E噩≈S Journal of the European Ceramic Society 23(2003)935-942 www.elsevier.com/locate/jeurceramsoc Co-extrusion of Al2O3/ZrO2 bi-phase high temperature ceramics with fine scale aligned microstructures C. Kaya,*, E.G. Butler, M. H. Lewisb a Interdisciplinary Research Centre(IRC) in Materials Processing and School of Metallurgy and Materials The University of Birmingham, Edgbaston, Birmtingham, B152TT, UK Department of Physics, Centre for Advanced Materials, University of Warwick, Coventry, CV4 7AL, UK Received I March 2002: received in revised form 24 June 2002: accepted 30 June 2002 Abstract Structural ceramic composites comprising continuous fibrillar microstructure are produced using sol-based technology which involves the extrusion, at room temperature, of a two-phase material(Al2O3/ZrO2) resulting in an aligned bi-phase structure which is then multiple co-extruded to reduce the lateral dimensions of the phases. Two sol-derived pastes of differing chemistry (y-AIOOH as alumina source and zirconia) are co-extruded in parallel, and layed-up in closed-packed linear array to form a heterogeneous macro-plug for subsequent extrusion with or without zirconia coating. The second and third extrusion steps produce a filament with markedly reduced lateral paste dimensions provided that the flow properties of the chemically different pastes are similar. The resulting extrudates in the form of continuous green monofilaments, are subsequently laid up in a mould where the structure is pressed and consolidated in desired shape, then pressureless sintered in air to form the multi-phase component. The developed process allows the microstructure to be controlled at a nanometer scale within each extruded filament and after the 3rd stage co- extrusion, each filament size within the final extrudate is reduced to a 65 um. C 2002 Elsevier Science Ltd. All rights reserved Keywords: Aligned microstructure: Al2O3; Extrusion; Sol-gel processes; ZrO2 1. Introduction Co-extrusion can be defined as the passing of two or more pastes through the same die to manufacture a In the last two decades, new materials are required for green body of uniform cross sectional area. It has been lightweight rigid structures with potential for high tem- widely used recently to produce multilayer ceramics, perature operation in the aerospace industry, as the multilayer tubes, alumina and Pbo-containing ferroic most developed alloy systems have reached a develop- ceramics, lead manganese niobate-lead titanate cera- ment limit in relation to high temperature stability and mics and silver palladium, zirconia and stainless steel deformation resistance. Monolithic ceramics can not be metal-ceramic pipes o and fine-scale alumina, mullite for such applications due to their low fractur and ZTA components. -14 One of the main advantages toughness and low resistance to thermal shock. Con- of using a co-extrusion technique is the significant tinuous fibre-reinforced ceramic matrix composites are reductions in the number of processing steps. It has the best solution for high-risk engineering components proven also that deliberately introduced weak interfaces in which high specific stifness and high temperature in a laminar structure suppresses the catastrophic fail nce are required -However, they are very ure, increases the fracture toughness and work of frac expensive to fabricate and current available fibres are ture by the operation of debonding and crack deflecting table only up to I200°C mechanisms. I5 The main objective of the present work is to demon- strate the feasibility of forming multiphase aligned Corresponding author. Tel. +44-121-4143537: fax: +44- fibrillar micostructures from nano-size sol particle pre- 4143441. cursors using an innovative multiple co-extrusion pro- E-mail address: c kaya(@ bham ac uk(C. Kaya) cess. For this aim, two sol-derived high solids-loading 0955-2219/03/S. see front matter C 2002 Elsevier Science Ltd. All rights reserved. PII:S0955-2219(02)00227-3
Co-extrusion of Al2O3/ZrO2 bi-phase high temperature ceramics with fine scale aligned microstructures C. Kayaa,*, E.G. Butlera , M.H. Lewisb a Interdisciplinary Research Centre (IRC) in Materials Processing and School of Metallurgy and Materials, The University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK bDepartment of Physics, Centre for Advanced Materials, University of Warwick, Coventry, CV4 7AL, UK Received 1 March 2002; received in revised form 24 June 2002; accepted 30 June 2002 Abstract Structural ceramic composites comprising continuous fibrillar microstructure are produced using sol-based technologywhich involves the extrusion, at room temperature, of a two-phase material (Al2O3/ZrO2) resulting in an aligned bi-phase structure which is then multiple co-extruded to reduce the lateral dimensions of the phases. Two sol-derived pastes of differing chemistry(g-AlOOH as alumina source and zirconia) are co-extruded in parallel, and layed-up in closed-packed linear array to form a heterogeneous macro-plug for subsequent extrusion with or without zirconia coating. The second and third extrusion steps produce a filament with markedlyreduced lateral paste dimensions provided that the flow properties of the chemicallydifferent pastes are similar. The resulting extrudates in the form of continuous green monofilaments, are subsequentlylaid up in a mould where the structure is pressed and consolidated in desired shape, then pressureless sintered in air to form the multi-phase component. The developed process allows the microstructure to be controlled at a nanometer scale within each extruded filament and after the 3rd stage coextrusion, each filament size within the final extrudate is reduced to 65 mm. # 2002 Elsevier Science Ltd. All rights reserved. Keywords: Aligned microstructure; Al2O3; Extrusion; Sol-gel processes; ZrO2 1. Introduction In the last two decades, new materials are required for lightweight rigid structures with potential for high temperature operation in the aerospace industry, as the most developed alloysystems have reached a development limit in relation to high temperature stabilityand deformation resistance. Monolithic ceramics can not be used for such applications due to their low fracture toughness and low resistance to thermal shock. Continuous fibre-reinforced ceramic matrix composites are the best solution for high-risk engineering components in which high specific stiffness and high temperature tolerance are required.15 However, theyare very expensive to fabricate and current available fibres are stable onlyup to 1200 C. Co-extrusion can be defined as the passing of two or more pastes through the same die to manufacture a green bodyof uniform cross sectional area. It has been widelyused recentlyto produce multilayer ceramics,6 multilayer tubes,7 alumina and PbO-containing ferroic ceramics,8 lead manganese niobate-lead titanate ceramics and silver palladium,9 zirconia and stainless steel metal-ceramic pipes10 and fine-scale alumina, mullite and ZTA components.1114 One of the main advantages of using a co-extrusion technique is the significant reductions in the number of processing steps. It has proven also that deliberatelyintroduced weak interfaces in a laminar structure suppresses the catastrophic failure, increases the fracture toughness and work of fracture bythe operation of debonding and crack deflecting mechanisms.15 The main objective of the present work is to demonstrate the feasibilityof forming multiphase aligned fibrillar micostructures from nano-size sol particle precursors using an innovative multiple co-extrusion process. For this aim, two sol-derived high solids-loading 0955-2219/03/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved. PII: S0955-2219(02)00227-3 Journal of the European Ceramic Society23 (2003) 935–942 www.elsevier.com/locate/jeurceramsoc * Corresponding author. Tel.: +44-121-4143537; fax: +44-121- 4143441. E-mail address: c.kaya@bham.ac.uk (C. Kaya).
C. Kaya et al /Journal of the European Ceramic Society 23(2003)935-94 pastes of alumina and zirconia are co-extruded in par- The seeded sol was first stirred magnetically for 10 h allel, and layed-up in closed-packed linear array to form and then ultrasonic agitation was employed at 15 kHz a heterogeneous macro-plug for subsequent extrusion for 3 h for further dispersion of any particle agglomer provided that the flow properties of the chemically dif- ates which might be present. 6 The final sol composition ferent pastes are similar. i.e., boehmite +2 wt. seeding powder+I wt. gly cerol+I wt. celacol was ball-mixed for 2 days using high purity zirconia balls. Before and after ball-milling 2. Experimental wor the zirconia balls were weighted to ensure that there was no contamination resulting from the milling media. The 2.1. Paste preparation from seeded boehmite (y- mixed seeded sol was then vacuum filtered in order to A1OOH)sol obtain a gel structure. The resulting soft white gel was further compacted using pressure filtration apparatus to Boehmite (r-AlOOH)sol(Remal A20, Remet corp, squeeze out the excess water, and obtain an extrudable USA) was used as the alumina source. The sol ha Paste I average particle size and solids-loading of 40 nm and 20 wt%, respectively. The as received sol is stable at a pl 2.2. Zirconia sol-paste preparation value of 4. The received boehmite sol was seeded with 2 wt% of the total mass using ultrafine a-Al2O3(30 nm, Zirconia sol was prepared using ultrafine and high emi gh purity polishing powder) purity zirconia powders(average grain size is 30 nm, VP powders. The flow chart for the seeding process, fol- zirconia, Degussa Ltd, Germany) with the addition of 3 lowed by paste preparation is given in Fig. 1. To seed mol% yttria. Kinetically stable and well dispersed zir the boehmite sol, the seeding powders was first dis- conia sol having 20 wt solids-loading was prepared persed in distilled water and then added into boehmite by the addition of small amount of zirconia to the sol. Glycerol (1 wt %)and celacol (I wt %)were also water, while the suspension was magnetically stirred added in order to minimise the surface roughness of the The best pH value in order to obtain the maximum sta extruded and increase the green strength, respectively. bility was found to be 8.5 and this ph value was ma Celacol (1 wt %)was first dissolved in water at 85C tained using ammonia. I wt. glycerol and celacol were and then added to the seeded sol in order keep it plasti- added to prepared zirconia sol. Cyclohexanone cally deformable during the multiple extrusion stage. ( C6H1oO)(I wt %)and I wt% boehmite sol were also added in order to minimise the shear-thickening effect The flow chart for zirconia gel and paste preparation is iven in Fig. 2. Each paste was then laboratory scale extrusion apparatus with an extrusion reduction ratio of 4: 1. The ram velocity chosen was very low(0.5 or I mm/min )to minimise edge-tearing defects Dispersed in water, ultrasonic that might form at higher velocities. Flow behaviour of each paste was determined using a computer program linked to the Instron test machine. Rheological beha Boehmite(AIOOH viour of each material. i.e. boehmite and zirconia was controlled to be similar for multiple extrusion Ball-milling for 2 The rheological behaviour of boehmite and zirconia Kinetically stable sol-derived pastes was characterised by a capillary rheometer 18-20 Three dies with different L/D(length/ Vacuum filtering diameter)ratios were used. The ram speeds ranged from 24 to 0.5 mm/min. and the load at each-pre-set velocity vas recorded. The paste rheology was characterise 40-45 wt so solids using the Benbow-Bridgwater relationship Pressure filtration P=2Ln(Do/D(oo +aI)+4(L/D(To +B,)(1) Extrusion where P is the pressure, Do and D are the diameters of the barrel and the die, respectively, L is the length of the Fig. 1. Flow chart for the preparation of boehmite sol-derived past die land and v is the extrudate velocity. The paste
pastes of alumina and zirconia are co-extruded in parallel, and layed-up in closed-packed linear array to form a heterogeneous macro-plug for subsequent extrusion provided that the flow properties of the chemicallydifferent pastes are similar. 2. Experimental work 2.1. Paste preparation from seeded boehmite (- AlOOH) sol Boehmite (g-AlOOH) sol (Remal A20, Remet corp., USA) was used as the alumina source. The sol has average particle size and solids-loading of 40 nm and 20 wt.%, respectively. The as received sol is stable at a pH value of 4. The received boehmite sol was seeded with 2 wt.% of the total mass using ultrafine a-Al2O3 (30 nm, BDH Chemical, UK., high puritypolishing powder) powders. The flow chart for the seeding process, followed bypaste preparation is given in Fig. 1. To seed the boehmite sol, the seeding powders was first dispersed in distilled water and then added into boehmite sol. Glycerol (1 wt.%) and celacol (1 wt.%) were also added in order to minimise the surface roughness of the extruded and increase the green strength, respectively. Celacol (1 wt.%) was first dissolved in water at 85 C and then added to the seeded sol in order keep it plasticallydeformable during the multiple extrusion stage. The seeded sol was first stirred magneticallyfor 10 h and then ultrasonic agitation was employed at 15 kHz for 3 h for further dispersion of anyparticle agglomerates which might be present.16 The final sol composition i.e., boehmite+2 wt.% seeding powder+1 wt.% glycerol+1 wt.% celacol was ball-mixed for 2 days using high purityzirconia balls. Before and after ball-milling, the zirconia balls were weighted to ensure that there was no contamination resulting from the milling media. The mixed seeded sol was then vacuum filtered in order to obtain a gel structure. The resulting soft white gel was further compacted using pressure filtration apparatus to squeeze out the excess water, and obtain an extrudable paste.17 2.2. Zirconia sol-paste preparation Zirconia sol was prepared using ultrafine and high purityzirconia powders (average grain size is 30 nm, VP zirconia, Degussa Ltd., Germany) with the addition of 3 mol% yttria. Kinetically stable and well dispersed zirconia sol having 20 wt.% solids-loading was prepared bythe addition of small amount of zirconia to the water, while the suspension was magneticallystirred. The best pH value in order to obtain the maximum stabilitywas found to be 8.5 and this pH value was maintained using ammonia. 1 wt.% glycerol and celacol were added to prepared zirconia sol. Cyclohexanone (C6H10O) (1 wt.%) and 1 wt.% boehmite sol were also added in order to minimise the shear-thickening effect. The flow chart for zirconia gel and paste preparation is given in Fig. 2. Each paste was then extruded using a laboratoryscale extrusion apparatus with an extrusion reduction ratio of 4:1. The ram velocitychosen was very low (0.5 or 1 mm/min.) to minimise edge-tearing defects that might form at higher velocities. Flow behaviour of each paste was determined using a computer program linked to the Instron test machine. Rheological behaviour of each material, i.e., boehmite and zirconia was controlled to be similar for multiple extrusion. 2.3. Paste rheology The rheological behaviour of boehmite and zirconia sol-derived pastes was characterised bya capillary rheometer.1820 Three dies with different L/D (length/ diameter) ratios were used. The ram speeds ranged from 124 to 0.5 mm/min. and the load at each-pre-set velocity was recorded. The paste rheologywas characterised using the Benbow–Bridgwater relationship:21 P ¼ 2Lnð Þ Do=D o þ 1Vm ð Þþ 4ð Þ L=D o þ 1Vm ð Þð1Þ where P is the pressure, Do and D are the diameters of the barrel and the die, respectively, L is the length of the Fig. 1. Flow chart for the preparation of boehmite sol-derived paste. die land and V is the extrudate velocity. The paste 936 C. Kaya et al. / Journal of the European Ceramic Society 23 (2003) 935–942
C. Kaya et al /Journal of the European Ceramic Society 23(2003)935-94 I wt% Boehmite 1 wt Glycer I wt %o Celacol Sol derived rsed in water Paste A AL,O paste B 卡ZrO2 ronia powder(30 nm) + Dispersed in water 16 mm days milling for 2 16 mm First Feedrod ronia sol B Vacuum filtering 40-45 wt solids Each monofilament size First Extrudate Pressure filtration Sol-derived past 4 mm Extrusion Second Feedrod Fig. 2. Flow cha he preparation of zirconia sol-derived paste parameters that describes the paste flow during extru- sion are o. the die entry yield stress; To the die wall shear stress; a, and B, the die entry and die land velocity Each monofilament size Second Extrudate coefficients. The m and n are the die entry and die land velocity exponents; these are used to describe non-linear pressure versus velocity data 2.4. Multiphase green body formation by multiple co- Third Feedrod extrusion The flow chart for multiphase green body formation using multiple co-extrusion, is shown in Fig. 3. As Each monofilament size shown in the monofilament extrusion section, square shaped (4x 4 mm) boehmite and zirconia monofilaments were first extruded separately. Then a total of 16 mono- Fig 3. Schematic representation of multiphase component fabrication filaments( 8 boehmite and 8 zirconia) were un-coated or dip coated with zirconia before they are layed-up in a square die. After the first co-extrusion, each co-extruded conia filaments were first produced with zirconia interface filament included 16 filaments and filament size was These multiple co-extruded and plastically deformable zir reduced from 4 to I mm using a 4 mm square die. 22 For conia coated filaments were layered (two or three multiple the second co-extrusion, the same process was repeated co-extruded filaments in each layer) in a rectangular die and the twice co-extruded filament included 256 mono- (70x 12x50 mm)and then pressed using an applied load filaments(128 boehmite and 128 alumina) having a of 10 kN. If two layers of multiple co-extruded filaments filament size of 250 um. After the third stage co-extru- were pressed in the die, the final test plaque layers, the sion, each extruded filament contained 4096 filaments and 24 546 filaments whilst in the case of three layers, the filament size of 62.5 um. During each step of this process, number of filaments within the test plaque was 36864 a zirconia interface was applied, using dip coating. Manu- facture of the multiphase consolidated test plaques with 2.5. Monofilament coating with ZrOz zirconia interfaces using die pressing was carried out, as shown in the multiphase green body formation section in In order to prepare a low solids -loading zirconia sus Fig 3. For this aim, 3rd stage co-extruded boehmite/zir- pension suitable for coating, Degussa zirconia powders
parameters that describes the paste flow during extrusion are o the die entryyield stress; o the die wall shear stress; , and , the die entryand die land velocity coefficients. The m and n are the die entryand die land velocityexponents; these are used to describe non-linear pressure versus velocitydata. 2.4. Multiphase green body formation by multiple coextrusion The flow chart for multiphase green bodyformation using multiple co-extrusion, is shown in Fig. 3. As shown in the monofilament extrusion section, square shaped (44 mm) boehmite and zirconia monofilaments were first extruded separately. Then a total of 16 mono- filaments (8 boehmite and 8 zirconia) were un-coated or dip coated with zirconia before theyare layed-up in a square die. After the first co-extrusion, each co-extruded filament included 16 filaments and filament size was reduced from 4 to 1 mm using a 4 mm square die.22 For the second co-extrusion, the same process was repeated and the twice co-extruded filament included 256 mono- filaments (128 boehmite and 128 alumina) having a filament size of 250 mm. After the third stage co-extrusion, each extruded filament contained 4096 filaments and a filament size of 62.5 mm. During each step of this process, a zirconia interface was applied, using dip coating. Manufacture of the multiphase consolidated test plaques with zirconia interfaces using die pressing was carried out, as shown in the multiphase green bodyformation section in Fig. 3. For this aim, 3rd stage co-extruded boehmite/zirconia filaments were first produced with zirconia interface. These multiple co-extruded and plasticallydeformable zirconia coated filaments were layered (two or three multiple co-extruded filaments in each layer) in a rectangular die (701250 mm3 ) and then pressed using an applied load of 10 kN. If two layers of multiple co-extruded filaments were pressed in the die, the final test plaque contained 24 546 filaments whilst in the case of three layers, the number of filaments within the test plaque was 36 864. 2.5. Monofilament coating with ZrO2 In order to prepare a low solids-loading zirconia suspension suitable for coating, Degussa zirconia powders Fig. 2. Flow chart for the preparation of zirconia sol-derived paste. Fig. 3. Schematic representation of multiphase component fabrication bymultiple co-extrusion. C. Kaya et al. / Journal of the European Ceramic Society 23 (2003) 935–942 937
C. Kaya et al. /Journal of the European Ceramic Society 23(2003)935-942 (VP zirconia, 30 nm)were dispersed in distilled water. 3. Results and discussion Kinetically stable, well dispersed zirconia suspensions containing 5-10 wt. zirconia powders were obtained Fig. 5 shows the flow behaviour of the chemically at a pH value of 3. Then, the green extrudates were different boehmite and zirconia pastes during monofila immersed in an ammonia based solution, consisting of ment extrusion using a constant speed of 0. 5 mm/min an ammonium salt of polymethacrylic acid (Versicol The success of the multiple extrusion relies on the KA21, PH: 9, Allied Colloids, UK and Versicol E10) rheology of the two pastes being matched. The sol djusted by pure ammonia solution to pH 11.5, thus derived boehmite and zirconia pastes(see Figs. I and 2) creating a negative surface charge and improving the show similar rheological behaviour, as shown in Fig. 5. wetting of the extrudates surface. This stage maximised If one paste is softer than the other, the resultant dif- the electrostatic attraction between the extrudates sur- ferences in velocities cause the formation of non-discrete face and the positively charged zirconia particles in the shapes. The phenomenon of phase migration occurs coating sol(see Fig. 4) when the water flows preferentially through the powder tem under an applied stress. 8 If phase migration 2. 6. Drying and sintering takes place during extrusion(due to unwell mixture of each of the constituents within the paste or extrusion at The consolidated multiphase green samples, i.e., mul- low deformation rates etc. ) the load at each ram speed tiple extruded and pressed, were first kept in a humidity can not be accurately determined. Boehmite and zirco- controlled chamber(from 80 to 55% relative humidity) nia sol-derived pastes display a very stable paste flow for I day to allow the residual water to remove slowly behaviour as shown in the load-distance curves in Fig. 5. from the green body, thus preventing the formation of The shear stress versus shear rate curves for boehmite any internal cracks. This was followed by 1 day drying and zirconia pastes(for a constant L/D: 16) are shown in normal air. The dried green body compacts were in Fig. 6(a). As shown in Fig. 6(a)the gradient of the pressureless sintered at 1400C for 2 h in air using a curves decrease as the shear rate increases, indicating 3C/min heating and cooling rates(the green samples that both pastes are shear thinning and can be co- were first heated to 110C using a 2C/min heating extruded using similar extrusion parameters. The rheol rate for 2 h, then 600C using 3C/min heating rate for ogy of the pastes can be described as non-Newtonian 2 h in order to burn out any residual organic materials Herschel-Bulkley flow which follows a power law func- that might be present within the body) tion indicating the existance of a critical yield stress which must be exceeded for flow to occur. 26 Measured 2.7. Other characterisation techniques pressure drops as a function of extrusion velocity for a constant L/D: 16 for boehmite and zirconia sol-derived Microstructural examinations on green and sintered pastes are shown in Fig. 6(b)confirming the presence of samples were carried out using a Field Emission Gun very similar flow behaviour of the pastes. The rheologi SEM(FEG SEM FX-4000, Jeol Ltd Japan) and sin- cal behaviour of ceramic pastes is mainly determined by tered densities were measured using Archimedes tech- the powder size, morphology, size distribution, powder nique. Interface behaviour in terms of the crac loading and binder rheology. As explained in Figs. I deflection/arrest was characterised using a crack propa- gation test. 23,24 A linear intercept technique was used to measure the average grain size on polished and ther- mally etched surfaces. 25 8 A203 Zro2 Surface Pre-treatment 2 7 Filament Zro Fig. 5. Load-distance curves of boehmite and zirconia during mono- Fig. 4. Schematic representation of dip-coating technic
(VP zirconia, 30 nm) were dispersed in distilled water. Kineticallystable, well dispersed zirconia suspensions containing 5–10 wt.% zirconia powders were obtained at a pH value of 3. Then, the green extrudates were immersed in an ammonia based solution, consisting of an ammonium salt of polymethacrylic acid (Versicol KA21, pH: 9, Allied Colloids, UK and Versicol E10) adjusted bypure ammonia solution to pH 11.5, thus creating a negative surface charge and improving the wetting of the extrudates surface. This stage maximised the electrostatic attraction between the extrudates surface and the positivelycharged zirconia particles in the coating sol (see Fig. 4). 2.6. Drying and sintering The consolidated multiphase green samples, i.e., multiple extruded and pressed, were first kept in a humidity controlled chamber (from 80 to 55% relative humidity) for 1 dayto allow the residual water to remove slowly from the green body, thus preventing the formation of anyinternal cracks. This was followed by1 daydrying in normal air. The dried green bodycompacts were pressureless sintered at 1400 C for 2 h in air using a 3 C/min. heating and cooling rates (the green samples were first heated to 110 C using a 2 C/min. heating rate for 2 h, then 600 C using 3 C/min. heating rate for 2 h in order to burn out anyresidual organic materials that might be present within the body). 2.7. Other characterisation techniques Microstructural examinations on green and sintered samples were carried out using a Field Emission Gun SEM (FEG SEM FX-4000, Jeol Ltd. Japan) and sintered densities were measured using Archimedes technique. Interface behaviour in terms of the crack deflection/arrest was characterised using a crack propagation test.23,24 A linear intercept technique was used to measure the average grain size on polished and thermallyetched surfaces.25 3. Results and discussion Fig. 5 shows the flow behaviour of the chemically different boehmite and zirconia pastes during monofilament extrusion using a constant speed of 0.5 mm/min. The success of the multiple extrusion relies on the rheologyof the two pastes being matched. The solderived boehmite and zirconia pastes (see Figs. 1 and 2) show similar rheological behaviour, as shown in Fig. 5. If one paste is softer than the other, the resultant differences in velocities cause the formation of non-discrete shapes. The phenomenon of phase migration occurs when the water flows preferentiallythrough the powder system under an applied stress.18 If phase migration takes place during extrusion (due to unwell mixture of each of the constituents within the paste or extrusion at low deformation rates etc.) the load at each ram speed can not be accuratelydetermined. Boehmite and zirconia sol-derived pastes displaya verystable paste flow behaviour as shown in the load-distance curves in Fig. 5. The shear stress versus shear rate curves for boehmite and zirconia pastes (for a constant L/D: 16) are shown in Fig. 6(a). As shown in Fig. 6(a) the gradient of the curves decrease as the shear rate increases, indicating that both pastes are shear thinning and can be coextruded using similar extrusion parameters. The rheologyof the pastes can be described as non-Newtonian Herschel–Bulkleyflow which follows a power law function indicating the existance of a critical yield stress which must be exceeded for flow to occur.26 Measured pressure drops as a function of extrusion velocityfor a constant L/D: 16 for boehmite and zirconia sol-derived pastes are shown in Fig. 6(b) confirming the presence of verysimilar flow behaviour of the pastes. The rheological behaviour of ceramic pastes is mainlydetermined by the powder size, morphology, size distribution, powder loading and binder rheology.19 As explained in Figs. 1 Fig. 5. Load-distance curves of boehmite and zirconia during monoFig. 4. Schematic representation of dip-coating technique. filament extrusion. 938 C. Kaya et al. / Journal of the European Ceramic Society 23 (2003) 935–942
C. Kaya et al. /Journal of the European Ceramic Society 23(2003)935-942 8 Shear Rate,(s1)×104 力sg (a) Extrudate Velocity, (m/s)x104 000834 1500 as a function of extrudate velocity for a constant L/D: 16 for boehmite and zirconia sol-derived pastes. Fig. 7. SEM images of co-extruded boehmite/zirconia multiphase component, showing the presence of cracks within the zirconia phase (light phase)if the rheological behaviour of two pastes are not similar and 2, both pastes used in this work are water based and and the difference in sintering shrinkage is about 7%.(a) cross sectional contains very small amount of additives as binder or and(b) longitudinal view(sample was sintered at 1400C for 2 h ) solvent. therefore, it can be concluded that the rheolo- gical behaviour of each paste is dependant on the part the interface when zirconia shows "shear thickening cle characteristics such as size, shape and loading as well behaviour during extrusion and the difference in linear as the chosen processing technique of the pastes. The sintering shrinkage between these two phases is about developed technique for the preparation of sol-derived 7% as shown in Fig. 7(a). The presence of extensive pastes provides homogeneously mixed pastes suitable cracks(caused by the difference in sintering shrinkage) for co-extrusion under the same extrusion parameters. perpendicular to the extrusion direction within the zir- Otherwise, it is not possible to prepare a paste from conia filaments is also evident from the longiditunal nano-size particles using conventional paste preparation section of co-extruded sintered filament shown in echniques Fig. 7(b). If the right rheology, drying cycle and sinter- The other critical parameter is the drying behaviour ing shrinkage are optimised using the necessary addi of the pastes. Fig. 7(a) shows a cross-sectional SEM tions (see Figs. 1 and 2), crack free co-extruded micrograph, indicating the effect of the rheology filaments are produced as shown in Fig 8. Fig 8(a)and matching of the two pastes on the microstructure of co-(b)show the cross-sectional and longidutional micro extruded component. For successful co-extrusion, both structure of a 2nd stage co-extruded alumina/ zirconia bi rheology, the drying and the sintering shrinkage of the phase filament after sintering at 1400C for 2 h In order pastes should be controlled in order to control the to optimise the linear sintering shrinkage of zirconia, 2 internal stresses and cracking. Fig. 7(a) shows a lst wt. coarse zirconia powders(300 nm)were added to the stage co-extruded alumina/zirconia microstructure after main composition as described in Fig. 2, so that the dif- sintering at 1400C for 2 h Significant crack formation ference in sintering shrinkage between these two phases is is visible within the zirconia phase (light phase) also controlled to be less than 3%. Crack formation within
and 2, both pastes used in this work are water based and contains verysmall amount of additives as binder or solvent, therefore, it can be concluded that the rheological behaviour of each paste is dependant on the particle characteristics such as size, shape and loading as well as the chosen processing technique. of the pastes. The developed technique for the preparation of sol-derived pastes provides homogeneouslymixed pastes suitable for co-extrusion under the same extrusion parameters. Otherwise, it is not possible to prepare a paste from nano-size particles using conventional paste preparation techniques. The other critical parameter is the drying behaviour of the pastes. Fig. 7(a) shows a cross-sectional SEM micrograph, indicating the effect of the rheology matching of the two pastes on the microstructure of coextruded component. For successful co-extrusion, both rheology, the drying and the sintering shrinkage of the pastes should be controlled in order to control the internal stresses and cracking. Fig. 7(a) shows a 1st stage co-extruded alumina/zirconia microstructure after sintering at 1400 C for 2 h. Significant crack formation is visible within the zirconia phase (light phase) also at the interface when zirconia shows ‘‘shear thickening’’ behaviour during extrusion and the difference in linear sintering shrinkage between these two phases is about 7% as shown in Fig. 7(a). The presence of extensive cracks (caused bythe difference in sintering shrinkage) perpendicular to the extrusion direction within the zirconia filaments is also evident from the longiditunal section of co-extruded sintered filament shown in Fig. 7(b). If the right rheology, drying cycle and sintering shrinkage are optimised using the necessaryadditions (see Figs. 1 and 2), crack free co-extruded filaments are produced as shown in Fig. 8. Fig. 8(a) and (b) show the cross-sectional and longidutional microstructure of a 2nd stage co-extruded alumina/zirconia biphase filament after sintering at 1400 C for 2 h. In order to optimise the linear sintering shrinkage of zirconia, 2 wt.% coarse zirconia powders (300 nm) were added to the main composition as described in Fig. 2, so that the difference in sintering shrinkage between these two phases is controlled to be less than 3%. Crack formation within Fig. 6. (a) Shear stress–shear rate and (b) measured pressure changes as a function of extrudate velocityfor a constant L/D: 16 for boehmite and zirconia sol-derived pastes. Fig. 7. SEM images of co-extruded boehmite/zirconia multiphase component, showing the presence of cracks within the zirconia phase (light phase) if the rheological behaviour of two pastes are not similar and the difference in sintering shrinkage is about 7%. (a) cross sectional and (b) longitudinal view (sample was sintered at 1400 C for 2 h.). C. Kaya et al. / Journal of the European Ceramic Society 23 (2003) 935–942 939
C. Kaya et al /Journal of the European Ceramic Society 23(2003)935-94 Z i 50w m 六5 Fig 9. SEM images of 3rd stage co-extruded alumina/zirconia com- ponent, showing the fine microstructure of zirconia phase and dense Fig. 8. SEM images of co-extruded boehmite/zirconia multiphase microstructure of a-alumina(A: a-alumina, Z: zirconia). Sample sin- tered at I400°for2h interface if the rheological behaviour of two pastes is controlled to similar and the difference in sintering shrinkage is <3%.(a)cross average grain size of alumina and zirconia within the sectional and(b) longitudinal view(sample was sintered at 1400Cfor co-extruded and sintered bi-phase component was determined to be 1.6 and 0.45 um, respectively. The conia microstructure contains only equiaxed grains the any phase or at the interface is avoided, as shown in whilst the majority of grains in the alumina matrix are Fig. 8(a)and(b). It should also be noted from Fig8 near equiaxed with the presence of some elongated ones. that there is a shape distortion in alumina and zirconia as shown in Fig 9. These fine scale microstructures are due to difference in water removal behaviour of the expected to provide good mechanical and thermo- pastes during extrusion as they have different particle mechanical properties. Figs. 7-9 show the micro- shapes and packing densities in the paste form. How- structural features in assessing the effectiveness of the ever, this is considered not to be a critical issue provid- co-extrusion technique in order to produce di-phasic ing the each filament within the co-extruded structure is aligned microstructures and it can be concluded from conunuous these micrographs that the rheological matching of the Fig 9 demonstrates a 3rd stage co-extruded alumina/ two different pastes coupled with differential sintering zirconia microstructure with the absence of any inter or shrinkage's play the most critical role for a successful intragranular porosity as well as cracks after sintering at co-extrusion Shape distortion of each filament during 1400C for 2 h. One of the main objectives of the pre- co-extrusion is also noted as shown in Figs. 8 and 9 ent work is to control the fine-scale sintered micro- This distortion in shape can be eliminated using lower structure of the each phase present in the final sintered extrusion reduction ratios or extrusion rates component using ultrafine starting powders (<100 nm) The microstructures of zirconia coated monofilament Fig 9 shows the very fine and dense microstructures of using dip-coating are shown in Fig. 10. Homogeneous zirconia (in TZP form by the addition of 3 mol%Y203) coating around an alumina rod in green state is evident and a-alumina after sintering at 1400oC for 2 h. The from the micrograph shown Fig 10(a) and its thickness
the anyphase or at the interface is avoided, as shown in Fig. 8(a) and (b). It should also be noted from Fig. 8 that there is a shape distortion in alumina and zirconia due to difference in water removal behaviour of the pastes during extrusion as theyhave different particle shapes and packing densities in the paste form. However, this is considered not to be a critical issue providing the each filament within the co-extruded structure is continuous. Fig. 9 demonstrates a 3rd stage co-extruded alumina/ zirconia microstructure with the absence of anyinter or intragranular porosityas well as cracks after sintering at 1400 C for 2 h. One of the main objectives of the present work is to control the fine-scale sintered microstructure of the each phase present in the final sintered component using ultrafine starting powders (<100 nm). Fig. 9 shows the veryfine and dense microstructures of zirconia (in TZP form bythe addition of 3 mol% Y2O3) and a-alumina after sintering at 1400 C for 2 h. The average grain size of alumina and zirconia within the co-extruded and sintered bi-phase component was determined to be 1.6 and 0.45 mm, respectively. The zirconia microstructure contains onlyequiaxed grains whilst the majorityof grains in the alumina matrix are near equiaxed with the presence of some elongated ones, as shown in Fig. 9. These fine scale microstructures are expected to provide good mechanical and thermomechanical properties. Figs. 7–9 show the microstructural features in assessing the effectiveness of the co-extrusion technique in order to produce di-phasic aligned microstructures and it can be concluded from these micrographs that the rheological matching of the two different pastes coupled with differential sintering shrinkage’s playthe most critical role for a successful co-extrusion. Shape distortion of each filament during co-extrusion is also noted as shown in Figs. 8 and 9. This distortion in shape can be eliminated using lower extrusion reduction ratios or extrusion rates. The microstructures of zirconia coated monofilament using dip-coating are shown in Fig. 10. Homogeneous coating around an alumina rod in green state is evident from the micrograph shown Fig. 10(a) and its thickness Fig. 8. SEM images of co-extruded boehmite/zirconia multiphase component, showing the absence of cracks within the phases or at the interface if the rheological behaviour of two pastes is controlled to be similar and the difference in sintering shrinkage is <3%. (a) cross sectional and (b) longitudinal view (sample was sintered at 1400 C for 2 h). Fig. 9. SEM images of 3rd stage co-extruded alumina/zirconia component, showing the fine microstructure of zirconia phase and dense microstructure of a-alumina (A: a-alumina, Z: zirconia). Sample sintered at 1400 C for 2 h. 940 C. Kaya et al. / Journal of the European Ceramic Society 23 (2003) 935–942
C. Kaya et al /Journal of the European Ceramic Society 23(2003)935-94 sae 1i:9 Fig. 10. SEM images of a zirconia coated monofilament, showing(a)the homogeneous zirconia layer around the filament, (b)the thickness of the zirconia layer to be about 10 um and(c) the propagation of an indenter-induced crack at the alumina/ zirconia interface(A: alumina and z: zirconia) is about 10 um as shown in Fig. 10(b). These pictures the two pastes, drying and sintering shrinkage are both clearly show the effectiveness of the dip-coating process optimised and controlled in order to eliminate the crack which relies on the strong electrostatic attraction formation within the filaments or at the interface. 3rd between the coating sol particles and extrudate surface. co-extrusion provides a filament size of about 60-70 um The propagation of an indenter induced crack on a co- and these plastically deformable co-extruded filaments extruded filament with zirconia interface is shown in in green state are layed up in desired shape mold to Fig. 10(c). It is shown that the zirconia interface created produce net shape components. The final pressureless between alumina and zirconia phases within the co- sintered (1400C for 2 h)alumina and zirconia micro- extruded filament is able to deflect the crack along the structures within the co-extruded matrix are very fine boundary. (the average grain size of alumina and zirconia are determined to be 1.6 and 0.45 um, respectively) and pore free. This processing technique allows us to pro 4. Conclusions duce damage-tolerant components with the application of zirconia interface during co-extrusion or lay up stages This work introduces a new method of fabricating in the final mold diphasic ceramic microstructures with controlled phase dimensionality and anisotropy provided by thin paralle arrays(microlaminae) of 2 thermodynamically-compa- Acknowledgements tible ceramic pastes (boehmite/zirconia). Multiphase alumina/zirconia components were produced from This project is supported by the European Commis nano-size sol-derived pastes using co-extrusion. The sion under the contract number BRITE- EURAM feasibility of forming multiphase aligned fibrilar micro- BRPR-CT 97-0609. Project partners; University of structures using the innovative co-extrusion process Warwick (UK), Morgan Materials Technology, M-T developed is shown to be achiaviable if the rheology of (UK), Centro de estudios e Investigaciones, CEIT
is about 10 mm as shown in Fig. 10(b). These pictures clearlyshow the effectiveness of the dip-coating process which relies on the strong electrostatic attraction between the coating sol particles and extrudate surface. The propagation of an indenter induced crack on a coextruded filament with zirconia interface is shown in Fig. 10(c). It is shown that the zirconia interface created between alumina and zirconia phases within the coextruded filament is able to deflect the crack along the boundary. 4. Conclusions This work introduces a new method of fabricating diphasic ceramic microstructures with controlled phase dimensionalityand anisotropyprovided bythin parallel arrays (microlaminae) of 2 thermodynamically-compatible ceramic pastes (boehmite/zirconia). Multiphase alumina/zirconia components were produced from nano-size sol-derived pastes using co-extrusion. The feasibilityof forming multiphase aligned fibrilar microstructures using the innovative co-extrusion process developed is shown to be achiaviable if the rheologyof the two pastes, drying and sintering shrinkage are both optimised and controlled in order to eliminate the crack formation within the filaments or at the interface. 3rd co-extrusion provides a filament size of about 60–70 mm and these plasticallydeformable co-extruded filaments in green state are layed up in desired shape mold to produce net shape components. The final pressureless sintered (1400 C for 2 h) alumina and zirconia microstructures within the co-extruded matrix are veryfine (the average grain size of alumina and zirconia are determined to be 1.6 and 0.45 mm, respectively) and pore free. This processing technique allows us to produce damage-tolerant components with the application of zirconia interface during co-extrusion or layup stages in the final mold. Acknowledgements This project is supported bythe European Commission under the contract number BRITE- EURAM, BRPR- CT 97- 0609. Project partners; Universityof Warwick (UK), Morgan Materials Technology, M2 T (UK), Centro de Estudios e Investigaciones, CEIT Fig. 10. SEM images of a zirconia coated monofilament, showing (a) the homogeneous zirconia layer around the filament, (b) the thickness of the zirconia layer to be about 10 mm and (c) the propagation of an indenter-induced crack at the alumina/zirconia interface (A: alumina and Z: zirconia). C. Kaya et al. / Journal of the European Ceramic Society 23 (2003) 935–942 941
Journal of the European Ceramic Society 23(2003) 935-942 (Spain), Ecole des Mines de Paris(France)and Indus- 12. Kaya. C, Butler, E. G. and Lewis, M. H. Microstructural ria de Turbo Propulsores (Spain) are sincerely controlled mullite ceramics from monophasic and diphasic sol- acknowledged for their contribution. Mr R. Huzzard is derived pastes. J. Mater. Sci. (in press) acknowledged for the assistance with rheological mea 13. Kaya. C. and Butler, E. G, Innovative processing of multiphase high temperature ceramics, Mid-term report to European Co- ission, Conract no: BRPR-CT97-069 December 1999 14. Kaya. C, Butler, E. G. and Lewis, M. H, unpublished work References 5. Clegg, w.C. Kendall. C, Alford, N M. Button, T.w. and Birch- way to make tough ceramics. Nature. 1990. 34 1. Lewis. M. H. Daniel.A. M.. Cham berlain. A. Pharaoh. M. w. 16. Kaya, C and Butler, E. G, Innovative processing of multiphase Cain, G. Microstructure-property relationships in silicate-matrix high temperature ceramics, Final report to European Commission omposites. J. Microscopy, 1993, 169, 109-12 Conract no: BRPR-CT97-069 December 200 3. Sutherland. S. Plucknett. K. P. and Lewis. M. H. High tem. Birmingham. June 19 perature mechanical and thermal stability of silicate matrix com- 18. Huzzard, R.J. and Black burn. S, Influence of solids loading on posites. J. Composites Eng, 1995. 5. 1367-137 4. Evans, A. G. Domergue, J-M. agagginL 49-56. for relating the tensile constitutive behaviour of ceramic-matrix 19. Draper, O, Blackburn, S, Dolman, G, Smalley, K. and grif- omposites to constituent properties. J. Am. Ceram. Soc., 1994 fiths, A, A comparison of paste rheology and extrudate strength 77,1425-1435 with respect to binder formulation and forming technique. Mater 5. Tu, W, Lange, F. F. and Evans, A. G, Concept for Processing Tech, 1999, 92-93, 141-146. olerant ceramic composite with "strong" interfaces. J. Am 20. Huzzard, R J and Blackburn, S, Slip flow in concentarted alu- 6. Shannon, T. and Blackburn, S, The production of alumina/zir- mina suspensions. Powder Technology, 1998 21. Benbow, J. and Bridgwater, J, Paste Flow and Extrusion. Clar- nia laminated composites by co-extrusion. Ceramic Engineering endon press. Oxford. 199 ld Science Proceedings, 1995, 16. 1115-1 120. ler E. g. and 7. Liang, Z and Blackburn, S, Co-extrusion of multilayered tubes cterisation of mullite(Nextel 720M)fibre-reinforced mullite In Proceedings of The Better Ceramics Through Processing, ed matrix composites from hydrothermally processed mullite pre- J. Yeomans and J. Binner. The Institute of Materials. London Ceramic Matrix Com 1998,pp.109115 MC 4), ed. w. Krenkel, R. Naslain and H. Schneider. wiley 8. Hoy, W, Barda, A. Griffith, M VCH, Weinheim, Germany, 2001, pp. 639-644 fabrication of ceramics by cowt and Halloran, J. w., MIcro- 23. Wendorff. J. Janssen. R. and Claussen. N. Platinum as a weak 1998,81,152-158. interphase for fiber-reinforced oxide-matrix composites. J. Am. 9. Crumm, A. T. and Halloran, J. W. Fabrication of micro- 24. Kaya, C, Kaya, F, Boccaccini, A.R. and Chawla,KK,Fab- Ceran.Soc.,1998,81,2738-2740. onfigured multicomponent ceramics. J. Am. Ceram Soc., 1998. 81,1053-1057 rication and characterisation of ni-coated carbon fibre- reinforced 10. Chen. Z. Ikeda. K. Murakami. T and Takeda. T. Extrusion of alumina ceramic matrix composites using electrophoretic deposi netal-ceramic composite pipes. J. Am. Ceram. Soc., 2000, 83 tion. Acta Materialia. 2001. 49. 1189-1197 1081-1086. 25. Mendelson, M. I, Average grain size in polycrystalline ceramics. 11. Kaya, C and Butler, E. G, Plastic forming and microstructural J.Am. Ceran.Soc.,1969,52,443-446 development of a-alumina ceramics from highly compacted green 26. Whorlow, R. W, Rheological Techniques. Ellis Horwood, Lon- odies using extrusion. J. Eur. Ceram. Soc. 2002.22 1917-19
(Spain), Ecole des Mines de Paris (France) and Industria de Turbo Propulsores (Spain) are sincerely acknowledged for their contribution. Mr. R. Huzzard is acknowledged for the assistance with rheological measurements. References 1. Lewis, M. H., Daniel, A. M., Chamberlain, A., Pharaoh, M. W., Cain, G. Microstructure-propertyrelationships in silicate-matrix composites. J. Microscopy, 1993, 169, 109–121. 2. Evans, A. G. and Marshall, D. B., The mechanical behaviour of ceramic matrix composites. Acta Metall, 1989, 37, 2567–2583. 3. Sutherland, S., Plucknett, K. P. and Lewis, M. H., High temperature mechanical and thermal stabilityof silicate matrix composites. J. Composites Eng., 1995, 5, 1367–1374. 4. Evans, A. G., Domergue, J.-M. and Vagaggini, E., Methodology for relating the tensile constitutive behaviour of ceramic-matrix composites to constituent properties. J. Am. Ceram. Soc., 1994, 77, 1425–1435. 5. Tu, W., Lange, F. F. and Evans, A. G., Concept for damagetolerant ceramic composite with ‘‘strong’’ interfaces. J. Am. Ceram. Soc., 1996, 79, 417–424. 6. Shannon, T. and Blackburn, S., The production of alumina/zirconia laminated composites byco-extrusion. Ceramic Engineering and Science Proceedings, 1995, 16, 1115–1120. 7. Liang, Z. and Blackburn, S., Co-extrusion of multilayered tubes. In Proceedings of The Better Ceramics Through Processing, ed. J. Yeomans and J. Binner. The Institute of Materials, London, 1998, pp. 109–115. 8. Hoy, W., Barda, A., Griffith, M. and Halloran, J. W., Microfabrication of ceramics byco-extrusion. J. Am. Ceram. Soc., 1998, 81, 152–158. 9. Crumm, A. T. and Halloran, J. W., Fabrication of microconfigured multicomponent ceramics. J. Am. Ceram. Soc., 1998, 81, 1053–1057. 10. Chen, Z., Ikeda, K., Murakami, T. and Takeda, T., Extrusion of metal-ceramic composite pipes. J. Am. Ceram. Soc., 2000, 83, 1081–1086. 11. Kaya, C. and Butler, E. G., Plastic forming and microstructural development of a-alumina ceramics from highlycompacted green bodies using extrusion. J. Eur. Ceram. Soc., 2002, 22, 1917–1926. 12. Kaya, C., Butler, E. G. and Lewis, M. H., Microstructurally controlled mullite ceramics from monophasic and diphasic solderived pastes. J. Mater. Sci. (in press). 13. Kaya, C. and Butler, E. G., Innovative processing of multiphase high temperature ceramics, Mid-term report to European Commission, Conract no: BRPR-CT97–069, December 1999. 14. Kaya, C., Butler, E. G. and Lewis, M. H., unpublished work. 15. Clegg, W. C, Kendall, C, Alford, N.M, Button, T.W. and Birchall, J.D., A simple way to make tough ceramics. Nature, 1990, 347, 455–457. 16. Kaya, C. and Butler, E. G., Innovative processing of multiphase high temperature ceramics, Final report to European Commission, Conract no: BRPR-CT97–069, December 2001. 17. Kaya, C., Processing and Properties of Alumina Fibre-Reinforced Mullite Ceramic Matrix Composites. PhD Thesis, The University of Birmingham, June 1999. 18. Huzzard, R. J. and Blackburn, S., Influence of solids loading on aqueous injection moulding paste. Brit. Ceram. Trans, 1999, 98, 49–56. 19. Draper, O., Blackburn, S., Dolman, G., Smalley, K. and Grif- fiths, A., A comparison of paste rheologyand extrudate strength with respect to binder formulation and forming technique. Mater. Processing Tech., 1999, 92–93, 141–146. 20. Huzzard, R. J. and Blackburn, S., Slip flow in concentarted alumina suspensions. Powder Technology, 1998, 97, 118–123. 21. Benbow, J. and Bridgwater, J., Paste Flow and Extrusion. Clarendon Press, Oxford, 1993. 22. Kaya, C., Butler, E. G. and Lewis, M. H., Processing and characterisation of mullite (Nextel 720TM) fibre-reinforced mullite matrix composites from hydrothermally processed mullite precursors. In High Temperature Ceramic Matrix Composites (HTCMC 4), ed. W. Krenkel, R. Naslain and H. Schneider. WileyVCH, Weinheim, Germany, 2001, pp. 639–644. 23. Wendorff, J., Janssen, R. and Claussen, N., Platinum as a weak interphase for fiber-reinforced oxide-matrix composites. J. Am. Ceram. Soc., 1998, 81, 2738–2740. 24. Kaya, C., Kaya, F., Boccaccini, A. R. and Chawla, K. K., Fabrication and characterisation of ni-coated carbon fibre-reinforced alumina ceramic matrix composites using electrophoretic deposition. Acta Materialia, 2001, 49, 1189–1197. 25. Mendelson, M. I., Average grain size in polycrystalline ceramics. J. Am. Ceram. Soc., 1969, 52, 443–446. 26. Whorlow, R. W., Rheological Techniques. Ellis Horwood, London, 1992. 942 C. Kaya et al. / Journal of the European Ceramic Society 23 (2003) 935–942
