C MATERIALIA Pergamon cta mater.4902001)1189-1197 www.elsevier.com/locate/actamat FABRICATION AND CHARACTERISATION OF NI-COATED CARBON FIBRE-REINFORCED ALUMINA CERAMIC MATRIX COMPOSITES USING ELECTROPHORETIC DEPOSITION C KAYA,21, F. KAYA A R BOCCACCINI and K. K CHAWLA and tratesrials. The Universit of birmingham. Edgbaston. Baminghpapm. B15 2 n. sK. M et lurgacar ay Material Engineering Department, Yildiz Technical University, Besiktas, Istanbul, 80750 Turkey, Department of Materials, Imperial College of Science, Technology and Medicine, Prince Consort Road, London, SW7 2BP UK and"Department of Materials and Mechanical Engineering, University of Alabama at Birmingham, Birmingham. AL 35294 USA Received 26 May 2000: received in revised form 21 December 2000: accepted 27 December 2000) Abstract-The present study explores the feasibility of fabricating Ni-coated carbon fibre-reinforced alu mina ceramic matrix composites via a single-infiltration electrophoretic deposition(EPD) process perfo ned in vacuum. The nano-size boehmite sol was seeded using nano-size 8-alumina powder in order to ntrol the final sintered microstructure and then characterised using transmission electron microscopy differential thermal and thermogravimetric analysis (DTA/TG) and X-ray disc centrifuge system(Bl- DC) in order to determine the sol microstructure, phase transformation temperatures and particle siz (also degree of agglomeration), respectively. An EPD manufacturing cell for fabrication of Ni-coated oon fibre reinforced alumina matrix cor tes was designed and experiments were conducted under acuum (first time to date), resulting in full deposition of the sol material throughout the voids ithin/between the fibre tows. Composites with high green density (67%0 theoretical density) were pro duced using an applied voltage of 15v d. c. and deposition time of 400 s. The sintered dens pressureless sintering at 1250C for 2 h was 91% theoretical density Crack path propagation tes that the metallic Ni coating was able to provide a weak interface, as an indenter induced cr he alumina matrix was deflected and arrested at the Ni interface. c 200/ Acta Materialia Ine by Elsevier Science Ltd. All rights reserved. Keywords: Electrophoretic deposition: Composites: Nickel: Interface: Microstructure 1 INTRODUCTION ered by several hundred degrees. A commercial sol can be seeded with isostructural seeds in order to A sol is generally defined as a colloidal dispersi ower crystallisation temperature and enhance of sol (or colloidal) processing route has many modifiers/seeds in terms of grain size, pore size and advantages, such as greater purity, higher homogen- pore size distribution. Boehmite(y-AlOOH)sol ity and ultrafine (5-100 nm)particle size distri- one of the ideal candidate materials to manufacture bution, in comparison to conventional ceramic pow- high quality alumina base ceramic components with der manufacturing processes. The main goal of this controlled tinal sintered microstructure, as it con technique is to achieve an ultra homogeneous or tains highly sinter-active ceramic particles on a atomic scale mixing of different chemical compo- ometer scale. Without seeding, however, a commer nents. The high surface area to volume ratio of a cial or hydrothermally produced boehmite sol ceramic sol makes the material usually highly sin- requires very high sintering temperatures(1600C) ter-active, thus sintering temperatures can be low- for complete densification. This is due to the large and extensive pore network that develops during the reconstructive transformation to the final stable t To whom all correspondence should be addressed. Fax: phase a-Al2O3, according to the dehydration and +44-121-4143441. high temperature phase transformation of boehm- E-mail address: c kaya@bhamac uk(C. Kaya) te[3]: 1359-6454/01/$20.00@ 2001 Acta Materialia Inc. Published by Elsevier Science Ltd. All rights reserved PI:S1359-6454(01)00018-0
Acta mater. 49 (2001) 1189–1197 www.elsevier.com/locate/actamat FABRICATION AND CHARACTERISATION OF Ni-COATED CARBON FIBRE-REINFORCED ALUMINA CERAMIC MATRIX COMPOSITES USING ELECTROPHORETIC DEPOSITION C. KAYA1, 2†, F. KAYA1 , A. R. BOCCACCINI3 and K. K. CHAWLA4 1 Interdisciplinary Research Centre (IRC) For High Performance Applications and School of Metallurgy and Materials, The University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK, 2 Metallurgical and Material Engineering Department, Yildiz Technical University, Besiktas, Istanbul, 80750 Turkey, 3 Department of Materials, Imperial College of Science, Technology and Medicine, Prince Consort Road, London, SW7 2BP UK and 4 Department of Materials and Mechanical Engineering, University of Alabama at Birmingham, Birmingham, AL 35294, USA ( Received 26 May 2000; received in revised form 21 December 2000; accepted 27 December 2000 ) Abstract—The present study explores the feasibility of fabricating Ni-coated carbon fibre-reinforced alumina ceramic matrix composites via a single-infiltration electrophoretic deposition (EPD) process performed in vacuum. The nano-size boehmite sol was seeded using nano-size δ-alumina powder in order to control the final sintered microstructure and then characterised using transmission electron microscopy, differential thermal and thermogravimetric analysis (DTA/TG) and X-ray disc centrifuge system (BIXDC) in order to determine the sol microstructure, phase transformation temperatures and particle size (also degree of agglomeration), respectively. An EPD manufacturing cell for fabrication of Ni-coated carbon fibre reinforced alumina matrix composites was designed and experiments were conducted under vacuum (first time to date), resulting in full deposition of the sol material throughout the voids within/between the fibre tows. Composites with high green density (67% theoretical density) were produced using an applied voltage of 15 V d.c. and deposition time of 400 s. The sintered density after pressureless sintering at 1250°C for 2 h was 91% theoretical density. Crack path propagation test showed that the metallic Ni coating was able to provide a weak interface, as an indenter induced crack within the alumina matrix was deflected and arrested at the Ni interface. 2001 Acta Materialia Inc. Published by Elsevier Science Ltd. All rights reserved. Keywords: Electrophoretic deposition; Composites; Nickel; Interface; Microstructure 1. INTRODUCTION A sol is generally defined as a colloidal dispersion of very fine solid particles in the range of 10 nm to 2 µm in a liquid medium, where the suspension is sustained indefinitely by Brownian motion. The use of sol (or colloidal) processing route has many advantages, such as greater purity, higher homogeneity and ultrafine (5–100 nm) particle size distribution, in comparison to conventional ceramic powder manufacturing processes. The main goal of this technique is to achieve an ultra homogeneous or atomic scale mixing of different chemical components. The high surface area to volume ratio of a ceramic sol makes the material usually highly sinter-active, thus sintering temperatures can be low- † To whom all correspondence should be addressed. Fax: 44-121-4143441. E-mail address: c.kaya@bham.ac.uk (C. Kaya) 1359-6454/01/$20.00 2001 Acta Materialia Inc. Published by Elsevier Science Ltd. All rights reserved. PII: S13 59-6454(01)00018-0 ered by several hundred degrees. A commercial sol can be seeded with isostructural seeds in order to lower crystallisation temperature and enhance densification with refined microstructure through solid-state epitaxy [1, 2]. The final sintered microstructure can be controlled using these modifiers/seeds in terms of grain size, pore size and pore size distribution. Boehmite (γ-AlOOH) sol is one of the ideal candidate materials to manufacture high quality alumina base ceramic components with controlled final sintered microstructure, as it contains highly sinter-active ceramic particles on a nanometer scale. Without seeding, however, a commercial or hydrothermally produced boehmite sol requires very high sintering temperatures (>1600°C) for complete densification. This is due to the large and extensive pore network that develops during the reconstructive transformation to the final stable phase α-Al2O3, according to the dehydration and high temperature phase transformation of boehmite [3]:
KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX Y-AlOOH -C rAl2O3→6Al2O3 Al M-ALO where w is the weight of charged particles deposited Therefore, the final sintered microstructure of alu- per unit area of electrode, C, the solids-loading of the nina derived from pure boehmite is very porous after suspension, Eo the permittivity of vacuum, e, the rela- even sintering at 1600oC for a long time(6 h), if the tive permittivity of the liquid, S the zeta potential of sol is not seeded [2]. On the contrary, when boehmite medium, E the applied potential, L the distance sol is seeded with crystallographically suitable modi fiers, high-density alumina components with con between the two electrodes and t the deposition time trolled microstructure are achievable at relatively low From this equation, it is clear that for a given suspen- sion, the zeta potential, dielectric constant, viscosit sintering temperatures(1100-1300C)[2, 4, 5]. Seed- and solids-loading are the critical factors determining been tried [2, 4, 5]. Because the addition of seeds to the EPD behaviour. In order to obtain high hetero- boehmite gels enhances the e-o transformation, com. coagulated-particle EPD rates, high E(water is ideal) plete densification of a-Al_O, occurs at temperatures and s values, together with low viscosity, are neces- as low as 1180%C with a grain size of only 0.43 um. sary as particle mobility within the suspension will Electrophoretic deposition (EPD) is as a novel, be enhanced. However, to produce a homogeneous relatively simple, high forming-rate technique for pro- green infiltrated microstructure, it is also crucial to lucing ceramic components [6-18 This infiltration optimise the solids-loading of the suspension without process relies on the presence of small charged par- The main objective of this study is to show the ticles in a liquid. l.e. a sol, which, on the aplication critical steps in producing Ni-coated carbon fibre- of an electric field, will move and deposit on an reinforced alumina matrix composites using the EPD oppositely charged electrode. This technique requires only low-cost equipment and offers new possibilities technique. The experiments were carried out in vac- for the design of ceramics monoliths or fibre- uum. EPD parameters in terms of applied voltage and reinforced composites with more uniform microstruc- deposition time were examined and optimised. More- tures [19]. It has been established that the EPD pro- sol with nanosize 8-alumina powder is described in cess can be utilised to infiltrate woven or non-woven fibre ceramic preforms [10, 13, 17, 19, 20). This tech- order to lower the sintering temperature of alumina. nique allows the ceramic medium to effectively fill Crack deflection behaviour at the ductile ni interface the inter- and intra-tow regions of fibres of small was examined crack path propagation test on diameters, which may be in close proximity (to the sintered composite sample point of touching in some cases). The requirement for full infiltration of the fibre preforms is that the 2. EXPERIMENTAL WORK infiltrating ceramic be in the form of a sol, e. g. nanos zed particles in suspension, to enable them to pen- 2.1.Materials etrate in between the closely spaced fibres. Commer- A commercially available boehmite(y-AIOOH)sol cial silica sols have been utilised with success to (Remet corp, USA, Remal A20) having 40 nm aver- produce SiC fibre reinforced silica matrix system [21] age particle size was used as the alumina source.The and in woven stainless steel fibre mat reinforced glass sol contains 20 wt% solids-loading and the boehmite [22, 23]. Moreover, commercial boehmite sol has particles are in the lath shape. The as-received been used in a previous work to fabricate metal fibre boehmite sol was seeded with 0.5 wt% nanosize reinforced alumina matrix composites [17] (13 nm)8-alumina(Aluminium Oxide C, Degussa gener be ine eoced ss noificane y he te ceanic Ag Germany and d - alumina (BDi i h eo scal s uk) sitional homogeneity and stability of the starting col- mina and 0.5% a-alumina. The seeding powder was lodal suspension and by the EPD fabrication para- first dispersed in distilled water, then the dispersion meters. To obtain a uniformly infiltrated green was added to the boehmite sol whilst this was stirred microstructure, EPD requires a kinetically stable, magnetically. Finally, the seeded boehmite sol was well-dispersed suspension having the highest possible ball-mixed for 12 h using high purity Tzp balls in a solids-loading but a relatively low viscosity, which plastic container. affects the particle electrophoretic mobility, and Nickel coated carbon fibres(Inco spp, IncofiberM, hence, deposition efficiency [9]. Furthermore, the 12K50, UK) were used as reinforcement. These fibres of the electrodefibre is infuenced strongly by the were in the form of continuous tows of nickel coated single carbon fibres. Ni was deposited using a gas process time, electrode separation and applied poten- plating technology. Fibre diameter and nickel coating tial, according to the following equation [15] us and had values of 10-15 an
1190 KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX gAlOOH → 400° CgAl2O3 → 800° CdAl2O3 → 1000° Cq (1) Al2O3 → 11001200° CaAl2O3 Therefore, the final sintered microstructure of alumina derived from pure boehmite is very porous after even sintering at 1600°C for a long time (>6 h), if the sol is not seeded [2]. On the contrary, when boehmite sol is seeded with crystallographically suitable modi- fiers, high-density alumina components with controlled microstructure are achievable at relatively low sintering temperatures (1100–1300°C) [2, 4, 5]. Seeding with γ-Al2O3, α-Al2O3 and α-Fe2O3 particles has been tried [2, 4, 5]. Because the addition of seeds to boehmite gels enhances the θ–α transformation, complete densification of α-Al2O3 occurs at temperatures as low as 1180°C with a grain size of only 0.43 µm. Electrophoretic deposition (EPD) is as a novel, relatively simple, high forming-rate technique for producing ceramic components [6–18]. This infiltration process relies on the presence of small charged particles in a liquid, i.e. a sol, which, on the application of an electric field, will move and deposit on an oppositely charged electrode. This technique requires only low-cost equipment and offers new possibilities for the design of ceramics monoliths or fibrereinforced composites with more uniform microstructures [19]. It has been established that the EPD process can be utilised to infiltrate woven or non-woven fibre ceramic preforms [10, 13, 17, 19, 20]. This technique allows the ceramic medium to effectively fill the inter- and intra-tow regions of fibres of small diameters, which may be in close proximity (to the point of touching in some cases). The requirement for full infiltration of the fibre preforms is that the infiltrating ceramic be in the form of a sol, e.g. nanosized particles in suspension, to enable them to penetrate in between the closely spaced fibres. Commercial silica sols have been utilised with success to produce SiC fibre reinforced silica matrix system [21] and in woven stainless steel fibre mat reinforced glass [22, 23]. Moreover, commercial boehmite sol has been used in a previous work to fabricate metal fibre reinforced alumina matrix composites [17]. In general, the properties of the sintered ceramic matrix will be influenced significantly by the compositional homogeneity and stability of the starting colloidal suspension and by the EPD fabrication parameters. To obtain a uniformly infiltrated green microstructure, EPD requires a kinetically stable, well-dispersed suspension having the highest possible solids-loading but a relatively low viscosity, which affects the particle electrophoretic mobility, and hence, deposition efficiency [9]. Furthermore, the number of charged particles deposited per unit area of the electrode/fibre is influenced strongly by the process time, electrode separation and applied potential, according to the following equation [15]: W 2 3 Ci 0rz 1 h E L t (2) where w is the weight of charged particles deposited per unit area of electrode, Ci the solids-loading of the suspension, 0 the permittivity of vacuum, r the relative permittivity of the liquid, ζ the zeta potential of the particles, η the viscosity of the suspension medium, E the applied potential, L the distance between the two electrodes and t the deposition time. From this equation, it is clear that for a given suspension, the zeta potential, dielectric constant, viscosity and solids-loading are the critical factors determining the EPD behaviour. In order to obtain high heterocoagulated-particle EPD rates, high (water is ideal) and ζ values, together with low viscosity, are necessary as particle mobility within the suspension will be enhanced. However, to produce a homogeneous green infiltrated microstructure, it is also crucial to optimise the solids-loading of the suspension without causing flocculation [9]. The main objective of this study is to show the critical steps in producing Ni-coated carbon fibrereinforced alumina matrix composites using the EPD technique. The experiments were carried out in vacuum. EPD parameters in terms of applied voltage and deposition time were examined and optimised. Moreover, the seeding process of a commercial boehmite sol with nanosize δ-alumina powder is described in order to lower the sintering temperature of alumina. Crack deflection behaviour at the ductile Ni interface was examined using crack path propagation test on sintered composite samples. 2. EXPERIMENTAL WORK 2.1. Materials A commercially available boehmite (γ-AlOOH) sol (Remet corp, USA, Remal A20) having 40 nm average particle size was used as the alumina source. The sol contains 20 wt% solids-loading and the boehmite particles are in the lath shape. The as-received boehmite sol was seeded with 0.5 wt% nanosize (13 nm) δ-alumina (Aluminium Oxide C, Degussa AG, Germany) and α-alumina (BDH Chemicals, UK) powders. The seeding material contains 99.5% δ-alumina and 0.5% α-alumina. The seeding powder was first dispersed in distilled water, then the dispersion was added to the boehmite sol whilst this was stirred magnetically. Finally, the seeded boehmite sol was ball-mixed for 12 h using high purity TZP balls in a plastic container. Nickel coated carbon fibres (Inco spp, Incofiber, 12K50, UK) were used as reinforcement. These fibres were in the form of continuous tows of nickel coated single carbon fibres. Ni was deposited using a gas plating technology. Fibre diameter and nickel coating were very homogeneous and had values of 10–15 and
KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX 0.5 um, respectively. The nickel coating provided weight gain per millisecond during the deposition excellent conductivity which is essential for EPD, as process, i.e. in real time. The dimensions of the anode well as ease of fibre hand and adequate wett-(25 mmX25 mm) were half of the cathode dimen ability. These fibres have been used recently as sions (50 mmX50 mm) in order to eliminate th reinforcement in borosilicate glass matrix com- 'edge effect which may give an inhomogeneous posites [201 thickness from the centre to the edges of the anode 22. EPD The EPD-prepared green body specimens containing about 25-30 vol% fibre loading were dried under An in situ EPD cell was designed in order to infil- humidity controlled atmosphere for one day and left trate the Ni-coated carbon fibre tows with the boehm- in normal air for another day before being press ite sol. The tows were unidirectionally aligned in a ureless sintered at 1250C for 2 h under nitrogen grooved perspex frame. EPD experiments were car- atmosphere ried out under vacuum. The Epd cell used is sche- 23. Microstructural characterisation matically shown in The distance between each tow was chosen to be in the range 0. 20.4 mm. Nickel To prepare green and sintered fibre reinforced coated carbon fibres held in the frame were used as CMC samples for cross-sectional scanning electron the deposition electrode( cathode). Two stainless steel microscopy (SEM), the specimens were placed in a plates on either side of the anode served as the posi- vacuum chamber and vacuum-impregnated with Epo- ive(anode) electrodes. After the fibre preform was fix resin. Impregnated green and sintered CMC placed in the sol, the system was vacuum degassed samples were left to harden overnight and then cut to remove any entrapped air, and then the cell elec- into slices using a diamond saw. a high resolution trodes were connected to a 0-60V d c power supply. scanning electron microscope(Field Emission Gun, EPD was performed subsequently under constant FEG SEM, Hitachi S-4000, Japan) was employed to voltage conditions (5, 10, 15 and 20 V) using varying characterise the various microstructural features of leposition times(from 50 to 500 s). An electrode sep- the infiltrated and sintered composite bodies, includ aration distance of 15 mm was used in all experi- ing: grain shape and size; porosity distribution and ments. Under the applied electric field, the very fine location; ductile interface, deposit thickness and infil- boehmite particles possessing a net positive surface tration of the matrix into the fibre architecture on both charge, as determined from the electrophoretic green and sintered samples. mobility data(see below ), migrated towards the nega- A Phillips CM 20 transmission electron microscop tive electrode, i.e. the Ni coated carbon fibre tows. (TEM) was used to observe and characterise the sol The particles infiltrated the fibre tows and deposited particle shape, size and degree of agglomeration, until a sufficient matrix thickness, which enveloped well as the nano-scale particle-particle interactions the fibre tows, was achieved. The fibre preform acting TEM chemical analysis of the sintered specimen was as the electrode was connected to a balance linked to then conducted, using a JEOL 4000 FX TEM a computer. The EPD apparatus is able to record the equipped with energy dispersive X-ray analysis. Pow der samples of the material deposited in between the layers of carbon fibre in each of the EPD-infiltrated CMPUTER green compacts were extracted. These samples were then subjected to differential thermal analysis (DtA) Vacuum Chamber Digital balance in order to determine the phase transformation tem- peratures. Other samples of this powder were calcined ((+) at given temperatures for 2 h and then analysed using X-ray( CuKo radiation) powder diffraction to identify the phases present. Finally, in order to characterise the interfacial behaviour of the composite produced under optimised EPD conditions, the crack path observation technique [24] was used on sintered and polished samples. 3. RESULTS AND DISCUSSION igure 2 shows a bright-field TEM micrograph of the spatial arrangement of the boehmite particles in Elcctrode suspension. With reference to the boehmite particles Ni coated fibers Electrode shown in the picture, the lath morphology of the boehmite particles is evident from the arrowed planar Fig. 1. Schematic diagram of the custom-built vacuum in situ and side views. The modal particle size of 40 nm is lectrophoretic deposition(EPD) cell incorporating Ni-coated so seen with an indicated size range of 2060nm carbon fibres as the deposition electrode Particle size analysis (cumulative mass distribution
KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX 1191 0.5 µm, respectively. The nickel coating provided excellent conductivity which is essential for EPD, as well as ease of fibre handling and adequate wettability. These fibres have been used recently as reinforcement in borosilicate glass matrix composites [20]. 2.2. EPD An in situ EPD cell was designed in order to infiltrate the Ni-coated carbon fibre tows with the boehmite sol. The tows were unidirectionally aligned in a grooved perspex frame. EPD experiments were carried out under vacuum. The EPD cell used is schematically shown in Fig. 1. The distance between each tow was chosen to be in the range 0.2–0.4 mm. Nickel coated carbon fibres held in the frame were used as the deposition electrode (cathode). Two stainless steel plates on either side of the anode served as the positive (anode) electrodes. After the fibre preform was placed in the sol, the system was vacuum degassed to remove any entrapped air, and then the cell electrodes were connected to a 0–60 V d.c. power supply. EPD was performed subsequently under constant voltage conditions (5, 10, 15 and 20 V) using varying deposition times (from 50 to 500 s). An electrode separation distance of 15 mm was used in all experiments. Under the applied electric field, the very fine boehmite particles possessing a net positive surface charge, as determined from the electrophoretic mobility data (see below), migrated towards the negative electrode, i.e. the Ni coated carbon fibre tows. The particles infiltrated the fibre tows and deposited until a sufficient matrix thickness, which enveloped the fibre tows, was achieved. The fibre preform acting as the electrode was connected to a balance linked to a computer. The EPD apparatus is able to record the Fig. 1. Schematic diagram of the custom-built vacuum in situ electrophoretic deposition (EPD) cell incorporating Ni-coated carbon fibres as the deposition electrode. weight gain per millisecond during the deposition process, i.e. in real time. The dimensions of the anode (25 mm25 mm) were half of the cathode dimensions (50 mm50 mm) in order to eliminate the ‘edge effect’ which may give an inhomogeneous thickness from the centre to the edges of the anode. The EPD-prepared green body specimens containing about 25–30 vol% fibre loading were dried under humidity controlled atmosphere for one day and left in normal air for another day before being pressureless sintered at 1250°C for 2 h under nitrogen atmosphere. 2.3. Microstructural characterisation To prepare green and sintered fibre reinforced CMC samples for cross-sectional scanning electron microscopy (SEM), the specimens were placed in a vacuum chamber and vacuum-impregnated with Epo- fix resin. Impregnated green and sintered CMC samples were left to harden overnight and then cut into slices using a diamond saw. A high resolution scanning electron microscope (Field Emission Gun, FEG SEM, Hitachi S-4000, Japan) was employed to characterise the various microstructural features of the infiltrated and sintered composite bodies, including: grain shape and size; porosity distribution and location; ductile interface, deposit thickness and infiltration of the matrix into the fibre architecture on both green and sintered samples. A Phillips CM 20 transmission electron microscope (TEM) was used to observe and characterise the sol particle shape, size and degree of agglomeration, as well as the nano-scale particle–particle interactions. TEM chemical analysis of the sintered specimen was then conducted, using a JEOL 4000 FX TEM equipped with energy dispersive X-ray analysis. Powder samples of the material deposited in between the layers of carbon fibre in each of the EPD-infiltrated green compacts were extracted. These samples were then subjected to differential thermal analysis (DTA) in order to determine the phase transformation temperatures. Other samples of this powder were calcined at given temperatures for 2 h and then analysed using X-ray (CuKα radiation) powder diffraction to identify the phases present. Finally, in order to characterise the interfacial behaviour of the composite produced under optimised EPD conditions, the crack path observation technique [24] was used on sintered and polished samples. 3. RESULTS AND DISCUSSION Figure 2 shows a bright-field TEM micrograph of the spatial arrangement of the boehmite particles in suspension. With reference to the boehmite particles shown in the picture, the lath morphology of the boehmite particles is evident from the arrowed planar and side views. The modal particle size of 40 nm is also seen with an indicated size range of 20–60 nm. Particle size analysis (cumulative mass distribution
l19 KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX Table 1. The effect of the vacuum atmosphere on the green and sintered densities(in theoretical density, TD) of alumina matrix composite fabricated by EPD, The samples were sintered at 1250 C for 2h Processing ro Green density(% TD) Sintered density(S PD in air EPD under vacuum green body. Thus, all EPD experiments were carried out under a high vacuum in order to obtain full depo sition of the sol material throughout the voids within the fibre mat. Two different EPD experiments were Fig. 2. Bright-field TEM micrograph showing the parti initially carried out in air and under vacuum usin show the surface and the side view of the lath-shape boehmite constant applied voltage of 10 V for 4 min with a con stant electrode separation of 15 mm in order to explore the effect of vacuum. Results are presented in Table I. It was found that vacuum EPD provided 9o)of the boehmite sol indicated that 100% of the a higher degree of deposition by eliminating the effect formed as a result of the electrolysis of Fig.3. It was also found that the boehmite sol used water. Under vacuum, very fine boehmite particles 油 hin the suspension. The graph of particle size dis- g逃的 deep into the inter/intra: fibre tows,fl as there were no big heteroflocculated agglomerates ing all the voids, resulting in the formation of high- uality, dense green(and sintered) composites. The tribution shows no extreme large particle sizes (Fig. maximum green and sintered densities were 54 and 75% of theoretical density (TD) for EPD experiments In situ EPD system has been developed recently carried out in air, respectively, whilst vacuum EPD [19] and successfully applied to produce different process provided green and sintered density values of fibre-reinforced composites, such as alumina fibre- 67 and 84%TD, respectively. These results confirmed reinforced mullite [9], mullite fibre-reinforced mullite the effectiveness of the vacuum environment 25], woven stainless steel fibre-reinforced silica [131 and nickel coated carbon fibre-reinforced borosilicate d, Figure 4 shows the particle electrophoretic mobility ata for the nano-size aqueous boehmite sol as a func glass composites[20]. For the first time in this work, tion of sol pH. From these data, it is clear that the however, in situ EPD experiments were carried out boehmite particles are positively charged below ph under vacuum(see Fig. 1), in order to eliminate the 9.5 and negatively charged above this point. At the undesirable formation and entrapment of bubbles working ph value of 4, therefore, positively charged within the deposit due to the electrolysis (evolution boehmite particles will move and deposit on to the of gases)of the aqueous sol dispersion medium. The negative electrode(fibres )under an applied d.c.volt (100 nm)particle material, in form of a sol, would The microstructure of uncoated and Ni-coated car penetrate deep into the inter/intra-fibre tows regions, filling the voids and thus providing a dense composite Diameter, nm Fig 3. X-ray disc centrifuge(BI-XDC)particle size distribution Fig. 4. Particle electrophoretic mobility data for Remal A20 (in %o cumulative mass, smaller than) of boehmite sol for a boehmite suspension. The suspension solids-loading is solids-loading of 2 wt%. Note that 100% of the total boehmite 0.01 wt% of the dispersion medium. Note that the boehmite particles are smaller than 60 nm, showing the absence of big particles have positive surface charge at the working pH value eteroflocculated chains within the suspension
1192 KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX Fig. 2. Bright-field TEM micrograph showing the particle shape and size of the used Remal A20 boehmite sol. Arrows show the surface and the side view of the lath-shape boehmite particles. %) of the boehmite sol indicated that 100% of the total particles were smaller than 60 nm, as shown in Fig. 3. It was also found that the boehmite sol used in this work was kinetically stable and well dispersed, as there were no big heteroflocculated agglomerates within the suspension. The graph of particle size distribution shows no extreme large particle sizes (Fig. 3). In situ EPD system has been developed recently [19] and successfully applied to produce different fibre-reinforced composites, such as alumina fibrereinforced mullite [9], mullite fibre-reinforced mullite [25], woven stainless steel fibre-reinforced silica [13] and nickel coated carbon fibre-reinforced borosilicate glass composites [20]. For the first time in this work, however, in situ EPD experiments were carried out under vacuum (see Fig. 1), in order to eliminate the undesirable formation and entrapment of bubbles within the deposit due to the electrolysis (evolution of gases) of the aqueous sol dispersion medium. The main purpose of the EPD process is that ultra fine (100 nm) particle material, in form of a sol, would penetrate deep into the inter/intra-fibre tows regions, filling the voids and thus providing a dense composite Fig. 3. X-ray disc centrifuge (BI-XDC) particle size distribution (in % cumulative mass, smaller than) of boehmite sol for a solids-loading of 2 wt%. Note that 100% of the total boehmite particles are smaller than 60 nm, showing the absence of big heteroflocculated chains within the suspension. Table 1. The effect of the vacuum atmosphere on the green and sintered densities (in % theoretical density, TD) of alumina matrix composites fabricated by EPD. The samples were sintered at 1250°C for 2 h Sintered density (% Processing route Green density (% TD) TD) EPD in air 54 75 EPD under vacuum 67 84 green body. Thus, all EPD experiments were carried out under a high vacuum in order to obtain full deposition of the sol material throughout the voids within the fibre mat. Two different EPD experiments were initially carried out in air and under vacuum using constant applied voltage of 10 V for 4 min with a constant electrode separation of 15 mm in order to explore the effect of vacuum. Results are presented in Table 1. It was found that vacuum EPD provided a higher degree of deposition by eliminating the effect of gases formed as a result of the electrolysis of water. Under vacuum, very fine boehmite particles can penetrate deep into the inter/intra-fibre tows, filling all the voids, resulting in the formation of highquality, dense green (and sintered) composites. The maximum green and sintered densities were 54 and 75% of theoretical density (TD) for EPD experiments carried out in air, respectively, whilst vacuum EPD process provided green and sintered density values of 67 and 84%TD, respectively. These results confirmed the effectiveness of the vacuum environment. Figure 4 shows the particle electrophoretic mobility data for the nano-size aqueous boehmite sol as a function of sol pH. From these data, it is clear that the boehmite particles are positively charged below pH 9.5 and negatively charged above this point. At the working pH value of 4, therefore, positively charged boehmite particles will move and deposit on to the negative electrode (fibres) under an applied d.c. voltage. The microstructure of uncoated and Ni-coated carFig. 4. Particle electrophoretic mobility data for Remal A20 boehmite suspension. The suspension solids-loading is 0.01 wt% of the dispersion medium. Note that the boehmite particles have positive surface charge at the working pH value of 4
KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX 1193 bon fibres, see Fig. 5(a)and(b), respectively, showed that the metallic Ni coating around the fibres was very △5V homogeneous and with a uniform thickness of about 0.5 um. The carbon fibres were 10-15 um in diam- When EPD is used as a forming technique fo MCs, it is possible to use either constant current or 3 constant voltage conditions. Constant current con- ditions would result in a continually increasing 。=9=3=a882 age due to the increasing resistance of the deposit as it grows in thickness and mass, thus constant voltage conditions were used in this work. Voltages higher than 20V were not used in these experiments in order Electrophoretic Deposition Time(s) to prevent gas bubbles being incorporated within the deposited ceramic matrix. The results from experi- -5V mental trials using the in situ EPD cell design under 000 vacuum are shown in Fig. 6. The graphs in Fig. 6(a) g weight and thickness as a function of deposition time 3300 experiments were performed for duration of up to 200 500 s, as this gave a deposit thickness of about 3 100 o um, which was enough to produce a composite 5 with an acceptable green density. The deposit thick- ness increased with increasing deposition time, as the amount of particles deposited on to the fibres preform Electrophoretic Deposition Time(s) increased. When aqueous based sols are used in EPD 3 experiments, one problem associated with this is the 3 (c) 0 3.2 um electrophoretic thickness as a function of deposition time for different applied voltages. In(c), the deposit formation rate as a function of EPD time under optimised applied voltage of electrolysis of the water. Higher voltages resulted in rapid deposit formation, but also in the undesirable formation and entrapment deposit due to the electrolysis of the aqueous sol dis- persion medium, while low voltages reduced the ele trolysis, but they also needed higher deposition times Thus, a compromise had to be found and voltage and 231pm deposition time were optimised Figure 6a shows that the eight al Most linear with increasing deposition time(up to Fig. 5. FEG SEM micrographs, showing the microstructure of 500 s)and voltage.An oltage of 20 V seems that the metallic Ni coating around the fibres is very homo. ideal according to the I app of deposited material geneous and it has a uniform thickness of 0.5 um. The carbon as shown in Fig. 6(a). However, the deposited matrix fibres are 10-15 um in diameter a result of the gas
KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX 1193 bon fibres, see Fig. 5(a) and (b), respectively, showed that the metallic Ni coating around the fibres was very homogeneous and with a uniform thickness of about 0.5 µm. The carbon fibres were 10–15 µm in diameter. When EPD is used as a forming technique for CMCs, it is possible to use either constant current or constant voltage conditions. Constant current conditions would result in a continually increasing voltage due to the increasing resistance of the deposit as it grows in thickness and mass, thus constant voltage conditions were used in this work. Voltages higher than 20 V were not used in these experiments in order to prevent gas bubbles being incorporated within the deposited ceramic matrix. The results from experimental trials using the in situ EPD cell design under vacuum are shown in Fig. 6. The graphs in Fig. 6(a) and (b) show the results of electrophoretic deposit weight and thickness as a function of deposition time for different applied voltages, respectively. EPD experiments were performed for duration of up to 500 s, as this gave a deposit thickness of about 660 µm, which was enough to produce a composite with an acceptable green density. The deposit thickness increased with increasing deposition time, as the amount of particles deposited on to the fibres preform increased. When aqueous based sols are used in EPD experiments, one problem associated with this is the Fig. 5. FEG SEM micrographs, showing the microstructure of (a) uncoated and (b) Ni-coated carbon fibres. It can be seen that the metallic Ni coating around the fibres is very homogeneous and it has a uniform thickness of 0.5 µm. The carbon fibres are 10–15 µm in diameter. Fig. 6. Graphs of the (a) electrophoretic deposit weight and (b) electrophoretic thickness as a function of deposition time for different applied voltages. In (c), the deposit formation rate as a function of EPD time under optimised applied voltage of 15 V is shown. electrolysis of the water. Higher voltages resulted in rapid deposit formation, but also in the undesirable formation and entrapment of bubbles within the deposit due to the electrolysis of the aqueous sol dispersion medium, while low voltages reduced the electrolysis, but they also needed higher deposition times. Thus, a compromise had to be found and voltage and deposition time were optimised. Figure 6a shows that the increase in weight is almost linear with increasing deposition time (up to 500 s) and voltage. An applied voltage of 20 V seems ideal according to the amount of deposited material, as shown in Fig. 6(a). However, the deposited matrix microstructure is porous as a result of the gases
KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX evolved at the electrode surface(due to the electro- (up to 150 s) and then it started to decrease with lytic decomposition of the aqueous medium at this increasing deposition time. The decrease in depo relatively high voltage), becoming trapped in the sition rate was attributed to the increase in the resist leposit. Porosity dependence on the applied voltages ance of the deposit, as the current diminishes due to is shown in Table 2. As the voltage was increased the increase in deposit thickness and the removal of from 15 to 20 V, the volume fraction of surface charged boehmite particles from the sol. As a result porosity was increased more than twice (see Table of the decrease in the potential drop across the sus- 2)but there was no significant increase in material pension, the velocity and deposition rate deposition(see Fig 6a). However, applied voltages of charged colloidal boehmite particles also decreased 5 and 10 V showed similar trends and no significant [191 increase in deposit weight was observed. Thus, an The simultaneous decrease in the cell current is due applied voltage of 15 v which gave the maximum to the increasing thickness of the deposit, which in chosen. Under applied voltages of 5 and 10V, the the electrophoretic cell, ie, is given by lo urrent in EPD deposit weight(and the highest final TD)was turn increases the total cell resistance. The ci deposition rate was very slow due to the low particle mobility in the suspension, and the quality of the =(V-VaR deposits was poor. The thickness of selected electro- phoretic deposits, measured from the fibre surface, was plotted against the deposition time for different where Vc is the applied constant voltage of the cell voltages in Fig. 6(b). The deposition weight and Va the instantaneous voltage drop across the deposit, thickness as functions of deposition time and voltage which is equal to the product of the instantaneous cur- followed a very similar trend, as shown in both Fig. rent and the electrical resistance of the deposit, and 6(a) and(b). When using applied voltages of 5 and Rs the electrical resistance of the sol, which can be 10V, the deposition weight and thickness did not considered to be constant. The rate of current change significantly with time, indicating that the decrease is a function of the physical and chemical applied electric field was too low to influence the properties of the deposit, since the density, porosity, movement of the boehmite sol particles and cause thickness and chemical nature of the deposit control adequate deposition. However, higher voltages of 15 changes in the deposit resistance. According to equ and 20v produced deposition thickness which ation (3), the EPd deposit thickness will reach ppeared nearly linear over the time period studied. maximum limiting value when Va=Ve, since i=0. It It was also noted that increasing the applied voltage was found that EPD at times greater than 400 s pro- from 15 to 20 V did not increase both the deposition duced very little change in the EPD deposition rate weight and thickness. This was possibly due to the (see also Fig. 6(a)and (b)). Thus, for the nickel- early formation of a deposition layer on the outer sur- coated carbon fibre/boehmite system used here, depo- face of the fibre mat which inhibited further particle sition times shorter than about 4 min are the most deposition through the fibre mat, when an applied practical and economical oltage of 20 V was used. It was concluded that there Electrophoretic deposition parameters were was an optimum applied voltage of 15V for the parti- optimised in order to achieve fully infiltrated nickel cular sol and fibre systems used in this work coated carbon fibre-reinforced alumina CMC pre- In situ (real time) deposit weight measurements forms with the minimum amount of excess material provided highly reliable data to determine the depo- being present in the outer regions of the preform In sition rate accurately. Fig. 6(c)shows, for example, the present study, the optimum deposition voltage and the rate of deposit formation as a function of time time were determined as 15 V and 400 s, respectively, for a constant optimised applied voltage of 15 V. The for full infiltration. The full infiltration supplied an deposition rate was normalised for the weight of the EPD deposit thickness of >650 um. USing these para- Ni-coated carbon fibre mat tested. The rate of EPD meters, it was possible to produce fully infiltrated per gram of fibre per second. This figure shows that beneficial as the fibre mats then have a reduced pa p is given in grams of electrophoretic deposit formed mats with only a thin(100 um)excess layer. This was the rate of deposition was very high at the beginning pensity to form large cracks during the drying stage This cracking is due to the differential shrinkage of Table 2. Porosity(%)dependence on the applie during ep the gel network which generates tensile stresses at the The composite sample was sintered at 1250 C for 2 h the deposits. These stresses may reach values higher than the mechanical resistance of the (from density gel, especially with thicker sections, leading to the measurements) initiation of cracks [22 The key issue in limiting the incorporation of a col lodal matrix into a reinforcing fibre mat using EPD is the tightness of the fibre architecture. The nicke coated carbon fibre mats aligned in a perspex frame used in this study had so
1194 KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX evolved at the electrode surface (due to the electrolytic decomposition of the aqueous medium at this relatively high voltage), becoming trapped in the deposit. Porosity dependence on the applied voltages is shown in Table 2. As the voltage was increased from 15 to 20 V, the volume fraction of surface porosity was increased more than twice (see Table 2) but there was no significant increase in material deposition (see Fig. 6a). However, applied voltages of 5 and 10 V showed similar trends and no significant increase in deposit weight was observed. Thus, an applied voltage of 15 V which gave the maximum EPD deposit weight (and the highest final TD) was chosen. Under applied voltages of 5 and 10 V, the deposition rate was very slow due to the low particle mobility in the suspension, and the quality of the deposits was poor. The thickness of selected electrophoretic deposits, measured from the fibre surface, was plotted against the deposition time for different voltages in Fig. 6(b). The deposition weight and thickness as functions of deposition time and voltage followed a very similar trend, as shown in both Fig. 6(a) and (b). When using applied voltages of 5 and 10 V, the deposition weight and thickness did not change significantly with time, indicating that the applied electric field was too low to influence the movement of the boehmite sol particles and cause adequate deposition. However, higher voltages of 15 and 20 V produced deposition thickness which appeared nearly linear over the time period studied. It was also noted that increasing the applied voltage from 15 to 20 V did not increase both the deposition weight and thickness. This was possibly due to the early formation of a deposition layer on the outer surface of the fibre mat which inhibited further particle deposition through the fibre mat, when an applied voltage of 20 V was used. It was concluded that there was an optimum applied voltage of 15 V for the particular sol and fibre systems used in this work. In situ (real time) deposit weight measurements provided highly reliable data to determine the deposition rate accurately. Fig. 6(c) shows, for example, the rate of deposit formation as a function of time for a constant optimised applied voltage of 15 V. The deposition rate was normalised for the weight of the Ni-coated carbon fibre mat tested. The rate of EPD is given in grams of electrophoretic deposit formed per gram of fibre per second. This figure shows that the rate of deposition was very high at the beginning Table 2. Porosity (%) dependence on the applied voltage during EPD. The composite sample was sintered at 1250°C for 2 h Porosity, % Applied voltage (V) (from density measurements) 5 17 10 14 15 9 20 19 (up to 150 s) and then it started to decrease with increasing deposition time. The decrease in deposition rate was attributed to the increase in the resistance of the deposit, as the current diminishes due to the increase in deposit thickness and the removal of charged boehmite particles from the sol. As a result of the decrease in the potential drop across the suspension, the velocity and deposition rate of the charged colloidal boehmite particles also decreased [19]. The simultaneous decrease in the cell current is due to the increasing thickness of the deposit, which in turn increases the total cell resistance. The current in the electrophoretic cell, ic, is given by [26] ic (VcVd)/Rs (3) where Vc is the applied constant voltage of the cell, Vd the instantaneous voltage drop across the deposit, which is equal to the product of the instantaneous current and the electrical resistance of the deposit, and Rs the electrical resistance of the sol, which can be considered to be constant. The rate of current decrease is a function of the physical and chemical properties of the deposit, since the density, porosity, thickness and chemical nature of the deposit control changes in the deposit resistance. According to equation (3), the EPD deposit thickness will reach a maximum limiting value when VdVc, since ic0. It was found that EPD at times greater than 400 s produced very little change in the EPD deposition rate (see also Fig. 6(a) and (b)). Thus, for the nickelcoated carbon fibre/boehmite system used here, deposition times shorter than about 4 min are the most practical and economical. Electrophoretic deposition parameters were optimised in order to achieve fully infiltrated nickelcoated carbon fibre-reinforced alumina CMC preforms with the minimum amount of excess material being present in the outer regions of the preform. In the present study, the optimum deposition voltage and time were determined as 15 V and 400 s, respectively, for full infiltration. The full infiltration supplied an EPD deposit thickness of >650 µm. Using these parameters, it was possible to produce fully infiltrated mats with only a thin (100 µm) excess layer. This was beneficial as the fibre mats then have a reduced propensity to form large cracks during the drying stage. This cracking is due to the differential shrinkage of the gel network which generates tensile stresses at the surface of the deposits. These stresses may reach values higher than the mechanical resistance of the gel, especially with thicker sections, leading to the initiation of cracks [22]. The key issue in limiting the incorporation of a colloidal matrix into a reinforcing fibre mat using EPD is the tightness of the fibre architecture. The nickelcoated carbon fibre mats aligned in a perspex frame used in this study had some very tight regions,
KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX especially at the fibre-fibre contact points. The final the matrix, which resulted from the contact with the green and sintered densities and, hence, the mechan- diamond saw during the very slow-speed cutting ical properties of the final composites made from operation. Dense composite samples were obtained these fabrics, will be poor if these regions are not by pressureless sintering at 1250C for 2 h under N2 fully infiltrated with matrix material during EPD. The atmosphere. A fully infiltrated and dense sintered FEG-SEM pictures shown in Fig. 7 represent the microstructure is shown in Fig 8. Sintered densities green microstructures of composites produced under of about 91% of TD were measured by using the optimised EPD parameters. Figure 7(a) shows a SEM Archimedes technique. This result confirms that micrograph of an EPD-formed green body, containing present processing approach is very effective in 30 vol% fibre loading. High green densities (about obtaining dense CMCs samples at relatively low sin- 61% of TD) were achieved at 15 V for 400 s Green tering temperatures and avoiding the cost-intensive density was measured by dividing the mass of the hot-pressing fabrication procedure, which has been ample by its geometrically determined volume. It the common practise in previous studies on fibre- can also be seen in Fig. 7(b) that the Ni-coated carbon reinforced CMCs [27 fibre preform was fully infiltrated with the boehmite A crack path propagation test was performed on sol. Even regions where the Ni-coated fibres were sintered samples to characterise the Ni interface nearly touching each other were fully impregnated by behaviour in terms of crack deflection and debonding. the nano-size boehmite powders in a very short time, Fig. 9(a) shows the interactions between an indenter- i.e. 400 s, leading to high-quality green bodies. The induced crack created within the alumina matrix and effectiveness of the EPD providing full deposition the ductile Ni interface. The crack first interacts wit between two fibres with a separation of 400-500 nm the Ni coating and then it is deflected at the interface is clearly visible in Fig. 7(b). It must be noted that One can conclude from the crack path in Fig. 9(a) these green samples were not polished in order to that a weak bonding between the Ni coating and the avoid damaging the ductile Ni interface. Thus, some alumina matrix exists. Thus, no significant chemical cutting effects are visible on the carbon fibres and reaction between the Ni interface and the alumin matrix at the sintering temperature, i.e. 1250.C, has occurred, as shown in Fig. 9(b). From this picture, it is also evident that the Ni coating provides a useful weak interface between carbon fibres and alumina matrix, resulting in crack deflection and debonding These mechanisms should result in an overall pseudo-ductile fracture mode of the composite since catastrophic failure is expected to be eliminated by the operation of crack deflection and pull-out mec anisms [28 X-ray diffraction patterns for the powders extracted from the regions between each layer of the nickel 7.5 um coated carbon fibre preforms after calcination at 1250C for 2 h are shown in Fig. 10. All EPd formed mples produced pure a-alumina peaks, with no additional peaks corresponding to non-transformed 273pr Fig. 7. FEG SEM micrographs of EPD- infiltrated unsintered Ni-coated carbon fibre reinforced alumina matrix composite containing 30 vol% fibre loading. The fibre preform was infil- trated using an applied voltage of 15 V for 400 s Both(a) high Fig. 8. FEG SE uum EPD-infiltrated N and(b) low magnification micrographs show that the Ni-coated coated Carbon d alumina matrix composite con- carbon fibre preform was fully infiltrated with the boehmite taining 30 vol% after sintering at 1250%C for 2 h sol. Even regions where the Ni-coated fibres were nearly touch- in nitrogen atmosphere. The micrograph shows the full depo- ing each other have been fully impregnated by the nanosize sition of the boehmite matrix into the fibre preform and a dense boehmite particles, leading to high-quality green bodies. microstructure after
KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX 1195 especially at the fibre–fibre contact points. The final green and sintered densities and, hence, the mechanical properties of the final composites made from these fabrics, will be poor if these regions are not fully infiltrated with matrix material during EPD. The FEG-SEM pictures shown in Fig. 7 represent the green microstructures of composites produced under optimised EPD parameters. Figure 7(a) shows a SEM micrograph of an EPD-formed green body, containing 30 vol% fibre loading. High green densities (about 61% of TD) were achieved at 15 V for 400 s. Green density was measured by dividing the mass of the sample by its geometrically determined volume. It can also be seen in Fig. 7(b) that the Ni-coated carbon fibre preform was fully infiltrated with the boehmite sol. Even regions where the Ni-coated fibres were nearly touching each other were fully impregnated by the nano-size boehmite powders in a very short time, i.e. 400 s, leading to high-quality green bodies. The effectiveness of the EPD providing full deposition between two fibres with a separation of 400–500 nm is clearly visible in Fig. 7(b). It must be noted that these green samples were not polished in order to avoid damaging the ductile Ni interface. Thus, some cutting effects are visible on the carbon fibres and Fig. 7. FEG SEM micrographs of EPD-infiltrated unsintered Ni-coated carbon fibre reinforced alumina matrix composite containing 30 vol% fibre loading. The fibre preform was infiltrated using an applied voltage of 15 V for 400 s. Both (a) high and (b) low magnification micrographs show that the Ni-coated carbon fibre preform was fully infiltrated with the boehmite sol. Even regions where the Ni-coated fibres were nearly touching each other have been fully impregnated by the nanosize boehmite particles, leading to high-quality green bodies. the matrix, which resulted from the contact with the diamond saw during the very slow-speed cutting operation. Dense composite samples were obtained by pressureless sintering at 1250°C for 2 h under N2 atmosphere. A fully infiltrated and dense sintered microstructure is shown in Fig. 8. Sintered densities of about 91% of TD were measured by using the Archimedes technique. This result confirms that the present processing approach is very effective in obtaining dense CMCs samples at relatively low sintering temperatures and avoiding the cost-intensive hot-pressing fabrication procedure, which has been the common practise in previous studies on fibrereinforced CMCs [27]. A crack path propagation test was performed on sintered samples to characterise the Ni interface behaviour in terms of crack deflection and debonding. Fig. 9(a) shows the interactions between an indenterinduced crack created within the alumina matrix and the ductile Ni interface. The crack first interacts with the Ni coating and then it is deflected at the interface. One can conclude from the crack path in Fig. 9(a) that a weak bonding between the Ni coating and the alumina matrix exists. Thus, no significant chemical reaction between the Ni interface and the alumina matrix at the sintering temperature, i.e. 1250°C, has occurred, as shown in Fig. 9(b). From this picture, it is also evident that the Ni coating provides a useful weak interface between carbon fibres and alumina matrix, resulting in crack deflection and debonding. These mechanisms should result in an overall ‘pseudo-ductile’ fracture mode of the composite since catastrophic failure is expected to be eliminated by the operation of crack deflection and pull-out mechanisms [28]. X-ray diffraction patterns for the powders extracted from the regions between each layer of the nickel coated carbon fibre preforms after calcination at 1250°C for 2 h, are shown in Fig. 10. All EPD formed samples produced pure α-alumina peaks, with no additional peaks corresponding to non-transformed Fig. 8. FEG SEM micrograph of vacuum EPD-infiltrated Nicoated Carbon fibre reinforced alumina matrix composite containing 30 vol% fibre loading after sintering at 1250°C for 2 h in nitrogen atmosphere. The micrograph shows the full deposition of the boehmite matrix into the fibre preform and a dense composite microstructure after sintering
KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX powder samples obtained from the dried boehmite precursor sols exhibited major endothermic peaks at 100 and 250C, corresponding to the removal of water from boehmite. As seen from the dta curve in Fig. ll, first boehmite transformation from x AlOOH to y-Al2O3 has been completed at 500C and the other phase transformations of transitional alum inas(H-8-0)to a-alumina, as shown in Eq(1), take place at 800, 1050 and 1250C, respectively. The final a-alumina formation is evident with a major 29μm peak at 1250C. The weight lost of boehmite as a 185 um function of temperature is also seen from the TGA ured(starting and finishing weights of the sample tested were 16.2 and 12.8 mg, respectively) 4. CONCLUSIONS EPD was used under vacuum as a novel, quick and cost-effective processing technique to explore the possibility of fabricating nickel coated carbon fibre reinforced alumina matrix composites, which could be densified by a pressureless sintering route. Boehm- ite sol seeded with nano-size d-alumina powders was found to be co matrix after firing for 2 h at 1250C. Unidirectionally Fig 9. FEG SEM micrograph of EPD-infiltrated Ni-coated car- aligned Ni coated carbon fibre-reinforced alumina propagation test(sample sintered at 1250C for 2 h under nitro- matrix composites were produced using an optimised gen atmosphere)showing: (a)the propagation of an indenter- single-stage EPD technique. The process parameters induced crack along the interface between the Ni coating and in terms of applied d. c. voltage and deposition time hatrix, proving that the crack is deflected and arrested at the were found to be 15 V and 400 s, respectively, for interface, (b)the presence of a weak interface between the duc. tile Ni coating and the alumina matrix(Ni: nickel coating, M ull infiltration of the fibre mats used. vacuum alumina matrix and C: carbon fibre) environment during EPD was proven to be effective in reducing the formation of gas bubbles within the deposited green body, resulting in a high degree of excess alumina. Thus, Fig. 10 confirms that the a- deposition and high final density of the component alumina matrix can be produced from seeded boehm- produced. Under vacuum, ultra fine boehmite par ite sol using low sintering temperatures, as low as ticles within the sol could easily penetrate deep into 1250C, in agreement with literature reports [2, 4, 5]. the inter/intra-fibre tows regions, filling the voids and The DTA traces(heating rate: 10C/min) for the thus providing a dense composite Crack path propa "2 Theta Fig. 10. X-ray(CuKo) diffraction patterns for EPD deposited boehmite material, showing the final a-alumina matrix structure after sintering at 1250C for 2 h
1196 KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX Fig. 9. FEG SEM micrograph of EPD-infiltrated Ni-coated carbon fibre reinforced alumina matrix composite after crack path propagation test (sample sintered at 1250°C for 2 h under nitrogen atmosphere) showing: (a) the propagation of an indenterinduced crack along the interface between the Ni coating and matrix, proving that the crack is deflected and arrested at the interface, (b) the presence of a weak interface between the ductile Ni coating and the alumina matrix (Ni: nickel coating, M: alumina matrix and C: carbon fibre). excess alumina. Thus, Fig. 10 confirms that the α- alumina matrix can be produced from seeded boehmite sol using low sintering temperatures, as low as 1250°C, in agreement with literature reports [2, 4, 5]. The DTA traces (heating rate: 10°C/min) for the Fig. 10. X-ray (CuKα) diffraction patterns for EPD deposited boehmite material, showing the final α-alumina matrix structure after sintering at 1250°C for 2 h. powder samples obtained from the dried boehmite precursor sols exhibited major endothermic peaks at 100 and 250°C, corresponding to the removal of water from boehmite. As seen from the DTA curve in Fig. 11, first boehmite transformation from γ- AlOOH to γ-Al2O3 has been completed at 500°C and the other phase transformations of transitional aluminas (γ–δ–θ) to α-alumina, as shown in Eq. (1), take place at 800, 1050 and 1250°C, respectively. The final α-alumina formation is evident with a major peak at 1250°C. The weight lost of boehmite as a function of temperature is also seen from the TGA curve. A total weight loss of about 20.98% was measured (starting and finishing weights of the sample tested were 16.2 and 12.8 mg, respectively). 4. CONCLUSIONS EPD was used under vacuum as a novel, quick and cost-effective processing technique to explore the possibility of fabricating nickel coated carbon fibrereinforced alumina matrix composites, which could be densified by a pressureless sintering route. Boehmite sol seeded with nano-size δ-alumina powders was found to be converted completely to α-alumina matrix after firing for 2 h at 1250°C. Unidirectionally aligned Ni coated carbon fibre-reinforced alumina matrix composites were produced using an optimised single-stage EPD technique. The process parameters in terms of applied d.c. voltage and deposition time were found to be 15 V and 400 s, respectively, for full infiltration of the fibre mats used. Vacuum environment during EPD was proven to be effective in reducing the formation of gas bubbles within the deposited green body, resulting in a high degree of deposition and high final density of the component produced. Under vacuum, ultra fine boehmite particles within the sol could easily penetrate deep into the inter/intra-fibre tows regions, filling the voids and thus providing a dense composite. Crack path propa-
KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX DTA TGA Temperature,C Fig. 11. DTA and TGA traces(heating rate 10-C/min) for the boehmite sol, showing the phase transformation ation test indicates that the ductile Ni interface was 10 R. Butler, E. G. and Ponton able to deflect the cracks created within the matrix sses.1995.10.1215 proving the presence of a weak interface. Thus, the Ponton, C. B. Ceram Transactions. vol. 8 composite produced in this work should show dam- merican Ceramic Society, Westerville, OH, 1998, pp age-tolerant' behaviour under the operation of crack deflection and fibre pull-out mechanisms. Further 12. Kaya, C. Trusty, P. A and Ponton, C B, in Better Cer- characterisation of the mechanical performance of mics Through Processing, eds J Binner and J. Yeomans, these composites, also at high-temperatures, is the 13. Kaya, C, Trusty, P. A. and Boccaccini, A.R., J.Eur focus of current research 14. Zhang, Z, Huang, Y. and Liang, Z, J. Am. Ceram. Soc cknowledgements-Professors M H. Loretto and I R are acknowledged for the provision of laboratory fa 15. Hirat Nishimoto, A and Ishihara. Y, Nippon Sera- tsu roi 992),108. sity of Birmingham, respectively. The authors would like to 16. Harbach, F and Nienburg,H,J.Eur. Ceram. Soc,1998 8,675 thank Dr Less Renny (Inco Speciality Powder Product, UK) 17. Boccaccini, A.R. and Trusty, P. A, J. Mater. Sci, 1998, for supplying the fibres. 18. Sarkar, P and Nicholson, P. S, J. Am. Ceram. Soc., 1996. REFERENCES 9(8),1987 1. Kilbride. I. P and Barker. A. J, Br. Ceram. Trans., 1994. 19. Kaya, C, PhD thesis, The University of Birmingham, 93(5),187 birmingham, UK 1999. Kumagai, M. and Messing. G. L.,J. Am. Ceram. Soc., 20. Kaya, C, Boccaccini, A. R and Chawla, K. K, J.Am Ceram.Soc,2000,83(8),1885 3. Kirk, R.E. and Othmer, D. F.(eds ) Encyclopedia of 21. Iston, T.J. Ponton, C.B., Marquis, P. M. and Butler,E. Chemical Technology, voL. 2. Wiley, New York, 1978, G, in Proceeding of the Third Euro Ceramics, eds P. pp.249 Duran and J. F. Fernandez, Madrid, Spain, vol. 1. 1993 ai,M. and Messing. G. L.,J. Am. Ceram. Soc., p.419 22. Trusty, P. A and Boccaccini, A R, Appl. Compos Mater. 5. McArdle, J. L. and Messing. G. L, J. Am. Ceram. Soc. 1986,69(7),C98 23. Boccaccini, A. R, Ovenstone, J. and Trusty, P. A, Appl. 6. Nicholson, P. S, Sarkar, P. and Huang, X, J. Mater. Sci. Compos Mater., 1997, 4(3), 145 1993,28 4. Morgan, P. E. D. and Marshall, D. B, J. Am. Cera. Soc. 7. Nicholson, P S, Sarkar, P and Huang, X. in Proceedings 1995,78(6,1553. Science and Technology of Zirconia VS, ed. S. P. S. 25. Kaya, C, Gu, X, Aldawery, I and Butler, E. G, submitted adwal. M. J. Bannister. R. H. J. Hannink, Technomic for publication, July 1999 993,p.503 6. Illston, T.J., MPhil thesis, The University of Birmingham, 8. Nicholson, P.S., Sarkar, P and Huang. X. J. An. Ceram. Birmingham, UK, 1995 oc,1992,75(10),2907 7. Prewo, K. M.J. Mater. Sci. 1982. 17. 3549 9. Kaya, C, Trusty, P. A and Ponton, C.B., Br. Ceram. 28. Marshall, D. B. and Evans, A. G.,J. Am. Ceram. Soc., Trans.,1998,81(10),48. 1986.68.225
KAYA et al.: FABRICATION AND CHARACTERISATION OF ALUMINA CERAMIC MATRIX 1197 Fig. 11. DTA and TGA traces (heating rate 10°C/min) for the boehmite sol, showing the phase transformation temperatures and also the weight lost as a function of temperature. gation test indicates that the ductile Ni interface was able to deflect the cracks created within the matrix, proving the presence of a weak interface. Thus, the composite produced in this work should show ‘damage-tolerant’ behaviour under the operation of crack deflection and fibre pull-out mechanisms. Further characterisation of the mechanical performance of these composites, also at high-temperatures, is the focus of current research. Acknowledgements—Professors M.H. Loretto and I.R. Harris are acknowledged for the provision of laboratory facilities at the IRC and School of Metallurgy & Materials, at the University of Birmingham, respectively. The authors would like to thank Dr Less Renny (Inco Speciality Powder Product, UK) for supplying the fibres. REFERENCES 1. Kilbride, I. P. and Barker, A. J., Br. Ceram. Trans., 1994, 93(5), 187. 2. Kumagai, M. and Messing, G. L., J. Am. Ceram. Soc., 1984, 67(11), C230. 3. Kirk, R. E. and Othmer, D. F. (eds.), Encyclopedia of Chemical Technology, vol. 2. Wiley, New York, 1978, pp. 249. 4. Kumagai, M. and Messing, G. L., J. Am. Ceram. Soc., 1985, 68(9), 500. 5. McArdle, J. L. and Messing, G. L., J. Am. Ceram. Soc., 1986, 69(7), C98. 6. Nicholson, P. S., Sarkar, P. and Huang, X., J. Mater. Sci., 1993, 28. 7. Nicholson, P. S., Sarkar, P. and Huang, X. in Proceedings of Science and Technology of Zirconia VS, ed. S. P. S. Badwal, M. J. Bannister, R. H. J. Hannink, Technomic, 1993, p. 503. 8. Nicholson, P. S., Sarkar, P. and Huang, X., J. Am. Ceram. Soc., 1992, 75(10), 2907. 9. Kaya, C., Trusty, P. A. and Ponton, C. B., Br. Ceram. Trans., 1998, 81(10), 48. 10. Trusty, P. A., Boccaccini, A. R., Butler, E. G. and Ponton, C. B., Mater. Manuf. Processes, 1995, 10, 1215. 11. Kaya, C., Trusty, P. A. and Ponton, C. B., Ceramic Processing Science, Ceramic Transactions, vol. 83. The American Ceramic Society, Westerville, OH, 1998, pp. 399. 12. Kaya, C., Trusty, P. A. and Ponton, C. B., in Better Ceramics Through Processing, eds J. Binner and J. Yeomans, British Ceramic Proceedings, vol. 56. 1998, pp. 93. 13. Kaya, C., Trusty, P. A. and Boccaccini, A. R., J. Eur. Ceram. Soc., 1999, 19, 2859. 14. Zhang, Z., Huang, Y. and Liang, Z., J. Am. Ceram. Soc., 1994, 77(7), 1946. 15. Hirata, Y., Nishimoto, A. and Ishihara, Y., Nippon Seramikkusu Kyokai Gakujutsu Ronbunshi, 1991, 99(2), 108. 16. Harbach, F. and Nienburg, H., J. Eur. Ceram. Soc., 1998, 18, 675. 17. Boccaccini, A. R. and Trusty, P. A., J. Mater. Sci., 1998, 33, 933. 18. Sarkar, P. and Nicholson, P. S., J. Am. Ceram. Soc., 1996, 79(8), 1987. 19. Kaya, C., PhD thesis, The University of Birmingham, Birmingham, UK, 1999. 20. Kaya, C., Boccaccini, A. R. and Chawla, K. K., J. Am. Ceram. Soc., 2000, 83(8), 1885. 21. Illston, T. J., Ponton, C. B., Marquis, P. M. and Butler, E. G., in Proceeding of the Third Euro Ceramics, eds P. Duran. and J. F. Fernandez, Madrid, Spain, vol. 1. 1993, p. 419. 22. Trusty, P. A. and Boccaccini, A. R., Appl. Compos. Mater., 1998, 5(4), 207. 23. Boccaccini, A. R., Ovenstone, J. and Trusty, P. A., Appl. Compos. Mater., 1997, 4(3), 145. 24. Morgan, P. E. D. and Marshall, D. B., J. Am. Ceram. Soc., 1995, 78(6), 1553. 25. Kaya, C., Gu, X., Aldawery, I. and Butler, E. G., submitted for publication, July 1999. 26. Illston, T. J., MPhil thesis, The University of Birmingham, Birmingham, UK, 1995. 27. Prewo, K. M., J. Mater. Sci., 1982, 17, 3549. 28. Marshall, D. B. and Evans, A. G., J. Am. Ceram. Soc., 1986, 68, 225