《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_j.1151-2916.2003.tb03463.x[1]

High-Temperature Water Vapor Effects ournal ⊥ m Cera. Sot,868]1272-8102003) High-Temperature Stability of Sic-Based Composites i High-Water-Vapor-Pressure Environments Karren L. More, Peter F. Tortorelli,*and Larry R.Walker Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6064 Naren Miriyala, Jeffrey R Price, and Mark van Solar Turbines, Inc., San Diego, California Microstructural characterization of boron-containing si fibers)and fibers having compositional improvements(such as reinforced Sic composites exposed at high temperature in Hi-NicalonTM) have also been characterized ,y In general, the high-water-vapor-pressure environments was used to deter- newer SiC fibers tend to be closer to stoichiometric SiC compo mine surface recession rates and to understand the controlling sition and have fewer impurities. Chemical-vapor-deposited degradation processes under these conditions. Results showed(CVD) boron nitride is currently the preferred fiber-matrix inter- that composite degradation was controlled by a series of facial component used by CFCC manufacturers since BN has eactions involving the formation of silica, boria, borosilicate improved oxidation resistance compared with carbon. At room glass, and gaseous products. Comparison of results (from temperature, BN fiber coatings allow for the fiber-matrix debond characterization of composites exposed at 1200.C and 1.5 atm ing characteristic of tough composites, but at elevated tempera- of H,o in a laboratory furnace and in the combustion zone of tures in water-containing environments, BN undergoes rapid a gas turbine) showed that these reactions were common to egradation that can negatively impact the long-term stability and both exposure conditions and, consequently, there was little usefulness of the composite. Although the thin bn fiber coating effect of gas velocity on degradation rates of boron-containing (0.4 Hm)represent a low percentage constituent of the total SiC/SiC composite materials. material of the bulk composite(90 vol% of the composites, accelerated high-temperature applications. Improvements in the ceramic oxidation of these materials in high-H-O environments was ex fibers, interfacial coatings, and matrix processing have resulted in ted to be similar to that observed for monolithic sic hanced resistance to oxidation at temperatures up to at least liners show that this is not the case 22 To further the use of these testing of these materials in gas turbine engine hot-section com- materials in high-temperature applications, the issue of under onents, there is relatively little understanding of the oxidation tanding composite degradation has become an extremely impor mechanisms for SiC-based CFCCs in high-pressure steam envi tant one. Several different composite processing routes have been nts typically experience used to produce SiC/BN/SiC CFCC materials for hot-section To develop a complete understanding of the oxidation behavior components in engines, thereby resulting in different matrix of Sic-based CFCCs in water-containing environments, the con- compositions and microstructures. As a result, it has become tributions of different bulk composite microstructures and individ mperative to fully evaluate the degradation mechanisms associ- ual composite constituents(fibers, fiber-matrix interface coatings, ated with the use of various Sic/BN/SiC CFCC materials in nd matrix) to overall CFCC stability must be understood. Silicon high-water-vapor-containing environments and to develop an un- carbide fibers typically comprise 20-40 vol% of the Sic-based derstanding of the primary issues associated with the long-term CFCCs. The microstructural stability and oxidation behavior of stability of several different commercially available composites of ceramic-grade(CG)Nicalon TM (Si-C-O)fibers have been evalu- this type. To this end, this paper describes recent work on the ated.5,As a result of poor strength retention of CG NicalonTMat characterization of several commercially available SiC/BN/SIC temperatures >1000oC, SiC fibers having improved thermal sta- opposites exposed to high water-va ressures in a h emperature bility via cost-effective processing methods(such as Tyranno SiC pressure laboratory-scale exposure facility. In ular. oxidation mechanisms for the various fCCS were aluated based on the composite processing method used and the arting composition Of primary importance was as the contributio E.J. Opila-contributing editor of bn and B-containing compounds to the overall rate of com- osite degradation in these high-H2O oxidizing environments Results obtained from laboratory exposures are compared with similarly processed materials exposed in long-term engine tests at the Symposi Materials at the131 Annual meet吗号 of The Minerals, Il. Experimental Procedure Secretary rsr ehe gy efciency da d rehe wable bnergy onf ct of ndus ras iech. Two different types of Sic/BN/SiC composites were evaluated Resources progra contract D0OR22725 with UT-Battel in this study. The first was a composite with continuous Hi- Nicalon M SiC fiber reinforcement, CVD BN interfacial coati 1272

High-Temperature Stability of SiC-Based Composites in High-Water-Vapor-Pressure Environments Karren L. More,* Peter F. Tortorelli,* and Larry R. Walker Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6064 Naren Miriyala, Jeffrey R. Price,* and Mark van Roode Solar Turbines, Inc., San Diego, California 92186 Microstructural characterization of boron-containing SiC￾reinforced SiC composites exposed at high temperature in high-water-vapor-pressure environments was used to deter￾mine surface recession rates and to understand the controlling degradation processes under these conditions. Results showed that composite degradation was controlled by a series of reactions involving the formation of silica, boria, borosilicate glass, and gaseous products. Comparison of results (from characterization of composites exposed at 1200°C and 1.5 atm of H2O in a laboratory furnace and in the combustion zone of a gas turbine) showed that these reactions were common to both exposure conditions and, consequently, there was little effect of gas velocity on degradation rates of boron-containing SiC/SiC composite materials. I. Introduction RECENT advances in the development and manufacture of SiC-based continuous-fiber-reinforced ceramic-matrix com￾posites (CFCCs) have led to the use of these materials in several high-temperature applications.1–4 Improvements in the ceramic fibers, interfacial coatings, and matrix processing have resulted in composites with relatively good mechanical properties and en￾hanced resistance to oxidation at temperatures up to at least 1200°C. However, even with the recent implementation and testing of these materials in gas turbine engine hot-section com￾ponents,1–4 there is relatively little understanding of the oxidation mechanisms for SiC-based CFCCs in high-pressure steam envi￾ronments typically experienced in a turbine combustor. To develop a complete understanding of the oxidation behavior of SiC-based CFCCs in water-containing environments, the con￾tributions of different bulk composite microstructures and individ￾ual composite constituents (fibers, fiber-matrix interface coatings, and matrix) to overall CFCC stability must be understood. Silicon carbide fibers typically comprise 20–40 vol% of the SiC-based CFCCs. The microstructural stability and oxidation behavior of ceramic-grade (CG) Nicalon™ (Si-C-O) fibers have been evalu￾ated.5,6 As a result of poor strength retention of CG Nicalon™ at temperatures
1000°C, SiC fibers having improved thermal sta￾bility via cost-effective processing methods (such as Tyranno SiC fibers)7 and fibers having compositional improvements (such as Hi-Nicalon™) have also been characterized.8,9 In general, the newer SiC fibers tend to be closer to stoichiometric SiC compo￾sition and have fewer impurities. Chemical-vapor-deposited (CVD) boron nitride is currently the preferred fiber-matrix inter￾facial component used by CFCC manufacturers since BN has improved oxidation resistance compared with carbon.10,11 At room temperature, BN fiber coatings allow for the fiber-matrix debond￾ing characteristic of tough composites, but at elevated tempera￾tures in water-containing environments, BN undergoes rapid degradation that can negatively impact the long-term stability and usefulness of the composite.12 Although the thin BN fiber coatings (0.4 m) represent a low percentage constituent of the total material of the bulk composite (5 vol%), these coatings can have a significant effect on composite oxidation behavior.13 The deleterious effect of water vapor on the oxidation rate and SiO2 growth of SiC ceramics has been well documented.14–17 More recently, work has been conducted at higher water-vapor pressures (
1 atm) more typical of those in stationary gas turbines.18–21 Based on these results and the fact that SiC (fibers and matrix) comprises
90 vol% of the composites, accelerated oxidation of these materials in high-H2O environments was ex￾pected to be similar to that observed for monolithic SiC.20,21 However, observations after field-testing of CFCC combustor liners show that this is not the case.22 To further the use of these materials in high-temperature applications, the issue of under￾standing composite degradation has become an extremely impor￾tant one. Several different composite processing routes have been used to produce SiC/BN/SiC CFCC materials for hot-section components in engines, thereby resulting in different matrix compositions and microstructures. As a result, it has become imperative to fully evaluate the degradation mechanisms associ￾ated with the use of various SiC/BN/SiC CFCC materials in high-water-vapor-containing environments and to develop an un￾derstanding of the primary issues associated with the long-term stability of several different commercially available composites of this type. To this end, this paper describes recent work on the characterization of several commercially available SiC/BN/SiC composites exposed to high water-vapor pressures in a high￾temperature, high-pressure laboratory-scale exposure facility. In particular, oxidation mechanisms for the various CFCCs were evaluated based on the composite processing method used and the starting composition. Of primary importance was the contribution of BN and B-containing compounds to the overall rate of com￾posite degradation in these high-H2O oxidizing environments. Results obtained from laboratory exposures are compared with similarly processed materials exposed in long-term engine tests. II. Experimental Procedure Two different types of SiC/BN/SiC composites were evaluated in this study. The first was a composite with continuous Hi￾Nicalon™ SiC fiber reinforcement, CVD BN interfacial coatings, E. J. Opila—contributing editor Manuscript No. 186662. Received September 13, 2002; approved April 2, 2003. Presented at the Symposium on Water Vapor Effects on Oxidation of High￾Temperature Materials at the 131st Annual Meeting of The Minerals, Metals & Materials Society (TMS), Seattle, WA, February 18–20, 2002. This research was sponsored by the U.S. Department of Energy, Assistant Secretary for Energy Efficiency and Renewable Energy, Office of Industrial Tech￾nologies, as part of the Distributed Energy Resources Program under contract DE-AC05-00OR22725 with UT-Battelle LLC. *Member, American Ceramic Society. High-Temperature Water Vapor Effects J. Am. Ceram. Soc., 86 [8] 1272–81 (2003) 1272 journal

High-Temperature Stability of sic-Based Composites in High-Water-Vapor-Pressure Emvironmer 273 trix processed using the isothermal chemical vapo II. Results ation(CVi) technique at GE Power Systems Composites, (Newark, DE).The second type of SiC/BN/SIC composite The two SiC/SiC composites compared in this study had very material also contained Hi-Nicalon TM fibers and CVd BN inter- different starting microstructures. The CVI composites were char facial coatings, but had a SiC matrix fabricated using both a CVI acterized by a two-dimensional woven fiber structure with a(os processing step followed by matrix densification via a silicon fiber lay-up, as shown in the relatively low magnification BSI melt-infiltration(Mi) process similar to that described in Ref, 24 image in Fig. 1(a). In addition to the cvd bn interfacial coating The MI composites were manufactured by either Goodrich Cor sed for the laboratory-exposed coupons, a B-containing particu (Santa Fe Springs, CA) or GE Power Systems Composites, LLC filler was used within and between the fiber tows to minimize the siC matrix infiltration time The fiber-coating structure within each type of CFCC were exposed in a high-temperature, high- the fiber tows for the CVI SiC/BN/SiC composite is shown in Fig pressure exposure facility (referred to in text as the ORnl 1(b). The BN-coated fibrous preforms were densified with Sic by furnace), which has been described in detail elsewhere. 25 No the CVI process. The resulting CVI composite was%95% protective SiC seal coatings were applied to any surfaces of the omposite coupons exposed in the laboratory furnace To simulate A similar fiber lay-up was used for the MI SiC/BN/SIC the typical water-vapor pressure and maximum temperature of a composite, as shown in Fig. 2(a). The MI SiC/BN/SiC used a olar Turbines Centaur 50s engine combustor environment(use uniformly thin, continuous CVD BN coating, -0.4 um thick for long-term engine tests, as described below ), all laboratory round the fibers. To protect the BN-coated fibrous preform during exposures were conducted at 1200C in a slowly flowing(3 MI processing, a-1-2 um CVI SiC layer was applied around the cm/min) gas mixture of air+ 15%H,O at 10 atm. Each exposure fiber tows(see Fig. 2(b). This minimal CVI treatment left a fairly period was 500 h, after which the specimens were carefully pen structure for subsequent molten silicon MI processing while moved and selectively cut for microstructural analysis. A ill protecting the fibers and fiber coatings. After processing, the 20.3-0.4 cm thick cross section was cut from the end of each MI matrix was composed primarily of Si SiC and the composite coupon.The cut section was mounted (with cut face down)in was near full density Composite density differences between the epoxy and polished. After cutting, the same composite coupons two composites are evident when the bulk microstructure of the were then placed back in the furnace for additional exposures for times up to 3000 h in 500 h increments and another cross section was cut from each coupon after each additional 500 h e ure this way, a total of six cross sections (cut after 500, 1000, 1500 2000, 2500, and 3000 h) were cut from each exposed CVI and mI composite coupon for microstructural evaluation In addition to the laboratory furnace exposures, engine tests -eric Fiber tow vere conducted using Sic/BN/SiC CFCC combustor liners ma ufactured using both CVI-and Ml-processed composites in a Solar Turbines(San Diego, CA) SoLONOX Centaur 50S engine located at the Chevron engine test site in Bakersfield. CA. A full-scale set of cfCc combustor liners consisted of a 33 cm diameter inner liner(either a SiC/BN/SiC MI liner produced by Goodrich Corp Santa Fe Springs, CA, or a CVI liner produced by GE Power Systems Composites, LLC, Newark, DE)and a 76 cm diameter SiC/C/SiC CVI outer liner(manufactured by GE Power Systems Composites, LLC, Newark, DE). This paper includes results for the engine exposure of two inner liners from two separate engine tests the mi inner liner ran in an engine for 2758 h and the cvi inner combustor liner was exposed for 2266 h. The gas surface for each liner was coated with a protective CVD SiC seal coat(-200 and 215 um for the MI and CVI liners, respectively The engine-exposed CFCC combustor liners were exposed to a maximum temperature of "1200.C (in the hot spots associated with fuel impingement areas) and a gas velocity of 30 m/s Microstructural examinations were conducted on polished cross sections prepared after each of the six exposures (up to 3000 h in SiC(matrix) 500 h increments) conducted in the ornl furnace to measure BN fiber oxide product thickness and surface recession, and to asses coating ubsurface oxidation-induced damage to the different CCCs (specimen mass and dimensional changes were monitored but were not reliable quantitative indicators of composite degradation due to friability of oxide products and the nature of some ,6,c constituent interactions. ) Composite surface recession was deter lined microstructurally by measuring the thickness of composite Sic (fiber) material below the surface that was not affected by oxidation good material) and subtracting this amount from the as-processed composite coupon thickness. Results from these examinations were com 20 um t sed combustor liners. Aft-to-fore liner cross sections were en from both heavily damaged areas and areas that appeared damaged from the inner and outer liners. Electron probe microanalysis(EPMA) and backscatte were the analysis techni sections before and after exposure in the d to examin ine CFCC cross Fig.1.micr ure of CVI SiC/BN/SiC: (a) Bse image of bulk mace or Sol microstructure showing fiber lay-up and CVI SiC matrix and(b) Turbines Centaur 50S engine. These examinations were conducte magnification image showing the Bn fiber coating and boron-containing using a JEOL 733 Superprobe (JEOL USA, Inc, Peabody, MA) filler within fiber tows

and a SiC matrix processed using the isothermal chemical vapor infiltration (CVI) technique at GE Power Systems Composites, LLC (Newark, DE).23 The second type of SiC/BN/SiC composite material also contained Hi-Nicalon™ fibers and CVD BN inter￾facial coatings, but had a SiC matrix fabricated using both a CVI processing step followed by matrix densification via a silicon melt-infiltration (MI) process similar to that described in Ref. 24. The MI composites were manufactured by either Goodrich Corp. (Santa Fe Springs, CA) or GE Power Systems Composites, LLC. As-processed coupons (typically 2.5 cm  5.0 cm  0.3 cm) of each type of CFCC were exposed in a high-temperature, high￾pressure exposure facility (referred to in text as the ORNL furnace), which has been described in detail elsewhere.25 No protective SiC seal coatings were applied to any surfaces of the composite coupons exposed in the laboratory furnace. To simulate the typical water-vapor pressure and maximum temperature of a Solar Turbines Centaur 50S engine combustor environment (used for long-term engine tests, as described below), all laboratory exposures were conducted at 1200°C in a slowly flowing (3 cm/min) gas mixture of air  15% H2O at 10 atm. Each exposure period was 500 h, after which the specimens were carefully removed and selectively cut for microstructural analysis. A 0.30.4 cm thick cross section was cut from the end of each coupon. The cut section was mounted (with cut face down) in epoxy and polished. After cutting, the same composite coupons were then placed back in the furnace for additional exposures for times up to 3000 h in 500 h increments and another cross section was cut from each coupon after each additional 500 h exposure. In this way, a total of six cross sections (cut after 500, 1000, 1500, 2000, 2500, and 3000 h) were cut from each exposed CVI and MI composite coupon for microstructural evaluation. In addition to the laboratory furnace exposures, engine tests were conducted using SiC/BN/SiC CFCC combustor liners man￾ufactured using both CVI- and MI-processed composites in a Solar Turbines (San Diego, CA) SoLoNOx Centaur 50S engine located at the Chevron engine test site in Bakersfield, CA. A full-scale set of CFCC combustor liners consisted of a 33 cm diameter inner liner (either a SiC/BN/SiC MI liner produced by Goodrich Corp., Santa Fe Springs, CA, or a CVI liner produced by GE Power Systems Composites, LLC, Newark, DE) and a 76 cm diameter SiC/C/SiC CVI outer liner (manufactured by GE Power Systems Composites, LLC, Newark, DE). This paper includes results for the engine exposure of two inner liners from two separate engine tests; the MI inner liner ran in an engine for 2758 h and the CVI inner combustor liner was exposed for 2266 h.1 The gas-path surface for each liner was coated with a protective CVD SiC seal coat (200 and 215 m for the MI and CVI liners, respectively). The engine-exposed CFCC combustor liners were exposed to a maximum temperature of 1200°C (in the hot spots associated with fuel impingement areas) and a gas velocity of 30 m/s. Microstructural examinations were conducted on polished cross sections prepared after each of the six exposures (up to 3000 h in 500 h increments) conducted in the ORNL furnace to measure oxide product thickness and surface recession, and to assess subsurface oxidation-induced damage to the different CFCCs (specimen mass and dimensional changes were monitored but were not reliable quantitative indicators of composite degradation due to friability of oxide products and the nature of some constituent interactions.) Composite surface recession was deter￾mined microstructurally by measuring the thickness of composite material below the surface that was not affected by oxidation (good material) and subtracting this amount from the as-processed composite coupon thickness. Results from these examinations were compared with similar analyses performed on the engine￾exposed combustor liners. Aft-to-fore liner cross sections were taken from both heavily damaged areas and areas that appeared undamaged from the inner and outer liners. Electron probe microanalysis (EPMA) and backscatter electron (BSE) imaging were the primary analysis techniques used to examine CFCC cross sections before and after exposure in the ORNL furnace or Solar Turbines Centaur 50S engine. These examinations were conducted using a JEOL 733 Superprobe (JEOL USA, Inc., Peabody, MA). III. Results The two SiC/SiC composites compared in this study had very different starting microstructures. The CVI composites were char￾acterized by a two-dimensional woven fiber structure with a 0°/45° fiber lay-up, as shown in the relatively low magnification BSE image in Fig. 1(a). In addition to the CVD BN interfacial coating used for the laboratory-exposed coupons, a B-containing particu￾late filler was used within and between the fiber tows to minimize the SiC matrix infiltration time. The fiber-coating structure within the fiber tows for the CVI SiC/BN/SiC composite is shown in Fig. 1(b). The BN-coated fibrous preforms were densified with SiC by the CVI process. The resulting CVI composite was 85%–95% dense. A similar fiber lay-up was used for the MI SiC/BN/SiC composite, as shown in Fig. 2(a). The MI SiC/BN/SiC used a uniformly thin, continuous CVD BN coating, 0.4 m thick, around the fibers. To protect the BN-coated fibrous preform during MI processing, a 1–2 m CVI SiC layer was applied around the fiber tows (see Fig. 2(b)). This minimal CVI treatment left a fairly open structure for subsequent molten silicon MI processing while still protecting the fibers and fiber coatings. After processing, the MI matrix was composed primarily of Si  SiC and the composite was near full density. Composite density differences between the two composites are evident when the bulk microstructure of the Fig. 1. Microstructure of CVI SiC/BN/SiC: (a) BSE image of bulk microstructure showing fiber lay-up and CVI SiC matrix and (b) higher￾magnification image showing the BN fiber coating and boron-containing filler within fiber tows. August 2003 High-Temperature Stability of SiC-Based Composites in High-Water-Vapor-Pressure Environments 1273

1274 Journal of the American Ceramic Society-More et al Vol 86. No. 8 these constituents oxidizes at a different rate and different reaction products having very different product morphologies are formed An example of such is shown in Fig. 5. Once the CVI SiC layer around the fibers is breached. oxidation of the fiber tows is faster than that of the other constituents and occurs primarily via the Fiber tow dation along/through the bn interfaces. Figure 6 shows that the BN acts as a conduit for rapid transport of oxidant through the fiber tow and clearly oxidizes faster than the surrounding CVI SiC and Sic fibers. Within the tows, the fibers are completely consumed by the oxidized bn(reactions between boria, SiC, and silica.)The preferred BN oxidation path around individual fibers within a tow is shown clearly in Fig. 6. A fully consumed fiber tow serves as the path for the progression of the oxidation front further into the compos as shown in Fig. 7. Areas that are primarily Si SiC matrix will oxidize fairly rapidly but not as rapidly as th fiber tows since the tows contain the thin, but continuous, BN 100m A photograph of the SiC/BN/SiC MI inner liner after the 2758 h engine test is shown in Fig. 8. The "white damaged areas observed on the gas-path surface of the inner liner are regions of greater damage accumulation that corresponded directly with fuel impingement areas on the liner surface. These areas also corre- sponded with higher-temperature regions where the liner surface was subjected to temperatures 21200C Surface recession of the MI liner, measured microstructurally from several aft-to-fore cross sections, is summarized in Fig. 9 for three different areas visually BN fiber showing different degrees of surface damage. The three areas used coating to generate data for this plot are circled in Fig. 8. The line at "3000 um on the graph in Fig 9 represents the CVD SiC seal coat/MI composite interface. While the majority of the inner liner urface did not exhibit the extensive damage associated with the white regions(see Fig 8), the white areas on the liner surface were SiC fiber clearly characteristic of regions showing break-through in the thick CVD SIC seal coat due to accelerated sic recession in the combustor, albeit at a somewhat lower rate than that given by siC/S Robinson and Smialek for recession of cVd Sic (silica volatil- matrix 10 um ization by Si(OH) formation) in a high-pressure burner rig at 1200.C.The CVD SiC seal coat recession rate for combustor liners engine tested in a Solar Turbines Centaur 50S engine as 0 11 um/h at 1200C, which was determined directly from microstructural measurements of the present as well as previous Fig.2. Microstructure of MI SiC/BN/SiC:(a)BSE image of bulk engine-tested C VD SiC seal-coated liners. In the areas surround magnification SEM image of 0.4 Hm BN coatings around Hi-Nicalon M Sic seal coat was still intact but did show varying de fibers and thin Cvd SiC laver surrounding fiber tow urface recession. The underlying MI composite material in these less damaged regions did not exhibit any microstructural damage Within the white areas on the inner liner surface. damage to the VI and MI SiC/BN/SiC composites are compared; large, inter- underlying MI composite was evident. In less damaged areas connected matrix porosity is characteristic of the CVI composites, occasional break-through of the sic seal coat resulted in localized whereas MI composites are much denser ubsurface composite damage. This is shown in Fig. 10. Greater MI composite damage depths were observed when the CVD SiC seal coat was completely removed by volatilization to fully expose (I MI SiC/BNAiC large regions of the CFCC substrate. However, the MI composite The Mi composite coupons were exposed in the ornl urface damage was limited to the first or second fiber layer in the to 1.5 atm of H,o at 1200.C for a total of 3000 h during th white areas, even in the highly damaged regions. The typical MI icrostructural damage was extensive. Figure 3 shows a composite damage observed in large areas of Sic seal coat sion of images of the bulk MI composite recession and concurrent break-through is shown in Fig. ll. From the data presented in Fig silica scale formation as a function of exposure time. The lines 9, the maximum depth of microstructural damage observed on the drawn approximately parallel to the original coupon faces repre- inner liner was -300 um below the seal coat, which corresponded sent the total amount of surface recession(composite degradation with the depth of the second layer of fibers from the working that is, the depth to which the structure has been altered due to urface. These areas were associated with oxidation of many of the oxidation and associated constituent and product reactions(see on te constituents, including the si Sic matrix, the Cvd Sic around the fiber tows. and the bn coating and Sic fibers The surface of the mi sic/BN/sic before oxidation wa within the fiber tows. A SiO,+ borosilicate glass scale developed composed primarily of the Si SiC MI matrix, whether thinly on the surface(Fig. I1) covering a fiber tow(CVI SiC) or as large-scale matrix areas During the steady-state recession of pure CVD SiC in the between tows, as shown in Fig 4. The composite can be consid- high-gas-velocity, high-H2O-pressure environment typical of a ered as a complex layered structure of the different constituents turbine engine combustor 1O, volatilizes at the same rat Ite see Fig 4) which are oxidized as the oxidation front progresses SiC oxidizes, and thus, thick surface scale does not form these through the composi te structure the mi matrix(Si SiC results are consistent with observations made on the remaining sic normally oxidizes first, followed by the CvI SiC layer around the eal coat on the gas-path surface of the inner liner. A thick, porous fiber tows, which is subsequently followed by oxidation within the scale was not observed; however, a thin vitreous SiO, layer was ber tows(composed of Bn coatings around SiC fibers ). Each of present on the oxidized SiC surface 20, 21 In the case of the MI

CVI and MI SiC/BN/SiC composites are compared; large, inter￾connected matrix porosity is characteristic of the CVI composites, whereas MI composites are much denser. (1) MI SiC/BN/SiC The MI composite coupons were exposed in the ORNL furnace to 1.5 atm of H2O at 1200°C for a total of 3000 h. During this time, microstructural damage was extensive. Figure 3 shows a succes￾sion of images of the bulk MI composite recession and concurrent silica scale formation as a function of exposure time. The lines drawn approximately parallel to the original coupon faces repre￾sent the total amount of surface recession (composite degradation), that is, the depth to which the structure has been altered due to oxidation and associated constituent and product reactions (see below). The surface of the MI SiC/BN/SiC before oxidation was composed primarily of the Si  SiC MI matrix, whether thinly covering a fiber tow (CVI SiC) or as large-scale matrix areas between tows, as shown in Fig. 4. The composite can be consid￾ered as a complex layered structure of the different constituents (see Fig. 4) which are oxidized as the oxidation front progresses through the composite structure. Thus, the MI matrix (Si  SiC) normally oxidizes first, followed by the CVI SiC layer around the fiber tows, which is subsequently followed by oxidation within the fiber tows (composed of BN coatings around SiC fibers). Each of these constituents oxidizes at a different rate, and different reaction products having very different product morphologies are formed. An example of such is shown in Fig. 5. Once the CVI SiC layer around the fibers is breached, oxidation of the fiber tows is faster than that of the other constituents and occurs primarily via the oxidation along/through the BN interfaces. Figure 6 shows that the BN acts as a conduit for rapid transport of oxidant through the fiber tow and clearly oxidizes faster than the surrounding CVI SiC and SiC fibers. Within the tows, the fibers are completely consumed by the oxidized BN (reactions between boria, SiC, and silica.) The preferred BN oxidation path around individual fibers within a tow is shown clearly in Fig. 6. A fully consumed fiber tow serves as the path for the progression of the oxidation front further into the composite, as shown in Fig. 7. Areas that are primarily Si  SiC matrix will oxidize fairly rapidly,20 but not as rapidly as the fiber tows since the tows contain the thin, but continuous, BN.12 A photograph of the SiC/BN/SiC MI inner liner after the 2758 h engine test is shown in Fig. 8. The “white” damaged areas observed on the gas-path surface of the inner liner are regions of greater damage accumulation that corresponded directly with fuel impingement areas on the liner surface. These areas also corre￾sponded with higher-temperature regions where the liner surface was subjected to temperatures
1200°C. Surface recession of the MI liner, measured microstructurally from several aft-to-fore cross sections, is summarized in Fig. 9 for three different areas visually showing different degrees of surface damage. The three areas used to generate data for this plot are circled in Fig. 8. The line at 3000 m on the graph in Fig. 9 represents the CVD SiC seal coat/MI composite interface. While the majority of the inner liner surface did not exhibit the extensive damage associated with the white regions (see Fig. 8), the white areas on the liner surface were clearly characteristic of regions showing break-through in the thick CVD SiC seal coat due to accelerated SiC recession in the combustor, albeit at a somewhat lower rate than that given by Robinson and Smialek for recession of CVD SiC (silica volatil￾ization by Si(OH)4 formation) in a high-pressure burner rig at 1200°C.18 The CVD SiC seal coat recession rate for combustor liners engine tested in a Solar Turbines Centaur 50S engine was 0.11 m/h at 1200°C, which was determined directly from microstructural measurements of the present as well as previously engine-tested CVD SiC seal-coated liners.26 In the areas surround￾ing the white areas on the inner liner gas-path surface, the CVD SiC seal coat was still intact but did show varying degrees of surface recession. The underlying MI composite material in these less damaged regions did not exhibit any microstructural damage. Within the white areas on the inner liner surface, damage to the underlying MI composite was evident. In less damaged areas, occasional break-through of the SiC seal coat resulted in localized subsurface composite damage. This is shown in Fig. 10. Greater MI composite damage depths were observed when the CVD SiC seal coat was completely removed by volatilization to fully expose large regions of the CFCC substrate. However, the MI composite surface damage was limited to the first or second fiber layer in the white areas, even in the highly damaged regions. The typical MI composite damage observed in large areas of SiC seal coat break-through is shown in Fig. 11. From the data presented in Fig. 9, the maximum depth of microstructural damage observed on the inner liner was 300 m below the seal coat, which corresponded with the depth of the second layer of fibers from the working surface. These areas were associated with oxidation of many of the composite constituents, including the Si  SiC matrix, the CVD SiC around the fiber tows, and the BN coating and SiC fibers within the fiber tows. A SiO2  borosilicate glass scale developed on the surface (Fig. 11). During the steady-state recession of pure CVD SiC in the high-gas-velocity, high-H2O-pressure environment typical of a turbine engine combustor,16,18 SiO2 volatilizes at the same rate as SiC oxidizes, and thus, a thick surface scale does not form. These results are consistent with observations made on the remaining SiC seal coat on the gas-path surface of the inner liner. A thick, porous scale was not observed; however, a thin vitreous SiO2 layer was present on the oxidized SiC surface.20,21 In the case of the MI Fig. 2. Microstructure of MI SiC/BN/SiC: (a) BSE image of bulk microstructure showing fiber lay-up and dense MI matrix and (b) higher￾magnification SEM image of 0.4 m BN coatings around Hi-Nicalon™ fibers and thin CVD SiC layer surrounding fiber tows. 1274 Journal of the American Ceramic Society—More et al. Vol. 86, No. 8

August 2003 High-Temperature Stability of Sic-Based Composites in High-Water-Vapor-Pressure Emironments 275 As processed 1500h 2500h lica scale mm Fig 3. Cross-section BSE images of MI SiC/BN/SiC after exposure in a high-pressure furnace at 1200C and 1.5 atm of H2O for 0, 1500, and 2500 h showing damage to composite and Sio, scale development mposite material with bn as a constituent within the fiber tows, (2) CV SiC/BNAiC the borosilicate surface product (primarily a low-B-containing The CVI SiC/BNSiC composite was exposed in the ORNL orosilicate glass) was relatively stable in the combustor environ- furnace for a total of 1500 h at 1200 C and 1.5 atm of H,O before t volatility of boria based or uilibrium thermochemical considerations. 2 These observations extensive composite damage prevented reintroduction into the expo- sure facility. Figure 13 is an image of the coupon cross section after indicate that there is a low boria activity in the glass and/or the rate exposure for 1500 h at 1200C. Note the very thick, porous scale on of formation of the borosilicate glass was significantly greater than the volatilization rate the surface of the CVI SiC/BN/SiC. The as-oxidized morphology was quite different from that observed on the MI composite(compare Figure 12 is a higher-magnification image of the oxidation- Figs. 5 and 13). Fiber tows near the exposed surface were oxidized in associated with the various reactions are very simllar to phologies a manner similar to that observed for the MI SiC/BN/SiC, albeit more nduced microstructura bserved for the mi composite exposed at 1200.C to 1.5 atm of severely and rapidly, as shown in Fig. 14 (compare with Fig. 7)as a water vapor in the laboratory furnace(compare Figs result of substantial borosilicate glass formation in these areas The both cases. d was significantly accelerated because of the fiber tows oxidized more rapidly in the C vi composite compared with esence of the BN fiber coatings, which acted as preferred path the MI composite because of the presence of significant amounts of for the transport of oxidants well into the fiber tows B-containing phases(filler) within the fiber tows as well as greater amounts of BN. The amount of BN and B-containing phases present within the CVi composite fiber tows was >10 times that in the 0.4 um BN fiber coatings in the MI composite. In addition to the extensive surface oxidation, subsurface reaction within the large, interconnected porosity was observed in the CVI SiC/BN/SiC. During the first 500 h exposure, silica scales formed on the Cvi SiC matrix surrounding the subsurface poros- ity, as shown in Fig. 15(a). With additional exposure, the CVI SiC matrix around the subsurface porosity was breached. As shown Fig. 15(b), rapid degradation of the B-containing filler material and the cvd bn fiber coatings within the tows then occurred a manner similar to that described previously for the mi compos ite. However, because of the greater amount of boron-containing phases in the CVI composite, subsurface fiber tows were con sumed during this process and pools of glass were formed within these areas well below the exposed surface. Thus, for the CVI SIC/BN/SIC, oxidation was not simply limited to the exposed surface of the composite. Rapid CVI composite degradation was compounded by subsurface accumulation of amage due to greater amounts of boron(B-co filler and Fig4. Cross-section BSE image of as-processed MI SiC/BN/SiC show- bn fiber coatings) as well as a network connected ing complex layered structure of composite constituents

composite material with BN as a constituent within the fiber tows, the borosilicate surface product (primarily a low-B-containing borosilicate glass) was relatively stable in the combustor environ￾ment despite the expected significant volatility of boria based on equilibrium thermochemical considerations.12 These observations indicate that there is a low boria activity in the glass and/or the rate of formation of the borosilicate glass was significantly greater than the volatilization rate. Figure 12 is a higher-magnification image of the oxidation￾induced microstructural damage in the MI liner. The morphologies associated with the various reactions are very similar to what was observed for the MI composite exposed at 1200°C to 1.5 atm of water vapor in the laboratory furnace (compare Figs. 6 and 12). In both cases, damage was significantly accelerated because of the presence of the BN fiber coatings, which acted as preferred paths for the transport of oxidants well into the fiber tows. (2) CVI SiC/BN/SiC The CVI SiC/BN/SiC composite was exposed in the ORNL furnace for a total of 1500 h at 1200°C and 1.5 atm of H2O before extensive composite damage prevented reintroduction into the expo￾sure facility. Figure 13 is an image of the coupon cross section after exposure for 1500 h at 1200°C. Note the very thick, porous scale on the surface of the CVI SiC/BN/SiC. The as-oxidized morphology was quite different from that observed on the MI composite (compare Figs. 5 and 13). Fiber tows near the exposed surface were oxidized in a manner similar to that observed for the MI SiC/BN/SiC, albeit more severely and rapidly, as shown in Fig. 14 (compare with Fig. 7) as a result of substantial borosilicate glass formation in these areas. The fiber tows oxidized more rapidly in the CVI composite compared with the MI composite because of the presence of significant amounts of B-containing phases (filler) within the fiber tows as well as greater amounts of BN. The amount of BN and B-containing phases present within the CVI composite fiber tows was
10 times that in the 0.4 m BN fiber coatings in the MI composite. In addition to the extensive surface oxidation, subsurface reaction within the large, interconnected porosity was observed in the CVI SiC/BN/SiC. During the first 500 h exposure, silica scales formed on the CVI SiC matrix surrounding the subsurface poros￾ity, as shown in Fig. 15(a). With additional exposure, the CVI SiC matrix around the subsurface porosity was breached. As shown in Fig. 15(b), rapid degradation of the B-containing filler material and the CVD BN fiber coatings within the tows then occurred in a manner similar to that described previously for the MI compos￾ite. However, because of the greater amount of boron-containing phases in the CVI composite, subsurface fiber tows were con￾sumed during this process and pools of glass were formed within these areas well below the exposed surface. Thus, for the CVI SiC/BN/SiC, oxidation was not simply limited to the exposed surface of the composite. Rapid CVI composite degradation was compounded by subsurface accumulation of oxidation-induced damage due to greater amounts of boron (B-containing filler and BN fiber coatings) as well as a network of interconnected, subsurface porosity. Fig. 3. Cross-section BSE images of MI SiC/BN/SiC after exposure in a high-pressure furnace at 1200°C and 1.5 atm of H2O for 0, 1500, and 2500 h showing damage to composite and SiO2 scale development. Fig. 4. Cross-section BSE image of as-processed MI SiC/BN/SiC show￾ing complex layered structure of composite constituents. August 2003 High-Temperature Stability of SiC-Based Composites in High-Water-Vapor-Pressure Environments 1275

Journal of the American Ceramic Society-More et al Vol 86. No. 8 Scale due to matrix (Si+SiC)oxidation ale due to Si oxidation (SiC+BN) oxidation 100 Fig. 5. Cross-section BSE image of MI SiC/BN/SiC after exposure in high-pressure furnace for 500 h at 1200C and 1.5 atm of H,O showing different scale morphologies that form on the different composite constituents Areas of white surface damage were also observed on the cvi section and was associated with subsurface CVI composite dam- SiC/BN/SIC inner liner that was exposed for 2266 h in the gas age to a depth of -250 um urbine. Again, these regions were indicative of varying tempera- The nature of the CVI composite damage observed in areas tures across the liner's gas-path surface and a localized increase in where the seal coat had completely recessed ( via silica volatiliza- liner/composite damage associated with the fuel impingement tion) was similar to that observed for the MI SiC/BN/SiC. The the engine test, the temperature associated with the white surface the oxidation of CVD SiC surrounding subsurface pores in the CVI damage was assumed to be -1200oC based on temperature composite and(2)more extensive damage to the underlying CVi measurements made during the early stages of a previous engine composite in areas of Sic seal coat break-through. Figure 18 test. However, the temperature in some of the areas showing shows the subsurface oxidation within pores well below the maximum damage may be slightly higher than 1200.. A photo gas-path surface. The subsurface porosity within the bulk CVI aph of the gas-path surface of the CvI inner liner after engine composite(below gas-path surface) may be considered a slow testing is shown in Fig. 16(fuel impingement areas are designate gas-flow area. For this reason, oxidation of the SiC surfaces of the by the 12 small arrows). Many different areas of this liner were pores below the surface resulted in the formation and retention of mpared in cross section, but the primary areas used for compar a crystalline Sio,(cristobalite, as determined by transmission ison were from regions showing the maximum amount of white electron microscopy) scale(designated area I in Fig. 18)which urface damage(area I in Fig. 16) and a region showing minimal was structurally similar to the sio, formed during exposure of surface damage(area 3 in Fig. 16) where the CVD SiC seal coat CVD SiC in the ORNL furnace 2,2 Composite damage just below appeared to still be intact. The amount of surface recession areas of CVD SiC seal coat break-through(designated area 2 in easured microstructurally from these two areas is summarized in Fig. 18)was more extensive than in similar areas observed for the 16. Note that the black line at liner thickness =0 in the plot MI composite(see Fig 10). Considering that the CVI SiC/BN/SIC in Fig 17 corresponds to the CVD SiC seal coat/CVI SiC/BN/SiC inner liner had a slightly thicker as-processed CVD SiC seal coat interface. Area 3(minimum damage area) showed c100 (215 um for CVI VS -20 forM) and that the cⅥ recession of the SiC seal coat with no break-through, whereas area composite liner was exposed in an engine for 300 h less, the I exhibited seal coat break-through in the middle of the aft-to-fore oxidation behavior and total amount of damage accumulation observed for the CVI composite material were significantly worse B-Si-O-C B-Si-O ○ 100um

Areas of white surface damage were also observed on the CVI SiC/BN/SiC inner liner that was exposed for 2266 h in the gas turbine. Again, these regions were indicative of varying tempera￾tures across the liner’s gas-path surface and a localized increase in liner/composite damage associated with the fuel impingement areas. While no accurate temperature data were recorded during the engine test, the temperature associated with the white surface damage was assumed to be 1200°C based on temperature measurements made during the early stages of a previous engine test. However, the temperature in some of the areas showing maximum damage may be slightly higher than 1200°C. A photo￾graph of the gas-path surface of the CVI inner liner after engine testing is shown in Fig. 16 (fuel impingement areas are designated by the 12 small arrows). Many different areas of this liner were compared in cross section, but the primary areas used for compar￾ison were from regions showing the maximum amount of white surface damage (area 1 in Fig. 16) and a region showing minimal surface damage (area 3 in Fig. 16) where the CVD SiC seal coat appeared to still be intact. The amount of surface recession measured microstructurally from these two areas is summarized in Fig. 16. Note that the black line at liner thickness  0 in the plot in Fig. 17 corresponds to the CVD SiC seal coat/CVI SiC/BN/SiC interface. Area 3 (minimum damage area) showed 100 m recession of the SiC seal coat with no break-through, whereas area 1 exhibited seal coat break-through in the middle of the aft-to-fore section and was associated with subsurface CVI composite dam￾age to a depth of 250 m. The nature of the CVI composite damage observed in areas where the seal coat had completely recessed (via silica volatiliza￾tion) was similar to that observed for the MI SiC/BN/SiC. The major differences between the CVI and MI composites were (1) the oxidation of CVD SiC surrounding subsurface pores in the CVI composite and (2) more extensive damage to the underlying CVI composite in areas of SiC seal coat break-through. Figure 18 shows the subsurface oxidation within pores well below the gas-path surface. The subsurface porosity within the bulk CVI composite (below gas-path surface) may be considered a slow gas-flow area. For this reason, oxidation of the SiC surfaces of the pores below the surface resulted in the formation and retention of a crystalline SiO2 (cristobalite, as determined by transmission electron microscopy) scale (designated area 1 in Fig. 18) which was structurally similar to the SiO2 formed during exposure of CVD SiC in the ORNL furnace.20,21 Composite damage just below areas of CVD SiC seal coat break-through (designated area 2 in Fig. 18) was more extensive than in similar areas observed for the MI composite (see Fig. 10). Considering that the CVI SiC/BN/SiC inner liner had a slightly thicker as-processed CVD SiC seal coat (215 m for CVI vs 200 m for MI) and that the CVI composite liner was exposed in an engine for 300 h less, the oxidation behavior and total amount of damage accumulation observed for the CVI composite material were significantly worse Fig. 5. Cross-section BSE image of MI SiC/BN/SiC after exposure in high-pressure furnace for 500 h at 1200°C and 1.5 atm of H2O showing different scale morphologies that form on the different composite constituents. Fig. 6. BSE image of MI SiC/BN/SiC showing the accelerated oxidation along/through BN interfacial coatings during exposure in high-pressure furnace for 2000 h at 1200°C and 1.5 atm of H2O. Fig. 7. BSE cross-section image of the surface of MI SiC/BN/SiC composite showing several consumed fiber tows after exposure for 2000 h at 1200°C and 1.5 atm of H2O that oxidized and formed borosilicate glass. 1276 Journal of the American Ceramic Society—More et al. Vol. 86, No. 8

August 2003 High-Temperature Stability of Sic-Based Composites in High-Water-Vapor-Pressure Emironments 277 Fore GP32 IF Aft ↓ Injector Location Fig. 8. Photograph of the gas-path surface of an MI SiC/BN/SiC engine-tested inner combustor liner exposed in engine for 2758 h. The three areas circled were microstructurally evaluated using aft-to-fore cross sections. 3300 Borosilicate gl 3100 3000 Seal Coat 200 2700 Most damage 1100m 2600 Fig. 11. Area from engine-tested MI SIC/BN/SiC inner liner where CVD SiC seal coat had recessed completely over a fairly large area. Damage Fig 9. Surface recession measured from aft-to-fore cross sections circled in Fig 9. Areas of minimum and maximum damage are included in plot. remaining seal coat remaining sealcoat Borosilicate gla 每争 RAO Fig. 12. Higher-magnification BSE image of the surface damage formed Fig. 10. Area from engine-tested MI SiC/BN/SIC liner where CVD on MI SiC/BN/SiC inner liner after 2758 h engine test. Note stable red some damage and some break-through of CVD Sic borosilicate glass layer on surface. eal coat was observed. Localized damage to underlying CFCC was seal coat was fully recessed over much larger areas on the urface (large white areas similar to area I in Fig. 16), the than the mi composite for the actual time that the composite itself composite damage observed was relatively severe and was similar ombustor environment. When the CVD Sic to the laboratory-exposed surfaces of CVI Sic/BN/SiC(see Fig

than the MI composite for the actual time that the composite itself was exposed to the combustor environment. When the CVD SiC seal coat was fully recessed over much larger areas on the gas-path surface (large white areas similar to area 1 in Fig. 16), the composite damage observed was relatively severe and was similar to the laboratory-exposed surfaces of CVI SiC/BN/SiC (see Fig. Fig. 8. Photograph of the gas-path surface of an MI SiC/BN/SiC engine-tested inner combustor liner exposed in engine for 2758 h. The three areas circled were microstructurally evaluated using aft-to-fore cross sections. Fig. 9. Surface recession measured from aft-to-fore cross sections circled in Fig. 9. Areas of minimum and maximum damage are included in plot. Fig. 10. Area from engine-tested MI SiC/BN/SiC inner liner where CVD SiC seal coat showed some damage and some break-through of CVD SiC seal coat was observed. Localized damage to underlying CFCC was observed. Fig. 11. Area from engine-tested MI SiC/BN/SiC inner liner where CVD SiC seal coat had recessed completely over a fairly large area. Damage to underlying CFCC was observed affecting 1–2 fiber tow layers. Fig. 12. Higher-magnification BSE image of the surface damage formed on MI SiC/BN/SiC inner liner after 2758 h engine test. Note stable borosilicate glass layer on surface. August 2003 High-Temperature Stability of SiC-Based Composites in High-Water-Vapor-Pressure Environments 1277

1278 Journal of the American Ceramic Society-More et al Vol 86. No. 8 Borosilicate glass I mm Fig. 13. BSE image of CVI SiC/BN/SiC cross section after 1500 h at 1200C and 1.5 atm of H2O in high-pressure furnace. Note thick, highly voided glass ayer on surface of composite 14). The greater depth of microstructural damage created in the For BN or other boron-containing compounds CV composite was clearly due to the amount of bn and B-containing phases used within the fiber tows and matrix and the 2BN+2O2(g)=B2O4)+N2(g) oxidation associated with the large amount of subsurface, inter- connected porosity. The effect of greater amounts of B-containing 2BN+3H2O(g)=B2O3)+N24g)+3H2g) on the increased oxidation of the CVI composite is shown 19, where the thickness and amount of borosilicate product B,O, (0)+H o(g)= 2HBO (g formed around the fibers within the fiber tows were considerably greater near the gas-path surface for the Cvi liner compared with B,O (0)+ SiO,= borosilicate glass At the temperature and water-vapor pressures of study, Hao is the dominant oxidant of the Si-based materi (3) Composite Damage Depths indirectly observed by the microstructural examinations discussed Microstructural analysis was used to measure the total in this paper. (While such results cannot distinguish between e damage depth. The composite damage depth was deter reactions(1)and(2)for Si or reactions (3)and(4) for SiC, it is measuring the surface oxide product thickness plus the nown from other studies 6, 7 20 26 that H-o is the dominant of any subsurface reactions that altered the composite xidant of the Si-based materials under the present conditions For example, fully oxidized fiber tows and matrix areas were here are also additional constituent reactions with water vapor cluded as part of the subsurface microstructural damage. Accu- and the oxides formed. Not only are the SiC and Bn oxidation ulated damage depth should not be confused with composite reactions accelerated in the presence of water vapor, but reactant recession, which is measured by subtracting the amount of diffusivities through the subsequently formed borosilicate glass are ubsurface, nonoxidized composite material (remaining"good high, thereby further increasing the degradation rates of the composite material) from the as-processed(starting )composite composite.2 Impurities(or proprietary additives) present in these nickness. Composite damage lengths for both the CVI and MI commercial composites also increase the oxidation rates, partic composites are plotted in Fig. 20 as a function of exposure time in larly in the presence of water vapor. Besides the accelerated he ORNL furnace. For comparison, the silica product thickness or oxidation, excessive amounts of gaseous reaction products are CVD SiC exposed under similar conditions is also shown in Fig formed during the numerous concurrent reactions res 0. Note that, despite being mainly composed of SiC fibers and highly defective, nonprotective, surface products(as matrix, these composites had much greater degradation rates than Figs. 5 and 13). Reaction(5)occurs on the gas-path surfa pure CVD SiC. Furthermore, based on composite recession mea- surements, the rate of degradation for the CvI SiC/BN/SiC was significantly higher than that of the MI SiC/BN/Si Borosilicate glass IV. Discussion m various constituents of the subject composites result in vapo, Possible oxidation reactions involving oxygen and water For Si (present in the MI composite matrix only) Si + O2(g)= SiO2 Borosilicate gl Si+ 2H,o(g)= Sio,+ 2H,(g) SiC+:O2g)= Sio2+co(g) SiC+ 3H,o(g)=SiO,+ CO(g)+3H,(g) Fig 14. BSe image of several consumed fiber tows in CVI SIC/BN/SIC after exposure for 1500 h at 1200C and 1.5 atm of H? SiO2+ 2H- o(g)= si(oh)(g)

14). The greater depth of microstructural damage created in the CVI composite was clearly due to the amount of BN and B-containing phases used within the fiber tows and matrix and the oxidation associated with the large amount of subsurface, inter￾connected porosity. The effect of greater amounts of B-containing phases on the increased oxidation of the CVI composite is shown in Fig. 19, where the thickness and amount of borosilicate product formed around the fibers within the fiber tows were considerably greater near the gas-path surface for the CVI liner compared with the MI liner (see Fig. 12). (3) Composite Damage Depths Microstructural analysis was used to measure the total compos￾ite damage depth. The composite damage depth was determined by measuring the surface oxide product thickness plus the thickness of any subsurface reactions that altered the composite structure. For example, fully oxidized fiber tows and matrix areas were included as part of the subsurface microstructural damage. Accu￾mulated damage depth should not be confused with composite recession, which is measured by subtracting the amount of subsurface, nonoxidized composite material (remaining “good” composite material) from the as-processed (starting) composite thickness. Composite damage lengths for both the CVI and MI composites are plotted in Fig. 20 as a function of exposure time in the ORNL furnace. For comparison, the silica product thickness on CVD SiC exposed under similar conditions is also shown in Fig. 20. Note that, despite being mainly composed of SiC fibers and matrix, these composites had much greater degradation rates than pure CVD SiC. Furthermore, based on composite recession mea￾surements, the rate of degradation for the CVI SiC/BN/SiC was significantly higher than that of the MI SiC/BN/SiC. IV. Discussion The various constituents of the subject composites result in many possible oxidation reactions involving oxygen and water vapor. For Si (present in the MI composite matrix only): Si O2 g SiO2 (1) Si 2H2O g SiO2 2H2 g (2) For SiC: SiC 3 2 O2 g SiO2 CO g (3) SiC 3H2O g SiO2 CO g 3H2 g (4) SiO2 2H2O g Si OH 4 g (5) For BN or other boron-containing compounds: 2BN 3 2 O2 g B2O3 l N2 g (6) 2BN 3H2O g B2O3 l N2 g 3H2 g (7) B2O3 l H2O g 2HBO2 g (8) B2O3 l SiO2 borosilicate glass (9) At the temperature and water-vapor pressures of the current study, H2O is the dominant oxidant of the Si-based materi￾als.16,17,20,21 Most of these reactions were either directly or indirectly observed by the microstructural examinations discussed in this paper. (While such results cannot distinguish between reactions (1) and (2) for Si or reactions (3) and (4) for SiC, it is known from other studies16,17,20,21,26 that H2O is the dominant oxidant of the Si-based materials under the present conditions.) There are also additional constituent reactions with water vapor and the oxides formed. Not only are the SiC and BN oxidation reactions accelerated in the presence of water vapor, but reactant diffusivities through the subsequently formed borosilicate glass are high, thereby further increasing the degradation rates of the composite.12 Impurities (or proprietary additives) present in these commercial composites also increase the oxidation rates, particu￾larly in the presence of water vapor.17 Besides the accelerated oxidation, excessive amounts of gaseous reaction products are formed during the numerous concurrent reactions resulting in highly defective, nonprotective, surface products (as shown in Figs. 5 and 13). Reaction (5) occurs on the gas-path surfaces of the Fig. 13. BSE image of CVI SiC/BN/SiC cross section after 1500 h at 1200°C and 1.5 atm of H2O in high-pressure furnace. Note thick, highly voided glass layer on surface of composite. Fig. 14. BSE image of several consumed fiber tows in CVI SiC/BN/SiC after exposure for 1500 h at 1200°C and 1.5 atm of H2O in high-pressure furnace. 1278 Journal of the American Ceramic Society—More et al. Vol. 86, No. 8

High-Temperature Stability of sic-Based Composites in High-Water-Vapor-Pressure Emvironments 1279 Area 3 Fig. 17. Surface recession measurements comparing two sections taken from CvI SiC/BN/SIC inner liner measured from aft-to-fore cross sections shown in Fig. 18 combustor liners where silica volatilization results in velocity. dependent CVD SiC seal coat recession during engine testing Clearly, the amount of borosilicate that forms and its viscosity depend on the amount of BN(and additional B-containing phases) present in the starting material. The CVI SiC/BN/SiC contain significantly more boron than the MI SiC/BN/SiC. Therefore, when this composite is exposed at higher temperatures and H,O pressures, greater amounts of borosilicate glass can form and this increases the overall composite degradation rate(Fig. 20). The formation of significant amounts of gaseous reaction products during oxidation is evident by the formation of a highly defective borosilicate glass surface product(see bubblelike voids in Fig. 13) Also, the starting CVI composite has >10 vol% interconnected large-scale porosity which allows oxidation reactions to take place in areas well below the exposed surface of the composite. Pore surfaces are oxidized. and. once the Cvd SiC seal coat is breached, extensive borosilicate glass forms and consumes the fiber tows. The MI SiC/BN/SiC is near full density and, thus, little subsurface oxidation occurs Fig. 15. Cross-section BSE image of CVI SiC/BN/SiC after(a)500 h at 1200C and 1.5 atm of H,O in high-pressure furnace show The estimates of CVI and MI composite degradation rates CVI SiC around a large subsurface pore(arrows to SiO, scale) and btained from examination of the liners exposed in engines were (b)1500 h where CVI Sic has been breached and large pools of glass nearly identical to those measured for the same composites after formed in subsurface regions associated with pores due to accelerated exposure in the high-pressure laboratory furnace despite the fact attack of fiber tows that the gas velocity in the turbine combustor was several orders of magnitude higher. Table I summarizes the recession(oxidation) Area 3 Area 2 Area 4 1234 Area 1 I hermae hotograph of the gas-path surface of CVI SiC/BN/SiC inner liner after 2266 h engine test. Section I and section 3 were used for aft-to-fore ural analysis

combustor liners where silica volatilization results in velocity￾dependent CVD SiC seal coat recession during engine testing. Clearly, the amount of borosilicate that forms and its viscosity depend on the amount of BN (and additional B-containing phases) present in the starting material. The CVI SiC/BN/SiC contains significantly more boron than the MI SiC/BN/SiC. Therefore, when this composite is exposed at higher temperatures and H2O pressures, greater amounts of borosilicate glass can form and this increases the overall composite degradation rate (Fig. 20). The formation of significant amounts of gaseous reaction products during oxidation is evident by the formation of a highly defective borosilicate glass surface product (see bubblelike voids in Fig. 13). Also, the starting CVI composite has
10 vol% interconnected, large-scale porosity which allows oxidation reactions to take place in areas well below the exposed surface of the composite. Pore surfaces are oxidized, and, once the CVD SiC seal coat is breached, extensive borosilicate glass forms and consumes the fiber tows. The MI SiC/BN/SiC is near full density and, thus, little subsurface oxidation occurs. The estimates of CVI and MI composite degradation rates obtained from examination of the liners exposed in engines were nearly identical to those measured for the same composites after exposure in the high-pressure laboratory furnace despite the fact that the gas velocity in the turbine combustor was several orders of magnitude higher. Table I summarizes the recession (oxidation) Fig. 15. Cross-section BSE image of CVI SiC/BN/SiC after (a) 500 h at 1200°C and 1.5 atm of H2O in high-pressure furnace showing oxidation of CVI SiC around a large subsurface pore (arrows point to SiO2 scale) and (b) 1500 h where CVI SiC has been breached and large pools of glass formed in subsurface regions associated with pores due to accelerated attack of fiber tows. Fig. 16. Photograph of the gas-path surface of CVI SiC/BN/SiC inner liner after 2266 h engine test. Section 1 and section 3 were used for aft-to-fore microstructural analysis. Fig. 17. Surface recession measurements comparing two sections taken from CVI SiC/BN/SiC inner liner measured from aft-to-fore cross sections shown in Fig. 18. August 2003 High-Temperature Stability of SiC-Based Composites in High-Water-Vapor-Pressure Environments 1279

1280 Journal of the American Ceramic Society-More et al Vol 86. No. 8 400 Area 1200 CVI SiC/BN/SiC 1000 MI SICE 400 200 CVD SiC Area I 100ur Time(h) Fig. 20. Summary of composite damage lengths for CVI and MI Fig. 18. BSE image of CVI SiC/BN/SiC inner liner showing subsurface SiC/BN/SIC composites compared with large pores after engine exposures in the high-temperature, high-pressure furnace at 1200.C and 1.5 atm of H,o rates(as determined from microstructural analysis) after exposure in the ornl furnace and in the solar Turbines Centaur engine for It is important to note that although there is little effect of gas the two different types of composites used in this study as well as velocity on composite stability due to the nature of the particula for CVD SiC(measured from CVD SiC seal coats applied to composites). The rates for engine-exposed composites given in reactions th ol oxidation-induced degradation, this is not th Table I represent the maximum recession rates for the composite case for monolithic SiC. Work at high H,O pressures at both lot materials after the CVD SiC seal coat is breached (i.e, composit and high gas velocities showed that Sic recession followed coat.)The oxidation rates measured indicate that MI and CVI same order. However, the controlling mechanism under slow composite degradation and recession are not as sensitive to gas gas-flow conditions in the laboratory furnace(nonprotective velocity cor d with volatility-induced recession of monolithic oxide formation) was different from that observed for much CVD SiC or CVD SiC seal coats. 8, 19 The observation of similar higher gas velocities experienced in an engine combustor(silica microstrucutral degradation as well as the formation and stability of significant amounts of borosilicate glass on the MI and CVI composite surfaces after both engine and laboratory exposures umma suggests that at high water-vapor pressures, such case, the mechanism for composite degradation is likely similar for The oxidation behavior of two types of SiC/BN/SiC composites both types of exposures and that gas velocity plays little, if any, at high water-vapor pressures during laboratory and engine expo- role in the controlling reaction processes. Once the CVD SiC seal sure has been explained based on multiple surface reactions coat is breached by volatility-induced recession during engine involving the formation of silica, boria, borosilicate glass, and exposure, the recession rate of the underlying composite increase gaseous products. The higher degradation rate for the CVI SiC to the recession rate associated with borosilicate(damage)forma- BN/SiC composite is a result of (1)the increased amount of BN tion(which is the rate given in Table I ) This lack of a significant and B-containing phases in this composite and (2)the larger velocity effect is consistent with the multiple oxidation reactions amount of interconnected, subsurface porosity compared with the that lead to borosilicate glass formation. In both the laboratory and MI SiC/BN/SiC. The much faster degradation rates of both the MI engine exposures, the porous borosilicate glass that forms on the and CVI composites compared with CVD SiC can be attributed to omposite surface is ineffective as a barrier to further oxidation the presence of boron and the formation of borosilicate glass, during exposure. which significantly increases the oxidation rate of the primary composite constituent, SiC. The recession and damage accumula tion rates measured for both composites after high-water-vapor- pressure laboratory exposures were similar to those measured for the same composites manufactured as combustor liners and ex- posed in engine tests when CVD SiC seal coats were breached These results indicate a similar controlling mechanism for S Table L. Summary of Recession Rates ory Furna d cvI SiC/BN/SiC Composites and CVD sure in a High-Temperature, High-Pressure Labora velocity =3 cm/min)and an Engine Ce or(gas Velocity 30 m/s Recession rate(um/h) Material aboratory furnac Engine combust d CVD SIC 0.11 MI SIC/BN/SIC 0.20 Fig.19. BSE image of CVI SIC/BN/SiC inner liner of a damaged CVI SIC/BN/SIC 0.50 ing preferred ox determined from microstructural measurements after

rates (as determined from microstructural analysis) after exposure in the ORNL furnace and in the Solar Turbines Centaur engine for the two different types of composites used in this study as well as for CVD SiC (measured from CVD SiC seal coats applied to composites). The rates for engine-exposed composites given in Table I represent the maximum recession rates for the composite materials after the CVD SiC seal coat is breached (i.e., composite rates given do not include the recession rate for the CVD SiC seal coat.) The oxidation rates measured indicate that MI and CVI composite degradation and recession are not as sensitive to gas velocity compared with volatility-induced recession of monolithic CVD SiC or CVD SiC seal coats.18,19 The observation of similar microstrucutral degradation as well as the formation and stability of significant amounts of borosilicate glass on the MI and CVI composite surfaces after both engine and laboratory exposures suggests that at high water-vapor pressures, such as in the present case, the mechanism for composite degradation is likely similar for both types of exposures and that gas velocity plays little, if any, role in the controlling reaction processes. Once the CVD SiC seal coat is breached by volatility-induced recession during engine exposure, the recession rate of the underlying composite increases to the recession rate associated with borosilicate (damage) forma￾tion (which is the rate given in Table I.) This lack of a significant velocity effect is consistent with the multiple oxidation reactions that lead to borosilicate glass formation. In both the laboratory and engine exposures, the porous borosilicate glass that forms on the composite surface is ineffective as a barrier to further oxidation during exposure. It is important to note that although there is little effect of gas velocity on composite stability due to the nature of the particular reactions that control oxidation-induced degradation, this is not the case for monolithic SiC. Work at high H2O pressures at both low and high gas velocities showed that SiC recession followed paralinear kinetics17–19,21 and that recession rates can be of the same order.21 However, the controlling mechanism under slow gas-flow conditions in the laboratory furnace (nonprotective oxide formation) was different from that observed for much higher gas velocities experienced in an engine combustor (silica volatilization).18,19,21 V. Summary The oxidation behavior of two types of SiC/BN/SiC composites at high water-vapor pressures during laboratory and engine expo￾sure has been explained based on multiple surface reactions involving the formation of silica, boria, borosilicate glass, and gaseous products. The higher degradation rate for the CVI SiC/ BN/SiC composite is a result of (1) the increased amount of BN and B-containing phases in this composite and (2) the larger amount of interconnected, subsurface porosity compared with the MI SiC/BN/SiC. The much faster degradation rates of both the MI and CVI composites compared with CVD SiC can be attributed to the presence of boron and the formation of borosilicate glass, which significantly increases the oxidation rate of the primary composite constituent, SiC. The recession and damage accumula￾tion rates measured for both composites after high-water-vapor￾pressure laboratory exposures were similar to those measured for the same composites manufactured as combustor liners and ex￾posed in engine tests when CVD SiC seal coats were breached. These results indicate a similar controlling mechanism for SiC/ Fig. 18. BSE image of CVI SiC/BN/SiC inner liner showing subsurface oxidation within large pores after engine test. Fig. 19. BSE image of CVI SiC/BN/SiC inner liner of a damaged composite fiber tow showing preferred oxidation of BN fiber coatings and boron-containing filler within tow. Fig. 20. Summary of composite damage lengths for CVI and MI SiC/BN/SiC composites compared with CVD SiC as a function of time for exposures in the high-temperature, high-pressure furnace at 1200°C and 1.5 atm of H2O. Table I. Summary of Recession Rates for MI and CVI SiC/BN/SiC Composites and CVD SiC after Exposure in a High-Temperature, High-Pressure Laboratory Furnace (Gas Velocity
3 cm/min) and an Engine Combustor (Gas Velocity
30 m/s)† Material Recession rate (m/h) Laboratory furnace Engine combustor CVD SiC 0.06 0.11 MI SiC/BN/SiC 0.20 0.20 CVI SiC/BN/SiC 0.40 0.50 † Temperature  1200°C, pressure  10 atm, pH2O  1.5 atm. Rates were determined from microstructural measurements after testing. 1280 Journal of the American Ceramic Society—More et al. Vol. 86, No. 8

ust 2003 High-Temperature Stability of Sic-Based Composites in High-Water-Vapor-Pressure Emvironments BN/SiC ceramic composites exposed to both high and low gas IoR. A. Lowden, O. J. Schwarz, and K. L. More, "Improved Fiber Coatings for velocities at similar elevated H,O pressures. cag nasa o mp, e canda. ge te Poco,n Nitride islerhase in s Matrix Composites, "J. Am. Ceram Soc., 74[1]2482-88(1989) 12N. S. Jacobson, G. N. Morscher, D. R. Bryant, and R. E. Tressler,"High- Temperature Oxidation of Boron Nitride: Il, Boron Nitride Layers in Composites, We wish to thank J. R. Keiser. M. Howell and K.S. Trent for technical assistance JAm. Ceram.Soc,82间61473-82(1999 F. Tortorelli, and L. R. Walker, "Effects of High Water Vap tes. mater References 14B. E. Deal and A S. Grove, "General Relationship for the Thermal Oxidation of ala and J. R. Price. "The Evaluation of CFCC Liners After Field Testing in a Gas Turbine-Ir, ASME Paper 2000-GT-648 in Proceedings of ASME Turbo E J. Opila, "Oxidation Kinetics of Chemically Vapor Deposited Silicon Carbide in Wet Oxygen, " J. Am. Ceram Soc, 77 [3]730-36(1994) 16E. J. Opila and R. E Hann, "Paralinear Oxidation of CVD SiC in Water Vapor Program-Combustor Liner Development Summary", ASME Paper 2001-GT-512 in J Am Ceram. Soc., 80[11 197-205(1997) Proceedings of ASME Turbo Expo 2001. American Society of Mechanical Engineers, 17E. J. Opila, "Variation of the Oxidation Rate of Silicon ressure. Anm Cer 823]625-36(1999 SA. J. Dean, G.S. Corman, B Bagepalli, K L. Luthra, P S. DiMascio, and R.M. R. C. Robinson and J. L. Smialek, "SiC Recession Caused by SiO, Scale vola- Orenstein,"Design and Testing of CFCC Shroud and Combustor Components', tility under Combustion Conditions: I, Experimental Results and Empirical Model ASME Paper 99-G1-235 in Proceedings of ASME Turbo Erpo 1999. American J. Am. Ceram Soc., 82 [7) 1817-25(199 E.J. Opila, J. L. Smialek, R. C, Robinson, D. S. Fox, and N. S. Jacobsen, "SiC ognarelli, and cession Caused by SiO, Scale Volatility under Combustion Conditions: nics and Gaseous-Diffusion Model. J. dm Ceram. Soc. 82 Combustor and Shroud Applications", ASME Paper 2000-G 1826-34(1999 ASME Turbo Expo 2000. American Society of Mechanica New York, 20K. L. More, P. F. Tortorelli, J. R. Keiser, and M. K. Ferber " Observations of Accelerated Silicon Carbide Recession by Oxidation at High Water-Vapor Pressures, G. Simon and A. R. Bunsell"Mechanical and Structural Characterization of the Nicalon Silicon Carbide Fibre, J. Mater. Sci., 19, 3649-57(1984) P. F. Tortorelli and K. L. More, "Effects of High Water-Vapor Pressure 6. L. More, P. F. Tortorelli, and R. A. Lowden, "Oxidation-Induced Microstruc- Oxidation of Silicon Carbide at 1200oC, J. Am. Ceram. Soc., 86[8]1249-55(2003 dited by s. B. Newcomb and J. A. Little w. D. Brentnall, and J. R. Price, "Exposure of Ceramics and Ceramic Matrix 7K. Kumagawa, H. Yamaoka, M. Shibuya, and T. Yamamura, "Fabrication and Composites in Simulated and Actual Combustor Environments, " J. Eng. Gas rties of New Improved Si-M-C-(O)Tyranno Fibers, Ceram Turbines Power, 122, 212-18(2000). Eng.Proc,1965-72(1998 P Gray, GE Power Systems Composites, LLC, personal communication. Matrix Composites by 24K. L Luthra, R N Singh, ano Am Ceram Soc. Bull.72[7]79-8319M> Energy Filtering TEM, " J. Electron Microsc., 4 R. Keiser. M. Howell. J J. and R A Rosenberg, "Com rostructural evolutio Hi-Nicalon SiC Fibers Annealed and Crept in Various Oxygen Partial Pressure Proceedings of Corrosion/6. NACE International, Houston, TX, 1996 Atmospheres, J Mater. Sci, 35, 1153-64(2000) K. L. More and P. F, Tortorelli, unpublished data 1999

BN/SiC ceramic composites exposed to both high and low gas velocities at similar elevated H2O pressures. Acknowledgments We wish to thank J. R. Keiser, M. Howell, and K. S. Trent for technical assistance. References 1 N. Miriyala and J. R. Price, “The Evaluation of CFCC Liners After Field Testing in a Gas Turbine—II”; ASME Paper 2000-GT-648 in Proceedings of ASME Turbo Expo 2000. American Society of Mechanical Engineers, New York, 2000. 2 N. Miriyala, A. Fahme, and M. van Roode, “Ceramic Stationary Gas Turbine Program—Combustor Liner Development Summary”; ASME Paper 2001-GT-512 in Proceedings of ASME Turbo Expo 2001. American Society of Mechanical Engineers, New York, 2001. 3 A. J. Dean, G. S. Corman, B. Bagepalli, K. L. Luthra, P. S. DiMascio, and R. M. Orenstein, “Design and Testing of CFCC Shroud and Combustor Components”; ASME Paper 99-GT-235 in Proceedings of ASME Turbo Expo 1999. American Society of Mechanical Engineers, New York, 1999. 4 G. S. Corman, A. J. Dean, S. Brabetz, M. J. Brun, K. L. Luthra, L. Tognarelli, and M. Pecchioli, “Rig and Engine Testing of Melt Infiltrated Ceramic Composites for Combustor and Shroud Applications”; ASME Paper 2000-GT-638 in Proceedings of ASME Turbo Expo 2000. American Society of Mechanical Engineers, New York, 2000. 5 G. Simon and A. R. Bunsell, “Mechanical and Structural Characterization of the Nicalon Silicon Carbide Fibre,” J. Mater. Sci., 19, 3649–57 (1984). 6 K. L. More, P. F. Tortorelli, and R. A. Lowden, “Oxidation-Induced Microstruc￾tural Changes in Fiber-Reinforced SiC”; pp. 683–94 in Microscopy of Oxidation 3. Edited by S. B. Newcomb and J. A. Little. Institute of Materials, London, U.K., 1997. 7 K. Kumagawa, H. Yamaoka, M. Shibuya, and T. Yamamura, “Fabrication and Mechanical Properties of New Improved Si-M-C-(O) Tyranno Fibers,” Ceram. Sci. Eng. Proc., 19 [3] 65–72 (1998). 8 T. Shibayama, G. W. He, H. Takahashi, M. Kawasaki, and A. Kohyama, “Interface Structure Analysis of SiC Fiber Reinforced SiC Matrix Composites by Energy Filtering TEM,” J. Electron Microsc., 48 [6] 893–97 (1999). 9 G. W. He, T. Shibayama, and H. Takahashi, “Microstructural Evolution of Hi-Nicalon SiC Fibers Annealed and Crept in Various Oxygen Partial Pressure Atmospheres,” J. Mater. Sci., 35, 1153–64 (2000). 10R. A. Lowden, O. J. Schwarz, and K. L. More, “Improved Fiber Coatings for Nicalon/SiC Composites,” Ceram. Eng. Sci. Proc., 14 [7–8] 375–84 (1993). 11R. Naslain, O. Dugne, and A. Guette, “Boron Nitride Interphase in Ceramic￾Matrix Composites,” J. Am. Ceram. Soc., 74 [1] 2482–88 (1989). 12N. S. Jacobson, G. N. Morscher, D. R. Bryant, and R. E. Tressler, “High￾Temperature Oxidation of Boron Nitride: II, Boron Nitride Layers in Composites,” J. Am. Ceram. Soc., 82 [6] 1473–82 (1999). 13K. L. More, P. F. Tortorelli, and L. R. Walker, “Effects of High Water Vapor Pressures on the Oxidation of SiC-Based Fiber-Reinforced Composites,” Mater. Sci. Forum, 369–372, 385–94 (2001). 14B. E. Deal and A. S. Grove, “General Relationship for the Thermal Oxidation of Silicon,” J. Appl. Phys., 36, 3770–78 (1965). 15E. J. Opila, “Oxidation Kinetics of Chemically Vapor Deposited Silicon Carbide in Wet Oxygen,” J. Am. Ceram. Soc., 77 [3] 730–36 (1994). 16E. J. Opila and R. E. Hann, “Paralinear Oxidation of CVD SiC in Water Vapor,” J. Am. Ceram. Soc., 80 [1] 197–205 (1997). 17E. J. Opila, “Variation of the Oxidation Rate of Silicon Carbide with Water￾Vapor Pressure,” J. Am. Ceram. Soc., 82 [3] 625–36 (1999). 18R. C. Robinson and J. L. Smialek, “SiC Recession Caused by SiO2 ScaleVola￾tility under Combustion Conditions: I, Experimental Results and Empirical Model,” J. Am. Ceram. Soc., 82 [7] 1817–25 (1999). 19E. J. Opila, J. L. Smialek, R. C. Robinson, D. S. Fox, and N. S. Jacobsen, “SiC Recession Caused by SiO2 Scale Volatility under Combustion Conditions: II, Thermodynamics and Gaseous-Diffusion Model,” J. Am. Ceram. Soc., 82 [7] 1826–34 (1999). 20K. L. More, P. F. Tortorelli, J. R. Keiser, and M. K. Ferber, “Observations of Accelerated Silicon Carbide Recession by Oxidation at High Water-Vapor Pressures,” J. Am. Ceram. Soc., 83 [1] 211–13 (2000). 21P. F. Tortorelli and K. L. More, “Effects of High Water-Vapor Pressure on Oxidation of Silicon Carbide at 1200°C,” J. Am. Ceram. Soc., 86 [8] 1249–55 (2003). 22K. L. More, P. F. Tortorelli, M. K. Ferber, L. R. Walker, J. R. Keiser, N. Miriyala, W. D. Brentnall, and J. R. Price, “Exposure of Ceramics and Ceramic Matrix Composites in Simulated and Actual Combustor Environments,” J. Eng. Gas Turbines Power, 122, 212–18 (2000). 23P. Gray, GE Power Systems Composites, LLC, personal communication. 24K. L. Luthra, R. N. Singh, and M. K. Brun, “Toughened Silcomp Composites, Process and Preliminary Properties,” Am. Ceram. Soc. Bull. 72 [7] 79–85 (1993). 25J. R. Keiser, M. Howell, J. J. Williams, and R. A. Rosenberg, “Compatibility of Selected Ceramics with Steam-Methane Reformer Environments”; Paper No. 140 in Proceedings of Corrosion/96. NACE International, Houston, TX, 1996. 26K. L. More and P. F. Tortorelli, unpublished data 1999.
August 2003 High-Temperature Stability of SiC-Based Composites in High-Water-Vapor-Pressure Environments 1281