MIATERIAL TENGE ENGMEERIM ELSEVIER Materials Science and Engineering A244(1998)11-21 Sol-gel synthesis of ceramic matrix composites E D. Rodeghiero, B C. Moore, B.S. Wolkenberg, M. Wuthenow, O. K. Tse, EP Giannelis Department of Materials Science and Engineering, Cornell Uninersity, Ithaca, NY 14853-1501, US.A Abstract Sol-gel techniques have been used to produce various high temperature ceramic matrix composites including Ni/a-Al,O3, e/a-Al,O3, Ni/ZrO2, SiC(whisker)/a-Al,O3, and SiC(platelet)/a-Al,O,, as well as chemically modified versions of some of these ystems. In all cases, the composites have displayed uniform microstructures with a high degree of dispersion between the matrix nd reinforcement phases, a goal often not achieved when utilizing conventional powder mixing and processing techniques. The metal-ceramic composites investigated exhibit enhanced toughness and machinability as well as the potential for catalytic applications due to their novel fine-scale microstructure. Likewise, the Sic-reinforced alumina materials have been shown to be lighter, stiffer and tougher than pure alumina, without the use of the extreme hot-pressing temperatures and pressures needed by conventional powder processing approaches to produce the same results. o 1998 Elsevier Science S.A. All rights reserved Keywords: Sol-gel; Ceramic matrix composites; Microstructure 1. ntroduction processes, and as a result, virtually all sol-gel research since the late 1980s has been carried out in the thin Sol-gel processing received extensive attention in the film/coating area. However, this ignores sol-gels po- 1980s as literally hundreds of re- tential to play a supporting role in the synthesis of searchers sought after novel, low temperature methods monolithic structural ceramics and ceramic composites of producing common oxide ceramics such as silica, In other words, while the production of structural alumina, zirconia and titania in fully dense monolithic ceramics will most likely never be accomplished solely form [1]. Much of this excitement resulted from the by low temperature sol-gel techniques, the incorpora roduction of the first large-scale xerogels by Yoldas in tion of some sol-gel aspects into a broader synthesis 975 [2-5]. These self-supporting monolithic alumina scheme can nevertheless be highly beneficial. In fact, gels were highly porous (60-70%), but nevertheless many pioneers of the sol-gel community have felt this suggested the potential for producing fully dense ce- way from the very beginning. For instance, Roy et al. ramic components at reduced temperatures in a near- described their original goal in the sol-gel field as net-shape fashion. Due to the inherent fracture achievement of homogeneity on the finest possible associated with the drying and consolidation of bulk scale he production of mono-phasic glasses and gels, however, it later became accepted that sol-gel mono-phasic ceramic powders and precursors [7]. In- would instead be limited to a much smaller realm of deed, it was Roy who first brought sol-gel science to applications, namely the production of thin films, wh broad attention in the ceramics industry in the 1950s because of their planar geometry were not susceptible and 1960s for exactly this reason [2] to the formidable cracking problem of monoliths More recently, it is the incorporation of sol-gel Fortunately for the sol-gel community, the great techniques into the synthesis of ceramic matrix com- explosion in the microelectronics industry came during posites which seems especially appealing. In 1981, Rice the same time period. This translated into a lar and Becher demonstrated that ZrO,Al,O3 ceramic-ce demand for both thin film materials and thin film ramic composites produced through sol-gel approaches were superior to their ball milled, powder-derived coun- Corresponding author. Tel: +1 607 2556684: fax: +1 607 terparts in overall fracture properties [8,9]. In fact, in 2552365 this work the first demonstration of a simultaneous 0921-5093/98/S19.00 0 1998 Elsevier Science S.A. All rights reserved PIS0921-5093(9700821-6
Materials Science and Engineering A244 (1998) 11–21 Sol–gel synthesis of ceramic matrix composites E.D. Rodeghiero, B.C. Moore, B.S. Wolkenberg, M. Wuthenow, O.K. Tse, E.P. Giannelis * Department of Materials Science and Engineering, Cornell Uni6ersity, Ithaca, NY 14853-1501, USA Abstract Sol–gel techniques have been used to produce various high temperature ceramic matrix composites including Ni/a-Al2O3, Fe/a-Al2O3, Ni/ZrO2, SiC(whisker)/a-Al2O3, and SiC(platelet)/a-Al2O3, as well as chemically modified versions of some of these systems. In all cases, the composites have displayed uniform microstructures with a high degree of dispersion between the matrix and reinforcement phases, a goal often not achieved when utilizing conventional powder mixing and processing techniques. The metal–ceramic composites investigated exhibit enhanced toughness and machinability as well as the potential for catalytic applications due to their novel fine-scale microstructure. Likewise, the SiC-reinforced alumina materials have been shown to be lighter, stiffer and tougher than pure alumina, without the use of the extreme hot-pressing temperatures and pressures needed by conventional powder processing approaches to produce the same results. © 1998 Elsevier Science S.A. All rights reserved. Keywords: Sol–gel; Ceramic matrix composites; Microstructure 1. Introduction Sol–gel processing received extensive attention in the 1970s and early 1980s as literally hundreds of researchers sought after novel, low temperature methods of producing common oxide ceramics such as silica, alumina, zirconia and titania in fully dense monolithic form [1]. Much of this excitement resulted from the production of the first large-scale xerogels by Yoldas in 1975 [2–5]. These self-supporting monolithic alumina gels were highly porous (60–70%), but nevertheless suggested the potential for producing fully dense ceramic components at reduced temperatures in a nearnet-shape fashion. Due to the inherent fracture associated with the drying and consolidation of bulk gels, however, it later became accepted that sol–gel would instead be limited to a much smaller realm of applications, namely the production of thin films, which because of their planar geometry were not susceptible to the formidable cracking problem of monoliths [6]. Fortunately for the sol–gel community, the great explosion in the microelectronics industry came during the same time period. This translated into a large demand for both thin film materials and thin film processes, and as a result, virtually all sol–gel research since the late 1980s has been carried out in the thin film/coating area. However, this ignores sol–gel’s potential to play a supporting role in the synthesis of monolithic structural ceramics and ceramic composites. In other words, while the production of structural ceramics will most likely never be accomplished solely by low temperature sol–gel techniques, the incorporation of some sol–gel aspects into a broader synthesis scheme can nevertheless be highly beneficial. In fact, many pioneers of the sol–gel community have felt this way from the very beginning. For instance, Roy et al. described their original goal in the sol–gel field as achievement of ‘homogeneity on the finest possible scale’ in the production of mono-phasic glasses and mono-phasic ceramic powders and precursors [7]. Indeed, it was Roy who first brought sol–gel science to broad attention in the ceramics industry in the 1950s and 1960s for exactly this reason [2]. More recently, it is the incorporation of sol–gel techniques into the synthesis of ceramic matrix composites which seems especially appealing. In 1981, Rice and Becher demonstrated that ZrO2/Al2O3 ceramic–ceramic composites produced through sol–gel approaches were superior to their ball milled, powder-derived counterparts in overall fracture properties [8,9]. In fact, in this work the first demonstration of a simultaneous * Corresponding author. Tel.: +1 607 2556684; fax: +1 607 2552365. 0921-5093/98/$19.00 © 1998 Elsevier Science S.A. All rights reserved. PII S0921-5093(97)008 21-6
12 D. Rodeghiero et al / Materials Science and Engineering 4244(1998)11-21 increase in both fracture strength and fracture tough ness for the ZrO -reinforced Al,O3 system was re- ported. Rice attributed this beneficial behavior to the extreme homogeneity of the sol-gel derived com osites. In 1984. Hoffman et al. also demonstrated the extensive composite homogeneity and dispersion that could be achieved through sol-gel approaches by syn and CdS/SiO, (where the AgCl and Cds phases were in s fine-scale photosensitive composites such as AgCl/SiO crystalline form), to phase-separated phosphate and mixed oxide glasses such as CrPO4/SiO2, CePO/Sio and Nd, SIO2(where both the matrix and minor phases were amorphous)[10, 1l]. At the same time, Roy et al. were also using these same synthesis procedures to produce the first sol-gel derived metal-ceramic com- osites(including Cu/Al,O3, Ni/Al,O3, Cu/ZrO2 and 2-theta( degrees) Cu/SiO2)[12]. In fact, Hoffman et al. and Roy et al. Fi of XRD patterns for the processing of a 20/80 vol% collectively performed the most extensive work on the Ni al-ceramic composite: (a) dried, unreduced powder, sol-gel synthesis of ceramic matrix composites to the blet drogen reduced powder, and(c)1400C hot-pressed present day, thoroughly investigating over 30 different hemical systems. However, in their efforts no attention was given to high temperature consolidation, mechani- stiffer and tougher than pure alumina, these composites cal properties or structural applications. Finally, sol were consolidated at lower temperatures and pressures gel techniques have even been successfully used to than would have been required had conventional pow- roduce multilayer ceramic-ceramic composites [13]. In der processing techniques been used this work, the phenomenon of Liesegang band forma tion was used to produce two-dimensional precipitated CuCrOa layers in silica gels. The thickness and spacing 2. Experimental of these layers were shown to be tailorable, and the bands were also shown to survive sintering tempera- 2. 1. Metal-ceramic composite synthesis tures as high as 1100C, indicating that high tempera ure anisotropic composites could be produced To produce the Ni/alumina and Ni/zirconia metal In this paper, we review our synthesizing a ceramic composites, first a 0.15 M absolute ethanol range of different high temperature ceramic matrix composites using sol-gel techniques. These composites vary from being metal-ceramic in nature(e.g. Ni/ a Al2O3, Fe/a-Al2O3, etc. ) to ceramic-ceramic (e.g. SiC articulate reinforced a-AL, O3). Throughout this work, it is the physical and mechanical properties of the g Sic composites, the composite microstructures and the property: microstructure: synthesis relationships which are the elements of primary interest. The advantages 3 gained from using the sol-gel type syntheses in place of conventional powder mixing and processing are numer ous. For instance. in the case of the metal-ceramic composites, extremely fine (often nanoscale)mi- crostructure with a high degree of dispersion between the metal and ceramic phases have been produced. As a result, these composites exhibit enhanced toughness and durability as well as a simultaneous potential for catal- ysis applications. Likewise, the efforts at producing 2-theta(degree ol-gel derived, Sic-reinforced alumina composites have resulted in materials with highly uniform and Fig. 2 ce of XRD patterns for th Alo posite:(a) dried, uncal homogeneous morphologies without the presence of cined (b)900C air calcined powder, and(c)1750.C hot- Sic agglomerates. Furthermore, while being lighter, pressed pellet
12 E.D. Rodeghiero et al. / Materials Science and Engineering A244 (1998) 11–21 increase in both fracture strength and fracture toughness for the ZrO2-reinforced Al2O3 system was reported. Rice attributed this beneficial behavior to the extreme homogeneity of the sol–gel derived composites. In 1984, Hoffman et al. also demonstrated the extensive composite homogeneity and dispersion that could be achieved through sol–gel approaches by synthesizing various types of di-phasic gels ranging from fine-scale photosensitive composites such as AgCl/SiO2 and CdS/SiO2 (where the AgCl and CdS phases were in crystalline form), to phase-separated phosphate and mixed oxide glasses such as CrPO4/SiO2, CePO4/SiO2 and Nd2O3/SiO2 (where both the matrix and minor phases were amorphous) [10,11]. At the same time, Roy et al. were also using these same synthesis procedures to produce the first sol–gel derived metal–ceramic composites (including Cu/Al2O3, Ni/Al2O3, Cu/ZrO2 and Cu/SiO2) [12]. In fact, Hoffman et al. and Roy et al. collectively performed the most extensive work on the sol–gel synthesis of ceramic matrix composites to the present day, thoroughly investigating over 30 different chemical systems. However, in their efforts no attention was given to high temperature consolidation, mechanical properties or structural applications. Finally, sol– gel techniques have even been successfully used to produce multilayer ceramic–ceramic composites [13]. In this work, the phenomenon of Liesegang band formation was used to produce two-dimensional precipitated CuCrO4 layers in silica gels. The thickness and spacing of these layers were shown to be tailorable, and the bands were also shown to survive sintering temperatures as high as 1100°C, indicating that high temperature anisotropic composites could be produced. In this paper, we review our work in synthesizing a range of different high temperature ceramic matrix composites using sol–gel techniques. These composites vary from being metal–ceramic in nature (e.g. Ni/aAl2O3, Fe/a-Al2O3, etc.) to ceramic–ceramic (e.g. SiC particulate reinforced a-Al2O3). Throughout this work, it is the physical and mechanical properties of the composites, the composite microstructures and the property:microstructure:synthesis relationships which are the elements of primary interest. The advantages gained from using the sol–gel type syntheses in place of conventional powder mixing and processing are numerous. For instance, in the case of the metal–ceramic composites, extremely fine (often nanoscale) microstructures with a high degree of dispersion between the metal and ceramic phases have been produced. As a result, these composites exhibit enhanced toughness and durability as well as a simultaneous potential for catalysis applications. Likewise, the efforts at producing sol–gel derived, SiC-reinforced alumina composites have resulted in materials with highly uniform and homogeneous morphologies without the presence of SiC agglomerates. Furthermore, while being lighter, Fig. 1. Sequence of XRD patterns for the processing of a 20/80 vol.% Ni/a-Al2O3 metal–ceramic composite; (a) dried, unreduced powder, (b) 1000°C hydrogen reduced powder, and (c) 1400°C hot-pressed pellet. stiffer and tougher than pure alumina, these composites were consolidated at lower temperatures and pressures than would have been required had conventional powder processing techniques been used. 2. Experimental 2.1. Metal–ceramic composite synthesis To produce the Ni/alumina and Ni/zirconia metal– ceramic composites, first a 0.15 M absolute ethanol Fig. 2. Sequence of XRD patterns for the processing of a 20/80 vol.% SiC(whisker)/a-Al2O3 ceramic–ceramic composite; (a) dried, uncalcined powder, (b) 900°C air calcined powder, and (c) 1750°C hotpressed pellet
E D. Rodeghiero et al/ Materials Science and Engineering 4244(1998)11-21 13 EHT=20. 00 kV Photo No. =28 Detector= Q Fig. 3. Backscattered electron SEM micrograph of the microstructure of a 5/95 vol. Ni/a-AL,O, composite; (Ni= light contrast, a-Al, O,= dark solution of either aluminum isopropoxide, Altwo the gels were vacuum dried and OCH(CH3)213(Aldrich Chemical), or zirconium isopro and to 231 groun 230 E esh powder form with the use of an oxide isopropanol complex, Zr[OCH(CH3)2]4 agate and pestle (CH,),CHOH(Aldrich Chemical), was prepared and Reduction of the gro heated until boiling. Next an aqueous 0. 2 M solution of formed by placing the powders in a quartz tube furnace nickel formate dihydrate, Ni(CHO,)2 2H,O(Johnson and heating under flowing 99.99% hydrogen(20 cm3 Matthey), was prepared and added to the ethanol solu- min-)for I h at a temperature of 1000-1100C. The tion. The addition of the aqueous metal salt solution to role of this heat treatment was 2-fold. First, the metal the alkoxide solution typically caused immediate gela- salt was decomposed to its metallic state(either Ni or tion of the ceramic precursor. Nevertheless, the gel Fe). Second, the condensation reaction in the ceramic formed was stirred vigorously for at least an additional phase was driven to completion, forcing the elimination 10 min at 70oC, in order to ensure complete inter-d of all excess water and hydroxyls persion of the metallic and ceramic precursors. Natu The metal-ceramic powders which resulted after re- rally, the amount of metal salt solution added depended duction were then uniaxially hot-pressed at 1400oC and pon the final metal-ceramic composite compositio 10 MPa for 3 h in 0.5 in inside diameter a-Al,O3 dies desired. The gel was then transferred to crystallization This hot-pressing was carried out under a reducing gas dishes for drying at 100C for 24 h. Finally, the dried mixture of Co and Co2 flowing at a total rate of 10 gels were ground with an agate mortar and pestle to a cm min-. The oxygen partial pressure typically em- powder size of 230 mesh. ployed was 10-12 atm. The role of this CO atmosphere In the case of the Fe/a-Al2O3 composites, a 500 ml, was not to promote further reduction but rather to 0. 1 M solution containing an appropriate ratio of alu- prevent the occurrence of any reoxidation. (A thorough minum nitrate nonahydrate, Al(NO3)3 9H,O(Aldrich review of the thermodynamics of the Ni/Alyo system in Chemical), and ferric nitrate nonahydrate, regard to both the 1000oC hydrogen reduction and the Fe(NO3)3. 9H,O(Aldrich Chemical), was first synthe- 1400C CO/CO2 hot-pressing has been previously re- sized. Then while stirring, aqueous 1 M NaOH was ported [14]. The thermodynamic aspects of the other slowly added until the ph of the nitrate mixture sur- composite systems investigated here are very similar. passed 7. The gel which resulted was next centrifuged at Following hot-pressing, the sintered composite pellets 7000 rpm for I h and twice washed with deionized were removed from the dies with the use of a circular water. Then, after repeating the centrifuge/washing step diamond blade saw. High quality, optically smooth
E.D. Rodeghiero et al. / Materials Science and Engineering A244 (1998) 11–21 13 Fig. 3. Backscattered electron SEM micrograph of the microstructure of a 5/95 vol.% Ni/a-Al2O3 composite; (Ni=light contrast, a-Al2O3=dark contrast). solution of either aluminum isopropoxide, Al [OCH(CH3)2]3 (Aldrich Chemical), or zirconium isopropoxide isopropanol complex, Zr[OCH(CH3)2]4 · (CH3)2CHOH (Aldrich Chemical), was prepared and heated until boiling. Next an aqueous 0.2 M solution of nickel formate dihydrate, Ni(CHO2)2 · 2H2O (Johnson Matthey), was prepared and added to the ethanol solution. The addition of the aqueous metal salt solution to the alkoxide solution typically caused immediate gelation of the ceramic precursor. Nevertheless, the gel formed was stirred vigorously for at least an additional 10 min at 70°C, in order to ensure complete inter-dispersion of the metallic and ceramic precursors. Naturally, the amount of metal salt solution added depended upon the final metal–ceramic composite composition desired. The gel was then transferred to crystallization dishes for drying at 100°C for 24 h. Finally, the dried gels were ground with an agate mortar and pestle to a powder size of 230 mesh. In the case of the Fe/a–Al2O3 composites, a 500 ml, 0.1 M solution containing an appropriate ratio of aluminum nitrate nonahydrate, Al(NO3)3 · 9H2O (Aldrich Chemical), and ferric nitrate nonahydrate, Fe(NO3)3 · 9H2O (Aldrich Chemical), was first synthesized. Then while stirring, aqueous 1 M NaOH was slowly added until the pH of the nitrate mixture surpassed 7. The gel which resulted was next centrifuged at 7000 rpm for 1 h and twice washed with deionized water. Then, after repeating the centrifuge/washing step two more times, the gels were vacuum dried and ground to 230 mesh powder form with the use of an agate mortar and pestle. Reduction of the ground precursor gels was performed by placing the powders in a quartz tube furnace and heating under flowing 99.99% hydrogen (20 cm3 min−1 ) for 1 h at a temperature of 1000–1100°C. The role of this heat treatment was 2-fold. First, the metal salt was decomposed to its metallic state (either Ni or Fe). Second, the condensation reaction in the ceramic phase was driven to completion, forcing the elimination of all excess water and hydroxyls. The metal–ceramic powders which resulted after reduction were then uniaxially hot-pressed at 1400°C and 10 MPa for 3 h in 0.5 in. inside diameter a-Al2O3 dies. This hot-pressing was carried out under a reducing gas mixture of CO and CO2 flowing at a total rate of 10 cm3 min−1 . The oxygen partial pressure typically employed was 10−12 atm. The role of this CO atmosphere was not to promote further reduction but rather to prevent the occurrence of any reoxidation. (A thorough review of the thermodynamics of the Ni/Al/O system in regard to both the 1000°C hydrogen reduction and the 1400°C CO/CO2 hot-pressing has been previously reported [14]. The thermodynamic aspects of the other composite systems investigated here are very similar.) Following hot-pressing, the sintered composite pellets were removed from the dies with the use of a circular diamond blade saw. High quality, optically smooth
E D. Rodeghiero et al/ Materials Science and Engineering 4244(1998)11-21 EHT=10.00 KV 8 mm 30um Photo No =18 Detector= Q Fig. 4. Backscattered electron SEM micrograph of the microstructure of a 50/50 vol. Fe/a-Al20 site: (Fe= light contrast, Al,,= dark surfaces were prepared by flattening the pellet faces 2.3. Ceramic-ceramic composite synthesis with a 20 um metal bonded diamond wheel(Struers), coarse polishing with Sic paper, and then fine polishing The preparation of the Sic(particulate)-reinforced with 6 um diamond paste impregnated Texmet polish lumina composites consisted of the following steps ing cloth(Buehler). Finally, an aqueous ultrasonic First a 0.15m absolute ethanol solution of aluminum to remove pellet surface contamination ier polishing isopropoxide was again prepared and heated to boil- cleaning bath was utilized immediately ing. Next, an appropriate amount of either SiC whiskers(≈ I um diameter by≈l5 um long) or Sic 2. 2. Doped metal-ceramic composite synthesi platelets(0.5-5 um thick by 5-70 um diameter)(John In certain instances. some of the metal-ceramic com- son Matthey) was added to the ceramic precursor solu- osites were doped with additional phases or com tion while stirring. After 5 min, just enough water to pounds. Two of the more extensively investigated gel the precursor solution was added. The mixture was xamples were Ni/a-Al2O3 doped with ZrO2 and Ni/a left to stir continuously at 70C until the gel was Al,O, doped with Cr2O3. In the case of the Zro viscous enough to prevent SiC settling. The gel was then transferred to crystallization dishes and dried for pared by simply adding small amounts of the zirconium 24 h at 100C. Finally, the dried gels were delicately alkoxide to the initial aluminum isopropoxide solution. hand ground and sieved in the same fashion as the Reduction and hot-pressing were then carried out as metal-ceramic precursors. To fully transform the ce- described above. To produce the Cr,O, doped material, ramic gel to A2 O, the dried powders were calcined in a chromium formate salt (added to the nickel formate a quartz tube furnace at 900 C in air for 1 h. This was dihydrate solution) was used as the dopant source. then followed by uniaxial hot-pressing at 1750.C and Again, reduction and hot-pressing were carried out 35 MPa for 3 h. The 0.5 in. inside diameter dies used normally. In this case, however, even though the in this case consisted of high strength graphite(Poco chromium source was incorporated along with the N raphite, grade ZxF-5Q) and were enclosed in a the form of a salt, the employed hydrogen reduction chamber backfilled with argon for the duration of the temperature of 1000 C was not high enough to reduce high temperature exposure. Handling and post-process- the chromium to its metallic state and hence, an alu- ing of the sintered Sic/a-Al2O3 was performed in a mina-rich Al,O3/Cr,O similar fashion as described for the metal-ceramic duced as the ceramic phase composites
14 E.D. Rodeghiero et al. / Materials Science and Engineering A244 (1998) 11–21 Fig. 4. Backscattered electron SEM micrograph of the microstructure of a 50/50 vol.% Fe/a-Al2O3 composite; (Fe=light contrast, a-Al2O3=dark contrast). surfaces were prepared by flattening the pellet faces with a 20 mm metal bonded diamond wheel (Struers), coarse polishing with SiC paper, and then fine polishing with 6 mm diamond paste impregnated Texmet® polishing cloth (Buehler). Finally, an aqueous ultrasonic cleaning bath was utilized immediately after polishing to remove pellet surface contamination. 2.2. Doped metal–ceramic composite synthesis In certain instances, some of the metal–ceramic composites were doped with additional phases or compounds. Two of the more extensively investigated examples were Ni/a-Al2O3 doped with ZrO2 and Ni/aAl2O3 doped with Cr2O3. In the case of the ZrO2 modified Ni/a-Al2O3, the doped composites were prepared by simply adding small amounts of the zirconium alkoxide to the initial aluminum isopropoxide solution. Reduction and hot-pressing were then carried out as described above. To produce the Cr2O3 doped material, a chromium formate salt (added to the nickel formate dihydrate solution) was used as the dopant source. Again, reduction and hot-pressing were carried out normally. In this case, however, even though the chromium source was incorporated along with the Ni in the form of a salt, the employed hydrogen reduction temperature of 1000°C was not high enough to reduce the chromium to its metallic state, and hence, an alumina-rich Al2O3/Cr2O3 solid solution (ruby) was produced as the ceramic phase. 2.3. Ceramic–ceramic composite synthesis The preparation of the SiC(particulate)-reinforced alumina composites consisted of the following steps. First, a 0.15 M absolute ethanol solution of aluminum isopropoxide was again prepared and heated to boiling. Next, an appropriate amount of either SiC whiskers (:1 mm diameter by :15 mm long) or SiC platelets (0.5–5 mm thick by 5–70 mm diameter) (Johnson Matthey) was added to the ceramic precursor solution while stirring. After 5 min, just enough water to gel the precursor solution was added. The mixture was left to stir continuously at 70°C until the gel was viscous enough to prevent SiC settling. The gel was then transferred to crystallization dishes and dried for 24 h at 100°C. Finally, the dried gels were delicately hand ground and sieved in the same fashion as the metal–ceramic precursors. To fully transform the ceramic gel to Al2O3, the dried powders were calcined in a quartz tube furnace at 900°C in air for 1 h. This was then followed by uniaxial hot-pressing at 1750°C and 35 MPa for 3 h. The 0.5 in. inside diameter dies used in this case consisted of high strength graphite (Poco Graphite, grade ZXF-5Q) and were enclosed in a chamber backfilled with argon for the duration of the high temperature exposure. Handling and post-processing of the sintered SiC/a-Al2O3 was performed in a similar fashion as described for the metal–ceramic composites
D. Rodeghiero et al/ Materials Science and Engineering 4244(1998)11-21 2. 4. Characterization The density of all hot-pressed composites was deter mined by calculating the mass and volume of each X-ray diffraction(XRD) of both the precursor and pellet after cutting into a prismatic configuration. In heat treated powders as well as the sintered pellets wa addition, the relative or percentage density of each performed using a Scintag Pad X diffractometer with material was calculated by dividing the measured den CuKa radiation. Reduced metal-ceramic powders were sity by the theoretical density based on composite com- encapsulated in 1.0 mm diameter glass capillaries under position. The resulting number, always slightly argon to prevent reoxidation of the very small metallic < 100%, was used as an indicator of residual porosity particles present at this stage of processing In the case of the metal-ceramic composites, electrical Polished, sintered microstructures were investigated resistivity of the hot-pressed pellets was also measured with both optical and scanning electron microscopy with the use of a standard two-probe ohmmeter, in (SEM). The optical microscope used was an inverted order to analyze percolation of the metallic phase. Only PME Olympus instrument while the scanning electron approximate measurements needed to be performed microscope employed was a Leica 440 Stereoscan ma- since the quantitative difference between percolated and chine with both secondary and backscattered electron non-percolated readings was several orders of magni- imaging capabilities. The SEM was also occasionally tude sed to image composite fracture surfaces The elastic constants of the isotropic metal-ceramic composites were determined using an ultrasonic tech- nique previously reported [15]. a similar approach was adopted for the SiC/a-Al2O3 materials; however, due to the uniaxial nature of the hot-pressing employed, the Sic whiskers and platelets had a tendency to lie perpen- dicular to the hot-pressing direction, resulting in an- isotropic composites of the transverse isotropy' type symmetry [16]. This required the incorporation of a quipment was needed but measurements had to be performed in several additional directions including at platelet alignment. The exact mathematics of this ex tended analysis will not be presented here, but the reader is referred to the derivations of Neighbours and Schacher for more insight [17] fracture toughness testing was performed by cutting 25m the sintered composites into beams, machining chevron notches into the beams, and then breaking the speci mens to complete failure on an Instron Model 1125 mechanical testing instrument equipped with a three point bending fixture. A linear variable displacement transducer(RDP-Electrosense, model RDP D5/10G8) was used to measure beam load-point displacement while a piezoelectric transducer (Kistler Instrument, model 9301A) recorded specimen load under constant displacement rate conditions. Data in the form of load/ displacement curves were collected with the use of a computerized data acquisition system. The beam di- mensions used throughout the testing were 1.9 x 1.9 8.0 mm. and the chevron notches were machined with a Well wire saw (Ahlburg Technical Equipment, model 3242) using 220 um diameter diamond impreg nated steel wire. The included angle of the chevron notch was maintained at x94o while all other sample and notch dimensions were in compliance with the Fig. 5. Optical micrographs of the microstructure of a 20 /80 vol% SiC(whisker)/a-Al,O, composite;(a) pellet face perpendicular to hot. work by Wu[18]. A much more thorough explanation ressing direction and(b) pellet face parallel to hot-pressing direction; of the toughness testing apparatus and process used is SiC=light contrast, a-Al, O,=dark contrast currently in preparation [191
E.D. Rodeghiero et al. / Materials Science and Engineering A244 (1998) 11–21 15 2.4. Characterization X-ray diffraction (XRD) of both the precursor and heat treated powders as well as the sintered pellets was performed using a Scintag Pad X diffractometer with CuKa radiation. Reduced metal–ceramic powders were encapsulated in 1.0 mm diameter glass capillaries under argon to prevent reoxidation of the very small metallic particles present at this stage of processing. Polished, sintered microstructures were investigated with both optical and scanning electron microscopy (SEM). The optical microscope used was an inverted PME Olympus instrument while the scanning electron microscope employed was a Leica 440 Stereoscan machine with both secondary and backscattered electron imaging capabilities. The SEM was also occasionally used to image composite fracture surfaces. The density of all hot-pressed composites was determined by calculating the mass and volume of each pellet after cutting into a prismatic configuration. In addition, the relative or percentage density of each material was calculated by dividing the measured density by the theoretical density based on composite composition. The resulting number, always slightly B100%, was used as an indicator of residual porosity. In the case of the metal–ceramic composites, electrical resistivity of the hot-pressed pellets was also measured, with the use of a standard two-probe ohmmeter, in order to analyze percolation of the metallic phase. Only approximate measurements needed to be performed since the quantitative difference between percolated and non-percolated readings was several orders of magnitude. The elastic constants of the isotropic metal–ceramic composites were determined using an ultrasonic technique previously reported [15]. A similar approach was adopted for the SiC/a-Al2O3 materials; however, due to the uniaxial nature of the hot-pressing employed, the SiC whiskers and platelets had a tendency to lie perpendicular to the hot-pressing direction, resulting in anisotropic composites of the ‘transverse isotropy’ type symmetry [16]. This required the incorporation of a much more complex acoustic analysis. (No additional equipment was needed but measurements had to be performed in several additional directions including at least one direction oblique to the plane of whisker/ platelet alignment.) The exact mathematics of this extended analysis will not be presented here, but the reader is referred to the derivations of Neighbours and Schacher for more insight [17]. Fracture toughness testing was performed by cutting the sintered composites into beams, machining chevron notches into the beams, and then breaking the specimens to complete failure on an Instron Model 1125 mechanical testing instrument equipped with a threepoint bending fixture. A linear variable displacement transducer (RDP-Electrosense, model RDP D5/10G8) was used to measure beam load-point displacement while a piezoelectric transducer (Kistler Instrument, model 9301A) recorded specimen load under constant displacement rate conditions. Data in the form of load/ displacement curves were collected with the use of a computerized data acquisition system. The beam dimensions used throughout the testing were 1.9×1.9× 8.0 mm3 , and the chevron notches were machined with a Well® wire saw (Ahlburg Technical Equipment, model 3242) using 220 mm diameter diamond impregnated steel wire. The included angle of the chevron notch was maintained at :94° while all other sample and notch dimensions were in compliance with the work by Wu [18]. A much more thorough explanation of the toughness testing apparatus and process used is currently in preparation [19]. Fig. 5. Optical micrographs of the microstructure of a 20/80 vol.% SiC(whisker)/a-Al2O3 composite; (a) pellet face perpendicular to hotpressing direction and (b) pellet face parallel to hot-pressing direction; (SiC=light contrast, a-Al2O3=dark contrast)
D. Rodeghiero et al/ Materials Science and Engineering 4244(1998)11-21 Density and Youngs modulus data of sol-gel derived Ni/o-AlO3 metal-ceramic composites Ni/a-Al2O3 volume Relative density Young's modulus orrected Youngs modulus, (g cm-3) Eo (GPa) 9988 353 44555 9992 891b 214° The critical stress intensity factors were determined can be added simultaneously. A secondary approach through a work of fracture analysis. This involved was used for the synthesis of the Fe/a-Al2O3 com- neasuring the area under the load /displacement curve posites. In this case, a coprecipitation technique involv of each composite beam through integration and then ing the addition of Naoh to an aqueous solution of dividing by twice the fracture area to obtain the work aluminum and ferric nitrates was utilized. The benefits of fracture, ywof. Then, assuming plane strain, Kiwor was of this approach are simplicity and low cost calculated using the following equati To prepare the SiC-reinforced alumina materials, a K1wo={(2E)/(1-v2) imilar yet more complex route had to be utilized. First an appropriate amount of Sic whiskers or platelets was where E is the Youngs modulus of the composite and dispersed in a stirred aluminum isopropoxide solution. v is Poissons ratio(both determined through the acous- Enough water to gel the solution was next added tic testing). In the case of the SiC/a-Al,O3 composites, However, due to the relatively large mass of the Sic more than one value of e and v exist for reasons particulates, the gel initially formed was typically not previously discussed. Furthermore, it is not at all clear viscous enough to support the whiskers or platelets and which set of the anisotropic elastic values should be prevent settling. Hence, through stirring the mixture at substituted into Eq 20]. However, since the degree elevated temperature(&70oC), the viscosity of the gel of elastic anisotropy in the SiC/a-Al2O3 composites was was raised until the SiC particulates could no longer found to be rather small anyway, the value of e in the settle. Finally, drying was performed at a 100oC. This plane of the Sic whiskers and platelets (i.e. parallel to overall procedure was devised and first performed by the length of the chevron-notched beams)was the one J.J. Lannutti et al. in 1984 [21]. However, to our ed to determine the kiwor values knowledge, no high temperature consolidation or me- chanical property data was ever subsequently reported) The primary attribute of this sol-gel technique is its 3. Results and discussion excellent ability to homogeneously disperse the SiC particulates in a stirring liquid and then freeze' them To produce the metal-ceramic composites, our most into position, resulting in a very high degree of final widely used technique consisted of first preparing a composite uniformity and the complete elimination of metal alkoxide solution as previously discussed. To this SiC agglomerates, a challenge which is almost insur was then added an aqueous solution of a metal salt mountable using conventional powder mixing tech- which resulted in spontaneous hydrolysis and conden- niques sation of the ceramic precursor and incorporation of Fig. I depicts the typical crystallographic evolution le metal salt into the growing ceramic gel at or near of an alumina-based metal-ceramic composite. The the molecular level. Attributes of this approach are its XRD pattern in Fig. 1(a) corresponds to a 20/80 vol. mplicity, its very high homogeneity, and its high Ni/a-Al2O3 precursor gel in the dried state just after degree of chemical flexibility due to the large selection grinding. While the absence of ceramic phase reflections of metal alkoxides and metal salts available. This tech- is to be expected for the air-dried materials, the fact nique is also very favorable for doping, since small that there are also no apparent XRD features at amounts of additional metal alkoxides or metal salts ibutable to the nickel formate salt has significant
16 E.D. Rodeghiero et al. / Materials Science and Engineering A244 (1998) 11–21 Table 1 Density and Young’s modulus data of sol–gel derived Ni/a-Al2O3 metal–ceramic compositesa Ni/a-Al Relative density Young’s modulus, Porosity corrected Young’s modulus, 2O3 volume Density E0 (g cm (% of theoretical) E (GPa) (GPa) −3 composition ) — 390c 0/100 3.97b — 5/95 4.18 99.2 388 379 10/90 4.41 98.7 339 353 15/85 4.55 355 96.6 320 20/80 4.77 96.2 344 306 30/70 5.12 93.9 288 348 33/67 5.23 93.2 251 310 40/60 5.54 352 93.2 285 50/50 5.97 313 92.8 250 100/0 8.91b — 214c — a Adopted from [15]. b Reference [29]. c Reference [30]. The critical stress intensity factors were determined through a work of fracture analysis. This involved measuring the area under the load/displacement curve of each composite beam through integration and then dividing by twice the fracture area to obtain the work of fracture, gwof. Then, assuming plane strain, KIwof was calculated using the following equation: KIwof={(2E · gwof)/(1−n 2 )}1/2 (1) where E is the Young’s modulus of the composite and n is Poisson’s ratio (both determined through the acoustic testing). In the case of the SiC/a-Al2O3 composites, more than one value of E and n exist for reasons previously discussed. Furthermore, it is not at all clear which set of the anisotropic elastic values should be substituted into Eq. (1) [20]. However, since the degree of elastic anisotropy in the SiC/a-Al2O3 composites was found to be rather small anyway, the value of E in the plane of the SiC whiskers and platelets (i.e. parallel to the length of the chevron-notched beams) was the one used to determine the KIwof values. 3. Results and discussion To produce the metal–ceramic composites, our most widely used technique consisted of first preparing a metal alkoxide solution as previously discussed. To this was then added an aqueous solution of a metal salt which resulted in spontaneous hydrolysis and condensation of the ceramic precursor and incorporation of the metal salt into the growing ceramic gel at or near the molecular level. Attributes of this approach are its simplicity, its very high homogeneity, and its high degree of chemical flexibility due to the large selection of metal alkoxides and metal salts available. This technique is also very favorable for doping, since small amounts of additional metal alkoxides or metal salts can be added simultaneously. A secondary approach was used for the synthesis of the Fe/a-Al2O3 composites. In this case, a coprecipitation technique involving the addition of NaOH to an aqueous solution of aluminum and ferric nitrates was utilized. The benefits of this approach are simplicity and low cost. To prepare the SiC-reinforced alumina materials, a similar yet more complex route had to be utilized. First, an appropriate amount of SiC whiskers or platelets was dispersed in a stirred aluminum isopropoxide solution. Enough water to gel the solution was next added. However, due to the relatively large mass of the SiC particulates, the gel initially formed was typically not viscous enough to support the whiskers or platelets and prevent settling. Hence, through stirring the mixture at elevated temperature (:70°C), the viscosity of the gel was raised until the SiC particulates could no longer settle. Finally, drying was performed at :100°C. (This overall procedure was devised and first performed by J.J. Lannutti et al. in 1984 [21]. However, to our knowledge, no high temperature consolidation or mechanical property data was ever subsequently reported). The primary attribute of this sol–gel technique is its excellent ability to homogeneously disperse the SiC particulates in a stirring liquid and then ‘freeze’ them into position, resulting in a very high degree of final composite uniformity and the complete elimination of SiC agglomerates, a challenge which is almost insurmountable using conventional powder mixing techniques. Fig. 1 depicts the typical crystallographic evolution of an alumina-based metal–ceramic composite. The XRD pattern in Fig. 1(a) corresponds to a 20/80 vol.% Ni/a-Al2O3 precursor gel in the dried state just after grinding. While the absence of ceramic phase reflections is to be expected for the air-dried materials, the fact that there are also no apparent XRD features attributable to the nickel formate salt has significant
D. Rodeghiero et al/ Materials Science and Engineering 4244(1998)11-21 17 Table 2 Density and Youngs modulus data of sol-gel derived SiC(whisker)/a-Al2O3 ceramic-ceramic composites SiC(w)/a-AlO, volume Density Relative density Young's modulus gcm) C of theoretical) E⊥GPa) 3.22 Reference [30]- implications, for it indicates that the metal source is investigation is whether or not the Sic whiskers and homogeneously dispersed throughout the alumina gel platelets begin oxidizing during the 900oC air calcina- at a very fine scale and high level of dispersion. The tion. Theoretically, oxidation of Sic should take place XRD pattern for this same powder after reduction is in air at or near a temperature of 850C [25]. However, depicted in Fig. 1(b). Note that the Ni lll and 200 the glassy silica which forms as a result is known to peaks are now discernible but very broad. Typically, passivate the remaining Sic and deter further oxida the Ni particle size of the composite powders in this tion. Furthermore, the surrounding alumina gel in these reduced state is calculated (using Scherrer's equation composite powders might also help to delay the oxida [22] and the half-height width of the lll peak)to be on tion process. Indeed, the Sic whisker peaks shown in the order of 10-30 nm depending on the composite's Fig. 2 do not seem to be affected in any way by the composition. For the 20/80 vol. Ni/ a-Al2O3 material calcination. Unfortunately, a reducing atmosphere such depicted in Fig. 1(b), the Ni size is found to be 20 nm. as H2 or CO could not be used in the heat treatment of Interestingly, still no ceramic phase reflections are ob- the SiC/a-AlO3 precursor powders since only air was served despite the reduction temperature of 1000C. found to adequately remove all the residual organics This is because a-Al2O3 typically does not form from from the alumina gel. Interestingly, this problem is not either amorphous or metastable cubic alumina until a an issue in the 1000oC hydrogen reduction of the temperature of 1100-1200C [23]. Finally, Fig. I(c) metal-ceramic composites, most likely because of the displays the XRD pattern for the same composite after sheer presence of the metals. Indeed, Ni and Fe are oth the reduction and hot-pressing steps that known to be effective catalysts in gas phase reactions the Ni peaks are still present and now much sharper and therefore, facilitate the removal of organics fro due to coarsening. In addition, the a-Al2O3 phase is the matrix even in the absence of O2 also now present, accounting for all the additional Fig. 3 displays a backscattered electron SEM mi- observed peak crograph of the sintered microstructure of a 5/95 vol% Fig. 2 displays a similar crystallographic evolut Ni/a-Al2O3 composite. In most of the metal-ceramic for a 20/80 vol. SiC(whisker)/a-Al2O3 composite m composite systems investigated, the metal volume fr terial In Fig. 2(a), the XRD pattern corresponding to tion was varied from 5 to 50 vol % hence, the com- the dried and ground precursor powder shows only posite depicted in Fig 3 represents the lower extreme in reflections due to the Sic whiskers, which are found to terms of metal loading. Note both the uniformity and composed of the 3C polytype (or 'B-SiC ). The Sic high degree of dispersion in the microstructure, at platelets used in this work, on the other hand, were tributes which would be impossible to reproduce using found to be composed of numerous Sic polytypes more conventional powder mixing techniques. The mi- including 2H, 4H, 6H and some 3C. As can be seen in crostructure consists of isolated, equiaxed Ni particle Fig. 2(b), air calcination of the SiC(whisker )/a-Al2O3 embedded in the a-Al2O3 matrix. This is confirmed by precursor powder at 900oC leaves the XRD pattern resistivity measurements which find the composite to be rgely unchanged. However, following hot-pressing at insulating (i.e. the metallic phase is not continuous) 1750oC, the a-Al,O, phase is clearly present(Fig. 2(c), The average size of the isolated Ni particles, determined just as in the case of the metal-ceramic composites. using a linear intercept technique previously reported Interestingly, the Sic whiskers can be seen to maintain [26], is x0.5 um. Such a small particle size is unusual their 3C polytype through all high temperature treat- for a metallic Ni phase hot-pressed for 3 h at a temper ments. The Sic platelets, on the other hand, have been ature only x 50C below its melting point. Due to the observed to favor the 6H polytype after high tempera- fine dispersion of the metal and its relatively small ure processing, which is consistent with the observa- particle size even in the fully sintered form, these mate tions of others [24]. One aspect currently under rials are ideal candidates for catalytic applications
E.D. Rodeghiero et al. / Materials Science and Engineering A244 (1998) 11–21 17 Table 2 Density and Young’s modulus data of sol–gel derived SiC(whisker)/a-Al2O3 ceramic–ceramic composites SiC(w)/a-Al Density Young’s modulus, Relative density Young’s modulus, 2O3 volume E EÞ (GPa) (g cm (GPa) −3 composition (% of theoretical) ) 390 — b 3.97a 390b 0/100 5/95 3.88 98.7 383 365 10/90 3.83 98.3 402 372 20/80 3.75 98.2 369 397 450b 450b 100/0 3.22a — a Reference [29]. b Reference [30]. implications, for it indicates that the metal source is homogeneously dispersed throughout the alumina gel at a very fine scale and high level of dispersion. The XRD pattern for this same powder after reduction is depicted in Fig. 1(b). Note that the Ni 111 and 200 peaks are now discernible but very broad. Typically, the Ni particle size of the composite powders in this reduced state is calculated (using Scherrer’s equation [22] and the half-height width of the 111 peak) to be on the order of 10–30 nm depending on the composite’s composition. For the 20/80 vol.% Ni/a-Al2O3 material depicted in Fig. 1(b), the Ni size is found to be 20 nm. Interestingly, still no ceramic phase reflections are observed despite the reduction temperature of 1000°C. This is because a-Al2O3 typically does not form from either amorphous or metastable cubic alumina until a temperature of 1100–1200°C [23]. Finally, Fig. 1(c) displays the XRD pattern for the same composite after both the reduction and hot-pressing steps. Note that the Ni peaks are still present and now much sharper due to coarsening. In addition, the a-Al2O3 phase is also now present, accounting for all the additional observed peaks. Fig. 2 displays a similar crystallographic evolution for a 20/80 vol.% SiC(whisker)/a-Al2O3 composite material. In Fig. 2(a), the XRD pattern corresponding to the dried and ground precursor powder shows only reflections due to the SiC whiskers, which are found to be composed of the 3C polytype (or ‘b-SiC’). The SiC platelets used in this work, on the other hand, were found to be composed of numerous SiC polytypes including 2H, 4H, 6H and some 3C. As can be seen in Fig. 2(b), air calcination of the SiC(whisker)/a-Al2O3 precursor powder at 900°C leaves the XRD pattern largely unchanged. However, following hot-pressing at 1750°C, the a-Al2O3 phase is clearly present (Fig. 2(c)), just as in the case of the metal–ceramic composites. Interestingly, the SiC whiskers can be seen to maintain their 3C polytype through all high temperature treatments. The SiC platelets, on the other hand, have been observed to favor the 6H polytype after high temperature processing, which is consistent with the observations of others [24]. One aspect currently under investigation is whether or not the SiC whiskers and platelets begin oxidizing during the 900°C air calcination. Theoretically, oxidation of SiC should take place in air at or near a temperature of 850°C [25]. However, the glassy silica which forms as a result is known to passivate the remaining SiC and deter further oxidation. Furthermore, the surrounding alumina gel in these composite powders might also help to delay the oxidation process. Indeed, the SiC whisker peaks shown in Fig. 2 do not seem to be affected in any way by the calcination. Unfortunately, a reducing atmosphere such as H2 or CO could not be used in the heat treatment of the SiC/a-Al2O3 precursor powders since only air was found to adequately remove all the residual organics from the alumina gel. Interestingly, this problem is not an issue in the 1000°C hydrogen reduction of the metal–ceramic composites, most likely because of the sheer presence of the metals. Indeed, Ni and Fe are known to be effective catalysts in gas phase reactions, and therefore, facilitate the removal of organics from the matrix even in the absence of O2. Fig. 3 displays a backscattered electron SEM micrograph of the sintered microstructure of a 5/95 vol.% Ni/a-Al2O3 composite. In most of the metal–ceramic composite systems investigated, the metal volume fraction was varied from 5 to 50 vol.%; hence, the composite depicted in Fig. 3 represents the lower extreme in terms of metal loading. Note both the uniformity and high degree of dispersion in the microstructure, attributes which would be impossible to reproduce using more conventional powder mixing techniques. The microstructure consists of isolated, equiaxed Ni particles embedded in the a-Al2O3 matrix. This is confirmed by resistivity measurements which find the composite to be insulating (i.e. the metallic phase is not continuous). The average size of the isolated Ni particles, determined using a linear intercept technique previously reported [26], is :0.5 mm. Such a small particle size is unusual for a metallic Ni phase hot-pressed for 3 h at a temperature only :50°C below its melting point. Due to the fine dispersion of the metal and its relatively small particle size even in the fully sintered form, these materials are ideal candidates for catalytic applications
D. Rodeghiero et al/ Materials Science and Engineering 4244(1998)11-21 Nia-Al,O 3 20 enforcement content(vol %) Fig. 6. Fracture toughness data of the sol-gel derived composites as a function of reinforcement content. a backscattered electron SEM micrograph of a hot- ing approaches. Indeed, such agglomerates have been pressed 50/50 vol. Fe/a-AL, O, composite is shown in shown to be detrimental to the mechanical properties Fig. 4. As would be expected at higher metal contents, (particularly the fracture strength) of Sic-reinforced the metallic phase of this composite appears continuous ceramics[28]. Finally, virtually all of the whiskers in the and somewhat coarser. This continuity is confirmed by micrographs shown here are seen to be intact and the resistivity experiments which find the material to be undamaged, with good bonding between whisker and conducting. Furthermore, the transition from insulating matrix to conducting behavior for all of the metal-ceramic The density, relative density and Youngs modulus of composites investigated in this work occurs between a complete series of Ni/a-Al2O3 metal-ceramic com- metal fractions of 15 and 20 vol % This range agrees posites are listed in Table 1. Literature values for pure extremely well with percolation theory which predicts Ni and pure a-Al2O3 are also included for reference. As hree-dimensional continuity in random systems at would be expected, the density of the composites near a volume fraction of 17%[27]. However, the most creases with Ni content. More importantly, the relative important features of Fig. 4 are the extremely high densities of the cermets are all high, indicating an dispersion and uniformity which are present despite the effective sintering process. The Youngs modulus values now rather large metal volume fraction vary steadily between the pure Ni and a-Al2O3 bounds Fortunately, the excellent microstructures provided However, these numbers are somewhat less than those by the sol-gel techniques used in this work are not predicted by theoretical constructs such as the Hashin limited to the metal-ceramic composites. Fig. 5(a) Shtrikman upper and lower bounds to the Halpin-Tsai shows an optical micrograph of a hot-pressed 20 /80 model [31]. This is due to the effects of porosity on the vol. SiC(whisker)/a-Al,O, ceramic-ceramic com- Youngs modulus of the materials. Phani et al. studied posite. Since the view shown is of a pellet face perpen- several different materials and arrived at the following dicular to the hot-pressing direction, most of the Sic equation for the relationship between porosity and whiskers present are seen to lie flat; however, some elastic modulus equiaxed features due to whiskers perpendicular or at inclined angles to this pellet face are also visible. Fig E=Eo[l-PPn+I 5(b)shows an optical micrograph of the same com- where E is the actual Youngs modulus with porosity posite but of a face parallel to the hot-pressing direc- present, Eo is the theoretical modulus or modulus with- tion. As a result, most whiskers intersect the plane of out porosity, the quantity [1-Pl is the relative density the picture. Furthermore, the few whiskers which do lie (P is the percentage porosity), and the exponent n is an in the plane of the micrograph are oriented horizontally empirical constant commonly taken as 1 [32]. By virtue expected due to the hot-pressing direction in of Eq.(2), Table I also lists the Youngs modulus mple. The primary feature to notice in these figures values the Ni/a-Al2O3 composites would have if the lack of Sic whisker agglomerates which are a very porosity were present(Eo). These values are larger than common and serious problem when using powder mix the actual modulus values but more in line with theo-
18 E.D. Rodeghiero et al. / Materials Science and Engineering A244 (1998) 11–21 Fig. 6. Fracture toughness data of the sol–gel derived composites as a function of reinforcement content. A backscattered electron SEM micrograph of a hotpressed 50/50 vol.% Fe/a-Al2O3 composite is shown in Fig. 4. As would be expected at higher metal contents, the metallic phase of this composite appears continuous and somewhat coarser. This continuity is confirmed by the resistivity experiments which find the material to be conducting. Furthermore, the transition from insulating to conducting behavior for all of the metal–ceramic composites investigated in this work occurs between metal fractions of 15 and 20 vol.%. This range agrees extremely well with percolation theory which predicts three-dimensional continuity in random systems at or near a volume fraction of 17% [27]. However, the most important features of Fig. 4 are the extremely high dispersion and uniformity which are present despite the now rather large metal volume fraction. Fortunately, the excellent microstructures provided by the sol–gel techniques used in this work are not limited to the metal–ceramic composites. Fig. 5(a) shows an optical micrograph of a hot-pressed 20/80 vol.% SiC(whisker)/a-Al2O3 ceramic–ceramic composite. Since the view shown is of a pellet face perpendicular to the hot-pressing direction, most of the SiC whiskers present are seen to lie flat; however, some equiaxed features due to whiskers perpendicular or at inclined angles to this pellet face are also visible. Fig. 5(b) shows an optical micrograph of the same composite but of a face parallel to the hot-pressing direction. As a result, most whiskers intersect the plane of the picture. Furthermore, the few whiskers which do lie in the plane of the micrograph are oriented horizontally as expected due to the hot-pressing direction in the sample. The primary feature to notice in these figures is the lack of SiC whisker agglomerates which are a very common and serious problem when using powder mixing approaches. Indeed, such agglomerates have been shown to be detrimental to the mechanical properties (particularly the fracture strength) of SiC-reinforced ceramics [28]. Finally, virtually all of the whiskers in the micrographs shown here are seen to be intact and undamaged, with good bonding between whisker and matrix. The density, relative density and Young’s modulus of a complete series of Ni/a-Al2O3 metal–ceramic composites are listed in Table 1. Literature values for pure Ni and pure a-Al2O3 are also included for reference. As would be expected, the density of the composites increases with Ni content. More importantly, the relative densities of the cermets are all high, indicating an effective sintering process. The Young’s modulus values vary steadily between the pure Ni and a-Al2O3 bounds. However, these numbers are somewhat less than those predicted by theoretical constructs such as the Hashin– Shtrikman upper and lower bounds to the Halpin–Tsai model [31]. This is due to the effects of porosity on the Young’s modulus of the materials. Phani et al. studied several different materials and arrived at the following equation for the relationship between porosity and elastic modulus: E=E0 . [1−P] 2n+1 (2) where E is the actual Young’s modulus with porosity present, E0 is the theoretical modulus or modulus without porosity, the quantity [1−P] is the relative density (P is the percentage porosity), and the exponent n is an empirical constant commonly taken as 1 [32]. By virtue of Eq. (2), Table 1 also lists the Young’s modulus values the Ni/a-Al2O3 composites would have if no porosity were present (E0). These values are larger than the actual modulus values but more in line with theo-
E D. Rodeghiero et al/ Materials Science and Engineering 4244(1998)11-21 EHT=10. 00 KV WD= 11 m 300nm Photo No =72 Detector= SE Fig. 7. Secondary electron SEM micrograph demonstrating ductile phase crack bridging in a 5/95 vol. Nif/,, metal-ceramic composite; retical predictions, suggesting that the porosity of the ment comes from classical ductile phase crack bridging composites(even though low) is indeed having a slight [26, 35, 36]. For the Sic/a-Al2O3 composites, added effect toughness comes from numerous reinforcement/crack Table 2 shows the same data for the SiC(whisker )/a- interactions including the widely accepted crack deflec Al,O3 composites synthesized by the sol-gel technique. tion, interface debonding, crack bridging, and whisker, In the case of these materials, however, two values of platelet pullout mechanisms [37, 38]. A secondary e Youngs modulus exist due to transverse isotropy. electron SEM micrograph exhibiting ductile phase Ey corresponds to the elastic modulus in the plane of crack bridging is depicted in Fig. 7. The figure shows a the whiskers (i.e. perpendicular to the hot-pressing di- 5/95 vol. Ni/ a-Al2O3 composite into which a crack rection) while E, corresponds to the elastic modulus has been introduced by performing a Vickers hardness perpendicular to the whiskers (i.e. parallel to the hot- indentation (not shown). Note the three Ni particles pressing direction). Note that, indeed, E>E for each which span the crack plane and hence increase the of the composites studied, as expected. In regard to the overall toughness of the composite. In related work, sintering of the composites listed in Table 2, notice that Cr,O3 doping of the Ni/a-Al2O3 system has been found the relative density of all the materials is quite high. to produce further toughness enhancement for reasons This is significant since the achievement of comparable related to a strengthening of the metal-ceramic inter- densities when using conventional powder mixing tech- face [39]. Likewise, microstructural evidence corrobo- niques requires both higher hot-pressing temperatures rating toughness enhancement in the ceramic-ceramic and pressures [33]. Hence, not only does the sol-gel composites has also been found. Fig. 8, showing a chnique result in better Sic dispersion, but also better secondary electron SEM image of the fracture surface overall consolidation in the alumina matrix of a 10/90 vol. SiC(whisker)/a-AL,O, composite, dis- a plot of the fracture toughness of various metal-ce- plays some of the toughening mechanisms mentioned ramic and ceramic-ceramic alumina based composites for the SiC-reinforced materials such as interface as a function of reinforcement volume fraction is shown debonding and whisker pullout in Fig6. a Kle value for alumina(≈4MPa√m)from Interestingly, the magnitude of the toughness the literature is also included [34]. Note that both the creases reflected in Fig. 6 for each of the different metal-ceramic composites and the Sic-reinforced ma- composites can, in fact, be quantitatively reproduced terials display enhanced toughness compared to pure quite well on the basis of the theoretical toughening alumina. In the cermet systems, this toughness enhance mechanisms/models mentioned above. However, the
E.D. Rodeghiero et al. / Materials Science and Engineering A244 (1998) 11–21 19 Fig. 7. Secondary electron SEM micrograph demonstrating ductile phase crack bridging in a 5/95 vol.% Ni/a-Al2O3 metal–ceramic composite; (Ni=light contrast, a-Al2O3=dark contrast). retical predictions, suggesting that the porosity of the composites (even though low) is indeed having a slight effect. Table 2 shows the same data for the SiC(whisker)/aAl2O3 composites synthesized by the sol-gel technique. In the case of these materials, however, two values of the Young’s modulus exist due to transverse isotropy. E corresponds to the elastic modulus in the plane of the whiskers (i.e. perpendicular to the hot-pressing direction) while EÞ corresponds to the elastic modulus perpendicular to the whiskers (i.e. parallel to the hotpressing direction). Note that, indeed, E \EÞ for each of the composites studied, as expected. In regard to the sintering of the composites listed in Table 2, notice that the relative density of all the materials is quite high. This is significant since the achievement of comparable densities when using conventional powder mixing techniques requires both higher hot-pressing temperatures and pressures [33]. Hence, not only does the sol–gel technique result in better SiC dispersion, but also better overall consolidation in the alumina matrix. A plot of the fracture toughness of various metal–ceramic and ceramic–ceramic alumina based composites as a function of reinforcement volume fraction is shown in Fig. 6. A KIc value for alumina (:4 MPa m) from the literature is also included [34]. Note that both the metal–ceramic composites and the SiC-reinforced materials display enhanced toughness compared to pure alumina. In the cermet systems, this toughness enhancement comes from classical ductile phase crack bridging [26,35,36]. For the SiC/a-Al2O3 composites, added toughness comes from numerous reinforcement/crack interactions including the widely accepted crack deflection, interface debonding, crack bridging, and whisker/ platelet pullout mechanisms [37,38]. A secondary electron SEM micrograph exhibiting ductile phase crack bridging is depicted in Fig. 7. The figure shows a 5/95 vol.% Ni/a-Al2O3 composite into which a crack has been introduced by performing a Vickers hardness indentation (not shown). Note the three Ni particles which span the crack plane and hence increase the overall toughness of the composite. In related work, Cr2O3 doping of the Ni/a-Al2O3 system has been found to produce further toughness enhancement for reasons related to a strengthening of the metal–ceramic interface [39]. Likewise, microstructural evidence corroborating toughness enhancement in the ceramic–ceramic composites has also been found. Fig. 8, showing a secondary electron SEM image of the fracture surface of a 10/90 vol.% SiC(whisker)/a-Al2O3 composite, displays some of the toughening mechanisms mentioned for the SiC-reinforced materials such as interface debonding and whisker pullout. Interestingly, the magnitude of the toughness increases reflected in Fig. 6 for each of the different composites can, in fact, be quantitatively reproduced quite well on the basis of the theoretical toughening mechanisms/models mentioned above. However, the
E D. Rodeghiero et al/ Materials Science and Engineering 4244(1998)11-21 EHT=25, 00 KV 6 m Photo no, =3 Detector= SE Fig. 8. Secondary electron SEM micrograph demonstrating interface debonding and whisker pullout in a 10/90 vol. SiC(whisker)/-AlO ceramIc-ceramic composite nature of the sol-gel synthesis employed in preparing Acknowledgements these highly uniform composites has no doubt had a positive impact on the final mechanical properties as This work was supported by onr (grant no well. In particular, many researchers presently feel that N001492J 1526)and AFOSR (grant no. F49620 homogeneity and good dispersion are the most impor- 93 10235). EDR and okT acknowledge the support of tant factors in synthesizing ceramic matrix composites, DoD fellowships. BCM, BSw, and Mw participated as if good overall fracture behavior(i.e. both high fracture part of an undergraduate research program. This study ughness and high fracture strength) is to be obtained benefited from the use of MrL Central Facilities [40]. In this light, fracture strength experiments(along funded by the National Science Foundation(grant no ith creep measurements) constitute the next crucial DMR-9121654) effor, gation to be undertaken in connection with this References 4. Conclusions Brinker. D. R. Ulrich(Eds ) Better Ceramics Through Chemistry, North-Holland, New York, 1984 ol-gel techniques have been used to synthesize C.J. Brinker. G. W. Scherer. Sol-Gel Science. Academic Press. San Diego, CA, 1990, Pp. 11-12. ge variety of ceramic matrix composites. The fund 3 B.E. Yoldas, A transparent porous alumina, Am. Ceram. Soc. mental advantage of using such a synthesis approach Bul.54(3)(1975)286-288. has been the production of extremely uniform and 4 B.E. Yoldas, Alumina sol preparation from alkoxides, Am. disperse microstructures not achievable using conven Ceram.Soc.Bull54(3)(1975)289-290. 5 B.E. Yoldas, Alumina gels that form porous transparent Al,O tional processing techniques. The composites produced J. Mater.Sci.10(1975)1856-1860 have included both metal-ceramic and ceramic-ce- 6C. Brinker, G.w. Scherer, Sol-Gel Science, Academic Press, ramic materials, some carefully doped with additional San Diego, CA, 1990, pp. 839-880 phases (a task made easy with sol-gel techniques) [7R. Roy, s Komarneni, D M. Roy, in C.J. Brinker, D E. Clark, Finally, the materials have displayed favorable physical nd D.R. Ulrich(Eds ) Better Ceramics Through Chemistry and mechanical properties, some of which can be at North-Holland, New York, 1984, pp. 347-359 [8R w. Rice, Ceramic composites-processing challenges, Ceram. tributed to the nature in which they were synthesized Eng.Sci.Proc.2(7-8)(1981)493-508
20 E.D. Rodeghiero et al. / Materials Science and Engineering A244 (1998) 11–21 Fig. 8. Secondary electron SEM micrograph demonstrating interface debonding and whisker pullout in a 10/90 vol.% SiC(whisker)/a-Al2O3 ceramic–ceramic composite. nature of the sol–gel synthesis employed in preparing these highly uniform composites has no doubt had a positive impact on the final mechanical properties as well. In particular, many researchers presently feel that homogeneity and good dispersion are the most important factors in synthesizing ceramic matrix composites, if good overall fracture behavior (i.e. both high fracture toughness and high fracture strength) is to be obtained [40]. In this light, fracture strength experiments (along with creep measurements) constitute the next crucial investigation to be undertaken in connection with this effort. 4. Conclusions Sol–gel techniques have been used to synthesize a large variety of ceramic matrix composites. The fundamental advantage of using such a synthesis approach has been the production of extremely uniform and disperse microstructures not achievable using conventional processing techniques. The composites produced have included both metal–ceramic and ceramic–ceramic materials, some carefully doped with additional phases (a task made easy with sol–gel techniques). Finally, the materials have displayed favorable physical and mechanical properties, some of which can be attributed to the nature in which they were synthesized. Acknowledgements This work was supported by ONR (grant no. N0014 92 J 1526) and AFOSR (grant no. F49620 93 1 0235). EDR and OKT acknowledge the support of DoD fellowships. BCM, BSW, and MW participated as part of an undergraduate research program. This study benefited from the use of MRL Central Facilities funded by the National Science Foundation (grant no. DMR-9121654) References [1] C.J. Brinker, D.E. Clark, D.R. Ulrich (Eds.), Better Ceramics Through Chemistry, North-Holland, New York, 1984. [2] C.J. Brinker, G.W. Scherer, Sol–Gel Science, Academic Press, San Diego, CA, 1990, pp. 11–12. [3] B.E. Yoldas, A transparent porous alumina, Am. Ceram. Soc. Bull. 54 (3) (1975) 286–288. [4] B.E. Yoldas, Alumina sol preparation from alkoxides, Am. Ceram. Soc. Bull. 54 (3) (1975) 289–290. [5] B.E. Yoldas, Alumina gels that form porous transparent Al2O3, J. Mater. Sci. 10 (1975) 1856–1860. [6] C.J. Brinker, G.W. Scherer, Sol–Gel Science, Academic Press, San Diego, CA, 1990, pp. 839–880. [7] R. Roy, S. Komarneni, D.M. Roy, in C.J. Brinker, D.E. Clark, and D.R. Ulrich (Eds.), Better Ceramics Through Chemistry, North-Holland, New York, 1984, pp. 347–359. [8] R.W. Rice, Ceramic composites-processing challenges, Ceram. Eng. Sci. Proc. 2 (7–8) (1981) 493–508