ournal JAm.Cm.soe,82m23575-8301999 Fabrication and Testing of Oxide/Oxide Microcomposites with Monazite and hibonite as Interlayers Triplicane A Parthasarathy,,t Emmanuel Boakye, ",t Michael K. Cinibulk, ,t and Melvin D. Petry Materials and Manufacturing Directorate, AFRL/MLIN, Air Force Research Laborator Wright-Patterson AFB, Ohio 45433-7817 Oxide/oxide microcomposites were fabricated and tested to there is only indirect evidence suggesting that these coatings valuate the effectiveness of monazite(LaPO)and hibonite actually prove effective in producing tough ceramic compos CaLo ites. This has been largely due to difficulties relating to fabri- matrix composites. For interlayer thicknesses of 0.3-0.5 cation of composites with an oxide matrix that is sufficiently m,both interlayers showed evidence of crack deflection; dense to test the coating concept. Oxide/oxide composites with however, debond lengths in hibonite-coated specimens were a sufficiently porous matrix and no interfacial treatment have limited to just a small fraction of the fiber diameter. Mona- been shown to demonstrate good composite behavior. , 12This zite-coated specimens showed multiple matrix cracks and resents a serious bottleneck in new interface conce extensive debonding at the coating/matrix interface. Com- a quick and inexpensive method to test concepts for oxide/ posite strengths were relatively high for both coatings, con- oxide CMCs would be a great aid in the development of oxi- sidering the fiber strength degradation during processi dation-resistant interface control for oxide/oxide CMcs. re- The strengths were greater than the calculated matrix cent work 3, 14 has ed that microcomposites(SCS6/BN racking stresses. However, the mean strengths were not Sici or Nicalon/BN/SiC13)can be used to simulate CMC significantly different from those of the control specimens, mechanical behavior of larger ites. In the present work although coated composites had higher Weibull moduli. he feasibility of fabricating oxide/oxide microcomposites and The lack of difference in strength is attributed to porosity using them to evaluate interface concepts is examined in the matrix. The results imply that matrix density needs The objectives of this work were(1)to develop a method to to be >85% to evaluate novel interface strategies reliabl fabricate oxide-matrix microcomposites, (2)to evaluate the mi- crocomposite test as a procedure to probe the effectiveness of interface concepts for oxide/oxide composites, and (3)to use the test to examine the effectiveness of hibonite and monazite HE use of ceramic-matrix composites(CMCs)in high in sapphire-reinforced Al2O3-matrix composites. Because a temperature structural applications is at present limited by dense matrix was a prerequisite for a valid test, a fiber that oxidation resistance at intermediate temperatures(700 could withstand processing temperatures >1300C was re 900C)and at temperatures >1200C. The poor oxidation re uired: thus, sapphire was selected. From the known coeffi- sistance of carbon and BN (used as interlayers in these cor cient of thermal expansion(CTE) of sapphire and polycrys- posites) has resulted in investigations with the objective of talline Al2O3, 6 the fiber was predicted (using the model of dentifying alternatives. Two of the more extensively studied Budiansky et al. 17) to be in residual axial tension(175 MPa oxide interlayers are lanthanum phosphate(the mineral mona- the matrix to be in residual compression (-125 MPa), and the zite, LaPO4)and calcium hexaluminate(the mineral hibonite, radial stress at the interface to be under compression(-15 CaAl12O19). Both of these compounds were first suggested as sible interlayers in oxide/oxide composites by Morgan and II. Experiments Marshall, 1, 2 Monazite has been reported to result in crack deflection in indentation tests and in laminates, as well as in chosen for this work. preliminary studies used an older stock of bonds weakly to Al,O3 and the fact that it is stable with Al,O lower-grade (grade-D) fiber( that had periodic kinks along the at s1500.C make it an attractive material for interlayer in fiber length). These fibers showed significant fiber strengt oxide/oxide composites. The hexaluminate is attractive be- degradation during processing; hence, later studies used fibers ause it has structural similarities to graphite and BN, espe- of a higher grade(grade A), which had better retained strengt cially a high anisotropy in fracture resistance. The hexalum nate shows a high preference for basal cleavage with a fracture cessing steps. For similar reasons, the Al2O3 powder used to anisotropy of >5. Extensive work on the processing of fibe consolidate the matrix for the microcomposites had to be modi- coatings of calcium hexaluminate has been reported fied after preliminary trials found that matrix density was in- sufficient. The matrix powder was switched from A-1000TM While significant progress has been made in obtaining uni- Alcoa, Pittsburgh, PA)to a better sintering powder, TM- form coatings of monazite and hibonite on mor DARTM (Tai-Mei Chemicals, Japan), which yielded higher ma trix densities. Both powders were of 99.99% purity. Several interesting results were found from both types of matrices and fibers, and, thus, all of the results were reported here, specify A. Evans--contributing editor ing what fiber and matrix were used The coatings that were put on the fibers to study the effec- tiveness of interface design were monazite and hibonite. The procedures used in obtaining the fiber coatings and the char- sACm2图以2319 acterization of the coatings have been described elsewhere: 6, Ig however, a brief outline is given below along with the proce- UES, Inc, Dayton, Ohio 45432 dure used in the present work for consolidating 3575
Fabrication and Testing of Oxide/Oxide Microcomposites with Monazite and Hibonite as Interlayers Triplicane A. Parthasarathy,*,† Emmanuel Boakye,*,† Michael K. Cinibulk,*,† and Melvin D. Petry† Materials and Manufacturing Directorate, AFRL/MLLN, Air Force Research Laboratory, Wright-Patterson AFB, Ohio 45433-7817 Oxide/oxide microcomposites were fabricated and tested to evaluate the effectiveness of monazite (LaPO4) and hibonite (CaAl12O19) as interlayers in sapphire-reinforced Al2O3- matrix composites. For interlayer thicknesses of 0.3–0.5 µm, both interlayers showed evidence of crack deflection; however, debond lengths in hibonite-coated specimens were limited to just a small fraction of the fiber diameter. Monazite-coated specimens showed multiple matrix cracks and extensive debonding at the coating/matrix interface. Composite strengths were relatively high for both coatings, considering the fiber strength degradation during processing. The strengths were greater than the calculated matrix cracking stresses. However, the mean strengths were not significantly different from those of the control specimens, although coated composites had higher Weibull moduli. The lack of difference in strength is attributed to porosity in the matrix. The results imply that matrix density needs to be >85% to evaluate novel interface strategies reliably. I. Introduction THE use of ceramic-matrix composites (CMCs) in hightemperature structural applications is at present limited by oxidation resistance at intermediate temperatures (700°– 900°C) and at temperatures $1200°C. The poor oxidation resistance of carbon and BN (used as interlayers in these composites) has resulted in investigations with the objective of identifying alternatives. Two of the more extensively studied oxide interlayers are lanthanum phosphate (the mineral monazite, LaPO4) and calcium hexaluminate (the mineral hibonite, CaAl12O19). Both of these compounds were first suggested as possible interlayers in oxide/oxide composites by Morgan and Marshall.1,2 Monazite has been reported to result in crack deflection in indentation tests2 and in laminates,3 as well as in fiber pushout tests.2,4 This observation that monazite, LaPO4, bonds weakly to Al2O3 and the fact that it is stable with Al2O3 at #1500°C make it an attractive material for interlayer in oxide/oxide composites. The hexaluminate is attractive because it has structural similarities to graphite and BN, especially a high anisotropy in fracture resistance. The hexaluminate shows a high preference for basal cleavage with a fracture anisotropy of >5.5 Extensive work on the processing of fiber coatings of calcium hexaluminates has been reported recently.1,6–10 While significant progress has been made in obtaining uniform coatings of monazite and hibonite on monofilaments, there is only indirect evidence suggesting that these coatings actually prove effective in producing tough ceramic composites. This has been largely due to difficulties relating to fabrication of composites with an oxide matrix that is sufficiently dense to test the coating concept. Oxide/oxide composites with a sufficiently porous matrix and no interfacial treatment have been shown to demonstrate good composite behavior.11,12 This presents a serious bottleneck in testing new interface concepts. A quick and inexpensive method to test concepts for oxide/ oxide CMCs would be a great aid in the development of oxidation-resistant interface control for oxide/oxide CMCs. Recent work13,14 has suggested that microcomposites (SCS6/BN/ SiC14 or Nicalon/BN/SiC13) can be used to simulate CMC mechanical behavior of larger composites. In the present work, the feasibility of fabricating oxide/oxide microcomposites and using them to evaluate interface concepts is examined. The objectives of this work were (1) to develop a method to fabricate oxide-matrix microcomposites, (2) to evaluate the microcomposite test as a procedure to probe the effectiveness of interface concepts for oxide/oxide composites, and (3) to use the test to examine the effectiveness of hibonite and monazite in sapphire-reinforced Al2O3-matrix composites. Because a dense matrix was a prerequisite for a valid test, a fiber that could withstand processing temperatures >1300°C was required; thus, sapphire was selected. From the known coefficient of thermal expansion (CTE) of sapphire15 and polycrystalline Al2O3, 16 the fiber was predicted (using the model of Budiansky et al.17) to be in residual axial tension (∼175 MPa), the matrix to be in residual compression (∼125 MPa), and the radial stress at the interface to be under compression (∼15 MPa). II. Experiments c-axis sapphire fiber (Sapphikon, Inc., Milford, NH) was chosen for this work. Preliminary studies used an older stock of lower-grade (grade-D) fiber (that had periodic kinks along the fiber length). These fibers showed significant fiber strength degradation during processing; hence, later studies used fibers of a higher grade (grade A), which had better retained strength after heat treatments that simulated the microcomposite processing steps. For similar reasons, the Al2O3 powder used to consolidate the matrix for the microcomposites had to be modified after preliminary trials found that matrix density was insufficient. The matrix powder was switched from A-1000™ (Alcoa, Pittsburgh, PA) to a better sintering powder, TMDAR™ (Tai-Mei Chemicals, Japan), which yielded higher matrix densities. Both powders were of 99.99% purity. Several interesting results were found from both types of matrices and fibers, and, thus, all of the results were reported here, specifying what fiber and matrix were used. The coatings that were put on the fibers to study the effectiveness of interface design were monazite and hibonite. The procedures used in obtaining the fiber coatings and the characterization of the coatings have been described elsewhere;6,18 however, a brief outline is given below along with the procedure used in the present work for consolidating the matrix. A. Evans—contributing editor Manuscript No. 189965. Received August 10, 1998; approved June 25, 1999. Supported by U.S. Air Force under Contract No. F33615-96-C-5258. *Member, American Ceramic Society. † UES, Inc., Dayton, Ohio 45432 J. Am. Ceram. Soc., 82 [12] 3575–83 (1999) Journal 3575
Journal of the American Ceramic Society-Parthasarathy et al. ( Filament Coating 1450°C. because a fev ng trials at 1500C showed that The approach used to deposit coating on fibers was the use spread around the fiber, viz of a sol, which wetted the fiber and transformed to the required beaded up or was lost crystalline compound upon heating above a crystallization tem-(3) Matrix Density perature. A continuous vertical coater developed by Hay and Hermes% was used to obtain coatings on a continuous fiber To estimate matrix density, specimens of microcomposites The coating thickness in this process could be varied by ad were infiltrated and mounted in an epoxy matrix to obtai justing the viscosity of the sol and the fiber transportation rate, longitudinal cross sections. The images of these sections ob- the thickness increasing with both variables. However, neither tained using an SEM(Model 360 FE, Leica)were analyzed using ADOBE PHOTOSHOP 4.0TM(Adobe Systems, Inc, CA)and could be varied indefinitely. Therefore, multiple runs through NIH IMAGETM(Scion Corp, MD)analysis software. First, the mages were converted from gray-scale to binary(black/white R tSS. Typically, 10-15 passes were made, and the coatings mages. Analysis of the resulting images was performed to obtain the areal fraction of pores. Conversion of an image to a ere characterized for coverage, thickness, and morphol- binary image required the input of a threshold. This procedure Ising a scanning electron microscope (SEM; Model 360 Leica, Deerfield, IL) gave pore area fraction values that varied by <10%, depending The sol used to obtain monazite coatings was a 60 g/L rhab- of the threshold, the procedure gave very reproducible results dophane sol (LaPO4 xH2O). The rhabdophane sol was formed Isually within 1%)for the various regions of a in aqueous solution by reacting lanthanum nitrate and diam- made it necessary to select an optimal threshold. This value terization of the sol have been reported in earlier work. 18 The was determined using images of sapphire as control to yield 0% hibonite(CaAl2O1g) sols were prepared by addition of cal porosity, the minimum value required to get O% pore density in cium acetate to a diluted commercial Al,O, sol (DISPERAL sapphire was used as the threshold. With this procedure, pore area fractions were measured to be within +3%(one standard Sol 10/2, CONDEA Chemie GmbH, Hamburg, Germany). Fi- nal sol concentrations were between 50 and 110 He deviation). The results must be taken as an upper bound on the treatment at 1400%C for 2 h resulted in complete conversion to W porosity level, because matrix grain pullout during polishing the desired phase of CaAl12O19, with the hibonite grains ori- ented with their weak planes nearly parallel to the fiber surface (4 Composite Testing For further details on the coating procedure and characteriza- The composites were tested in a configuration similar to that tion(SEM, X-ray, TEM, etc )of the coatings on sapphire fi- suggested by ASTM for tensile tests on single filaments.The ers. the reader is referred to ref. 6 tested region(gauge length) of the microcomposite was 1 in (2.5 cm)long; the bare ends(filaments with no matrix on)were ( Composite Processing also I in. in length on either side. The bare ends were attached The microcomposites were processed by coating the fib to cardboard using a room-temperature fast-curing epoxy. The ith a slurry containing the matrix powder particles and then epoxy was cured overnight before testing. The specimens were intering the matrix in air at 1450.C for 2 h. The matrix slurr gripped at the regions of the cardboard away from the epoxy stained a mixture of the Al,O powder and a binder(B- Grip slippage was observed with some of the specimens(typi 73305, Ferro, Cleveland, OH)along with ethanol to control the ally the higher-strength ones), if the ends of the microcom- viscosity of the slurry. A typical slurry had a composition of 30 g of Al,O3, 25 mL of a binder solution, and 20 mL of ethanol The specimens were tested under tension using a universal and was mixed by ball milling for 24 h. Monazite-and hibo- testing machine(Model Sintec 20/G, MTS, Inc, MN)con- nite-coated sapphire filaments as well as the uncoated filaments trolled through a computer. The loading rate was fixed at 127 were cut into segments 3 in. (7.6 cm)long. Microcomposites of um/min. The compliance of the machine plus cardboard was e/coating/Al,O, and the control specimens of sapphire easured and subtracted from the raw data to check for non- Al, O, were made by pultrusion. a syringe tipped with an 18 gauge hypodermic needle was filled with an Al,O3 slurry, and inearity in the load-displacement traces. The tests were con- ducted to failure then a coated filament was inserted into the syringe. The slurr vas coated onto the filament through pultrusion by slowl Ill. Results and Analysis withdrawing the filament as the plunger was pushed from the other end. The ends(I in(2.5 cm)on either side) of the (Fiber Coatings filaments were wiped to remove the matrix and provide bare ends for better gripping during tensile testing. The slurry described in detail elsewhere. 6 The coatings were a gy are coated filaments were then heat-treated at 1450 C for 2 h in air after a single pass, but, with multiple s, coating thick to sinter the matrix. The fiber was usually off-center, but there nesses of nearly I um could be obtained. The hibonite grains was a minimum of 5 um of matrix around the thinnest regions that formed always had a special orientation relative to the fiber The average thickness of the sintered matrix was 10 um. This axis, with their basal planes aligned parallel to the fiber surface yielded a matrix volume fraction of 0.25: increasing it further (Note that this is the preferred orientation to effect deflection of was made difficult by the shrinkage associated with drying, a matrix crack runn rpendicular to the fiber axis. )The binder burnout, and sintering of the slurry. The stresses asso grade-D fibers were used for all trials using hibonite ciated with CTE mismatch, along with the fiber being off- For the monazite coatings, both grades of fiber were used center, resulted in specimens having a slight curvature with a SEM observations showed the coating thicknesses to be 0.25 radius of curvature of-300 mm. This curvature made it diffi- um for the 10-pass multiple coatings on the grade-D filaments and 0.7 um for the 15-pass coatings on the grade - A filaments The two sources of Al O, that were used in the experiments A SEM image of the surface of the coating after heat treatment 则MA9yhmp02mBh和mF1( The coating is d and polycrystalline, with a grain size of the order of I um lower-quality grade-D sapphire filaments were coated with the however, elongated wormlike gaps are present, possibly caused A1000 powder-based slurry, and the higher-quality grade filaments were coated with the TM-DAR powder-based slurry The TM-DAR powder yielded higher matrix densities than the A1000 powder. The sintering temperature was limited to 需一知览
(1) Filament Coating The approach used to deposit coating on fibers was the use of a sol, which wetted the fiber and transformed to the required crystalline compound upon heating above a crystallization temperature. A continuous vertical coater developed by Hay and Hermes19 was used to obtain coatings on a continuous fiber. The coating thickness in this process could be varied by adjusting the viscosity of the sol and the fiber transportation rate, the thickness increasing with both variables. However, neither could be varied indefinitely. Therefore, multiple runs through the fiber coater had to be used to obtain sufficiently thick coatings. Typically, 10–15 passes were made, and the coatings were fired in-line during each pass at 1400°C in air. The coatings were characterized for coverage, thickness, and morphology using a scanning electron microscope (SEM; Model 360 FE, Leica, Deerfield, IL). The sol used to obtain monazite coatings was a 60 g/L rhabdophane sol (LaPO4?xH2O). The rhabdophane sol was formed in aqueous solution by reacting lanthanum nitrate and diammonium hydrogen phosphtate. Further details on the characterization of the sol have been reported in earlier work.18 The hibonite (CaAl12O19) sols were prepared by addition of calcium acetate to a diluted commercial Al2O3 sol (DISPERAL Sol 10/2, CONDEA Chemie GmbH, Hamburg, Germany). Final sol concentrations were between 50 and 110 g/L. Heat treatment at 1400°C for 2 h resulted in complete conversion to the desired phase of CaAl12O19, with the hibonite grains oriented with their weak planes nearly parallel to the fiber surface. For further details on the coating procedure and characterization (SEM, X-ray, TEM, etc.) of the coatings on sapphire fibers, the reader is referred to Ref. 6. (2) Composite Processing The microcomposites were processed by coating the fibers with a slurry containing the matrix powder particles and then sintering the matrix in air at 1450°C for 2 h. The matrix slurry contained a mixture of the Al2O3 powder and a binder (B- 73305, Ferro, Cleveland, OH) along with ethanol to control the viscosity of the slurry. A typical slurry had a composition of 30 g of Al2O3, 25 mL of a binder solution, and 20 mL of ethanol, and was mixed by ball milling for 24 h. Monazite- and hibonite-coated sapphire filaments as well as the uncoated filaments were cut into segments 3 in. (7.6 cm) long. Microcomposites of sapphire/coating/Al2O3 and the control specimens of sapphire/ Al2O3 were made by pultrusion. A syringe tipped with an 18 gauge hypodermic needle was filled with an Al2O3 slurry, and then a coated filament was inserted into the syringe. The slurry was coated onto the filament through pultrusion by slowly withdrawing the filament as the plunger was pushed from the other end. The ends (1 in. (2.5 cm) on either side) of the filaments were wiped to remove the matrix and provide bare ends for better gripping during tensile testing. The slurrycoated filaments were then heat-treated at 1450°C for 2 h in air to sinter the matrix. The fiber was usually off-center, but there was a minimum of 5 mm of matrix around the thinnest regions. The average thickness of the sintered matrix was 10 mm. This yielded a matrix volume fraction of 0.25; increasing it further was made difficult by the shrinkage associated with drying, binder burnout, and sintering of the slurry. The stresses associated with CTE mismatch, along with the fiber being offcenter, resulted in specimens having a slight curvature with a radius of curvature of ∼300 mm. This curvature made it difficult to deposit multiple coatings of the matrix. The two sources of Al2O3 that were used in the experiments were A1000 powder with a mean particle size of 0.5 mm and TM-DAR powder with a mean particle size of 0.2 mm. Both had purity >99.99% with <10 ppm silicon and magnesium. The lower-quality grade-D sapphire filaments were coated with the A1000 powder-based slurry, and the higher-quality grade-A filaments were coated with the TM-DAR powder-based slurry. The TM-DAR powder yielded higher matrix densities than the A1000 powder. The sintering temperature was limited to 1450°C, because a few sintering trials at 1500°C showed that the monazite lost its uniform spread around the fiber, viz., beaded up or was lost through reaction. (3) Matrix Density To estimate matrix density, specimens of microcomposites were infiltrated and mounted in an epoxy matrix to obtain longitudinal cross sections. The images of these sections obtained using an SEM (Model 360 FE, Leica) were analyzed using ADOBE PHOTOSHOP 4.0™ (Adobe Systems, Inc., CA) and NIH IMAGE™ (Scion Corp., MD) analysis software. First, the images were converted from gray-scale to binary (black/white) images. Analysis of the resulting images was performed to obtain the areal fraction of pores. Conversion of an image to a binary image required the input of a threshold. This procedure gave pore area fraction values that varied by #10%, depending on the choice of the threshold. Although for any chosen value of the threshold, the procedure gave very reproducible results (usually within 1%) for the various regions of a specimen. This made it necessary to select an optimal threshold. This value was determined using images of sapphire as control to yield 0% porosity; the minimum value required to get 0% pore density in sapphire was used as the threshold. With this procedure, pore area fractions were measured to be within ±3% (one standard deviation). The results must be taken as an upper bound on the porosity level, because matrix grain pullout during polishing was unavoidable. (4) Composite Testing The composites were tested in a configuration similar to that suggested by ASTM† for tensile tests on single filaments. The tested region (gauge length) of the microcomposite was 1 in. (2.5 cm) long; the bare ends (filaments with no matrix on) were also 1 in. in length on either side. The bare ends were attached to cardboard using a room-temperature fast-curing epoxy. The epoxy was cured overnight before testing. The specimens were gripped at the regions of the cardboard away from the epoxy. Grip slippage was observed with some of the specimens (typically the higher-strength ones), if the ends of the microcomposite were not bare. The specimens were tested under tension using a universal testing machine (Model Sintec 20/G, MTS, Inc., MN) controlled through a computer. The loading rate was fixed at 127 mm/min. The compliance of the machine plus cardboard was measured and subtracted from the raw data to check for nonlinearity in the load–displacement traces. The tests were conducted to failure. III. Results and Analysis (1) Fiber Coatings Details on the hibonite coating structure and morphology are described in detail elsewhere.6 The coatings were 0.25 mm after a single pass, but, with multiple passes, coating thicknesses of nearly 1 mm could be obtained. The hibonite grains that formed always had a special orientation relative to the fiber axis, with their basal planes aligned parallel to the fiber surface. (Note that this is the preferred orientation to effect deflection of a matrix crack running perpendicular to the fiber axis.) The grade-D fibers were used for all trials using hibonite coatings. For the monazite coatings, both grades of fiber were used. SEM observations showed the coating thicknesses to be 0.25 mm for the 10-pass multiple coatings on the grade-D filaments and 0.7 mm for the 15-pass coatings on the grade-A filaments. A SEM image of the surface of the coating after heat treatment in air for 2 h at 1450°C is shown in Fig. 1(a); a cross section of a fractured fiber is shown in Fig. 1(b). The coating is dense and polycrystalline, with a grain size of the order of 1 mm; however, elongated wormlike gaps are present, possibly caused † “Standard Test Method for Tensile Strength and Young’s Modulus for HighModulus Single Filament Materials,” ASTM Designation D3379-75. American Society for Testing and Materials, West Conshohocken, PA. 3576 Journal of the American Ceramic Society—Parthasarathy et al. Vol. 82, No. 12
December 1999 Fabrication and Testing of Oxide/Oxide Microcomposites with Monazite and Hibonite as Interlayers (a) (b) Matrix TM-DARV Monazite Coating sapphire obtained on Fig. 1. SEM micrograph of the mona ed sapphire fiber after a 1450C, 2 h heat treatment tha hedule. (a) Monazite coating is nea 二吧邮 he matirx-processing axed, and coverage is nearly complete, but some gaps rem tting in cross section Is continuous with a thickness of nearly 0.5 um matrix quite likely prevented any further growth. Fracture sur- from shrinkage during sintering. No differences in coating faces of the grade-D/A1000 and grade-A/TM-DAR composites morphology are observed between the monazite coatings on the revealed that the fracture originated from the surface of the grade-D and grade-A sapphire filaments fiber in all of the control specimens. This could have resulted from matrix cracks propagated through the fiber (2) Microcomposites b) Hibonite: The fracture surfaces of the hibonite- (A Matrix Density: SEM images of the as-sintered sur- interface-containing microcomposites are shown in Fig. 5 faces of both types of matrices are shown in Fig. 2. The ma These composites consisted of grade-D sapphire filaments re- trices are not fully dense in either material; however, the TM- inforced with A1000 slurry matrix, and, thus, the fracture sur- DAR powder yields a denser matrix than the A1000 powder faces should be compared with those in Fig. 4(a). The lower Figure 3 shows SEM images of a longitudinal cross section of magnification image( Fig. 5(a)) shows that the fracture surface one of the specimens fabricated using the TM-DAR powder. takes a short step at the interface region, but there is no indi- Such images were used to determine the mean matrix density cation of debonding. At the higher magnifications( Fig. 5(b)), as detailed earlier. The sapphire regions, such as those shown the step at the interface region can be attributed to the easy in Fig 3(a), were used to calibrate the procedure. Matrix den- cleaving hibonite, providing a deflection of the crack. The sities after sintering thus obtained were-80% and -85%(with hibonite phase is recognized from its platelet-type morphology a standard deviation of *3%) for the composites made with (seen more clearly in Fig. 5(c), and the chemistry is confirmed A1000 and TM-DAR AL2O3 slurries, respectivel from EDS. These fracture surfaces suggest that hibonite does (B) Fracture Surfaces: (a) Control: The fractured sur- cleave readily in the presence of a matrix crack, that multip faces of the control specimens(uncoated fiber in a matrix) deflections occur within the hibonite coating, but that the de- showed that there was no influence of the interface on the flected crack is not sustained parallel to the fiber surface fo fracture path. Figure 4 shows the SEM images of the fractured distances longer than a few micrometers surfaces of (a) grade-D sapphire-reinforced A1000 matrix m (c) Monazite: The fracutre surfaces of the monazite crocomposite and(b) grade-A sapphire-reinforced TM-DAR containing composites showed extensive evidence for crack matrix composite. The images show that fracture in the matrix deflection in both grade-D/A1000(Fig. 6(a))and grade-A/TM- and in the fiber are coplanar in both composites. It is also clear DAR (Fig. 6(b)composites. It was often found that the matrix that there is no signif ificant growth of the fiber into the matrix, dislodged from the composite from one side, the in- contrary to what might have been expected from known grain terfacial region. As seen from Fig. 6(a), this partial dislodging owth kinetics of Al O3 at these temperatures. The fiber po of the matrix, resembling a banana peel very characteristi sibly grew up to a distance of one grain diameter(-0.5 um), of these composites. Some of this could be attributed to the but the pores at the interface and/or the low density of the slightly bent shape of the composite in the as-sintered state, but
from shrinkage during sintering. No differences in coating morphology are observed between the monazite coatings on the grade-D and grade-A sapphire filaments. (2) Microcomposites (A) Matrix Density: SEM images of the as-sintered surfaces of both types of matrices are shown in Fig. 2. The matrices are not fully dense in either material; however, the TMDAR powder yields a denser matrix than the A1000 powder. Figure 3 shows SEM images of a longitudinal cross section of one of the specimens fabricated using the TM-DAR powder. Such images were used to determine the mean matrix density, as detailed earlier. The sapphire regions, such as those shown in Fig. 3(a), were used to calibrate the procedure. Matrix densities after sintering thus obtained were ∼80% and ∼85% (with a standard deviation of ±3%) for the composites made with A1000 and TM-DAR Al2O3 slurries, respectively. (B) Fracture Surfaces: (a) Control: The fractured surfaces of the control specimens (uncoated fiber in a matrix) showed that there was no influence of the interface on the fracture path. Figure 4 shows the SEM images of the fractured surfaces of (a) grade-D sapphire-reinforced A1000 matrix microcomposite and (b) grade-A sapphire-reinforced TM-DAR matrix composite. The images show that fracture in the matrix and in the fiber are coplanar in both composites. It is also clear that there is no significant growth of the fiber into the matrix, contrary to what might have been expected from known grain growth kinetics of Al2O3 at these temperatures. The fiber possibly grew up to a distance of one grain diameter (∼0.5 mm), but the pores at the interface and/or the low density of the matrix quite likely prevented any further growth. Fracture surfaces of the grade-D/A1000 and grade-A/TM-DAR composites revealed that the fracture originated from the surface of the fiber in all of the control specimens. This could have resulted from matrix cracks propagated through the fiber. (b) Hibonite: The fracture surfaces of the hiboniteinterface-containing microcomposites are shown in Fig. 5. These composites consisted of grade-D sapphire filaments reinforced with A1000 slurry matrix, and, thus, the fracture surfaces should be compared with those in Fig. 4(a). The lower magnification image (Fig. 5(a)) shows that the fracture surface takes a short step at the interface region, but there is no indication of debonding. At the higher magnifications (Fig. 5(b)), the step at the interface region can be attributed to the easycleaving hibonite, providing a deflection of the crack. The hibonite phase is recognized from its platelet-type morphology (seen more clearly in Fig. 5(c)), and the chemistry is confirmed from EDS. These fracture surfaces suggest that hibonite does cleave readily in the presence of a matrix crack, that multiple deflections occur within the hibonite coating, but that the deflected crack is not sustained parallel to the fiber surface for distances longer than a few micrometers. (c) Monazite: The fracutre surfaces of the monazitecontaining composites showed extensive evidence for crack deflection in both grade-D/A1000 (Fig. 6(a)) and grade-A/TMDAR (Fig. 6(b)) composites. It was often found that the matrix dislodged from the composite from one side, exposing the interfacial region. As seen from Fig. 6(a), this partial dislodging of the matrix, resembling a banana peel, was very characteristic of these composites. Some of this could be attributed to the slightly bent shape of the composite in the as-sintered state, but Fig. 1. SEM micrograph of the monazite-coated sapphire fiber after a 1450°C, 2 h heat treatment that simulates the matirx-processing schedule. (a) Monazite coating is nearly equiaxed, and coverage is nearly complete, but some gaps remain. (b) Coating in cross section is continuous with a thickness of nearly 0.5 mm. Fig. 2. SEM micrographs showing density of the matrix obtained on sapphire fibers using two different Al2O3 powders, (a) A-1000 and (b) TM-DAR sintered at 1450°C for 2 h. Neither matrix is fully dense, but the matrix formed from TM-DAR powder is more dense. December 1999 Fabrication and Testing of Oxide/Oxide Microcomposites with Monazite and Hibonite as Interlayers 3577
Journal of the American Ceramic Society-Parthasarathy et al. Vol. 82. No. 12 (a) 10 00 Monazite Sapphire μn Sapphire 点 Fig.3.(a),(b) cross section of an epoxy-infiltrated and Fig. 4. surfaces of control microcomposites(no interfacial 1000 matrix and(b) TM-DAR matrix No significant mounted microcom icated using TM-DAR powder, shown at growth of pphire fiber into the matrix is seen. Pores at the wo different mag interface analysis to estimat nsities of the matrices. Sapphire regions from matrix porosity) presumably limit any such such as shown in(a), were used to calibrate the metho e grips could also result in ty of the tensile hoop residual stresses in the matrix could contribute to same order of magnitude, as has been pointed out by Morscher 7(a). The matrix cracks were periodic along the length of the et al i In the case of nicalon inforced VD sic matrix fiber, as seen from Fig. 6(b). The periodic spacing of the matrix diameter of the fiber gave rise to much bigger changes in of the order of a millimeter. Finally, the fracture origin( Fig. 7(b) was usually found to be from the surface of the fiber, as was the case with control specimens The lack of significant nonlinearity precludes the possibility When the exposed interface region was examined at higher of extracting interface properties through unload-reload hys- magnifications, it was found that debonding occurred at the teresis, as was possible with the Nicalon/SiC microcompos matrix/monazite interface in both the grade-D/A1000 and the only quantitative parameter that could grade-A/TM-DAR composites. Figure 8(a) shows the evaluation was the mean and Weibull modulus of the ultimate debonded surface of the Aiooo matrix composites. Particles of failure load of the fibers and microcomposites. Because the al cross-sectional area of the the monazite with some Al2O, particles still sticking to the ength of the specimen, and because the matrix was not ex- the matrix Al,O3 monazite. EDS confirmed the chemistry of monazite. The im pected to carry any significant load at failure, all loads were pression made by the Al,O3 matrix grains is clear in Fig. 8(b) divided by the fiber cross-sectional area for comparisons. The which shows the debonded surface on the fiber of a grade-A strengths thus calculated were plotted on a Weibull distribution TM-DAR composite plot as determined using the usual equations (C) Strengths of Fibers, Coated Fibers, and Micro ites: All the specimens tested exhibited a linear load displacement relationship until failure. Any deviation from lin- machine compliance was 4.5 um/N, and maximum resolut, P (1) on the compliance measured over 10 N was estimated to be 0. 4 Hm/N. Assuming that the matrix had dulus of 300 The data were sorted and ranked in order. and the GPa(modulus of 85% dense Al, O3 20)and that the fiber's was robability estimator given above was ted and plotted as 450 GPa, the composite modulus was 412.5 GPa. The differ- In -In(1-P)] versus In(o). From and intercept, ence in compliance between the composite and the fiber was and o were calculated calculated to be -0.9 um/N, close to the resolution of the (a) phire D/A1000-Matrix Micro-CMCs: Figure 9(a) measurement. This was further complicated by the fact that Weibull distribution of strengths of the sapphire D
tensile hoop residual stresses in the matrix could contribute to this also, as evidenced from matrix cracks, as shown in Fig. 7(a). The matrix cracks were periodic along the length of the fiber, as seen from Fig. 6(b). The periodic spacing of the matrix cracks was of the order of a millimeter. Finally, the fracture origin (Fig. 7(b)) was usually found to be from the surface of the fiber, as was the case with control specimens. When the exposed interface region was examined at higher magnifications, it was found that debonding occurred at the matrix/monazite interface in both the grade-D/A1000 and the grade-A/TM-DAR composites. Figure 8(a) shows the debonded surface of the A1000 matrix composites. Particles of the matrix Al2O3 grains (now dislodged) made impressions on the monazite with some Al2O3 particles still sticking to the monazite. EDS confirmed the chemistry of monazite. The impression made by the Al2O3 matrix grains is clear in Fig. 8(b), which shows the debonded surface on the fiber of a grade-A/ TM-DAR composite. (C) Strengths of Fibers, Coated Fibers, and Microcomposites: All the specimens tested exhibited a linear load– displacement relationship until failure. Any deviation from linearity was within the error of measurement of the apparatus; machine compliance was 4.5 mm/N, and maximum resolution on the compliance measured over 10 N was estimated to be ∼0.4 mm/N. Assuming that the matrix had a modulus of 300 GPa (modulus of 85% dense Al2O3 20) and that the fiber’s was 450 GPa, the composite modulus was 412.5 GPa. The difference in compliance between the composite and the fiber was calculated to be ∼0.9 mm/N, close to the resolution of the measurement. This was further complicated by the fact that slippage at the grips could also result in nonlinearity of the same order of magnitude, as has been pointed out by Morscher et al.14 In the case of Nicalon-reinforced CVD SiC matrix,13 the higher stiffness of the CVI SiC matrix and the smaller diameter of the fiber gave rise to much bigger changes in compliance. The lack of significant nonlinearity precludes the possibility of extracting interface properties through unload–reload hysteresis, as was possible with the Nicalon/SiC microcomposites.21 The only quantitative parameter that could be used for evaluation was the mean and Weibull modulus of the ultimate failure load of the fibers and microcomposites. Because the actual cross-sectional area of the composites varied along the length of the specimen, and because the matrix was not expected to carry any significant load at failure, all loads were divided by the fiber cross-sectional area for comparisons. The strengths thus calculated were plotted on a Weibull distribution plot as determined using the usual equations Pf = 1 − expF−S s so D m G Pf = S i − 0.5 N D (1) The data were sorted and ranked in ascending order, and the probability estimator given above was calculated and plotted as ln [−ln (1 − Pf )] versus ln (s). From the slope and intercept, m and so were calculated. (a) Sapphire D/A1000-Matrix Micro-CMCs: Figure 9(a) shows the Weibull distribution of strengths of the sapphire D Fig. 3. (a), (b) Longitudinal cross section of an epoxy-infiltrated and mounted microcomposite fabricated using TM-DAR powder, shown at two different magnifications. Similar images were used in an image analysis to estimate the densities of the matrices. Sapphire regions, such as shown in (a), were used to calibrate the method. Fig. 4. Fracture surfaces of control microcomposites (no interfacial layers) with (a) A1000 matrix and (b) TM-DAR matrix. No significant growth of the sapphire fiber into the matrix is seen. Pores at the interface (arising from matrix porosity) presumably limit any such growth. 3578 Journal of the American Ceramic Society—Parthasarathy et al. Vol. 82, No. 12
December 1999 Fabrication and Testing of Oxide/Oxide Microcomposites with Monazite and Hibonite as Interlayers ( Sapphire lonazitc-coated Ire Debonded surface Hibonite latrix Dislodged) 0 (b) (b) Matri TM-DAR Debonded Surfac Malrix dislodged Hibonite Fig. 6. SEM macrophotographs of the monazite-coated microcom Al000 powder and(b) TM-DAR powder In(b), the matrix falls of intermittently along the length of the microcomposite, revealing peri- Sapphire odic matrix cracks Hibonitc onazite composites showed significant debonding, the lack of difference in strength between the control and monazite-coate composite was puzzling. To determine if this was because of poor matrix density, a second set of control and monazite- coated samples were sintered at 1500oC for 2 h to improve matrix density. There was only a marginal improvement in matrix density, and essentially the same result was obtained Lumina/ ( Fig 9(d) with some loss in strength, which could be ratio- nalized easily as due to loss of fiber strength following the 1500° heat treatment ferent magnifications revealing crack deflections within the hiboni (b) Sapphire A/ Monazite/TM-DAR Micro-CMCs: The is layer(a)Deflection is continuous along the fiber/matrix interface, (b) sue regarding matrix density motivated the consideration of deflections occur within the hibonite layer, and (c)deflections consist other low-temperature-sintering Al2 O3 powders. The powder, of cleavage within the hibonite coating TM-DAR, which sinters to full density at 1350.C in I h, as per the manufacturer's information was selected for the subse- quent studies. The other concern was that the strain-to-failure fiber under three different conditions: (a)uncoated, (b) hibo. of the grade- D sapphire(selected initially based on availability) might not be sufficiently large to cause significant matrix tions of fiber were tested after they were heat-treated at 1450.C cracking before fiber failure. Thus, the fiber was switched to for 2 h to simulate the heat treatment used to sinter the matrix. the better-strength grade-A fiber. This fiber/matrix combina- Hibonite appeared to degrade the fiber slightly, while monazite tion was examined with uncoated and monazite-coated fibers had little effect. In both coated fibers. however. the weibull As seen earlier, the fiber strength changes during processing modulus was much enhanced. Figure 9(b) shows the strength precluded a direct comparison of the strengths of the control of the control and monazite-coated microcomposites compared composites to those of the interface-treated composites. The with the fiber strengths shown in Fig. 9(a). The strengths of fiber straied by testing the fibers under different conditions: (a) h degradation during coating and matrix processing oth composites were lowered after matrix processing, and the average strengths were not significantly different. However 1450, 2 h to simulate sintering effects,(b)15 passes(at the monazite-coated composites had a much-improved Weibull 1250 C)through fiber coater plus 1450C, 2 h to simulate modulus. Essentially, the same result was obtained for the hi- thermal shock during coating, and (c) as-coated fiber heat bonite-coated composites, as shown in Fig. 9(c). Because the treated at 1450 C, 2 h. The results are plotted in Fig. 10(a)
fiber under three different conditions: (a) uncoated, (b) hibonite-coated, and (c) monazite-coated. All of these three variations of fiber were tested after they were heat-treated at 1450°C for 2 h to simulate the heat treatment used to sinter the matrix. Hibonite appeared to degrade the fiber slightly, while monazite had little effect. In both coated fibers, however, the Weibull modulus was much enhanced. Figure 9(b) shows the strength of the control and monazite-coated microcomposites compared with the fiber strengths shown in Fig. 9(a). The strengths of both composites were lowered after matrix processing, and the average strengths were not significantly different. However, the monazite-coated composites had a much-improved Weibull modulus. Essentially, the same result was obtained for the hibonite-coated composites, as shown in Fig. 9(c). Because the monazite composites showed significant debonding, the lack of difference in strength between the control and monazite-coated composite was puzzling. To determine if this was because of poor matrix density, a second set of control and monazitecoated samples were sintered at 1500°C for 2 h to improve matrix density. There was only a marginal improvement in matrix density, and essentially the same result was obtained (Fig. 9(d)) with some loss in strength, which could be rationalized easily as due to loss of fiber strength following the 1500°C heat treatment. (b) Sapphire A/Monazite/TM-DAR Micro-CMCs: The issue regarding matrix density motivated the consideration of other low-temperature-sintering Al2O3 powders. The powder, TM-DAR, which sinters to full density at 1350°C in 1 h, as per the manufacturer’s information, was selected for the subsequent studies. The other concern was that the strain-to-failure of the grade-D sapphire (selected initially based on availability) might not be sufficiently large to cause significant matrix cracking before fiber failure. Thus, the fiber was switched to the better-strength grade-A fiber. This fiber/matrix combination was examined with uncoated and monazite-coated fibers. As seen earlier, the fiber strength changes during processing precluded a direct comparison of the strengths of the control composites to those of the interface-treated composites. The fiber strength degradation during coating and matrix processing was studied by testing the fibers under different conditions: (a) 1450°, 2 h to simulate sintering effects, (b) 15 passes (at 1250°C) through fiber coater plus 1450°C, 2 h to simulate thermal shock during coating, and (c) as-coated fiber heattreated at 1450°C, 2 h. The results are plotted in Fig. 10(a). Fig. 5. Fracture surface of hibonite-coated microcomposites at different magnifications revealing crack deflections within the hibonite layer. (a) Deflection is continuous along the fiber/matrix interface, (b) deflections occur within the hibonite layer, and (c) deflections consist of cleavage within the hibonite coating. Fig. 6. SEM macrophotographs of the monazite-coated microcomposites showing significant debonding in these specimens from (a) A1000 powder and (b) TM-DAR powder. In (b), the matrix falls off intermittently along the length of the microcomposite, revealing periodic matrix cracks. December 1999 Fabrication and Testing of Oxide/Oxide Microcomposites with Monazite and Hibonite as Interlayers 3579
80 Journal of the American Ceramic Society-Parthasarathy et al. Vol. 82. No. 12 (a) sapphire Fiber Fracture Impressions on Monazite Origin Of Matrix Grain Boundary Grooves e 200um 2 um Fig. 7. Intermediate magnifications of specimens in Fig. 9 reveal that agnifications, specimens in Fig 9 reveal (a) both circumferential and axial cracks form in the matrix and(b) oating/matrix interface. Grainlike feature debonding often(but not always) tends to be partial and not go sions of matrix Al2O, grains imprinted on around the fiber. See text for possible causes. sintering with thermal grooving that follows densification Although the A fibers were much stronger than the monazite or hibonite coatings? The following is a discussion ade-D fibers 1450 C, 2 h heat treatment, the monazite- that attempts to answer this question, with a focus on the mona oated fiber hs of the grades A and D were not very zite-coated different; in fact, the coated grade-A fibers were weaker. Fur Because the monazite-containing composites exhibit ther, the coating on sapphire A resulted in only a marginal debonding along with a high Weibull modulus, and the control increase in Weibull modulus as compared with the effect seen specimens have the same mean strength as the monazite-or in the coated grade-D fibers(Fig. 9(a)) hibonite-containing modulus Figure 10(b) shows a comparison of the strengths obtained one of the following must be true using the control and monazite-coated microcomposites. As (1) The increase in Weibull modulus is an indication of a seen with the A1000-matrix microcomposites, the monazite- good composite oated composites show nearly the same strength as the cor (2) The control specimens are also good composites trol composites but with a significant increase in the Weibull If the first statement is true, then the higher Weibull modulus modulus. If microcomposite strengths are compared with of the hibonite- coated specimens compared with that of the the corresponding fiber strengths, then the control shows control indicates that hibonite-coated composites are good more strength degradation than the monazite-coated fiber composites. If the latter is true, the effectiveness of hibonite composites a coating cannot be inferred from this work. To find an answer, let us consider a different criterion for a good composite ncing criterion for a good composite is one that verifies if the composite's ultimate strength exceeds the matrix In the case of monazite. microstructures of fractured cracking stress, because this implies that the fibers are pro- omposites(Figs 5-8)show clear evidence for debonding at tected from the matrix cracks. The predicted axial compression the coating/matrix interface. confirming that the interface be in the matrix(from CTE mismatch, estimated at -125 MPa tween monazite and matrix Al,O3 is sufficiently weak to makes it difficult(in situ experiments in the SEM are required) debond under axial loading of the composite in tension. Frac- to observe matrix cracks after unloading. In the absence of such tographs of hibonite-containing composites show evidence for information, one can attempt to estimate the matrix cracking crack deflection, but the debond crack is not sustained beyond stress and compare it with the ultimate strength. The simplest a few micrometers. In quantitative terms, the strengths of the approach is to the strain-to-failure of the matrix as the composites are high, but they are not different from the control failure criterion. The work of Nanjangud et al. 20 gives strength opposites. The only positive indication in either composite and modulus of Al,O, as a function of porosity. For a porosity relation to the control specimens is an increase in Weibull of 15%, the mean strength is-280 MPa, and the modulus is modulus. Is this sufficient proof-of-concept for further work 00 GPa, which yield a failure strain of 0.095%. This trans-
Although the grade-A fibers were much stronger than the grade-D fibers after a 1450°C, 2 h heat treatment, the monazitecoated fiber strengths of the grades A and D were not very different; in fact, the coated grade-A fibers were weaker. Further, the coating on sapphire A resulted in only a marginal increase in Weibull modulus as compared with the effect seen in the coated grade-D fibers (Fig. 9(a)). Figure 10(b) shows a comparison of the strengths obtained using the control and monazite-coated microcomposites. As seen with the A1000-matrix microcomposites, the monazitecoated composites show nearly the same strength as the control composites but with a significant increase in the Weibull modulus. If microcomposite strengths are compared with the corresponding fiber strengths, then the control shows more strength degradation than the monazite-coated fiber composites. IV. Discussion In the case of monazite, microstructures of fractured microcomposites (Figs. 5–8) show clear evidence for debonding at the coating/matrix interface, confirming that the interface between monazite and matrix Al2O3 is sufficiently weak to debond under axial loading of the composite in tension. Fractographs of hibonite-containing composites show evidence for crack deflection, but the debond crack is not sustained beyond a few micrometers. In quantitative terms, the strengths of the composites are high, but they are not different from the control composites. The only positive indication in either composite in relation to the control specimens is an increase in Weibull modulus. Is this sufficient proof-of-concept for further work on monazite or hibonite coatings? The following is a discussion that attempts to answer this question, with a focus on the monazite-coated specimens. Because the monazite-containing composites exhibit debonding along with a high Weibull modulus, and the control specimens have the same mean strength as the monazite- or hibonite-containing composites but lower Weibull modulus, one of the following must be true. (1) The increase in Weibull modulus is an indication of a good composite. (2) The control specimens are also good composites. If the first statement is true, then the higher Weibull modulus of the hibonite-coated specimens compared with that of the control indicates that hibonite-coated composites are good composites. If the latter is true, the effectiveness of hibonite as a coating cannot be inferred from this work. To find an answer, let us consider a different criterion for a good composite. A convincing criterion for a good composite is one that verifies if the composite’s ultimate strength exceeds the matrix cracking stress, because this implies that the fibers are protected from the matrix cracks. The predicted axial compression in the matrix (from CTE mismatch, estimated at ∼125 MPa) makes it difficult (in situ experiments in the SEM are required) to observe matrix cracks after unloading. In the absence of such information, one can attempt to estimate the matrix cracking stress and compare it with the ultimate strength. The simplest approach is to use the strain-to-failure of the matrix as the failure criterion. The work of Nanjangud et al.20 gives strength and modulus of Al2O3 as a function of porosity. For a porosity of 15%, the mean strength is ∼280 MPa, and the modulus is ∼300 GPa, which yield a failure strain of 0.095%. This transFig. 7. Intermediate magnifications of specimens in Fig. 9 reveal that (a) both circumferential and axial cracks form in the matrix and (b) debonding often (but not always) tends to be partial and not go all around the fiber. See text for possible causes. Fig. 8. At high magnifications, specimens in Fig. 9 reveal debonding occurs at the coating/matrix interface. Grainlike feature in (b) are actually impressions of matrix Al2O3 grains imprinted on monazite during sintering with thermal grooving that follows densification. 3580 Journal of the American Ceramic Society—Parthasarathy et al. Vol. 82, No. 12
December 1999 Fabrication and Testing of Oxide/Oxide Microcomposites with Monazite and Hibonite as Interlayers (a)2 CMC-Control Fiber t'.1450-C 2h 己 2. 25 GPa Fiberdmonazi 233GP m=11.1 2345 CMCMonazite LL"Lr, 1-0.500 L.5 10.5 0.5 Ln(Stress, GPa) Ln [ Stress, GPa CMC-Contro 1(]C.]rb 1.06 GPa 0 L 18 GP m=5,2 CMC-Monazite m=577 Fiber-Hibonite CMC-Hibo 84GPa 1.18GP m=13.1 sTiLL ⊥LLL -0.500.5 10.500.51 Ln [ stress, GP Ln Stress, GPa] Fig. 9. Results of tension tests were analyzed based on maximum load-to-failure divided by fiber cross-sectional area. Significant variations appeared(scatter)in the data. A Weibull form was chosen for analyses. Results are plotted on a Weibull plot, comparing the fibers, coated control microcomposites, hibonite-coated micro-CMCs, and monazite- coated micro-CMCs. All data are for the sapphire fiber, grade D. See text for lates into an effective stress(composite load divided by fib nalized based on the porosity of the matrix. The fiber can be a, as was done in all of the strength values reported in this protected from matrix cracks, because these cracks are blunted work)of 508 MPa; inclusion of the axial residual strain in the by the pores in the matrix, as shown schematically in Fig. 11 matrix(calculated using the axisymmetric model of BHE) yields an effective stress for matrix cracking of 860 MPa. Use terface are those that lie right next to the interface, but, if the of the thin-film-on-flat-substrate model of Beuth22 and Ye et size of such cracks is limited to the grain size of the matrix al. 23 yields channel cracking stresses of 940, 860, or 804 MPa grains(1.0 um), no significant reduction in strength is pre- for choices of 3, 4, or 5 for a dimensionless parameter that dicted. However, such cracks, if not deflected at the interface, depends on elastic parameters, geometry, and crack depth, re- introduce a new distribution of flaws in the control specimens, causing a lower Weibull modulus. The same cracks can be of 740 MPa for steady-state crack extension. These values for deflected at the matrix/coating interface in the case of monazite matrix cracking are all less than the experimental stress-to. or hibonite, thus preventing them from adding to the flaw failure for the control and coated composites(l GPa, Figs. 10 population. This would explain the retention of the higher and 11). These calculations suggest that the matrix cracks form Weibull moduli of the coated fibers in the composites below the maximum stress applied. Two other issues remain. First, why and how does the coat The above calculations imply that the fibers are protected ing improve the Weibull modulus? Second, why is the strength from the matrix cracks even in the control specimens. This of the fiber embedded in the matrix lower than the strength of conclusion is further supported by the remarkable similarity in the fiber by itself, both in the control and in the coated situa- the statistical behavior of microcomposites compared with the tions. The following is an attempt at rationalization of these corresponding uncoated or coated fiber strength distribution observations (Figs. 9, 10). The enhancement in Weibull modulus obtained The strength of a fiber is determined by the largest flaw by the coating is clearly retained when the matrix is added. within the gauge length tested. If the distribution of flaws in the This implies that the composite behavior is fiber-dominated, fiber is narrow, then the distribution of strength-limiting flaws supporting the suggestion that matrix cracks do not influence in multiple specimens is also narrow, yielding a higher Weibull the ultimate failure stress of the composites modulus. Thus, factors that tend to change the flaw distribution Protection of fibers in the control specimens can be ratio- nust influence the Weibull distribution, while factors that
lates into an effective stress (composite load divided by fiber area, as was done in all of the strength values reported in this work) of 508 MPa; inclusion of the axial residual strain in the matrix (calculated using the axisymmetric model of BHE17) yields an effective stress for matrix cracking of 860 MPa. Use of the thin-film–on–flat-substrate model of Beuth22 and Ye et al.23 yields channel cracking stresses of 940, 860, or 804 MPa for choices of 3, 4, or 5 for a dimensionless parameter that depends on elastic parameters, geometry, and crack depth, respectively. The model of BHE17,24 yields a stress on the fiber of 740 MPa for steady-state crack extension. These values for matrix cracking are all less than the experimental stress-tofailure for the control and coated composites (∼1 GPa, Figs. 10 and 11). These calculations suggest that the matrix cracks form below the maximum stress applied. The above calculations imply that the fibers are protected from the matrix cracks even in the control specimens. This conclusion is further supported by the remarkable similarlity in the statistical behavior of microcomposites compared with the corresponding uncoated or coated fiber strength distributions (Figs. 9, 10). The enhancement in Weibull modulus obtained by the coating is clearly retained when the matrix is added. This implies that the composite behavior is fiber-dominated, supporting the suggestion that matrix cracks do not influence the ultimate failure stress of the composites. Protection of fibers in the control specimens can be rationalized based on the porosity of the matrix. The fiber can be protected from matrix cracks, because these cracks are blunted by the pores in the matrix, as shown schematically in Fig. 11. The only cracks that produce a stress concentration at the interface are those that lie right next to the interface, but, if the size of such cracks is limited to the grain size of the matrix grains (∼1.0 mm), no significant reduction in strength is predicted. However, such cracks, if not deflected at the interface, introduce a new distribution of flaws in the control specimens, causing a lower Weibull modulus. The same cracks can be deflected at the matrix/coating interface in the case of monazite or hibonite, thus preventing them from adding to the flaw population. This would explain the retention of the higher Weibull moduli of the coated fibers in the composites. Two other issues remain. First, why and how does the coating improve the Weibull modulus? Second, why is the strength of the fiber embedded in the matrix lower than the strength of the fiber by itself, both in the control and in the coated situations. The following is an attempt at rationalization of these observations. The strength of a fiber is determined by the largest flaw within the gauge length tested. If the distribution of flaws in the fiber is narrow, then the distribution of strength-limiting flaws in multiple specimens is also narrow, yielding a higher Weibull modulus. Thus, factors that tend to change the flaw distribution must influence the Weibull distribution, while factors that Fig. 9. Results of tension tests were analyzed based on maximum load-to-failure divided by fiber cross-sectional area. Significant variations appeared (scatter) in the data. A Weibull form was chosen for analyses. Results are plotted on a Weibull plot, comparing the fibers, coated fibers, control microcomposites, hibonite-coated micro-CMCs, and monazite-coated micro-CMCs. All data are for the sapphire fiber, grade D. See text for discussions. December 1999 Fabrication and Testing of Oxide/Oxide Microcomposites with Monazite and Hibonite as Interlayers 3581
Journal of the American Ceramic Society-Parthasarathny et al. Vol. 82. No. 12 (b)2 CMC-C -Control 1 E Sapphirc'A'/TM-DAR OE 0 m=4.19 50°c2h 3.25 GPa SanphireavmonaIM-DaR m=5,19 Pa:m=8.13 1-0.500.511.5 -1-0.500.511.5 Ln[ stress, GPa l Ln[ Stress, GPa specimens. Monazite-coated composites retain more fiber strength, compared to the control. Coated specimens also exhibit higher Weibull moduli change the mean flaw size change the strength. The coating of barriers )are necessary for matrix densities 85% befor tiveness of a novel inter- tiple matrix cracks with debonding occurring at the monazite/ face strategy convincingly. Secon matrix interface. However, the interlayer-containing com- osites showed no strength improvement over the control terface t atments (except ed as diffusion/reaction specimens, although they had a significantly higher Weibull modulus. The lack of difference in strength was attributed to the porosity of the matrix. The results imply that any proof- ATRIX CRACKS of-concept for interlayers requires matrices that are denser than References Composites, Mater. Sci. Eng, A, A162, 15-25(1993) 2P. E D. Morgan and D. B. Marshall, "Ceramic Composites of Monazite and aWw.M a Yttrium Phosphate/Yttrium Aluminate Laminate, " J. Am. Ceram Soc., 78[1 Processing-Induced Flaw 3121-24(1995 D.-H. Kuo and W. M. Kriven, "Microscturcture and Mechanical Respon FIBER Aluminate Systems, "Ceram. Eng. Sci. Proc., 17 [4]233-40(199 SD. C. Hitchcock and L C. D. Jonghe. "Fracture Tous Sodium-beta-Alumina, " J. Am. Ceram Soc., 66[9]C-204-C-205(1983) phase Derived from Sol-Gel Fiber Coatings, "J Plumbite Fiber-Matrix bM. K. Cinibulk and Its are shown schematically in this figure and discussed in the text. Matrix cracks can be prevented from penetrating the fiber by interfa- M. K. Cinibulk, "Magnetoplumbite Compounds as a Fiber Coating in Oxide/ cial pores, thus explaining the higher than expected strengths of th Oxide Sci Proc. control specimens( Figs. 9 and 10). Only flaws in grains bonded to the nibulk "Microstructure and Mechanical Behavior of an Hibonite fiber affect microcomposite strength in control specimens, giving rise 633-41(1995) to some of the lower strengths seen in the Weibull plots(Figs. " M. K. Cinibulk, "Deposition of Oxide ber Cloths by Electro- 0) static Attraction, J. Anm. Ceram. Soc., 8
change the mean flaw size change the strength. The coating of monazite is known to cause grooves or cusps at the monazite/ fiber interface because of surface tension effects,2 as seen in Fig. 10(b). These cusps result in stress concentrations that can tend to favor the smaller flaws (whose crack tips are closer to the fiber surface) to grow prematurely compared with the longer flaws. This could cause an increase in Weibull modulus. The reasons for the observed lowering of the Weibull modulus in control specimens is not clear. The lower strengths of composites compared to fibers is partially explained by residual axial tension in the fiber. Interestingly, thermal shock is expected to have a negligible effect; for a thermal diffusivity, a, of 0.05 cm2 /s (for Al2O3 16), the cooling rate, f, required to generate a DT of 100°C in a cylinder with a radius of 60 mm is extremely high (∼5 × 105 °C/s), estimated using the relationship, DT 4 0.25fr2 /a. 25 V. Implications and Limitations of This Work There are two significant implications resulting from the above discussion. First, it is necessary to process composites with a density >85% before the effectiveness of a novel interface strategy can be demonstrated convincingly. Second, no interface treatments (except those used as diffusion/reaction barriers) are necessary for matrix densities #85%. The second statement is supported by the work of Janssen et al.26 A major limitation of this work is that it neglects the effect of pore size and distribution. Clearly, a composite with very large pores of the same volume fraction (∼15%) in the matrix behaves very different from that with a fine distribution. This work did not characterize the size and distribution of the pores in the composites studied, although the uniformity is likely of the order of the grain size. Designs that use porous matrix composites must consider the long-term deleterious effects of pore coarsening. VI. Summary The effectiveness of using hibonite and monazite as interface coatings for oxidation-resistant oxide-matrix composites was examined using microcomposites. Using sapphire monofilaments in an Al2O3 matrix as the control composites, the fractography and fracture strengths were compared. The hibonitecoated specimens showed evidence of crack deflection, but debond lengths were limited to a few micrometers. The monazite-coated specimens showed extensive debonding and multiple matrix cracks with debonding occurring at the monazite/ matrix interface. However, the interlayer-containing composites showed no strength improvement over the control specimens, although they had a significantly higher Weibull modulus. The lack of difference in strength was attributed to the porosity of the matrix. The results imply that any proofof-concept for interlayers requires matrices that are denser than 85%. References 1 P. E. D. Morgan and D. B. Marshall, “Functional Interfaces for Oxide/Oxide Composites,” Mater. Sci. Eng., A, A162, 15–25 (1993). 2 P. E. D. Morgan and D. B. Marshall, “Ceramic Composites of Monazite and Alumina,” J. Am. Ceram. Soc., 78 [6] 1553–63 (1995). 3 D.-H. Kuo and W. M. Kriven, “Characterization of Yttrium Phosphate and a Yttrium Phosphate/Yttrium Aluminate Laminate,” J. Am. Ceram. Soc., 78 [11] 3121–24 (1995). 4 D.-H. Kuo and W. M. Kriven, “Microscturcture and Mechanical Response of Lanthanum Phosphate/Yttrium Aluminate and Yttrium Phosphate/Yttrium Aluminate Systems,” Ceram. Eng. Sci. Proc., 17 [4] 233–40 (1996). 5 D. C. Hitchcock and L. C. D. Jonghe, “Fracture Toughness Anisotropy of Sodium-beta-Alumina,” J. Am. Ceram. Soc., 66 [9] C-204–C-205 (1983). 6 M. K. Cinibulk and R. S. Hay, “Textured Magnetoplumbite Fiber–Matrix Interphase Derived from Sol–Gel Fiber Coatings,” J. Am. Ceram. Soc., 79 [5] 1233–46 (1996). 7 M. K. Cinibulk, “Magnetoplumbite Compounds as a Fiber Coating in Oxide/ Oxide Composites,” Ceram. Eng. Sci. Proc., 15 [5] 721–29 (1994). 8 M. K. Cinibulk, “Microstructure and Mechanical Behavior of an Hibonite Interphase in Alumina-Based Composites,” Ceram. Eng. Sci. Proc., 16 [5] 633–41 (1995). 9 M. K. Cinibulk, “Deposition of Oxide Coatings on Fiber Cloths by Electrostatic Attraction,” J. Am. Ceram. Soc., 80 [2] 453–60 (1997). Fig. 10. Results obtained using monazite-coated sapphire grade-A-based microcomposites are shown, comparing fibers, control, and coated specimens. Monazite-coated composites retain more fiber strength, compared to the control. Coated specimens also exhibit higher Weibull moduli. Fig. 11. Plausible effects of matrix porosity in influencing test results are shown schematically in this figure and discussed in the text. Matrix cracks can be prevented from penetrating the fiber by interfacial pores, thus explaining the higher than expected strengths of the control specimens (Figs. 9 and 10). Only flaws in grains bonded to the fiber affect microcomposite strength in control specimens, giving rise to some of the lower strengths seen in the Weibull plots (Figs. 9 and 10). 3582 Journal of the American Ceramic Society—Parthasarathy et al. Vol. 82, No. 12
December 1999 Fabrication and Testing of Oxide/Oxide Microcomposites with Monazite and Hibonite as Interlayers 3583 IuM. K. Cinibulk, ""Effect of Precursors and Dopants on the Synthesis of Tows using Liquid Precursors: Monazite Coatings on Nextel 720, "J.Am Sol-Gel-Derived Calcium Hexaluminate, " J ran. Soc.,81[12]357-68 (1998) eren820912321-3109 R.S. Hay and EE. Hermes, " Coating Apparatus for Continuous Fibers C. Levi, J. Y. Yang, B. J. Dalgleish, F w. Zok, and A G. Evans,"Pro- U.S. Pat. No. 5217533, June 8, 1993 and Performance of an All-Oxide Ceramic Composite, "J. Amm. Ceram. s C. Nanjangud, R. Brezny, and D J. Green, "Strength and Youngs Modulus Behavior of a Partially Sintered Porous Alumina, "J. Am. Ceram. Soc. -C. tu, FF. Lange, and A G. Evans, "Concept for a Damage-Tolerant 78 [1 266-68(1995) Ceramic Composite withStrong"Interfaces, "J. A. Ceram. Soc., 79[2] 2J. Lamon, "Interfaces and Interfacial Mechanics: Influence on the Mechani- cal Behavior of Ceramic Matrix Composites(CMC), "J. Phys /V, 3 1607-16 13J. Lamon, F. Rebillat, and A G. Evans, "Microcomposite Test Procedure (1993) for Evaluating the Interface Properties of Ceramic Matrix Composites, "J.Am. 22J. L Beuth, "Cracking of Thin Bonded Films in Residual Tension, "Int J. 1AG.N. Morscher, I. M. Fernandez, and M.J. Purdy, "Determination of In- Solids struct,291657-75(199 2T. Ye, Z. Suo, and A G. Evans, "Thin Film Cracking and the roles of 4T. A. Parthasarathy, R erties to oating Thermoelastic Effects into Fiber/Matrix Compos- Nal.Bar.sand,70A4]227-80(1960 I6S. J. Schneider, Engineered Materials Handbook, Ceramics and Glasses, 23.DBAm. Ceram.soc,82[3]579-84(199) Vol 4. ASM Inter Reaction I'B. Budiansky, J w. Hu and A.G. Evans. "Matrix Fracture in EDBoakye,RSHay,and D Petry, "Contin[2] 167-89(1986) Fiber-Reinforced Ceramics, ". Mech. Phs. Sc ites 1L.: Manufacturing and Materials Development. Edited by A. G. Evans and nuous Coating of oxide Fiber Naslaon. American Ceramic Society, Westerville, OH, 1995
10M. K. Cinibulk, “Effect of Precursors and Dopants on the Synthesis of Sol–Gel-Derived Calcium Hexaluminate,” J. Am. Ceram. Soc., 81 [12] 3157–68 (1998). 11C. Levi, J. Y. Yang, B. J. Dalgleish, F. W. Zok, and A. G. Evans, “Processing and Performance of an All-Oxide Ceramic Composite,” J. Am. Ceram. Soc., 81 [8] 2077–86 (1998). 12W.-C. Tu, F. F. Lange, and A. G. Evans, “Concept for a Damage-Tolerant Ceramic Composite with “Strong” Interfaces,” J. Am. Ceram. Soc., 79 [2] 417–24 (1996). 13J. Lamon, F. Rebillat, and A. G. Evans, “Microcomposite Test Procedure for Evaluating the Interface Properties of Ceramic Matrix Composites,” J. Am. Ceram. Soc., 78 [2] 401–405 (1995). 14G. N. Morscher, J. M. Fernandez, and M. J. Purdy, “Determination of Interfacial Properties using a Single Fiber Microcomposite Test,” J. Am. Ceram. Soc., 79 [4] 1083–91 (1996). 15W. E. Tefft, “Elastic Constants of Synthetic Crystal Corundum,” J. Res. Natl. Bur. Stand., 70A [4] 227–80 (1960). 16S. J. Schneider, Engineered Materials Handbook, Ceramics and Glasses, Vol. 4. ASM International, Materials Park, OH, 1991. 17B. Budiansky, J. W. Hutchinson, and A. G. Evans, “Matrix Fracture in Fiber-Reinforced Ceramics,” J. Mech. Phys. Solids, 34 [2] 167–89 (1986). 18E. D. Boakye, R. S. Hay, and D. Petry, “Continuous Coating of Oxide Fiber Tows using Liquid Precursors: Monazite Coatings on Nextel 720,” J. Am. Ceram. Soc., 82 [9] 2321–31 (1999). 19R. S. Hay and E. E. Hermes, “Coating Apparatus for Continuous Fibers,” U.S. Pat. No. 5 217 533, June 8, 1993. 20S. C. Nanjangud, R. Brezny, and D. J. Green, “Strength and Young’s Modulus Behavior of a Partially Sintered Porous Alumina,” J. Am. Ceram. Soc., 78 [1] 266–68 (1995). 21J. Lamon, “Interfaces and Interfacial Mechanics: Influence on the Mechanical Behavior of Ceramic Matrix Composites (CMC),” J. Phys IV, 3 1607–16 (1993). 22J. L. Beuth, “Cracking of Thin Bonded Films in Residual Tension,” Int. J. Solids Struct., 29 1657–75 (1992). 23T. Ye, Z. Suo, and A. G. Evans, “Thin Film Cracking and the Roles of Substrate and Interface,” Int. J. Solids Struct., 29 [21] 2639–48 (1992). 24T. A. Parthasarathy, R. J. Kerans, and N. J. Pagano, “Effective Fiber Properties to Incorporate Coating Thermoelastic Effects into Fiber/Matrix Composite Models,” J. Am. Ceram. Soc., 82 [3] 579–84 (1999). 25W. D. Kingery, Introduction to Ceramics, Wiley, New York, 1960. 26R. Janssen, J. Wendorff, and N. Claussen, Fiber Reinforcement of Reaction Bonded Oxide Ceramics, Vol. 58, High-Temperature Ceramic-Matrix Composites II: Manufacturing and Materials Development. Edited by A. G. Evans and R. Naslaon. American Ceramic Society, Westerville, OH, 1995. h December 1999 Fabrication and Testing of Oxide/Oxide Microcomposites with Monazite and Hibonite as Interlayers 3583