J. Ant Cerum. Soc., 85 [3]595-602(2002) ournal Effects of Thermal Aging on the Mechanical Properties of a Porous-Matrix Ceramic Composite Eric A.v. Carelli* Science and Technology Center, Siemens-Westinghouse Power Corporation, Pittsburgh, Pennsylvania 15235 Hiroki Fujita, James Y. Yang, and Frank W. Zok Materials Department, University of California, Santa Barbara, California 93106 properties of an all-oxide fiber-reinforced composite follow a The present article focuses on changes in the mechanical gas turbine and at the burner outlet. In current long-term exposure(1000 h) at temperatures of 1000-1200 C or near their upper use temperature, even with in air. The composite of interest derives its damage tolerance imparted by the use of thermal barrier coatings, thereby from a highly porous matrix, precluding the need for an significant temperature elevations with these alloys interphase at the fiber-matrix boundary. The key issue in- future environmental and performance standards, it is anticipated olves the stability of the porosity against densification and the that the targeted temperature elevations in turbine components will associated implications for long-term durability of the compos- be accomplished through the use of continuous-fiber-reinforced ite at elevated temperatures. For this purpose, comparisons ceramic composites(CFCCs). Among the various ceramic com- are made in the tensile properties and fracture characteristics of a 2D woven fiber composite both along the fiber direction posites that have been developed to date, the ones that have and at 45 to the fiber axes before and after the aging try in the past few years are those made from all-oxide constitu- indentation and through the determination of the matrix non-oxide ones(e.g, SiC/SiC)is their superior resistance to oxidation under typical turbine engine conditions and hence their pled with classical laminate theory. The study reveals that, potential for long-term durability despite evidence of some strengthening of the matrix and the As part of a broad activity aimed at developing and assessing fber-matrix interfaces during aging, the key tensile properties oxide composites for use in future generations of gas turbin in the 0o clud engines, the present study focuses on changes in the mechanical are unchanged. This strengthening is manifested to a more properties of a candidate all-oxide CFCC following long-term exposure(1000 h)at temperatures of 1000-1200C. The compos- orientation, wherein the modulus and the tensile strength each ite material of interest derives its damage tolerance from a highly exhibit a twofold increase after the 1200%C aging treatment. It also results in a change in the failure mechanism, from one fiber-matrix boundary. Although the efficacy of this material involving predominantly matrix damage and interply delami- concept in enabling damage tolerance has been demonstrated, -o it nation to one which is dominated by fiber fracture. Addition- remains to be established whether the matrix pore structure is ally, salient changes in the mechanical response beyond the stable against densification and whether the desirable damage maximum load suggest the existence of an optimum matrix tolerant characteristics can be retained for extended time periods at attains a maximum. The implications for long-term durability the porous-matrix concept have been shown to exhibit severe of this class of composite are discussed. degradation in composite properties once the matrix densifies appreciably. The main matrix constituent in the present composite is mullite, L. Introduction in the form of a weakly bonded particulate network. This selection n industry has been under increased pres- is based on the sluggish sintering kinetics of mullite'at the upper keeping up with market demands for increased power output and s intended to form a contiguous particulate network that should be to red efficiency. These goals can be achieved in part through reductions under subsequent service conditions. The minor matrix constituen airfoils with attendant increases in the temperatures both within the alumina, present both in the form of particulates from a slurr and as a product of pyrolysis of an aqueous precursor solution Because of its more rapid sintering kinetics, alumina serves to bond the mullite particulates and the fibers together, thereby E. Lara-Curzio--contributing editor enhancing the matrix-dominated composite properties, e.g., inter laminar strength and off-axis in-plane strength. However, if the degree of sintering becomes excessive, the damage-tolerant char acteristics may be compromised. The challenge involves selection cript No. 187723 Received May 4, 2001; approved December 21, 2001 of the relative fractions and topologies of the two phases such that under the network of mullite particles remains contiguous and hence nrough both internal research funds at the Science and Technology Center and a prevents global nkage, yet the extent of bonding within this ubcontract to the University of California at Santa Barbara, and by a gift from NGK network is sufficient to impart the requisite matrix integrity for Member. American Ceramic Society. acceptable off-axis properties. These opposing requirements or 595
Effects of Thermal Aging on the Mechanical Properties of a Porous-Matrix Ceramic Composite Eric A. V. Carelli* Science and Technology Center, Siemens-Westinghouse Power Corporation, Pittsburgh, Pennsylvania 15235 Hiroki Fujita,* James Y. Yang, and Frank W. Zok* Materials Department, University of California, Santa Barbara, California 93106 The present article focuses on changes in the mechanical properties of an all-oxide fiber-reinforced composite following long-term exposure (1000 h) at temperatures of 1000–1200°C in air. The composite of interest derives its damage tolerance from a highly porous matrix, precluding the need for an interphase at the fiber–matrix boundary. The key issue involves the stability of the porosity against densification and the associated implications for long-term durability of the composite at elevated temperatures. For this purpose, comparisons are made in the tensile properties and fracture characteristics of a 2D woven fiber composite both along the fiber direction and at 45° to the fiber axes before and after the aging treatments. Additionally, changes in the state of the matrix are probed through measurements of matrix hardness by Vickers indentation and through the determination of the matrix Young’s modulus, using the measured composite moduli coupled with classical laminate theory. The study reveals that, despite evidence of some strengthening of the matrix and the fiber–matrix interfaces during aging, the key tensile properties in the 0°/90° orientation, including strength and failure strain, are unchanged. This strengthening is manifested to a more significant extent in the composite properties in the 45° orientation, wherein the modulus and the tensile strength each exhibit a twofold increase after the 1200°C aging treatment. It also results in a change in the failure mechanism, from one involving predominantly matrix damage and interply delamination to one which is dominated by fiber fracture. Additionally, salient changes in the mechanical response beyond the maximum load suggest the existence of an optimum matrix strength at which the fracture energy in the 45° orientation attains a maximum. The implications for long-term durability of this class of composite are discussed. I. Introduction THE power generation industry has been under increased pressure to reduce NOx emissions from gas turbine engines while keeping up with market demands for increased power output and efficiency. These goals can be achieved in part through reductions in the amount of film cooling of combustor liners and turbine airfoils with attendant increases in the temperatures both within the gas turbine and at the burner outlet.1,2 In current gas turbine engines, many of the superalloy-based components are operating at or near their upper use temperature, even with the benefits imparted by the use of thermal barrier coatings, thereby precluding significant temperature elevations with these alloys. To meet future environmental and performance standards, it is anticipated that the targeted temperature elevations in turbine components will be accomplished through the use of continuous-fiber-reinforced ceramic composites (CFCCs). Among the various ceramic composites that have been developed to date, the ones that have attracted the greatest attention within the power generation industry in the past few years are those made from all-oxide constituents. The main advantage of the oxide-based composites over non-oxide ones (e.g., SiC/SiC) is their superior resistance to oxidation under typical turbine engine conditions and hence their potential for long-term durability. As part of a broad activity aimed at developing and assessing oxide composites for use in future generations of gas turbine engines, the present study focuses on changes in the mechanical properties of a candidate all-oxide CFCC following long-term exposure (1000 h) at temperatures of 1000–1200°C. The composite material of interest derives its damage tolerance from a highly porous matrix, precluding the need for an interphase at the fiber–matrix boundary. Although the efficacy of this material concept in enabling damage tolerance has been demonstrated,3–6 it remains to be established whether the matrix pore structure is stable against densification and whether the desirable damagetolerant characteristics can be retained for extended time periods at the targeted service temperatures. Indeed, other CFCCs based on the porous-matrix concept have been shown to exhibit severe degradation in composite properties once the matrix densifies appreciably.7 The main matrix constituent in the present composite is mullite, in the form of a weakly bonded particulate network. This selection is based on the sluggish sintering kinetics of mullite3 at the upper use temperature for the fibers (1200°C for Nextel 720). This phase is intended to form a contiguous particulate network that should be immune from appreciable densification both during processing and under subsequent service conditions. The minor matrix constituent is alumina, present both in the form of particulates from a slurry and as a product of pyrolysis of an aqueous precursor solution.3,5 Because of its more rapid sintering kinetics, alumina serves to bond the mullite particulates and the fibers together, thereby enhancing the matrix-dominated composite properties, e.g., interlaminar strength and off-axis in-plane strength. However, if the degree of sintering becomes excessive, the damage-tolerant characteristics may be compromised. The challenge involves selection of the relative fractions and topologies of the two phases such that the network of mullite particles remains contiguous and hence prevents global shrinkage, yet the extent of bonding within this network is sufficient to impart the requisite matrix integrity for acceptable off-axis properties. These opposing requirements on the E. Lara-Curzio—-contributing editor Manuscript No. 187723. Received May 4, 2001; approved December 21, 2001. Funding for this work was provided by the Air Force Office of Scientific Research under Contract No. F49620-99-1-0259, by Siemens Westinghouse Power Corp. through both internal research funds at the Science and Technology Center and a subcontract to the University of California at Santa Barbara, and by a gift from NGK Corp. *Member, American Ceramic Society. J. Am. Ceram. Soc., 85 [3] 595–602 (2002) 595 journal
Journal of the American Ceramic Society-Carelli et al. Vol. 85. No. 3 matrix suggest an optimum state, dictated in part by the combina- air furnace for 1000 h at temperatures of either 1000, 1 100%,or tion of properties that are required in the application of interest. 1200C and subsequently tested in uniaxial tension at ambien In this study, comparisons are made between the tensile temperature, following the procedures outlined above. The furnace properties of a 2D woven CFCC both along the fiber direction was heated with resistance wire coated with a ceramic mixture of (0/90o) and at 45. to the fiber axes before and after high- aluminophosphate and alumina, and insulated with aluminosilicate temperature aging treatments. These orientations are selected to Either two or three tests were performed for most conditions elicit the fiber-dominated and matrix-dominated composite prop- representative fractured specimens were examined by bot rties. Examinations of the broken specimens by optical and agnification light microscopy and scanning electron microscopy. scanning electron microscopy are used to elucidate the role of The porosity both before and after aging was measured follow- aging in the fracture characteristics. Changes in the state of the ing ASTM Standard C20-92 Changes in the microstructure of the matrix are probed through two additional complementary methods: aged specimens were elucidated from SEM observations of pol ()measurement of matrix hardness using Vickers indentation, and ished samples. Matrix hardness measurements were also made on (ii) determination of the matrix Youngs mod using the these polished samples within the matrix-rich regions between th measured composite moduli coupled with classical laminate the- fiber tows, using Vickers indentation with a load of 300 g. This ory. Additionally, some comparisons are made with the retention load was selected to produce indentations that were no larger than in properties of a comparable porous-matrix composite with an half of the spacing between tows in all materials. Additionally, the aluminosilicate matrix indents were placed away from the processing-induced cracks (described below). At least 10 such measurements were made on IL. Materials and Test Procedures samples in each aged condition The matrix modulus of both pristine and aged specimens was The composite material consists of Nextel 720 fiber cloth in an inferred from the measured composite moduli in both the 0/90 8-harness satin weave and a porous matrix of mullite and alumi- and +45 orientations using laminate theory. Details of the theory na.5 The matrix was produced in two steps. In the first,an are described in the Appendix. queous slurry containing mullite and alumina particulates was vacuum-infiltrated into a stack of 12 fiber cloths. The matrix Ill. Experimental Results and Analysis particulates were "l um diameter MU-107 mullite(Showa Denko KK )and -0. 2 um diameter AKP-50 alumina(Sumitomo Chem ypical micrographs of polished cross sections of both cal). The slurry contained 80% mullite and 20% alumina, The rocessed specimen and one aged at 1200%C are shown in Fi packing density of the matrix powder foll infiltration was 1. A notable feature is the presence of a more-or-less regular 960%. The green panels were dried and sintered at 900oC for 2 h pattern of matrix cracks, caused by the constrained shrinkage of to promote the development of alumina bridges between the the matrix during drying of the green panels. In the as-processed mullite particles, thereby imparting some structural integrity to the composite, the cracks are concentrated in the matrix-rich regions matrix for subsequent processing. In the second step, the panel were impregnated with an alumina precursor solution(aluminum hydroxychloride) and pyrolyzed at 900C for 2 h. The volt yield of alumina on pyrolysis of this solution was #%. The(a)As-processed mpregnation and pyrolysis sequence was performed twice. The panels were given a final heat treatment at 1200.C for 2 h to stabilize the precursor-derived alumina and to enhance the integ ty of the alumina bridges. The panel dimensions were 200 mm X 130mm×3. I mm thic Panels were made with fibers oriented either parallel or at 4 to the panel edges, thus facilitating preparation of tensile speci mens in both the longitudinal(0°90°) and off-axis(±45°) orientations. The key physical properties of each of the tested summarized in Table I. The fiber content, , wa determined from knowledge of the cloth volume and the final plate ASTM Standard C20-92>P, was measured in accordance with A series of mechanical tests was performed to determine the tensile properties of the as-processed composite in both the 0%/900 and the +45 orientations. The properties were measured usin tandard dog-bone tensile specimens with a gauge length of 50 mm and a gauge width of 8 mm. The longitudinal strains were 1( b)1000 h at 1200C measured us d at room temperature at a displacement rate of an extensometer over a 25 mm gauge length. The 1.25 mm/min Either two or three tests were performed for each material the effects of thermal aging on the mechanical properties, tensile specimens in both orientations were heated in an Table L. Summary of Physical Properties of CFCC Panels b Matrix porosity (% Fiber volar designation orentation Initial Final 39.5 37.9 0.2mm ABC 0°/90° 38.3 40.2 37.7 After slurry infiltration and drying. After precursor impregnation and pyrolysis Fig. 1. SEM micrographs of as-processed and thermally aged composite backscatter imaging mode)
matrix suggest an optimum state, dictated in part by the combination of properties that are required in the application of interest. In this study, comparisons are made between the tensile properties of a 2D woven CFCC both along the fiber direction (0°/90°) and at 45° to the fiber axes before and after hightemperature aging treatments. These orientations are selected to elicit the fiber-dominated and matrix-dominated composite properties. Examinations of the broken specimens by optical and scanning electron microscopy are used to elucidate the role of aging in the fracture characteristics. Changes in the state of the matrix are probed through two additional complementary methods: (i) measurement of matrix hardness using Vickers indentation, and (ii) determination of the matrix Young’s modulus, using the measured composite moduli coupled with classical laminate theory.8 Additionally, some comparisons are made with the retention in properties of a comparable porous-matrix composite with an aluminosilicate matrix.7 II. Materials and Test Procedures The composite material consists of Nextel 720 fiber cloth in an 8-harness satin weave and a porous matrix of mullite and alumina.3,5 The matrix was produced in two steps. In the first, an aqueous slurry containing mullite and alumina particulates was vacuum-infiltrated into a stack of 12 fiber cloths. The matrix particulates were 1 m diameter MU-107 mullite (Showa Denko K.K.) and 0.2 m diameter AKP-50 alumina (Sumitomo Chemical). The slurry contained 80% mullite and 20% alumina. The packing density of the matrix powder following infiltration was 60%. The green panels were dried and sintered at 900°C for 2 h to promote the development of alumina bridges between the mullite particles, thereby imparting some structural integrity to the matrix for subsequent processing. In the second step, the panels were impregnated with an alumina precursor solution (aluminum hydroxychloride) and pyrolyzed at 900°C for 2 h. The volumetric yield of alumina on pyrolysis of this solution was 3%. The impregnation and pyrolysis sequence was performed twice. The panels were given a final heat treatment at 1200°C for 2 h to stabilize the precursor-derived alumina and to enhance the integrity of the alumina bridges. The panel dimensions were 200 mm 130 mm 3.1 mm thick. Panels were made with fibers oriented either parallel or at 45° to the panel edges, thus facilitating preparation of tensile specimens in both the longitudinal (0°/90°) and off-axis (45°) orientations. The key physical properties of each of the tested panels are summarized in Table I. The fiber content, f, was determined from knowledge of the cloth volume and the final plate dimensions. The porosity, p, was measured in accordance with ASTM Standard C20-92. A series of mechanical tests was performed to determine the tensile properties of the as-processed composite in both the 0°/90° and the 45° orientations. The properties were measured using standard dog-bone tensile specimens with a gauge length of 50 mm and a gauge width of 8 mm. The longitudinal strains were measured using an extensometer over a 25 mm gauge length. The tests were performed at room temperature at a displacement rate of 1.25 mm/min. Either two or three tests were performed for each material. To assess the effects of thermal aging on the mechanical properties, tensile specimens in both orientations were heated in an air furnace for 1000 h at temperatures of either 1000°, 1100°, or 1200°C and subsequently tested in uniaxial tension at ambient temperature, following the procedures outlined above. The furnace was heated with resistance wire, coated with a ceramic mixture of aluminophosphate and alumina, and insulated with aluminosilicate fiber. Either two or three tests were performed for most conditions. Representative fractured specimens were examined by both lowmagnification light microscopy and scanning electron microscopy. The porosity both before and after aging was measured following ASTM Standard C20-92. Changes in the microstructure of the aged specimens were elucidated from SEM observations of polished samples. Matrix hardness measurements were also made on these polished samples within the matrix-rich regions between the fiber tows, using Vickers indentation with a load of 300 g. This load was selected to produce indentations that were no larger than half of the spacing between tows in all materials. Additionally, the indents were placed away from the processing-induced cracks (described below). At least 10 such measurements were made on samples in each aged condition. The matrix modulus of both pristine and aged specimens was inferred from the measured composite moduli in both the 0°/90° and 45° orientations using laminate theory. Details of the theory are described in the Appendix. III. Experimental Results and Analysis Typical micrographs of polished cross sections of both an as-processed specimen and one aged at 1200°C are shown in Fig. 1. A notable feature is the presence of a more-or-less regular pattern of matrix cracks, caused by the constrained shrinkage of the matrix during drying of the green panels. In the as-processed composite, the cracks are concentrated in the matrix-rich regions Table I. Summary of Physical Properties of CFCC Panels Panel designation Fiber orientation Matrix porosity (%) Fiber volume Initial fraction (%) † Final‡ A 0°/90° 39.5 37.9 38.6 B 0°/90° 40.1 38.3 38.9 C 45° 40.2 37.7 38.5 † After slurry infiltration and drying. ‡ After precursor impregnation and pyrolysis and final sintering treatment. Fig. 1. SEM micrographs of as-processed and thermally aged composite specimens (viewed in backscatter imaging mode). 596 Journal of the American Ceramic Society—Carelli et al. Vol. 85, No. 3
March 2002 Efects of Thermal Aging on the Mechanical Properties of a Porous-Matrix Ceramic Composite 000hat1200c 1000hat1100°c 1000hat1000°c 000 0003 0.004 Tensile Strain 10 u0z0 (b) Aging Temperature(C) (c) As-Processed 48% Nextel 720 fiber (Jurf and Butner, 2000 3 mm 0°/90 Fig 3. Macrophotographs of the 0/90 tensile specimens in two orthog- 20040060080010001200 onal views: (a) as-processed, (b)after aging for 1000 h at 1100.C, and(c Aging Temperature(C) after aging for 1000 h at 1200C Fig. 2. Effects of thermal aging on the tensile properties of the 0/90 composite. The modulus was calculated from the slope of the initial linear strain is accommodated by additional opening displacement of the rtion of the stress-strain curve over a stress range of =50 MPa. racks such that the net average strain is approximately zero. This hypothesis is supported by the measurements of the composite porosity, which indicated no significant change after any of the between fiber tows. They arrest at the interface with the longitu- aging treatments. Higher-magnification SEM examinations of the dinal fibers (oriented perpendicular to the crack plane) and matrix microstructure did not reveal any other obvious changes penetrate only slightly into the transverse tows. Following aging, due to aging the pattern of matrix cracks remains essentially the same, with the Representative stress-strain curves for the as-processed and the exception that the cracks tend to grow into the transverse tows and aged specimens in the 0 /90 orientation are plotted in Fig. 2(a) their opening displacement increases somewhat(see, for example, The variations in the Youngs modulus, E, and the ultimate tensile the cracks on the right side of Fig. 1(b)). These features are strength, u, with aging temperature are summarized in Figs. 2(b) believed to be due to some matrix densification in the matrix and (c). The only significant change is the slight increase in th segments contained between the cracks along the direction per- modulus, from 60 GPa in the as-processed condition to 70 GPa pendicular to the cracks. Since this shrinkage is constrained by the after the 1200C aging treatment. The tensile strength and the (dense) fibers in the adjacent longitudinal tows, the shrinkage failure strain, Es, remained unchanged; the averages and standard
between fiber tows. They arrest at the interface with the longitudinal fibers (oriented perpendicular to the crack plane) and penetrate only slightly into the transverse tows. Following aging, the pattern of matrix cracks remains essentially the same, with the exception that the cracks tend to grow into the transverse tows and their opening displacement increases somewhat (see, for example, the cracks on the right side of Fig. 1(b)). These features are believed to be due to some matrix densification in the matrix segments contained between the cracks along the direction perpendicular to the cracks. Since this shrinkage is constrained by the (dense) fibers in the adjacent longitudinal tows, the shrinkage strain is accommodated by additional opening displacement of the cracks such that the net average strain is approximately zero. This hypothesis is supported by the measurements of the composite porosity, which indicated no significant change after any of the aging treatments. Higher-magnification SEM examinations of the matrix microstructure did not reveal any other obvious changes due to aging. Representative stress–strain curves for the as-processed and the aged specimens in the 0°/90° orientation are plotted in Fig. 2(a). The variations in the Young’s modulus, E, and the ultimate tensile strength, u, with aging temperature are summarized in Figs. 2(b) and (c). The only significant change is the slight increase in the modulus, from 60 GPa in the as-processed condition to 70 GPa after the 1200°C aging treatment. The tensile strength and the failure strain, εf , remained unchanged; the averages and standard Fig. 2. Effects of thermal aging on the tensile properties of the 0°/90° composite. The modulus was calculated from the slope of the initial linear portion of the stress–strain curve, over a stress range of 50 MPa. Fig. 3. Macrophotographs of the 0°/90° tensile specimens in two orthogonal views: (a) as-processed, (b) after aging for 1000 h at 1100°C, and (c) after aging for 1000 h at 1200°C. March 2002 Effects of Thermal Aging on the Mechanical Properties of a Porous-Matrix Ceramic Composite 597
598 Journal of the American Ceramic Society-Carelli et al. Vol. 85. No. 3 deviations from 14 tests are g.= 145+8 MPa and E=0.26 Following aging at 1200.C, the fiber tow failures in the 0/90 +.03%. Also shown for comparison in Fig. 2(c)is the retained specimens remained largely uncorrelated with one another at the strength of a comparable all-oxide ceramic composite consisting of macroscopic level(Fig. 3(c). However, there was a noticeable the same Nextel 720 fibers in an aluminosilicate matrix(in place ncrease in the correlation in the fiber failure sites within each tow of the mullite/alumina matrix used in the present study ), after the(Fig. 4(c). The most highly correlated failure sites appeared in same aging treatments. The strength of the aluminosilicate-basee small clusters, each containing perhaps 3-10 fibers. Additionall material decreases rapidly for aging temperatures beyond 1000%C the amount of matrix material remaining adhered to the fiber eportedly due to extensive densification of the matrix and an surfaces was significantly greater than that in the as-processed attendant loss in damage tolerance. material. These features clearly indicate that both the matrix and the In the 0/90 as-processed tensile specimens, the locations of fiber-matrix interface have been strengthened as a consequence of the the tow failures were uncorrelated with one another, as evident in aging treatment, thereby reducing somewhat the extent of damage the macrophotograph in Fig. 3(a). Indeed, the tow failure sites tolerance. Nevertheless, the effects do not appear to be sufficiently were offset by distances up to several centimeters along the large to noticeably alter the 0/90% composite strength. loading direction. larly, highly uncorrelated fiber fra In the +45 orientation, the effects of matrix strengthening on were obtained within each longitudinal tow. An example of a aging were more pronounced(Fig. 5). In all cases, the tensile broken tow near the fracture surface is shown in Fig. 4(a). a response was characterized by elastic-plastic behavior, reminis- articularly striking feature is the seemingly large lateral separa- cent of metal plasticity (albeit at lower levels of strain). The tion between adjacent fibers. This feature is somewhat misleading transition from elastic to plastic behavior was gradual and the in the sense that there are large longitudinal separations between ultimate tensile strength was controlled by a plastic instability the fiber fracture sites and hence many of the broken fibers within analogous to necking in metals, at an average strain of 0.32+ a broken tow are well outside the field of view when imaging the 0.03%, independent of aging treatment. By contrast, Youngs tow at even modest magnifications. These observations attest to modulus and the tensile strength increased dramatically following the efficacy of the matrix in mitigating stress concentrations aging, by as much as a factor of 2 at the highest aging temperature around fiber breaks and hence yielding damage tolerant behavior. This trend reaffirms that some strengthening of both the matrix and Higher-magnification SEM observations revealed only small amounts of matrix particulates remaining adhered to the fiber off-axis composite strength, such changes may be beneficial surface (Fig. 4(b). This result suggests that failure involves In the as-processed +45tensile specimens and the ones aged at debonding and sliding either at or very near the fiber-matrix temperatures up to 1100C, failure occurred mainly through the interface during fiber fracture, analogous to that in dense-matrix matrix and was accompanied by extensive interply delamination CFCCs with weak interphases. Similar features were observed on and fiber" scissoring, but with minimal fiber fracture(Figs. 6(a) the specimens that had been aged at either 1000 or 1100C(.g, and(b). A consequence of this"scissoring"is through-thi g.3(b) swelling in the region near the fracture surface. Following th 5 50 um 10m (d) 0 Fig. 4. SEM micrographs of the fracture surfaces of the 0/90 specimens: (a, b) in as-processed condition, and (c, d) after aging for 1000 h at 1200
deviations from 14 tests are u 145 8 MPa and εf 0.26% 0.03%. Also shown for comparison in Fig. 2(c) is the retained strength of a comparable all-oxide ceramic composite consisting of the same Nextel 720 fibers in an aluminosilicate matrix (in place of the mullite/alumina matrix used in the present study), after the same aging treatments.7 The strength of the aluminosilicate-based material decreases rapidly for aging temperatures beyond 1000°C, reportedly due to extensive densification of the matrix and an attendant loss in damage tolerance. In the 0°/90° as-processed tensile specimens, the locations of the tow failures were uncorrelated with one another, as evident in the macrophotograph in Fig. 3(a). Indeed, the tow failure sites were offset by distances up to several centimeters along the loading direction. Similarly, highly uncorrelated fiber fractures were obtained within each longitudinal tow. An example of a broken tow near the fracture surface is shown in Fig. 4(a). A particularly striking feature is the seemingly large lateral separation between adjacent fibers. This feature is somewhat misleading in the sense that there are large longitudinal separations between the fiber fracture sites and hence many of the broken fibers within a broken tow are well outside the field of view when imaging the tow at even modest magnifications. These observations attest to the efficacy of the matrix in mitigating stress concentrations around fiber breaks and hence yielding damage tolerant behavior. Higher-magnification SEM observations revealed only small amounts of matrix particulates remaining adhered to the fiber surface (Fig. 4(b)). This result suggests that failure involves debonding and sliding either at or very near the fiber–matrix interface during fiber fracture, analogous to that in dense-matrix CFCCs with weak interphases. Similar features were observed on the specimens that had been aged at either 1000° or 1100°C (e.g., Fig. 3(b)). Following aging at 1200°C, the fiber tow failures in the 0°/90° specimens remained largely uncorrelated with one another at the macroscopic level (Fig. 3(c)). However, there was a noticeable increase in the correlation in the fiber failure sites within each tow (Fig. 4(c)). The most highly correlated failure sites appeared in small clusters, each containing perhaps 3–10 fibers. Additionally, the amount of matrix material remaining adhered to the fiber surfaces was significantly greater than that in the as-processed material. These features clearly indicate that both the matrix and the fiber–matrix interface have been strengthened as a consequence of the aging treatment, thereby reducing somewhat the extent of damage tolerance. Nevertheless, the effects do not appear to be sufficiently large to noticeably alter the 0°/90° composite strength. In the 45° orientation, the effects of matrix strengthening on aging were more pronounced (Fig. 5). In all cases, the tensile response was characterized by elastic–plastic behavior, reminiscent of metal plasticity (albeit at lower levels of strain). The transition from elastic to plastic behavior was gradual and the ultimate tensile strength was controlled by a plastic instability analogous to necking in metals,4 at an average strain of 0.32 0.03%, independent of aging treatment. By contrast, Young’s modulus and the tensile strength increased dramatically following aging, by as much as a factor of 2 at the highest aging temperature. This trend reaffirms that some strengthening of both the matrix and the fiber–matrix interfaces occurs during aging. In the context of off-axis composite strength, such changes may be beneficial. In the as-processed 45° tensile specimens and the ones aged at temperatures up to 1100°C, failure occurred mainly through the matrix and was accompanied by extensive interply delamination and fiber “scissoring,” but with minimal fiber fracture (Figs. 6(a) and (b)). A consequence of this “scissoring” is through-thickness swelling in the region near the fracture surface. Following the Fig. 4. SEM micrographs of the fracture surfaces of the 0°/90° specimens: (a,b) in as-processed condition, and (c,d) after aging for 1000 h at 1200°C. 598 Journal of the American Ceramic Society—Carelli et al. Vol. 85, No. 3
March 2002 Efects of Thermal Aging on the Mechanical Properties of a Porous-Matrix Ceramic Composite 1000hat1200°C 1000hat1100°c As-processed 0 0.001 0.003 ens 9 Strength Modulus 200400600800 (b) Aging Temperature (C) 1000hat1200°c 1000hat1100°C (c) 3 mm 020.30. 050.60.7 Post-Localization Displacement (mm Fig. 6. Macrophotographs of the +45 tensile specimens in two orthog h at 1100°c,and(c Fig. 5. Effects of thermal aging on the tensile properties of the +45 after aging for 1000 h at 1200C features but with greater amounts of matrix on the fibers, consis- 1200C aging, there was a reduction in the extent of delamination. tent with the trend seen in the 0/90 orientation. Furthermore the failure mechanism now included extensive fiber Salient changes in the fracture properties in the t45orientation fracture(Fig. 6(c)). This feature is consistent with the increases in are revealed in the tensile response following strain localizatio the matrix and interface strengths due to aging, which improve the (beyond the load maximum). The load-displacement response in this effectiveness of load transfer from the matrix to the fibers and regime(Fig 5(c)) can be viewed as the traction law that would b hence increase the propensity for fiber fracture pertinent to the fracture process zone in a notched specimen and hence When viewed in the SEM (Fig. 7(a, b)), the fracture surfaces of provides information about the steady-state fracture energy. The the +45 as-processed specimens revealed somewhat greater results indicate that, as the matrix strength is initially increased, e. g amounts of matrix particulates adhered to the fiber surfaces than from the as-processed condition to the one obtained after 1000 h at that in the 0/90 orientation. It is surmised that these differences 1100oC, the stress-displacement response is simply shifted up to are related to the differences in the stress states at the fiber-matrix higher strength levels, commensurate with the increase in the ultimate interfaces in the two orientations, coupled with differences in the tensile strength. The fracture energy increases proportionately, by surfaces of the specimens aged at 1200C exhibited similar hS% However, as the matrix strength is increased further, e. g, after amounts of sliding that occur between adjacent fibers. The fracture 1200C aging, the stress-displacement response begins at a higher
1200°C aging, there was a reduction in the extent of delamination. Furthermore, the failure mechanism now included extensive fiber fracture (Fig. 6(c)). This feature is consistent with the increases in the matrix and interface strengths due to aging, which improve the effectiveness of load transfer from the matrix to the fibers and hence increase the propensity for fiber fracture. When viewed in the SEM (Fig. 7(a,b)), the fracture surfaces of the 45° as-processed specimens revealed somewhat greater amounts of matrix particulates adhered to the fiber surfaces than that in the 0°/90° orientation. It is surmised that these differences are related to the differences in the stress states at the fiber–matrix interfaces in the two orientations, coupled with differences in the amounts of sliding that occur between adjacent fibers. The fracture surfaces of the specimens aged at 1200°C exhibited similar features but with greater amounts of matrix on the fibers, consistent with the trend seen in the 0°/90° orientation. Salient changes in the fracture properties in the 45° orientation are revealed in the tensile response following strain localization (beyond the load maximum). The load–displacement response in this regime (Fig. 5(c)) can be viewed as the traction law that would be pertinent to the fracture process zone in a notched specimen and hence provides information about the steady-state fracture energy. The results indicate that, as the matrix strength is initially increased, e.g., from the as-processed condition to the one obtained after 1000 h at 1100°C, the stress–displacement response is simply shifted up to higher strength levels, commensurate with the increase in the ultimate tensile strength. The fracture energy increases proportionately, by 25%. However, as the matrix strength is increased further, e.g., after the 1200°C aging, the stress–displacement response begins at a higher Fig. 5. Effects of thermal aging on the tensile properties of the 45° composite. Fig. 6. Macrophotographs of the 45° tensile specimens in two orthogonal views: (a) as-processed, (b) after aging for 1000 h at 1100°C, and (c) after aging for 1000 h at 1200°C. March 2002 Effects of Thermal Aging on the Mechanical Properties of a Porous-Matrix Ceramic Composite 599