample holders(purity: 99.7%, OMG, France)specially designed. The heating and cooling rates, used at atmospheric pressure under ambient air, are, respectively, 150 and 100Ch-. The expe scale(Precisa Instruments AG, Switzerland)with an accuracy oo t ures are regularly interrupted to weigh the specimens using a ofl×10-mg 3.3. Characterization of specimens after exposur 9四苏 Post-exposure cyclic tensile tests are performed at RT on the 5 40 dog-bone specimens. A spring-loaded clip on-gauge is attached to the 25 mm long of the straight section of the samples to record displacement. The composites are tested up to failure in a servo controlled testing machine(INSTRON 1185)equipped with self Fig. 5. The effect of corrosion environments on the ultimate strength and on the aligning grips at a cross-head speed of 0.40+0.05%o min strain to failure of the C410 specimens exposed for 600h

380 L. Quemard et al. / Journal of the European Ceramic Society 27 (2007) 377–388 Fig. 4. Schematic of the high temperature furnace and the water vapor saturator at atmospheric pressure. reducers maintains a slight difference of pressure between the tube interior and the metallic vessel (Ptube − Pvessel = −3 kPa). This permits to minimize stresses on the tube and increase its airtightness. The uniform heating zone of the furnace is approx￾imately 120 mm long which is longer than the gage length of the specimens. A high temperature furnace associated with a water saturator is used to run the corrosion tests at atmospheric pressure (Fig. 4). The dry air flows through a heated water column in order to be saturated in steam before its introduction in the alumina tube (i.d.: 34 mm, purity: 99.7%, OMG, France) of the furnace. The temperature of the water in the column is slightly higher than the dewpoint corresponding to the desired water vapor partial pres￾sure. For example, an air/steam (90/10) gas mixture is obtained for a column temperature of 48 ◦C (dewpoint for PH2O = 10 kPa is 46 ◦C). Water content in the gas stream is monitored by mea￾suring the condensate in the gas exhaust daily and the amount of water in the stock which supplies the heated column. The uniform heating zone of the furnace is approximately 220 mm long. In both corrosion test equipments, the C410 specimens are oriented parallel to the gas flow and placed on alumina sample holders (purity: 99.7%, OMG, France) specially designed. The heating and cooling rates, used at atmospheric pressure under ambient air, are, respectively, 150 and 100 ◦C h−1. The expo￾sures are regularly interrupted to weigh the specimens using a scale (Precisa Instruments AG, Switzerland) with an accuracy of 1 × 10−2 mg. 3.3. Characterization of specimens after exposure Post-exposure cyclic tensile tests are performed at RT on the dog-bone specimens. A spring-loaded clip on-gauge is attached to the 25 mm long of the straight section of the samples to record displacement. The composites are tested up to failure in a servo controlled testing machine (INSTRON 1185) equipped with self aligning grips at a cross-head speed of 0.40 ± 0.05% min−1. After exposures, the test specimens are cut perpendicularly and parallel to the loading axis then polished for examina￾tion using an optical microscope. Moreover, the fractured sur￾faces are analyzed by scanning electron microscopy (SEM). The chemical composition of the oxide phase formed during the corrosion tests is determined on polished cross-sections of the composites using electron probe micro analysis (EPMA, CAMECA SX100). 4. Results 4.1. Post-exposure mechanical results The post-exposure mechanical properties of the C410 pre￾damaged specimens are determined using cyclic tensile tests at RT. The results are shown in Fig. 5 and reported in Table 2. Three C410 specimens are used to determine the mechanical properties up to failure of the two batches of the as-received material using cyclic tensile tests at RT. The failures of all the specimens occurred in the gauge section. Small brittle rupture areas and wide non-brittle rup￾Fig. 5. The effect of corrosion environments on the ultimate strength and on the strain to failure of the C410 specimens exposed for 600 h

L Quemard et al. Journal of the European Ceramic Society 27(2007)377-388 Table 2 Summary of C4 10 composites post-exposure tensile properties at RT Test condition Exposure time(h) Weight change(%) Failure UTS (MPa) Strain to failure(%) E(GPa) +44 BBB 6245 170 As-received( 21212112 304±45 0.46±0.1 252±4 368±11 0.73±0.02 86±21 315±20 a Presence of brittle rupture areas. b Brittle rupture. n-brittle rupture ture areas characterized by fiber pull-out, are observed on the tested in low steam-pressure environment are similar to the as- fractured surfaces(Fig. 6)of the specimens exposed in low received ones(Fig. 7). Thus, a non-linear stress-strain behavior steam-pressure environments(conditions A and B). Brittle rup- without a plateau is observed up to the ultimate failure of the ture areas are located at the edge of the bulk porosities filled specimens. According to this behavior induced by matrix crack with an oxide phase. The stress-strain curves of the specimens ing, a continuous damaging occurred in the composite up to its failure. Moreover, the width of the hysteresis loops is narrow and the residual strains after unloading are very low. This ind cates a high fiber-matrix load transfer, thus a strong interfacial shear stress. The retained mechanical properties of samples aged at low steam-pressure are similar to the as-received materials and C410 data base, indicating that the fibers are not dam- aged significantly(Fig. 5). Many brittle rupture areas, located A1MPa. Air sen(Nigo) MB·A3am(20 Strain ( Borosilicate Glass MP· Air Skans门0 MPa- Air Slar ISss0 Fig. 6. SEMfracture surfaces of C410specimens in an air/ steam(80/20) b o gas mixture for 616 h at 1200C and 0.1 MPa(a) and for 603 h at 1200 C and Fig. 7. Stress-strain curves of the C410 specimens of batch 1(a)and batch 2 I MPa(b) (b)obtained at RT after exposure at 1200C for 600 h in various envi

L. Quemard et al. / Journal of the European Ceramic Society 27 (2007) 377–388 381 Table 2 Summary of C410 composites post-exposure tensile properties at RT Test condition Batch Exposure time (h) Weight change (%) Failure UTS (MPa) Strain to failure (%) E (GPa) A 2 611 +3.4 B areasa 296 0.68 170 B 1 616 +2.4 B areas 310 0.65 220 C 2 601 +3.3 Bb 195 0.22 220 D 1 606 +4.4 B 229 0.34 210 E 2 609 +3.9 B 240 0.45 170 F 1 603 +5.9 B 97 0.08 180 As-received (average) 1 – – NBc 304 ± 45 0.46 ± 0.12 252 ± 4 As-received (average) 2 – – NB 368 ± 11 0.73 ± 0.02 186 ± 21 C410 data base15,16 – – NB 315 ± 20 0.5 220 ± 25 a Presence of brittle rupture areas. b Brittle rupture. c Non-brittle rupture. ture areas characterized by fiber pull-out, are observed on the fractured surfaces (Fig. 6) of the specimens exposed in low steam-pressure environments (conditions A and B). Brittle rup￾ture areas are located at the edge of the bulk porosities filled with an oxide phase. The stress–strain curves of the specimens Fig. 6. SEM fracture surfaces of C410 specimens exposed in an air/steam (80/20) gas mixture for 616 h at 1200 ◦C and 0.1 MPa (a) and for 603 h at 1200 ◦C and 1 MPa (b). tested in low steam-pressure environment are similar to the as￾received ones (Fig. 7). Thus, a non-linear stress–strain behavior without a plateau is observed up to the ultimate failure of the specimens. According to this behavior induced by matrix crack￾ing, a continuous damaging occurred in the composite up to its failure. Moreover, the width of the hysteresis loops is narrow and the residual strains after unloading are very low. This indi￾cates a high fiber–matrix load transfer, thus a strong interfacial shear stress. The retained mechanical properties of samples aged at low steam-pressure are similar to the as-received materials and C410 data base,10,11 indicating that the fibers are not dam￾aged significantly (Fig. 5). Many brittle rupture areas, located Fig. 7. Stress–strain curves of the C410 specimens of batch 1 (a) and batch 2 (b) obtained at RT after exposure at 1200 ◦C for 600 h in various environments

L. Quemard et al. Journal of the European Ceramic Sociery 27 (2007)377-388 the edge of the bulk porosities filled with an oxide phase, are relation observed on the fractured surfaces of the samples aged in high water vapor-pressure environments(Fig. 6). The high steam- ET=2VIEr ressure exposures affect the ultimate tensile strength (UTS)and the strain to failure of the C410 composites, leading to reductions where Vr is the fiber volume(34%0), Er the fiber modulus at RT of 25-70% and 25-80%0, respectively(Fig. 5) The variations of the secant modulus, shown in Fig 8, are at failure of a 2D woven composite considering that only the determined from the stress-strain curves at each load-unload fibers, oriented in the direction of the loading axis, bear the load loop to highlight the damage progression in the composites. at failure First, the initial modulus of the aged composites has to be The materials exposed in the environments F and C have mpared with the as-received material modulus correspond- the lower retained mechanical properties and their extrapolated ing to the stress used for the pre-damaging(150MPa). Thus, modulus at failure are significantly higher than the theoretical modulus of the as-received material corresponding to a stress of progression and an embrittlement of the composite st damage 50 MPa. This is consistent with the sealing of the pre-damaging The residual strains after unloading of the composites racks with an oxide phase. Nevertheless, increasing the stress exposed at low steam-pressure are strongly increased at a results in a decrease of the modulus of the stress sligl ghtly higher than 100 MPa, by comparison with the as- faster than for the as-received material. This corresponds to a received materials(Fig. 9). This could be related to the presence faster damage progression in these aged specimens, certainly of the brittle areas observed on the fractured surfaces. Indeed associated to local brittle ruptures of few tows. This results in a brittle rupture areas could occur at a stress of 100-150 MPa. premature failure except for the materials exposed at low steam- Thus, the load is transferred on the remaining unbroken fibers pressure(environments A and B in Fig. 5). The modulus at This could result in a sudden opening of the cracks with the failure, determined using a linear extrapolation of the measured extension of the debonded length in the non- or less-oxidized modulus( dotted lines in Fig 8), is compared with the theoret- areas, enhancing the sliding of the fibers. 22.23However, the cal modulus at failure ETh. It is calculated from the following mechanical properties at failure are maintained(Fig. 5). The residual strains after unloading of the strongly damaged com- posites exposed in the gh steam-pressure environments, show low or non-significant variations by comparison with the as- received material(Fig 9). This is in agreement with their brittle As Reserved Batch failure(Table 2)certainly caused by the increase of the interfa- -CoI MPa.AiE Meam (SO cial shear stress at the fiber/matrix interphase. I MIa.Air Stam (80/ Weight changes of C410 composites after exposures at 1200Cin various environments for 600 h are reported in Table 2 and shown in Fig. 10. All of the samples show significant weight gains between 2.5 and 6%0. attributed to the formation of an 50 100 150 200 250 300 350 400 oxide phase generated by the matrix oxidation at the surface Maximal Stress after Cycling(MPa) and in the bulk. After a time t, which depends on the exposure conditions, the weight gains reach a pseudo-steady state cor- ing to the sealing of the seal-coat a-As Renewed Batch 2 partial or total filling of the bulk porosities. The weight change -.I MPa-AirSkum (90/ID) kinetics before reaching the steady state can be characterized -OI MPa. AipStcum (5(150) I MPa. Air/Sleam (0/1oI by weight gain rates obtained from linear regressions(Fig. 11) Fig. ll shows that high-pressure environments enhance the ini- tial oxidation kinetics, thus reducing the time to achieve the sealing of the composite. Thus, by increasing the total pres- sure by a factor of 10, the linear weight gain rate is increased four-fold. In the atmospheric pressure environments, increas- .5(V,Ey ing steam-pressure results in higher weight gain rates whicl 50100150 250300350 are comparable, for the air/steam(50/50) gas mixture, to those Maximal Stress after Cycling(MPa) obtained in high-pressure environments. This is consistent with Fig8. Damage progression of C410 specimens from batch 1(a)and batch 2 previous works which showed that the silicon carbide oxida- o)during tensile cycling at RT after exposures at 600 C for 600 h in varie tion kinetic is dramatically enhanced in presence of eam14-16 environments(the dotted lines correspond to the modulus at failure extrapolated ( the Sic oxidation rate is proportional to PH2o with a power from (7)) law exponent of 1). Nevertheless, increasing the water vapor

382 L. Quemard et al. / Journal of the European Ceramic Society 27 (2007) 377–388 at the edge of the bulk porosities filled with an oxide phase, are observed on the fractured surfaces of the samples aged in high water vapor-pressure environments (Fig. 6). The high steam￾pressure exposures affect the ultimate tensile strength (UTS) and the strain to failure of the C410 composites, leading to reductions of 25–70% and 25–80%, respectively (Fig. 5). The variations of the secant modulus, shown in Fig. 8, are determined from the stress–strain curves at each load–unload loop to highlight the damage progression in the composites. First, the initial modulus of the aged composites has to be compared with the as-received material modulus correspond￾ing to the stress used for the pre-damaging (150 MPa). Thus, the exposed materials have an initial modulus higher than the modulus of the as-received material corresponding to a stress of 150 MPa. This is consistent with the sealing of the pre-damaging cracks with an oxide phase. Nevertheless, increasing the stress results in a decrease of the modulus of the aged composites faster than for the as-received material. This corresponds to a faster damage progression in these aged specimens, certainly associated to local brittle ruptures of few tows. This results in a premature failure except for the materials exposed at low steam￾pressure (environments A and B in Fig. 5). The modulus at failure, determined using a linear extrapolation of the measured modulus (dotted lines in Fig. 8), is compared with the theoret￾ical modulus at failure ETh. It is calculated from the following Fig. 8. Damage progression of C410 specimens from batch 1 (a) and batch 2 (b) during tensile cycling at RT after exposures at 600 ◦C for 600 h in various environments (the dotted lines correspond to the modulus at failure extrapolated from (7)). relation: ETh = 1 2VfEf (10) where Vf is the fiber volume (34%), Ef the fiber modulus at RT (Hi-Nicalon® fiber: 250 GPa) and ETh the theoretical modulus at failure of a 2D woven composite considering that only the fibers, oriented in the direction of the loading axis, bear the load at failure. The materials exposed in the environments F and C have the lower retained mechanical properties and their extrapolated modulus at failure are significantly higher than the theoretical modulus (Fig. 8). This is consistent with a very fast damage progression and an embrittlement of the composite. The residual strains after unloading of the composites exposed at low steam-pressure are strongly increased at a stress slightly higher than 100 MPa, by comparison with the as￾received materials (Fig. 9). This could be related to the presence of the brittle areas observed on the fractured surfaces. Indeed, brittle rupture areas could occur at a stress of 100–150 MPa. Thus, the load is transferred on the remaining unbroken fibers. This could result in a sudden opening of the cracks with the extension of the debonded length in the non- or less-oxidized areas, enhancing the sliding of the fibers.22,23 However, the mechanical properties at failure are maintained (Fig. 5). The residual strains after unloading of the strongly damaged com￾posites exposed in the high steam-pressure environments, show low or non-significant variations by comparison with the as￾received material (Fig. 9). This is in agreement with their brittle failure (Table 2) certainly caused by the increase of the interfa￾cial shear stress at the fiber/matrix interphase. 4.2. Weight changes Weight changes of C410 composites after exposures at 1200 ◦C in various environments for 600 h are reported inTable 2 and shown in Fig. 10. All of the samples show significant weight gains, between 2.5 and 6%, attributed to the formation of an oxide phase generated by the matrix oxidation at the surface and in the bulk. After a time t, which depends on the exposure conditions, the weight gains reach a pseudo-steady state cor￾responding to the sealing of the seal-coat microcracks and the partial or total filling of the bulk porosities. The weight change kinetics before reaching the steady state can be characterized by weight gain rates obtained from linear regressions (Fig. 11). Fig. 11 shows that high-pressure environments enhance the ini￾tial oxidation kinetics, thus reducing the time to achieve the sealing of the composite. Thus, by increasing the total pres￾sure by a factor of 10, the linear weight gain rate is increased four-fold. In the atmospheric pressure environments, increas￾ing steam-pressure results in higher weight gain rates which are comparable, for the air/steam (50/50) gas mixture, to those obtained in high-pressure environments. This is consistent with previous works which showed that the silicon carbide oxida￾tion kinetic is dramatically enhanced in presence of steam14–16 (the SiC oxidation rate is proportional to PH2O with a power law exponent of 1). Nevertheless, increasing the water vapor

L Quemard et al /Journal of the European Ceramic Society 27(2007)377-388 与四 O. MP- Air Meam() sMB- Ai sam rwg -trI MP- Ai Seam (b). 55 100 50 Maximal Stress after Cycling(MPa) Maximal Strain after Cycling(MPa) As Reserved Batch 2 0I MP- Air Sean Isl) 01M内-ASca(5 8需5 合MANm 050,10,150.20.250,30350.40450.50.550.60.650,7 Maximal Stress after Cycling(MPa) Maximal Strain after Cycling(MPa) Fig. 9. Variation of the residual strains after unloading of C410 specimens from batch 1(a) and batch 2(b)during tensile cycling at RT after exposure for 600h in pressure in high-pressure environments by a factor 2 does not ture of the different constituents SiC, B4 C and Si-B-C which are enhance significantly the weight gain rates. This variation of oxidized in sequence as the oxidation front progresses through rate is certainly slowed down because of a diffusion controlled the core. Thus, the SiC matrix layers oxidize first followed by regime in which oxidation kinetics are dependent on the rate at the boron components through the pre-damaging cracks and the which oxygen and water vapor are supplied to the surface bulk porosities(Fig. 12). Each layers oxidizes at a different rate cording to(i) the 4.3. Post-exposure observations of C410 composites products and glass, (ii) the quantity of generated sealing phase and(iii) the capability of oxygen to diffuse and the volatility of All of the exposed C410 specimens were examined in the glass which are dependent on its composition. Silica and boria gage region. The matrix final layer at the surface and in the bulks SiC. The material can be considered as a complex layered struc (82 e'SteamooIU) 045 MPa 0.IMPa D Fig. 10. Weight change kinetics of C410 composites exposed at 1200C for Fig. 11. Linear weight gain rates of C410 composites exposed at 1200C for 600 h in various environments 600h in various air/steam gas mixtures

L. Quemard et al. / Journal of the European Ceramic Society 27 (2007) 377–388 383 Fig. 9. Variation of the residual strains after unloading of C410 specimens from batch 1 (a) and batch 2 (b) during tensile cycling at RT after exposure for 600 h in various environments. pressure in high-pressure environments by a factor 2 does not enhance significantly the weight gain rates. This variation of rate is certainly slowed down because of a diffusion controlled regime in which oxidation kinetics are dependent on the rate at which oxygen and water vapor are supplied to the surface. 4.3. Post-exposure observations of C410 composites All of the exposed C410 specimens were examined in the gage region. The matrix final layer at the surface and in the bulk is SiC. The material can be considered as a complex layered struc￾Fig. 10. Weight change kinetics of C410 composites exposed at 1200 ◦C for 600 h in various environments. ture of the different constituents SiC, B4C and Si–B–C which are oxidized in sequence as the oxidation front progresses through the core. Thus, the SiC matrix layers oxidize first followed by the boron components through the pre-damaging cracks and the bulk porosities (Fig. 12). Each layers oxidizes at a different rate according to (i) the reactions (1)–(9) forming gaseous reaction products and glass, (ii) the quantity of generated sealing phase and (iii) the capability of oxygen to diffuse and the volatility of glass which are dependent on its composition. Silica and boria Fig. 11. Linear weight gain rates of C410 composites exposed at 1200 ◦C for 600 h in various air/steam gas mixtures

L. Quemard et al. Journal of the European Ceramic Sociery 27 (2007)377-388 Borosilicate glass Borosilicate glass B Fig. 12. SEM fracture surface of a C410 specimen exposed in an air/steam Borosilicate glass (50/50)gas mixture for 601 h at 1200C and 0. 1 MPa, then tested in tension at Rt showing the attacked SiC fibers embedded in a borosilicate glass at the edge of a sealed bulk porosity. react to form a liquid borosilicate glass, flowing more easily than pure silica, which seals the bulk porosities and the pre-damaging racks of the seal-coat (Fig. 12). The borosilicate is not protec-(b): ,.i tive enough and once the first matrix layer around the fibers is reached, the oxidation of the tows(fiber and Pyc interphase) Fig 13. Polished cross-sections of a C410 composite exposed at 1200C and occurs( Fig. 12). The SiC fiber oxidation kinetic can be dramat an air/steam(80/20) gas mixture for 603 h then tested in tension at RT. cally enhanced by contact with the boria-containing glass, in showing the damages at the surface(a) and in the bulk(b). agreement with the following reaction 24-26 4.4. Analysis of the oxides generated SiC(s)+ yB2O3()=2B203-uSioz(soul)+ u CO(g) (11) Polished cross-sections in the gage length of the exposed The widest damaged areas, characterized by the dissolving of specimens are analyzed by EPMA. The mean compositions of tows with a large amount of fibers embedded in the borosilicate the bulk borosilicate glass are obtained from 40 to 60 measure- glass, are observed on the composites exposed in high steam- ment points located in several filled bulk porosities(Fig. 14). All pressure environments(Figs. 12 and 13). These degradations are of the materials exposed for 600 h contain an oxide phase rich responsible for the strong decrease of their mechanical proper- in silica, with a boria molar content ranging from 5 to 20%. The ties(environments C and F in Fig. 5). Not only the presence of low boria content is due to(i) its high volatility at high tempera- steam accelerates the oxidation of Sic and boron components, ture in steam environments and (ii) its reaction with SiC to form but the high O2 and H2O(g) diffusivities through the borosilicate silica. Thus, the lowest boria content is obtained in the highest glass further increase the degradation rates of the composites steam-pressure environment which caused the more important constituents. Thus, in the highest steam-pressure environment damages in the material. Nevertheless, the oxide composition (1 MPa-air/steam(80/20)), the exposure for 600 h leads to in the other environments after 600 h exposure, seems to be not the total degradation of the seal-coat which is replaced by a clearly dependent on the total pressure and the water vapor con- thick oxide layer in contact with the fibers( Fig. 13). Impurities tent present in these industrial composites also increase the oxidation Fig. 15 highlights a localised boria gradient in the glass with rates, particularly in presence of water vapor. Finally, exces- a minimal boria content around the fibers attacked. This is due sive amounts of gaseous reaction products are formed, creating to formation of silica by oxidation of the SiC fibers, highly porosities and resulting in a non-protective borosilicate glas enhanced by boria(I1)

384 L. Quemard et al. / Journal of the European Ceramic Society 27 (2007) 377–388 Fig. 12. SEM fracture surface of a C410 specimen exposed in an air/steam (50/50) gas mixture for 601 h at 1200 ◦C and 0.1 MPa, then tested in tension at RT showing the attacked SiC fibers embedded in a borosilicate glass at the edge of a sealed bulk porosity. react to form a liquid borosilicate glass, flowing more easily than pure silica, which seals the bulk porosities and the pre-damaging cracks of the seal-coat (Fig. 12). The borosilicate is not protec￾tive enough and once the first matrix layer around the fibers is reached, the oxidation of the tows (fiber and PyC interphase) occurs (Fig. 12). The SiC fiber oxidation kinetic can be dramat￾ically enhanced by contact with the boria-containing glass, in agreement with the following reaction24–26: xSiC(s) + yB2O3(l) = zB2O3 − vSiO2(soul) + uCO(g) (11) The widest damaged areas, characterized by the dissolving of tows with a large amount of fibers embedded in the borosilicate glass, are observed on the composites exposed in high steam￾pressure environments (Figs. 12 and 13). These degradations are responsible for the strong decrease of their mechanical proper￾ties (environments C and F in Fig. 5). Not only the presence of steam accelerates the oxidation of SiC and boron components, but the high O2 and H2O(g) diffusivities through the borosilicate glass further increase the degradation rates of the composites constituents. Thus, in the highest steam-pressure environment (1 MPa—air/steam (80/20)), the exposure for 600 h leads to the total degradation of the seal-coat which is replaced by a thick oxide layer in contact with the fibers (Fig. 13). Impurities present in these industrial composites also increase the oxidation rates, particularly in presence of water vapor. Finally, exces￾sive amounts of gaseous reaction products are formed, creating porosities and resulting in a non-protective borosilicate glass. Fig. 13. Polished cross-sections of a C410 composite exposed at 1200 ◦C and 1 MPa in an air/steam (80/20) gas mixture for 603 h then tested in tension at RT, showing the damages at the surface (a) and in the bulk (b). 4.4. Analysis of the oxides generated Polished cross-sections in the gage length of the exposed specimens are analyzed by EPMA. The mean compositions of the bulk borosilicate glass are obtained from 40 to 60 measure￾ment points located in several filled bulk porosities (Fig. 14). All of the materials exposed for 600 h contain an oxide phase rich in silica, with a boria molar content ranging from 5 to 20%. The low boria content is due to (i) its high volatility at high tempera￾ture in steam environments and (ii) its reaction with SiC to form silica. Thus, the lowest boria content is obtained in the highest steam-pressure environment which caused the more important damages in the material. Nevertheless, the oxide composition in the other environments after 600 h exposure, seems to be not clearly dependent on the total pressure and the water vapor con￾tent. Fig. 15 highlights a localised boria gradient in the glass with a minimal boria content around the fibers attacked. This is due to formation of silica by oxidation of the SiC fibers, highly enhanced by boria (11)

L Quemard et al /Journal of the European Ceramic Society 27(2007)377-388 dation of the silicon carbide in contact(11). This, first increases he matrix oxidation, then the fiber reinforcement consump- tion during the long times of exposure. Thus, the fiber volume decreases and is replaced by a borosilicate phase which weakens the mechanical properties of C410. Indeed, firstly, diminishing the fiber volume by dissolving of Sic decreases the number of fibers bearing the load and, thus the uts(Figs. 5 and 16) Secondly, the attacked fibers are embedded in glass(stron fiber/oxide chemical bond at rT) which facilitates the propa gation of the matrix cracks and causes the ultimate strain and the uts to diminish Fig. 14. Mean compositions of the borosilicate glass sealing the bulk porositi of C410 composites exposed for 600h in various environments at 1200C. 5.2. Effect of the ratio PH,0/Po, Results obtained using EPma Fig. 16 highlights that, increasing the ratio PH,o/Po2,1.e 5. Discussion he water vapor content relatively to the oxygen content, reduces th significantly the UTS and strain to failure of the C410 material 5.1. Efect of the total p in both atmospheric pressure and high-pressure environment Thus, increasing the ratio PH,o/Po, by a factor 9 in atmospheric Fig. 16 shows that increasing the total pressure of the envi- pressure gas mixtures decreases the UTs and the strain to fail ronment leads to a significant reduction of the UTS and strain to ure by a factor of 1.5 and 3, respectively. These variations are failure of the C410 material. Thus, by increasing the total pres- enhanced in high-pressure environments, in agreement with the sure of an air/steam(80/20) gas mixture by a factor of 10, the effect of the total pressure previously described. Thus, UTS and UTS and strain to failure are divided by 3 and 8, respectively. strain to failure diminish by a factor 2.5 and 4 with the only This can be attributed to the enhancement of oxidation kinetics increase of PH,o/Po, by a factor 2. It is well known that the in high-pressure environments resulting in the increase of the presence of water vapor enhances the oxidation kinetic of silicon weight gain kinetics(Figs. 10 and 11). Indeed, the borosilicate glass, generated faster and in high quantity, accelerates the oxi- Oxide B As Revered Material (eN Des 1 as21 P H2o/0 SiC Fiber a AsRocerwed Materal (Awere orache 1 an2 60 10 Position along the axis z(arbitrary units) Fig. 15. EPMA line scanning profile(axis 2) in a filled bulk porosity of a C410 specimen exposed at 1200C and 0. 1 MPa for 601 h in an air/steam(50/50) gas Fig. 16. Effects of the environment on UTS(a)and strain to failure(b)of C410 mixture composites exposed at 1200C for 600h in various air/steam gas mixtures

L. Quemard et al. / Journal of the European Ceramic Society 27 (2007) 377–388 385 Fig. 14. Mean compositions of the borosilicate glass sealing the bulk porosities of C410 composites exposed for 600 h in various environments at 1200 ◦C. Results obtained using EPMA. 5. Discussion 5.1. Effect of the total pressure Fig. 16 shows that increasing the total pressure of the envi￾ronment leads to a significant reduction of the UTS and strain to failure of the C410 material. Thus, by increasing the total pres￾sure of an air/steam (80/20) gas mixture by a factor of 10, the UTS and strain to failure are divided by 3 and 8, respectively. This can be attributed to the enhancement of oxidation kinetics in high-pressure environments resulting in the increase of the weight gain kinetics (Figs. 10 and 11). Indeed, the borosilicate glass, generated faster and in high quantity, accelerates the oxi￾Fig. 15. EPMA line scanning profile (axis Z) in a filled bulk porosity of a C410 specimen exposed at 1200 ◦C and 0.1 MPa for 601 h in an air/steam (50/50) gas mixture. dation of the silicon carbide in contact (11). This, first increases the matrix oxidation, then the fiber reinforcement consump￾tion during the long times of exposure. Thus, the fiber volume decreases and is replaced by a borosilicate phase which weakens the mechanical properties of C410. Indeed, firstly, diminishing the fiber volume by dissolving of SiC decreases the number of fibers bearing the load and, thus the UTS (Figs. 5 and 16). Secondly, the attacked fibers are embedded in glass (strong fiber/oxide chemical bond at RT) which facilitates the propa￾gation of the matrix cracks and causes the ultimate strain and the UTS to diminish. 5.2. Effect of the ratio PH2O/PO2 Fig. 16 highlights that, increasing the ratio PH2O/PO2 , i.e. the water vapor content relatively to the oxygen content, reduces significantly the UTS and strain to failure of the C410 material in both atmospheric pressure and high-pressure environments. Thus, increasing the ratio PH2O/PO2 by a factor 9 in atmospheric pressure gas mixtures decreases the UTS and the strain to fail￾ure by a factor of 1.5 and 3, respectively. These variations are enhanced in high-pressure environments, in agreement with the effect of the total pressure previously described. Thus, UTS and strain to failure diminish by a factor 2.5 and 4 with the only increase of PH2O/PO2 by a factor 2. It is well known that the presence of water vapor enhances the oxidation kinetic of silicon Fig. 16. Effects of the environment on UTS (a) and strain to failure (b) of C410 composites exposed at 1200 ◦C for 600 h in various air/steam gas mixtures

86 L. Quemard et al Journal of the European Ceramic Sociery 27 (2007)377-388 carbide at high temperature 4-16(8). This effect is attributed to morphological analyses, the highest fiber volume consumption the higher solubility of water in the borosilicate glass relatively associated with the highest formed oxide volume are obtained in to oxygen by creation of Si-OH bonds. It is generally agreed the high steam-pressure environments. The theoretical approach that the Sic oxidation rate is proportional to PH, o with a power is not applied to the composite exposed at atmospheric pres- law exponent of 1 15 16 The oxidation of the boron components sure in the air/steam (50/50) gas mixture because its modulus is also increased in high steam-pressure environments leading to is higher than the as-received material one (not pre-damaged) an acceleration of the C410 matrix oxidation kinetic. The faster This could be due to a large amount of fibers strongly bonded generation of the liquid boria-containing glass facilitates the sili- together by the borosilicate glass. Finally, the measurement con carbide dissolving(10), thus the progression of the oxidation of the strain could be slightly under-estimated because of the front in the bulk composite. This results in (i) an acceleration un-homogeneous damage across the section(only one spring of the fibers oxidation/dissolving kinetic and (ii) of loaded clip on-gauge used) the amount of fibers strongly bonded together with rigid bridge of glass(or embedded in glass), which cause embrittlement and 6. Conclusions weakening of the composite(Figs. 5 and 16). The second effect of steam on silicon and boron -based ceramics is the increase of The study of SiC(o/[Si, C, B](m) specimens exposed for silica and boria volatilities leading to linear material recession periods up to 600 h at 1200C in slow-flowing air-steam gas recession rates k have the following dependence ia-19d linear mixtures, in an atmospheric pressure furnace and in a high- rates. It is shown that the linear oxide volatilization an pressure furnace, have shown several results. The material shows no significant degradation of its retained mechanical proper- kaP (12) ties(strength and fracture strain)after exposures at atmospheric pressure for up to 600 h in air-steam(90/10)and(80/20) gas mixtures. However, for a 600 h exposure in a 50% steam- where PH o is the water vapor partial pressure(MPa), PTot the total pressure(MPa)and u the velocity of the gases(ms) containing environment at atmospheric pressure, the material losses over 40% of its strength and 70% of its strain to fail- Despite the fact that the volatility of B2 O3 is very important ure. In an air-steam(80/20) gas mixture at 1 MPa, strength at high temperature and in steam environments 4-6.8, 12, 1,3 it is and fracture strain reductions higher than 65 and 80%, respec- not high enough to avoid the dissolving reaction of SiC(Fig. 13). tively, are observed after exposure for up to 600h. Steam- This can be attributed to(i) the presence of silica which reduces pressure has a key role in the mechanism of degradation at the volatility of 03 in the borosilicate glass and (i) the velocity 1200C and similar damages observed in high-pressure environ- of the gases which is highly decreased in the bulk composite, ments with low-steam-content can be obtained in atmospheric thus diminishing the volatilization of B2 03, in agreement with pressure environments with high-steam-content. Microstruc tural evidence of the damages includes (i) the generation of 5.3. Quantification of the corrosion damages a borosilicate glass sealing the bulk porosities, (ii) the multi layered matrix oxidation and (iii) the oxidation/dissolving of the fibers which are strongly bonded together and embedded A theoretical approach, based on the retained mechani- in the solid glass at RT. High steam-pressure environments cal properties (UTS and elastic modulus)obtained at RT, is enhance the matrix oxidation and the generation of a liquid developed to quantify the corrosion damages in the compos- borosilicate glass through the pre-damaging cracks and the bulk ite(Appendix A). The results, shown in Fig. 17, show that the porosities. B203, contained in the borosilicate(5-20% molar materials aged at low steam-pressure are less affected than the after 600 h exposures), may first react with SiC matrix layers to composites tested at high steam-pressure In agreement with the form silica before reaching the fibers and causes embrittlement and weakening of the composites. Moreover, under mechanical S, CBI 100% .Bull Porosity loading, microcracks can open or connect initially closed bulk porosities, thus enhancing the diffusion of oxygen and steam through the material. This phenomenon should be taken into account to predict the durability of SiC(n/Si, C, BI(m) compos The service conditions of an aeronautic engine including the combination of high temperatures and high steam-pressures, possess a high potential for severe degradation of the mechanical properties in combustor components made of SiC(/Si, C, B](m) composites. Nevertheless, the combustor components made of C410 material possess thicker seal-coat than the machined dog bone specimens used in this study, thus increasing their life Fig17. Quantification of the corrosion damages in the C410 material exposed durability. Environmental barrier coatings may be required at 1200C for 600h in various environments, based on a theoretical approach long-term use of SiC(o/[Si, C, B]m) materials in high tempera- and the post-exposure mechanical properties ture combustion environments

386 L. Quemard et al. / Journal of the European Ceramic Society 27 (2007) 377–388 carbide at high temperature14–16 (8). This effect is attributed to the higher solubility of water in the borosilicate glass relatively to oxygen by creation of Si OH bonds. It is generally agreed that the SiC oxidation rate is proportional to PH2O with a power law exponent of 1.15,16 The oxidation of the boron components is also increased in high steam-pressure environments leading to an acceleration of the C410 matrix oxidation kinetic. The faster generation of the liquid boria-containing glass facilitates the sili￾con carbide dissolving (10), thus the progression of the oxidation front in the bulk composite. This results in (i) an acceleration of the fibers oxidation/dissolving kinetic and (ii) an increase of the amount of fibers strongly bonded together with rigid bridges of glass (or embedded in glass), which cause embrittlement and weakening of the composite (Figs. 5 and 16). The second effect of steam on silicon and boron-based ceramics is the increase of silica and boria volatilities leading to linear material recession rates. It is shown that the linear oxide volatilization and linear recession rates kl have the following dependence17–19: klαPH2O v1/2 P1/2 Tot (12) where PH2O is the water vapor partial pressure (MPa), PTot the total pressure (MPa) and v the velocity of the gases (m s−1). Despite the fact that the volatility of B2O3 is very important at high temperature and in steam environments,4–6,8,12,1,3 it is not high enough to avoid the dissolving reaction of SiC (Fig. 13). This can be attributed to (i) the presence of silica which reduces the volatility of B2O3 in the borosilicate glass and (ii) the velocity of the gases which is highly decreased in the bulk composite, thus diminishing the volatilization of B2O3, in agreement with (12). 5.3. Quantification of the corrosion damages A theoretical approach, based on the retained mechani￾cal properties (UTS and elastic modulus) obtained at RT, is developed to quantify the corrosion damages in the compos￾ite (Appendix A). The results, shown in Fig. 17, show that the materials aged at low steam-pressure are less affected than the composites tested at high steam-pressure. In agreement with the Fig. 17. Quantification of the corrosion damages in the C410 material exposed at 1200 ◦C for 600 h in various environments, based on a theoretical approach and the post-exposure mechanical properties. morphological analyses, the highest fiber volume consumption associated with the highest formed oxide volume are obtained in the high steam-pressure environments. The theoretical approach is not applied to the composite exposed at atmospheric pres￾sure in the air/steam (50/50) gas mixture because its modulus is higher than the as-received material one (not pre-damaged). This could be due to a large amount of fibers strongly bonded together by the borosilicate glass. Finally, the measurement of the strain could be slightly under-estimated because of the un-homogeneous damage across the section (only one spring￾loaded clip on-gauge used). 6. Conclusions The study of SiC(f)/[Si, C, B](m) specimens exposed for periods up to 600 h at 1200 ◦C in slow-flowing air-steam gas mixtures, in an atmospheric pressure furnace and in a high￾pressure furnace, have shown several results. The material shows no significant degradation of its retained mechanical proper￾ties (strength and fracture strain) after exposures at atmospheric pressure for up to 600 h in air-steam (90/10) and (80/20) gas mixtures. However, for a 600 h exposure in a 50% steam￾containing environment at atmospheric pressure, the material losses over 40% of its strength and 70% of its strain to fail￾ure. In an air-steam (80/20) gas mixture at 1 MPa, strength and fracture strain reductions higher than 65 and 80%, respec￾tively, are observed after exposure for up to 600 h. Steam￾pressure has a key role in the mechanism of degradation at 1200 ◦C and similar damages observed in high-pressure environ￾ments with low-steam-content can be obtained in atmospheric pressure environments with high-steam-content. Microstruc￾tural evidence of the damages includes (i) the generation of a borosilicate glass sealing the bulk porosities, (ii) the multi￾layered matrix oxidation and (iii) the oxidation/dissolving of the fibers which are strongly bonded together and embedded in the solid glass at RT. High steam-pressure environments enhance the matrix oxidation and the generation of a liquid borosilicate glass through the pre-damaging cracks and the bulk porosities. B2O3, contained in the borosilicate (5–20% molar after 600 h exposures), may first react with SiC matrix layers to form silica before reaching the fibers and causes embrittlement and weakening of the composites. Moreover, under mechanical loading, microcracks can open or connect initially closed bulk porosities, thus enhancing the diffusion of oxygen and steam through the material. This phenomenon should be taken into account to predict the durability of SiC(f)/[Si, C, B](m) compos￾ites. The service conditions of an aeronautic engine including the combination of high temperatures and high steam-pressures, possess a high potential for severe degradation of the mechanical properties in combustor components made of SiC(f)/[Si, C, B](m) composites. Nevertheless, the combustor components made of C410 material possess thicker seal-coat than the machined dog￾bone specimens used in this study, thus increasing their life durability. Environmental barrier coatings may be required for long-term use of SiC(f)/[Si, C, B](m) materials in high tempera￾ture combustion environments