《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_Electrophoretic Infiltration Deposition

J.Am. Ceran.So,901063-1070(2007) DOl:10.l11551-2916.2007.01504.x c 2007 The American Ceramic Society urna Constant Current Electrophoretic Infiltration Deposition of Fiber Reinforced Ceramic Composites Yahua Bao and Patrick S. nichols Ceramic Engineering Research Group, Department of Materials Science and Engineering. McMaster Univers Hamilton, Ontario Canada LSS 4L7 Modified electrophoretic infiltration deposition(EPID)under constant voltage(<10 V) was applied across the electrodes constant current conditions is used to fabricate non-conductive- minimize hydrogen evolution. A single fabric composite formed fiber-reinforced composites from an ethanol suspension, Parti- several minutes. Most EPD samples were very thin(one fiber les are infiltrated through a fiber preform by an electric field fabric, or a few fiber tows, thick). To obtain thick composites and are deposited on the front of the electrode and"backfill(multiple fiber fabrics or fiber tows), electrophoretically deposit through the fiber preform. a uniform morphology is achieved at ed fabrics were stacked and then pressure filtrated, i.e., EPD was the optimum deposition rate. The constant current EPID process is modeled as capillary d limits the application of EPd for complex shapes. To avoid infiltration electrophoresis. Particles "stream"the fiber preform the hydrolysis, a non-aqueous suspension is preferred for EPD due to the repulsive interaction between the fiber filaments and however, a thick composite may not be formed. the particles as both have the same-sign surface charge. Electro- The EPD cell can be modified for electrophoretic infiltration osmotic flow makes no contribution to deposit yield as the net of non-conductive fibers. The latter, as a preform, are attached How across a closed capillary cross section is zero. Hamaker's on the front of an electrode and charged particles in a non- law is extended to electrophoretic infiltration; however, the total aqueous suspension are electrophoretically infiltrated through deposit yield is controlled by particle electrophoresis outside the by an electric field. If the fibers and the particles have the same- capillaries due to the much lower electric field in the suspension. sign surface charge, they repel each other. Thus, the particles The deposit thickness increases linearly with time under opti first reach the electrode surface and deposit, and then "backfill mum current conditions. Too high a deposition rate promotes between the fiber filaments until the matrix extends to the outer entrapment in the depositing green bod fiber preform surface. -This process will be termed"electro- phoretic infiltration deposition(EPID) . Introduction Ohkawa and Elsner patented the fabrication of non-con- BER-REINFORCED ceramic matrix composites(CMCs)are ductive fiber-reinforced composites by constant-voltage EPID promising, light-weight structural materials that combine A high-constant voltage(320 V) was applied to an acetone sus- high-temperature strength, improved fracture toughness, dam- pension and EPid was conducted for 10-20 min. a thick de- problem as complete infiltration of the matrix into the fiber tows method. The green composite density was v. sa microstructure age tolerance, and thermal shock resistance but fabrication is a posit was obtained. Strecker et al. 4 published a crostructure, a reliable, simple, and cost-effective processing method must be developed that completely infiltrates the fibers respectively, the percentage theoretical density of the green com- th matrix precursors posite is approximately 30%, indicating that the green body Several techniques have been used to introduce matrices into prepared in this way is very porous, i. e, not well infiltrated infiltration including sol-gel and polymer precursor approach voltage is initially consumed in the suspension and hence the and electrophoretic deposition(EPD). Both CVI and solu- ultant electric field is very large and the resultant high initial tion infiltration have low penetration efficiency and hence it is deposition rate decreases the efficiency of infiltration, and high- faw concentration is inevitable. Recently, Stoll et al.- reported difficult to obtain a dense green body; it is also difficult by pres- successful fabrication of Nextel 720/A1,O3 composites with up sure filtration. Vibration-assisted infiltration has been reported to fabricate fiber-reinforced porous matrix composites Howev- to four fiber fabrics by EPId under constant voltage due to the same surface charge between the fibers and Al,O3 particles. But er, high-vibration rates can possibly damage the weak layer thick samples still needed to be pressure filtrated coating Since the 1990s, Epd has been explored to fabricate fiber In the present work, the electrophoretic deposition thro reinforced ceramic composites. Conventional wisdon conductive fibers is compared with electrophoretic infiltration suggest that conductive fibers would best serve as the de deposition through non-conductive fibers. It is demonstrated electrode, i. e, the matrix particles deposit directly on t that constant-current EPID can successfully fabricate fiber-re- Carbon fibers. carbon-coated SiC fiber ickel-coated car- inforced composites. The EPId process is modeled via a porous bon fibers, stainless-steel fibers, and polyoyrrole-coated polyethylene board. Under optimum constant current condi- fibers"have been used. Most experiments were conducted in an tons, the deposit yield increases linearly with deposition time. aqueous suspension under a constant voltage. In this case, a low niform fiber composites are achieved with non-conductive fi- ers. Conductive fibers suffer electric field shielding, which se- verely interferes with the deposition process. F. Zok--contributing editor II. Experimental Procedure cript No. 21706. Received April 15, 2006, approved October 30, 2006. TM-DAR alumina (Taimei Chemicals, Tokyo, particle size, 0. 1 um) was dispersed in ethanol with a 1063

Constant Current Electrophoretic Infiltration Deposition of Fiber￾Reinforced Ceramic Composites Yahua Bao and Patrick S. Nicholson**,w Ceramic Engineering Research Group, Department of Materials Science and Engineering, McMaster University, Hamilton, Ontario, Canada L8S 4L7 Modified electrophoretic infiltration deposition (EPID) under constant current conditions is used to fabricate non-conductive- fiber-reinforced composites from an ethanol suspension. Parti￾cles are infiltrated through a fiber preform by an electric field and are deposited on the front of the electrode and ‘‘backfill’’ through the fiber preform. A uniform morphology is achieved at the optimum deposition rate. The constant current EPID process is modeled as capillary infiltration electrophoresis. Particles ‘‘stream’’ the fiber preform due to the repulsive interaction between the fiber filaments and the particles as both have the same-sign surface charge. Electro￾osmotic flow makes no contribution to deposit yield as the net flow across a closed capillary cross section is zero. Hamaker’s law is extended to electrophoretic infiltration; however, the total deposit yield is controlled by particle electrophoresis outside the capillaries due to the much lower electric field in the suspension. The deposit thickness increases linearly with time under opti￾mum current conditions. Too high a deposition rate promotes air entrapment in the depositing green body. I. Introduction FIBER-REINFORCED ceramic matrix composites (CMCs) are promising, light-weight structural materials that combine high-temperature strength, improved fracture toughness, dam￾age tolerance, and thermal shock resistance, but fabrication is a problem as complete infiltration of the matrix into the fiber tows and fabrics is difficult. To design properties via controlled mi￾crostructures, a reliable, simple, and cost-effective processing method must be developed that completely infiltrates the fibers with matrix precursors. Several techniques have been used to introduce matrices into fiber preforms, i.e., chemical vapor infiltration (CVI),1 solution infiltration including sol–gel and polymer precursor approach￾es,1 pressure filtration (PF),2 vibration-assisted infiltration,3,4 and electrophoretic deposition (EPD).5–15 Both CVI and solu￾tion infiltration have low penetration efficiency and hence it is difficult to obtain a dense green body; it is also difficult by pres￾sure filtration. Vibration-assisted infiltration has been reported to fabricate fiber-reinforced porous matrix composites. Howev￾er, high-vibration rates can possibly damage the weak layer coating pre-introduced on the fiber surface. Since the 1990s, EPD has been explored to fabricate fiber￾reinforced ceramic composites.16 Conventional wisdom would suggest that conductive fibers would best serve as the deposition electrode, i.e., the matrix particles deposit directly on the fibers. Carbon fibers,5 carbon-coated SiC fibers,6–9 nickel-coated car￾bon fibers,10,11 stainless-steel fibers,12–15 and polyoyrrole-coated fibers17 have been used. Most experiments were conducted in an aqueous suspension under a constant voltage. In this case, a low, constant voltage (o10 V) was applied across the electrodes to minimize hydrogen evolution. A single fabric composite formed in several minutes. Most EPD samples were very thin (one fiber fabric, or a few fiber tows, thick). To obtain thick composites (multiple fiber fabrics or fiber tows), electrophoretically deposit￾ed fabrics were stacked and then pressure filtrated, i.e., EPD was combined with PF.18,19 PF, however, complicates the processing and limits the application of EPD for complex shapes. To avoid the hydrolysis, a non-aqueous suspension is preferred for EPD; however, a thick composite may not be formed. The EPD cell can be modified for electrophoretic infiltration of non-conductive fibers. The latter, as a preform, are attached on the front of an electrode and charged particles in a non￾aqueous suspension are electrophoretically infiltrated through by an electric field. If the fibers and the particles have the same￾sign surface charge, they repel each other. Thus, the particles first reach the electrode surface and deposit, and then ‘‘backfill’’ between the fiber filaments until the matrix extends to the outer fiber preform surface.19–25 This process will be termed ‘‘electro￾phoretic infiltration deposition (EPID).’’ Ohkawa and Elsner22 patented the fabrication of non-con￾ductive fiber-reinforced composites by constant-voltage EPID. A high-constant voltage (320 V) was applied to an acetone sus￾pension and EPID was conducted for 10–20 min. A thick de￾posit was obtained. Streckert et al. 24 published a microstructure of a thick Nicalon-HVR-fabric/SiC composite fabricated by this method. The green composite density was 0.8 g/cm3 . Consider￾ing that the fiber and SiC density are 2.35 g/cm3 and 3.20 g/cm3 , respectively, the percentage theoretical density of the green com￾posite is approximately 30%, indicating that the green body prepared in this way is very porous, i.e., not well infiltrated. Macro-voids exist in the composite due to inefficient particle infiltration. Virtually no particles infiltrate the fiber tows. Most voltage is initially consumed in the suspension and hence the resultant electric field is very large and the resultant high initial deposition rate decreases the efficiency of infiltration, and high- flaw concentration is inevitable. Recently, Stoll et al. 25 reported successful fabrication of Nextel 720/Al2O3 composites with up to four fiber fabrics by EPID under constant voltage due to the same surface charge between the fibers and Al2O3 particles. But thick samples still needed to be pressure filtrated. In the present work, the electrophoretic deposition through conductive fibers is compared with electrophoretic infiltration deposition through non-conductive fibers. It is demonstrated that constant-current EPID can successfully fabricate fiber-re￾inforced composites. The EPID process is modeled via a porous polyethylene board. Under optimum constant current condi￾tions, the deposit yield increases linearly with deposition time. Uniform fiber composites are achieved with non-conductive fi- bers. Conductive fibers suffer electric field shielding, which se￾verely interferes with the deposition process. II. Experimental Procedure TM-DAR alumina (Taimei Chemicals, Tokyo, Japan, particle size, 0.1 mm) was dispersed in ethanol with a polyethyleneimine F. Zok—contributing editor **Fellow, American Ceramic Society. w Author to whom correspondence should be addressed. e-mail: nicholsn@mcmaster.ca Manuscript No. 21706. Received April 15, 2006; approved October 30, 2006. Journal J. Am. Ceram. Soc., 90 [4] 1063–1070 (2007) DOI: 10.1111/j.1551-2916.2007.01504.x r 2007 The American Ceramic Society 1063

Journal of the American Ceramic Society--Bao and Nicholson Vol. 90. No. 4 Cathode Anode 0.8 E0.6 Beaker 0.4 Unidirectional fiber preform for back-filling 0.2 0.0 Insulator 0.1 Fig. 1. Schematic of the electrophoretic infiltration deposition cell for PEI concentration(wt%) fabrication of fiber-reinforced ceramic matrix composites. Fig. 2. Electrophoretic mobility of alumina with polyethyleneimine (PED addition. (PE dispersant(M.w. 10000, Polysciences, Warrington, PA) protonated with glacial acetic acid. The PEl concentration was(Fig. 2). The PEI also acts as a binder for ceramic processing. 26 determined by optimizing the electrophoretic mobility of the 0.5 wt% PEI was used to prepare the alumina suspensions alumina via a Pals Zeta potential analyzer(Brookhaven Instru for EPID ments, Holtsville, NY). Thirty volume percent alumina in eth anol with 0.5 wt% pei was ball milled for 24 h and then diluted to I vol% ension for EPD/EPID. A I vol% alumina sus- (2) Electrophoretic Deposition on Conductive Polypyrrole- pension at pH 4 adjusted with 0. 1M HCl was also used Coated Mullite/Alumina Fibers a submicron layer of conductive polypyrrole was coated onto When the fiber bundle was small. i.e. less than one tow well desized mullite/alumina(Nextel720)fibers(3M, St Paul, MN infiltrated fiber/alumina green opposites were formed. 17 via an in situ polymerization method. With polypyrrole-coated However, particles clogged the bundle surface when more fibers as the cathode and a circular stainless-steel screen as the than eight tows were involved. Figure 3 shows a cross section anode(diameter 2 cm), EPD was conducted at 10-30 V(con- of such a composite synthesized by EPd using a 0.1 mA stant voltage)or 0.05-0.2 mA(constant current) for several mi- constant current. Most alumina particles deposit around the nutes. The deposited Nextel 720 fibers were dried for 12 h in outer layer of the fiber bundle and no particles locate in the saturated ethanol atmosphere and then in air. After drying, th enter of the bundle composite was heated at 900C for 2 h and then vacuum infil trated with epoxy resin to polish for SEM characterization. Figure I shows a schematic of the EPID cell. Desized Nextel (3) Constant-Current Electrophoretic infiltration deposition 20 fibers were aligned in one direction and mounted in a plastic into Non-Conductive Mullite/Alumina Fibers Figure 4 shows a cross section of an epoxy-infiltrated fiber/alu- 25 mmx5 mmx3 mm. The opposite surface of the cathode mina, green composite fabricated by EPID(0.07 mA/cm- for 10 was covered with a plastic plate. The outer surface of the fiber- h)using a I vol% alumina suspension dispersed with 0.5 wt% bundle preform contacted the alumina suspension and alumina PEL. Submicron alumina particles particles were infiltrated there under a constant current (0.01-0.3 A/ cm). The inter-electrode distance was 2 cm. The process was continued until the fiber preform was fully backfilled with the alumina powder Central cavity a pre-weighed porous polyethylene(PE) board(Porex, Fair- with burn GA, 25 mm x 25 mm x 6 mm)with a pore size of 15-45 deposition um and 44% porosity was used as a model porous preform. This replaced the fiber preform in the EPId cell(Fig. 1). Voltage values during deposition were collected by a computer with an A/D acquisition board. A large volume of alumina suspension was used to maintain a constant concentration. The suspension conductivity was measured with a conductivity meter(Model 4100, Man-Tech Associates, Toronto, on) before and after EPID. After deposition, the surface deposit was peeled away and dried in air. The pe board was reweighed after drying. The weight gain is that of deposited alumina within. The green com- posite was vacuum infiltrated with epoxy resin to polish for SEM characterization Il. Results (1) Electrophoretic Mobility of Alumina with a PEl Dispersant Fig 3. Microstructure of conductive polypyrrole-coated Nextel 720

(PEI) dispersant (M.W. 10 000, Polysciences, Warrington, PA) protonated with glacial acetic acid. The PEI concentration was determined by optimizing the electrophoretic mobility of the alumina via a Pals Zeta potential analyzer (Brookhaven Instru￾ments, Holtsville, NY). Thirty volume percent alumina in eth￾anol with 0.5 wt% PEI was ball milled for 24 h and then diluted to 1 vol% suspension for EPD/EPID. A 1 vol% alumina sus￾pension at pH 4 adjusted with 0.1M HCl was also used. A submicron layer of conductive polypyrrole was coated onto desized mullite/alumina (Nextelt 720) fibers (3M, St. Paul, MN) via an in situ polymerization method.17 With polypyrrole-coated fibers as the cathode and a circular stainless-steel screen as the anode (diameter 2 cm), EPD was conducted at 10–30 V (con￾stant voltage) or 0.05–0.2 mA (constant current) for several mi￾nutes. The deposited Nextel 720 fibers were dried for 12 h in a saturated ethanol atmosphere and then in air. After drying, the composite was heated at 9001C for 2 h and then vacuum infil￾trated with epoxy resin to polish for SEM characterization. Figure 1 shows a schematic of the EPID cell. Desized Nextel 720 fibers were aligned in one direction and mounted in a plastic mold attached to a metal plate (cathode). The fiber preform was 25 mm
5 mm
3 mm. The opposite surface of the cathode was covered with a plastic plate. The outer surface of the fiber￾bundle preform contacted the alumina suspension and alumina particles were infiltrated there under a constant current (0.01–0.3 mA/cm2 ). The inter-electrode distance was 2 cm. The process was continued until the fiber preform was fully backfilled with the alumina powder. A pre-weighed porous polyethylene (PE) board (Porex, Fair￾burn GA, 25 mm
25 mm
6 mm) with a pore size of 15–45 mm and 44% porosity was used as a model porous preform. This replaced the fiber preform in the EPID cell (Fig. 1). Voltage values during deposition were collected by a computer with an A/D acquisition board. A large volume of alumina suspension was used to maintain a constant concentration. The suspension conductivity was measured with a conductivity meter (Model 4100, Man-Tech Associates, Toronto, ON) before and after EPID. After deposition, the surface deposit was peeled away and dried in air. The PE board was reweighed after drying. The weight gain is that of deposited alumina within. The green com￾posite was vacuum infiltrated with epoxy resin to polish for SEM characterization. III. Results (1) Electrophoretic Mobility of Alumina with a PEI Dispersant Alumina particles are positively charged in ethanol. The elec￾trophoretic mobility increased on PEI dispersant addition (Fig. 2). The PEI also acts as a binder for ceramic processing.26 0.5 wt% PEI was used to prepare the alumina suspensions for EPID. (2) Electrophoretic Deposition on Conductive Polypyrrole￾Coated Mullite/Alumina Fibers When the fiber bundle was small, i.e., less than one tow, well￾infiltrated fiber/alumina green mini-composites were formed.17 However, particles clogged the bundle surface when more than eight tows were involved. Figure 3 shows a cross section of such a composite synthesized by EPD using a 0.1 mA constant current. Most alumina particles deposit around the outer layer of the fiber bundle and no particles locate in the center of the bundle. (3) Constant-Current Electrophoretic Infiltration Deposition into Non-Conductive Mullite/Alumina Fibers Figure 4 shows a cross section of an epoxy-infiltrated fiber/alu￾mina, green composite fabricated by EPID (0.07 mA/cm2 for 10 h) using a 1 vol% alumina suspension dispersed with 0.5 wt% PEI. Submicron alumina particles are uniformly infiltrated + Beaker Anode Cathode Unidirectional fiber preform for back-filling Stirrer Suspension Polymer Insulator _ Fig. 1. Schematic of the electrophoretic infiltration deposition cell for fabrication of fiber-reinforced ceramic matrix composites. 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 0.1 0.2 0.3 0.4 0.5 0.6 PEI concentration (wt%) Mobility (µm/s)/(V/cm) Fig. 2. Electrophoretic mobility of alumina with polyethyleneimine (PEI) addition. 1mm Central cavity with no deposition Fig. 3. Microstructure of conductive polypyrrole-coated Nextel 720 fiber/Al2O3 composite by electrophoretic deposition with an 8-tow fiber bundle. 1064 Journal of the American Ceramic Society—Bao and Nicholson Vol. 90, No. 4

April 2007 Constant Current EPID of Fiber-Reinforced Ceramic Composites 1065 SoY Fig 4. Cross section of a fiber/alumina green composite fabricated by electrophoretic infiltration deposition through the inter-fiber voids to form a well-infiltrated board thickness. For current > 2.0 mA, the porous board with few macro-pores r uniform, fiber/alumina covered with deposit in 60 min composites were produc a I vol% alumina susp Figure 9 tracks the voltage change during epld. at 0.3 mA at pH 4. For a large uced using a >0.3 he voltage change is almost linear with time. When the curren deposits in the preform >0.6 mA, the voltage increases to a maximum(1000 V). Figure 10 shows the deposited alumina volume over the total pore vol- ume of the porous board. The maximum deposition occurs at a (4) EPID of Al,O3 into a Model Porous Media(PE)Under urrent of 0.3 mA. After EPID for 7h, alumina powder occupies Constant Current condition 53%of the PE-board pore volume, i.e., the infiltrated alumina Figure 5 shows a pe board, infiltrated with alumina for 3 h u in the pe board pores has 53% green density, the density of ing 0.3 mA current. The interface of the board /deposited alu- a monolithic alumina green body produced by EPD mina is clear(the right side of the pe board is attached to the The deposited alumina volume percentage decreases with cathode). Alumina particles first infiltrate the inter-connected creasing current. When the current >0.6 mA, the deposited pores to deposit on the front of the cathode. They then build alumina only occupies 45% of the pore volume. Too low a cur- backwards through the pores toward the board/ suspension in- rent(0. 1 mA)requires more than 40 h to infiltrate fully. This terface, i.e, the front pores do not clog as per conductive fibers. decreases the deposit density as concurrent sedimentation occurs Figure 6 shows an alumina deposit in a 10-um pore. It is well in the porous board. filtrated with alumina Figure 7 shows a plot of deposited alumina versus depo time t(current 0.3 mA). For t270 min, the outer surface is (1) Analysis of the electric Field Inside a Conductive Fiber overed with deposit until fully covered Bundle Figure 8 shows the effect of current value on the pyrrole-coated eight after 60 min. The deposited-alumina weight the fiber bundle is less than one tow fiber arly below 2.0 mA if the deposit thickness is less eight tows is used. central voids are left Suspension/PE Interface board interface Electrode Position (b) Fig. 5. Porous polyethylene(PE) board infiltrated with alumina particles

through the inter-fiber voids to form a well-infiltrated matrix with few macro-pores. Similar uniform, fiber/alumina green composites were produced using a 1 vol% alumina suspension at pH 4. For a large current ( 0.3 mA/cm2 ), less alumina deposits in the preform. (4) EPID of Al2O3 into a Model Porous Media (PE) Under Constant Current Conditions Figure 5 shows a PE board, infiltrated with alumina for 3 h us￾ing 0.3 mA current. The interface of the board/deposited alu￾mina is clear (the right side of the PE board is attached to the cathode). Alumina particles first infiltrate the inter-connected pores to deposit on the front of the cathode. They then build backwards through the pores toward the board/suspension in￾terface, i.e., the front pores do not clog as per conductive fibers. Figure 6 shows an alumina deposit in a 10-mm pore. It is well infiltrated with alumina. Figure 7 shows a plot of deposited alumina versus deposition time t (current 0.3 mA). For to270 min, the weight gain is linear with time, i.e.,10 mg/min. When t4270 min, the outer surface is covered with deposit until fully covered. Figure 8 shows the effect of current value on the deposit weight after 60 min. The deposited-alumina weight increases linearly below 2.0 mA if the deposit thickness is less than the board thickness. For current 2.0 mA, the porous board is covered with deposit in 60 min. Figure 9 tracks the voltage change during EPID. At 0.3 mA, the voltage change is almost linear with time. When the current 0.6 mA, the voltage increases to a maximum (1000 V). Figure 10 shows the deposited alumina volume over the total pore vol￾ume of the porous board. The maximum deposition occurs at a current of 0.3 mA. After EPID for 7 h, alumina powder occupies 53% of the PE-board pore volume, i.e., the infiltrated alumina in the PE board pores has 53% green density, the density of a monolithic alumina green body produced by EPD. The deposited alumina volume percentage decreases with in￾creasing current. When the current 0.6 mA, the deposited alumina only occupies 45% of the pore volume. Too low a cur￾rent (0.1 mA) requires more than 40 h to infiltrate fully. This decreases the deposit density as concurrent sedimentation occurs in the porous board. IV. Discussion (1) Analysis of the Electric Field Inside a Conductive Fiber Bundle Al2O3 particles deposit well on conductive, polypyrrole-coated Nextel 720 fibers when the fiber bundle is less than one tow fiber. When a fiber bundle of eight tows is used, central voids are left Fig. 4. Cross section of a fiber/alumina green composite fabricated by electrophoretic infiltration deposition. Electrode Position Suspension/PE board interface (a) (b) Interface 2mm 200µm Fig. 5. Porous polyethylene (PE) board infiltrated with alumina particles. April 2007 Constant Current EPID of Fiber-Reinforced Ceramic Composites 1065

1066 Journal of the American Ceramic Society--Bao and Nicholson Vol. 90. No. 4 Al2O3 2.5 E品 2 0 0 EPID Current(mA) Fig 8. Effect of the electrophoretic infiltration deposition(EPID) Fig. 6. Infiltration inside a I rent on the weight of deposited alumina after 60 m in the deposit. These voids cannot be eliminated by adjusting the (1 o=0 when y=0(bottom of the slit); EPD parameters, e.g., constant current or constant voltage (2)o =0 when y= a(top of the slit); modes. During EPD with conductive fiber bundles, pa (3)φ=φ。 when x=0( origin of the slit) first deposit on the outer fiber surfaces, thus clogging the bi (4)φ→0as This clogging effect is further enhanced by the electric field The solution of the two-dimensional Laplace equation gives shielding effect inside the conductive-fiber bundle as follows the potential as According to electrostatics, the electric potential distributio follows the Poisson equation (1) Thus, the electric field in the x direction at y= a/2 where o is the electric potential, E is the permittivity, and pe is the volume charge density. When pe=0. this reduces to the Laplace equation, i.e E=-2(9=S The electric field Ex declines drastically inside the slit(Fig. 12) In two-dimensional Cartesian coordinates this is The field penetration depth is only double the slit size(2a) eter of 10 um)forming a square array with a filament distance +ee 30 um, the fiber volume concentration is 8.5 vol %. As all fibers are grounded(they are the cathode), the outer square array (2 In two dimensions, the boundary conditions for an infinitely um away from the outside fibers) has a potential of 10 V(an long conductive slit with pore size, "a', are(Fig. 11) ode). Figure 13 shows a plot of the potential in this conductive fiber array via FEMLab 3. 0(COMSOL, Burlington, MA). Fig- 3.5 ire 14 shows the electric field distribution along the central line in the x direction. No electric field exists inside the fiber array 1200 0.90mA 2.5 1000 060mA 1.5 400 A 0.2mA 0.5 200 0.1m 600 250 750 1000 Fig. 7. Deposited alumina weight versus electrophoretic infiltration 9. Voltage change versus electrophoretic infiltration deposition e for different cell current values

in the deposit. These voids cannot be eliminated by adjusting the EPD parameters, e.g., constant current or constant voltage modes. During EPD with conductive fiber bundles, particles first deposit on the outer fiber surfaces, thus clogging the bundle. This clogging effect is further enhanced by the electric field shielding effect inside the conductive-fiber bundle as follows: According to electrostatics, the electric potential distribution follows the Poisson equation: H2 f ¼ 1 e rE (1) where f is the electric potential, e is the permittivity, and rE is the volume charge density. When rE 5 0, this reduces to the Laplace equation, i.e.: H2 f ¼ 0 (2) In two-dimensional Cartesian coordinates, this is q2 f qx2 þ q2 f qy2 ¼ 0 (3) In two dimensions, the boundary conditions for an infinitely long conductive slit with pore size, ‘‘a’’, are (Fig. 11): (1) f 5 0 when y 5 0 (bottom of the slit); (2) f 5 0 when y 5 a (top of the slit); (3) f 5 fo when x 5 0 (origin of the slit); (4) f-0 as x-N. The solution of the two-dimensional Laplace equation gives the potential as27 fðx; yÞ ¼ 4fo p X1 m¼1;3;5... 1 m sin mpy a empx=a (4) Thus, the electric field in the x direction at y 5 a/2 is Ex ¼ qfðx; a 2Þ qx ¼ 4fo a X1 m¼1;3;5... sin mp 2 empx=a (5) The electric field Ex declines drastically inside the slit (Fig. 12). The field penetration depth is only double the slit size (2a). For a conductive fiber bundle (assuming fibers with a diam￾eter of 10 mm) forming a square array with a filament distance of 30 mm, the fiber volume concentration is 8.5 vol%. As all fibers are grounded (they are the cathode), the outer square array (20 mm away from the outside fibers) has a potential of 10 V (an￾ode). Figure 13 shows a plot of the potential in this conductive fiber array via FEMLab 3.0 (COMSOL, Burlington, MA). Fig￾ure 14 shows the electric field distribution along the central line in the x direction. No electric field exists inside the fiber array Al2O3 20µm PE Fig. 6. Infiltration inside a 10-mm pore. 0 0.5 1 1.5 2 2.5 3 3.5 0 200 400 600 800 Time (min.) Infiltrated Alumina Weight (g) Fig. 7. Deposited alumina weight versus electrophoretic infiltration deposition time. 0 0.5 1 1.5 2 2.5 3 0 0.5 1.5 EPID Current (mA) Deposited alumina (g) 1 Fig. 8. Effect of the electrophoretic infiltration deposition (EPID) cur￾rent on the weight of deposited alumina after 60 min. 0 200 400 600 800 1000 1200 0 250 500 750 1000 Time (min.) Voltage (V) 0.3mA 0.60mA 0.90mA 0.2mA 0.40mA 0.1mA Fig. 9. Voltage change versus electrophoretic infiltration deposition time for different cell current values. 1066 Journal of the American Ceramic Society—Bao and Nicholson Vol. 90, No. 4

April 2007 Constant Current EPID of Fiber-Reinforced Ceramic Composites 1067 1.0 0.8 ￥王 u0.4 40 000.51.0152.0253.03.54.04.55.0 x/a Current(mA) Fig 12. Electric field Ex along the center of the two-dimensional slit. 10. Infiltrated alumina volume percentage versus deposition cur H modifies the particle surface charge and that of the fibers rendering the latter and the former to have the same sign surface even though the fiber concentration is 8.5 vol%. In fact, the field charge, i. e, both are positively charged. Thus, electrosteric(or penetration depth is only double the fiber spacing. The EPD driving force is the electric field, i.e., no field electrostatic)repulsion develops between the fibers and the par- iving force. During deposition, the particles initially deposi between the fibers on the surface of the outer conductive fibers. If the deposit re- or an ionically-stabilized susp sistance is the suspension resistance, the particles deposit on between charged surfaces consists of van der Waals attraction the outer fibers and their surface becomes insulating, i.e., they and electrical double-layer repulsion. As the filament diameter exert no"shielding " contribution. Thus, the electric field car (12 um)is far larger than the particle size(0. I um), the fiber fil- now penetrate further and particles deposit on the inside fibers ament is assumed to be an infinitely large, fat surface. Assuming However, if the fiber bundle is large, the outer fiber layer be that the fiber is pure alumina, the van der Waals interaction, VA comes clogged before the inside voids are filled and a central between an alumina filament and cavity is inevitable. This explains why no particles deposit at the fiber bundle center when the conductive-fiber bundle is large Galor et al.2 reported deep electrophoretic infiltration of olloidal silica particles into a porous conductive carbon sul rate(48 vol% porosity with an average pore size of 60 um) here rp is the particle radius(0.05 um), A is the Hamaker con They, however, reported a maximum weight gain of 3%. A tant for alumina across ethanol (3.37 x 10-J), and d is the suming that the theoretical density of the carbon substrate is paration distance between the fiber and particle. The electrical double-layer interaction potential between a pore volume. They argue that full infiltration can be achieved if spherical alumina particle in an ionically stabilized suspension ive full infiltration due to the electric field shielding effect. Prac- tically, it is concluded that conductive- fibers cannot aid the VR=2nErp-psr formation of dense fiber-reinforced ceramics by EPD or EPID (2) Repulsive Interaction Between the Fiber Filaments and the particles Electrophoretic infiltration deposition involves p:(1) electrophoresis of particles in the suspension, (2 ation the particles through the fiber preform, and (3) tion on the electrode behind the fiber preform s) particke. infiltrate, they must pass through the fiber preform to deposit on the back electrode and as the process continues, they build into ess. To maintain this backfilling process, no particle should de- osit in the fiber perform on the way through. The oxide fibers will also adsorb the PEl (or H)when dipped into the alumina uspension; thus, electrosteric (or electrostatic) interaction oc- curs between the fibers and the particles via the co-adsorbed P (or H)and influences the efficiency of the backfilling. PEl or (0.0) Fig. 13. Potential distribution in a conductive fiber array( white circles Fig. 11. Boundary conditions of a two-dimensional slit

even though the fiber concentration is 8.5 vol%. In fact, the field penetration depth is only double the fiber spacing. The EPD driving force is the electric field, i.e., no field, no driving force. During deposition, the particles initially deposit on the surface of the outer conductive fibers. If the deposit re￾sistance is  the suspension resistance, the particles deposit on the outer fibers and their surface becomes insulating, i.e., they exert no ‘‘shielding’’ contribution. Thus, the electric field can now penetrate further and particles deposit on the inside fibers. However, if the fiber bundle is large, the outer fiber layer be￾comes clogged before the inside voids are filled and a central cavity is inevitable. This explains why no particles deposit at the fiber bundle center when the conductive-fiber bundle is large (Fig. 3). Galor et al. 28 reported deep electrophoretic infiltration of colloidal silica particles into a porous conductive carbon sub￾strate (48 vol% porosity with an average pore size of 60 mm). They, however, reported a maximum weight gain of 3%. As￾suming that the theoretical density of the carbon substrate is 2.25 g/cm3 , their deposited SiO2 only occupies 5.5 vol% of the pore volume. They argue that full infiltration can be achieved if surface deposition is avoided. However, it is impossible to ache￾ive full infiltration due to the electric field shielding effect. Prac￾tically, it is concluded that conductive-fibers cannot aid the formation of dense fiber-reinforced ceramics by EPD or EPID. (2) Repulsive Interaction Between the Fiber Filaments and the Particles Electrophoretic infiltration deposition involves three steps: (1) electrophoresis of particles in the suspension, (2) infiltration of the particles through the fiber preform, and (3) particle deposi￾tion on the electrode behind the fiber preform. When particles infiltrate, they must pass through the fiber preform to deposit on the back electrode and as the process continues, they build into the fiber preform from back to front, i.e., a ‘‘backfilling’’ proc￾ess. To maintain this backfilling process, no particle should de￾posit in the fiber perform on the way through. The oxide fibers will also adsorb the PEI (or H1) when dipped into the alumina suspension; thus, electrosteric (or electrostatic) interaction oc￾curs between the fibers and the particles via the co-adsorbed PEI (or H1) and influences the efficiency of the backfilling. PEI or H1 modifies the particle surface charge and that of the fibers, rendering the latter and the former to have the same sign surface charge, i.e., both are positively charged. Thus, electrosteric (or electrostatic) repulsion develops between the fibers and the par￾ticles via their double-layer overlap. The particles thus ‘‘stream’’ between the fibers. For an ionically-stabilized suspension, the interaction energy between charged surfaces consists of van der Waals attraction and electrical double-layer repulsion. As the filament diameter (12 mm) is far larger than the particle size (0.1 mm), the fiber fil￾ament is assumed to be an infinitely large, flat surface. Assuming that the fiber is pure alumina, the van der Waals interaction, VA, between an alumina filament and an alumina particle is29 VA ¼ Arp 6D 1 þ D 2rp þ D þ D rp ln D 2rp þ D 
(6) where rp is the particle radius (0.05 mm), A is the Hamaker con￾stant for alumina across ethanol (3.37
1020 J 30), and D is the separation distance between the fiber and particle. The electrical double-layer interaction potential between a spherical alumina particle in an ionically stabilized suspension and an alumina filament (assumed flat) is31,32 VR ¼ 2perpzpzf
ln 1 þ expðkDÞ 1 expðkDÞ  ðz 2 p þ z 2 fÞ 2zpzf ln 1½  expð2kDÞ ( ) (7) 40 45 50 55 0 0.2 0.4 0.6 0.8 1 Current (mA) Vol% pores Fig. 10. Infiltrated alumina volume percentage versus deposition cur￾rent after complete infiltration. x y φ= φo a φ= 0 (0,0) φ= 0 (0,a) (∞, 0) y=a/2 x=a (∞, a) Fig. 11. Boundary conditions of a two-dimensional slit. 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 x/a Ex=4φo/a Fig. 12. Electric field Ex along the center of the two-dimensional slit. Fig. 13. Potential distribution in a conductive fiber array (white circles are fibers). April 2007 Constant Current EPID of Fiber-Reinforced Ceramic Composites 1067

1068 Journal of the American Ceramic Society--Bao and Nicholson Vol. 90. No. 4 on the other, and then current is passed, particles will infiltrate the medium and deposit on the back electrode. For flat sub- strates, Hamaker's lawdescribes the deposit yield as Y mobility, E is the electric field strength, S the electrode surface rea,cs the colloidal particle concentration in the suspension and t the tim For electrophoretic infiltration deposition of a flat board with parallel capillaries directed along the electric field to an electrode of opposite sign, this equation can be modified NVncs dsdt I/N(VEP+),ds dr ⅹ(m) x104 where Ip is the particle velocity, VEP the particle electrophoretic elocity, Vi the liquid velocity, S the capillary cross-section Fig 14. Electric field(E) along the center line of the fiber bundle area, and n the number of capillaries. When the suspension volume is large, cs is constant, i.e.: Y= Nc,// VEP dS+Ncs/VidS the debye length Following the procedure developed by Wang k can be estimated from the suspension conductivity. It is found that K Now, Is VL ds=0 due to net zero liquid flow across the ension with a conductivity of 2.5 cross section in a closed capillary. Thus, assuming that the field S/cm. Assuming that the fiber has the same zeta potential as alumina particle, then otal inter- action potential energy (VT=VA+VR between the fiber fila- Y =(NCSHEPEXS)r (11) ment and particle is estimated as per Fig. 15 It is clear that the electrical double-layer repulsive interaction where k= NCSHEPExS between the particles and fiber is significant and opposes particle F deposition as they pass. The maximum energy barrier is a 60kT at 10 nm and the equilibrium ation distance >100 nm k=CsHEPEx PA (12 Thus, particles can be considered" lubricated, as they diffuse the fiber preform and deposit on the back electrode where P is the porosity and d is the surface area According to Eq (11), the deposit yield depends on the elec- tric field in the suspension. The electric field, E, between two (3) Modeling EPID of Matrix Particles into a Non- parallel plates is Conductive, Porous Substrate A non-conductive porous medium(e.g. a fiber preform) can be modeled as a collection of parallel cylindrical micro-capillaries in the electric field direction. If an electrode is placed on one side of the porous medium and a suspension with the other electrode where /is the current, R is the suspension resistance, inter-electrode distance. Now.r is 70 S is the electrode a,os is the suspension con- ivity, andf is the t depending on the suspen sion volume electrode ition and the cell design. If allel electrode plates, then f= 1. If the suspension fills all around the electrodes, /<l. sc I fA where A is the current density During EPID, the suspension conductivity is assumed to re- main unchanged (<+3%); thus, os is constant at 2.5 HS/cm, the Separation Distance(nm) value measured before epid. the electric field is related exclu sively to the current density and cell coefficient, f. Fig 15. Total interaction energy as a function of the separation dis- Dividing the epid cell into two sub-cells. i.e. inside tance between a fiber filament and a particle capillaries and outside, the sub-cell coefficient is I inside

where e is the dielectric permittivity of ethanol, xp and xf are the zeta potential of the particle and fiber, respectively, and k1 is the Debye length. Following the procedure developed by Wang30 k1 can be estimated from the suspension conductivity. It is found that k1 is 61 nm for an alumina suspension with a conductivity of 2.5 mS/cm. Assuming that the fiber has the same zeta potential as an alumina particle, then zp 5 zf 5 80 mV. Thus, the total inter￾action potential energy (VT 5 VA1VR) between the fiber fila￾ment and particle is estimated as per Fig. 15. It is clear that the electrical double-layer repulsive interaction between the particles and fiber is significant and opposes particle deposition as they pass. The maximum energy barrier is 60 kT at 10 nm and the equilibrium separation distance 4100 nm. Thus, particles can be considered ‘‘lubricated,’’ as they diffuse the fiber preform and deposit on the back electrode. (3) Modeling EPID of Matrix Particles into a Non￾Conductive, Porous Substrate A non-conductive porous medium (e.g. a fiber preform) can be modeled as a collection of parallel cylindrical micro-capillaries in the electric field direction. If an electrode is placed on one side of the porous medium and a suspension with the other electrode on the other, and then current is passed, particles will infiltrate the medium and deposit on the back electrode. For flat sub￾strates, Hamaker’s law33 describes the deposit yield as; Y ¼ Z t 0 mEPEScs dt (8) where Y is the deposit mass, mEP the particle electrophoretic mobility, E is the electric field strength, S the electrode surface area, cs the colloidal particle concentration in the suspension, and t the time. For electrophoretic infiltration deposition of a flat board with parallel capillaries directed along the electric field to an electrode of opposite sign, this equation can be modified: Y ¼ Z t 0 Z S NVpcs dS dt ¼ Z t 0 Z S NðVEP þ VLÞcs dS dt (9) where Vp is the particle velocity, VEP the particle electrophoretic velocity, VL the liquid velocity, S the capillary cross-section area, and N the number of capillaries. When the suspension volume is large, cs is constant, i.e.: Y ¼ Ncst Z S VEP dS þ Ncs Z S VL dS (10) Now, R S VL dS ¼ 0 due to net zero liquid flow across the cross section in a closed capillary. Thus, assuming that the field, Ex, is uniform inside the capillary, Eq. (10) reduces to Y ¼ ðNcsmEPExSÞt ¼ kt (11) where k ¼ NcsmEPEX S: For a porous medium, k can be written as k ¼ csmEPEX PA (12) where P is the porosity and A is the surface area. According to Eq. (11), the deposit yield depends on the elec￾tric field in the suspension. The electric field, E, between two parallel plates is E ¼ IR L (13) where I is the current, R is the suspension resistance, and L is the inter-electrode distance. Now, R is R ¼ f L ssS (14) where S is the electrode surface area, ss is the suspension con￾ductivity, and f is the cell coefficient depending on the suspen￾sion volume, electrode, size and position, and the cell design. If the suspension just fills the rectangular space between two par￾allel electrode plates, then f 5 1. If the suspension fills all around the electrodes, fo1. So, E ¼ f I ssS ¼ f L ss (15) where L is the current density. During EPID, the suspension conductivity is assumed to re￾main unchanged (o73%); thus, ss is constant at 2.5 mS/cm, the value measured before EPID. The electric field is related exclu￾sively to the current density and cell coefficient, f. Dividing the EPID cell into two sub-cells, i.e., inside the capillaries and outside, the sub-cell coefficient is 1 inside the Fig. 14. Electric field (Ex) along the center line of the fiber bundle shown in Fig. 13. −40 −30 −20 −10 0 10 20 30 40 50 60 70 0 50 100 150 200 Separation Distance (nm) VT/kT Fig. 15. Total interaction energy as a function of the separation dis￾tance between a fiber filament and a particle. 1068 Journal of the American Ceramic Society—Bao and Nicholson Vol. 90, No. 4

April 2007 Constant Current EPID of Fiber-Reinforced Ceramic Composites Table I. Suspension and Electrophoretic Infiltration Deposi- For uniform Epid under constant current conditions. the tion Parameters in the Capillaries and in the Suspension deposit thickness is proportional to the time of EPlD, i.e 中a=C+kI(Rdep-Rsus-e甲p)t (18) capillaries suspension Suspension concentration cs(g/mL) 0.040.04 where k is a constant. Raep depends on the type of stabilizing Suspension conductivity os (uS/cm) agent in the suspension 0. If the deposit is uniform and the specific resistance remains Electrophoretic mobility HEp um.s (V/cm) constant, the applied voltage will increase linearly with time. At 0.3 mA, the observed voltage change is almost linear with time Sub-cell coefficient indicating that the deposit is uniform and its thickness increases Current density A(mA/cm linearly with time. This corresponds to the linear weight gain Electric field E(V/cm) with time(Fig. 7). The observed initial voltage drop may be re- Deposition/electrophoretic rate(mg/min) 23 lated to the preliminary suspension infiltration of the pores. For This increase is related to highly resistant entrapped air in the capillaries, but 0.6 mA, the voltage; hence the outside sub-cell coefficient can be determined deposited alumina only occupies 45% of the pore volume, ind between the same electrodes without intervening porous cating that air bubbles are trapped in the deposit. Clearly, the board in the cell, i.e., f=0.52. The electric field across the sus- deposition rate must be controlled to optimize the green density pension is 10 V/cm, and hence the EPID suspension paramete This is why constant current must be used for the EPID process. are as summarized in Table i According Eq.(ID). the deposition rate in V. Conclusions Conductive fibers are not suitable for fabrication of fiber-re- particle electrophoretic migration rate in an open suspension is inforced CMCs by electrophoretic deposition due to the electric 12 mg/min. The latter value is close to the measured deposition field shielding effect. The field penetration depth is only double he fiber filament opening distance and a central cavity inevit particle electrophoretic migration rate in the open suspension ably results when a large bundle of conductive fibers is used. (outside of the capillaries), i.e., all particles electrophoresing te A modified EPd cell can be used to fabricate non-conductive- the board outer surface enter the capillaries and infiltrate faster fiber-reinforced CMCs by EPlD. Particles are infiltrated into a there than in the suspension. This finding is not surprising as fiber preform to deposit on an electrode behind the preform and particles are""streamed"by the repulsion from the pore surfaces then backfill the preform. Thus, the front clogging encountered once they enter the porous board. This phenomenon also ex- in direct EPD is avoided. The repulsive interaction between the plains why no deleterious surface deposition(clogging) occurs fibers and the particles promotes particle streaming through until the porosity is fully filled. If the sub-cell coefficient outside the fiber preform, resulting in dense, uniform green composit ual the infiltration rate by EPID Constant-current EPID is modeled as capillary infiltration In an EPD cell with the cathode as the deposition electrode electrophoresis. Owing to the zero flow across a closed capillary the applied potentia s consumed in three steps:(a)a cross section, no effect of electro-osmotic flow is expected. The tential drop at each electrode(cathode and anode), (b)an ohmic infiltration deposition yield is proportional to the electric field tential drop over the suspension, and (c)a potential drop over inside the capillary; however, the total EPID yield is found to be the deposit on the cathode. In an EPID cell, on the other hand controlled by particle electrophoresis outside the capillaries, i. e, the potential drop over the suspension can be divided into two (I the potential drop across the open suspension outside the in the open suspension, due to the much lower electric field capillaries(between the anode and board outer surface), and (2) entrapment in the deposit decrease green density due to air the suspension in the capillaries, i.e aa= Voel+ IRdepd+IRsus-cap(d-d)+IRsusd2 References +Voe (16) K.K. Chawla, Ceramic Matrix Composites. Chapman, New York, 1993 where Voe and voe are the potential drop at the anode and cathode, respectively, I is the current flowing through the cell, Oxide Matrix Composite Reinforced with Continuous Oxide Fibers "/Am. Ce defined as the resistivity divided by the surface area), Rsus the J.J. Haslam, K.E. Berroth, and F. F. Lange. "Processing and Properties of an specific resistance of the suspension over the board outer surface ance of the <a Oxide Composite with a Porous Matrix, "J. Eu. Ceram Soc. 20 15]607-18 in the capillaries, d the deposit thickness, di the board thickness, Fibre Mats with Sic Powder suspensions"s in , th conference of the efCrarean spe erammic Society, September 9-13, 2001. Edited by A.R. Boccaccini, Trans Tech Neglecting the observed small change of suspension conduct- 1. MacLaren. M. H. Lewis and C. B. Ponton,"Electro sition Infiltration of 2-D Woven Sic Fibl ivity(<+3%), Rsus and Rsus-cap are constant, and hence Mullite Comp J. Eur. Ceran. Soc., 17[13] 1545-50(1997) utler and C. B. Ponton. ""Novel Tech- da=C+I( Rdep-Rsus-cap)d (17) T J Illston, C. B. Ponton. P where C is a constant. If Rdep is constant during deposition, the eposition of Silica/Alumina Colloids for the Manufacture of Cme's":pp applied voltage will increase linearly with deposit growth into Ceramic Engineering and Science Proceedings, Proceedings of the 18th Ar Advanced Ceramic Materials-B. Part 2(of the capillaries. 9-14.1993,1994.15(5)

capillaries, but o1 outside. Thus, the electric field differs inside and outside the capillaries. Under a constant current of 0.3 mA, the electric field in the capillaries is 44 V/cm. The suspension resistance is calculated via the applied current and recorded voltage; hence the outside sub-cell coefficient can be determined between the same electrodes without an intervening porous board in the cell, i.e., f 5 0.52. The electric field across the sus￾pension is 10 V/cm, and hence the EPID suspension parameters are as summarized in Table I. According Eq. (11), the deposition rate in the capillaries should be 23 mg/min, i.e., much higher than the measured value (10 mg/min). However, according to Hamaker’s equation, the particle electrophoretic migration rate in an open suspension is 12 mg/min. The latter value is close to the measured deposition rate, which indicates that the deposition is controlled by the particle electrophoretic migration rate in the open suspension (outside of the capillaries), i.e., all particles electrophoresing to the board outer surface enter the capillaries and infiltrate faster there than in the suspension. This finding is not surprising as particles are ‘‘streamed’’ by the repulsion from the pore surfaces once they enter the porous board. This phenomenon also ex￾plains why no deleterious surface deposition (clogging) occurs until the porosity is fully filled. If the sub-cell coefficient outside the capillaries is 1, then the electrophoretic migration rate must equal the infiltration rate. In an EPD cell with the cathode as the deposition electrode, the applied potential, fa, is consumed in three steps: (a) a po￾tential drop at each electrode (cathode and anode), (b) an ohmic potential drop over the suspension, and (c) a potential drop over the deposit on the cathode. In an EPID cell, on the other hand, the potential drop over the suspension can be divided into two: (1) the potential drop across the open suspension outside the capillaries (between the anode and board outer surface), and (2) the suspension in the capillaries, i.e. fa ¼ Hfc el þ IRdepd þ IRsuscapðd1 dÞ þ IRsusd2 þ Hfa el (16) where Vfel a and Vfel c are the potential drop at the anode and cathode, respectively, I is the current flowing through the cell, Rdep the specific resistance of the deposit (specific resistance is defined as the resistivity divided by the surface area), Rsus the specific resistance of the suspension over the board outer surface and the anode, Rsus–cap the specific resistance of the suspension in the capillaries, d the deposit thickness, d1 the board thickness, and d2 the distance between the board outer surface and the anode. Neglecting the observed small change of suspension conduct￾ivity (o73%), Rsus and Rsus–cap are constant, and hence fa ¼ C þ IðRdep RsuscapÞd (17) where C is a constant. If Rdep is constant during deposition, the applied voltage will increase linearly with deposit growth into the capillaries. For uniform EPID under constant current conditions, the deposit thickness is proportional to the time of EPID, i.e. fa ¼ C þ kIðRdep RsuscapÞt (18) where k is a constant. Rdep depends on the type of stabilizing agent in the suspension.34 If the deposit is uniform and the specific resistance remains constant, the applied voltage will increase linearly with time. At 0.3 mA, the observed voltage change is almost linear with time, indicating that the deposit is uniform and its thickness increases linearly with time. This corresponds to the linear weight gain with time (Fig. 7). The observed initial voltage drop may be re￾lated to the preliminary suspension infiltration of the pores. For a current 0.6 mA, the voltage jumps to a maximum (1000 V). This increase is related to highly resistant entrapped air in the deposit. The presence of the latter was confirmed by the max￾imum PE board weight gain versus the deposition current. Max￾imum deposition (53 vol%) occurs at a current of 0.3 mA with an increase in linear voltage. When the current 0.6 mA, the deposited alumina only occupies 45% of the pore volume, indi￾cating that air bubbles are trapped in the deposit. Clearly, the deposition rate must be controlled to optimize the green density. This is why constant current must be used for the EPID process. V. Conclusions Conductive fibers are not suitable for fabrication of fiber-re￾inforced CMCs by electrophoretic deposition due to the electric field shielding effect. The field penetration depth is only double the fiber filament opening distance and a central cavity inevit￾ably results when a large bundle of conductive fibers is used. A modified EPD cell can be used to fabricate non-conductive- fiber-reinforced CMCs by EPID. Particles are infiltrated into a fiber preform to deposit on an electrode behind the preform and then backfill the preform. Thus, the front clogging encountered in direct EPD is avoided. The repulsive interaction between the fibers and the particles promotes particle streaming through the fiber preform, resulting in dense, uniform green composites by EPID. Constant-current EPID is modeled as capillary infiltration electrophoresis. Owing to the zero flow across a closed capillary cross section, no effect of electro-osmotic flow is expected. The infiltration deposition yield is proportional to the electric field inside the capillary; however, the total EPID yield is found to be controlled by particle electrophoresis outside the capillaries, i.e., in the open suspension, due to the much lower electric field therein. High deposition rates decrease green density due to air entrapment in the deposit. References 1 K. K. Chawla, Ceramic Matrix Composites. Chapman, New York, 1993. 2 F. F. Lange and K. T. Miller, ‘‘Pressure Filtraion: Consolidation Kinetics and Mechanics,’’ Am. Ceram. Soc. Bull., 66, 1498–504 (1987). 3 M. G. Holmquist and F. F. Lange, ‘‘Processing and Properties of a Porous Oxide Matrix Composite Reinforced with Continuous Oxide Fibers,’’ J. Am. Cer￾am. Soc., 86 [10] 1733–40 (2003). 4 J. J. Haslam, K. E. Berroth, and F. F. Lange, ‘‘Processing and Properties of an All-Oxide Composite with a Porous Matrix,’’ J. Eu. Ceram. Soc., 20 [5] 607–18 (2000). 5 K. Moritz and E. Muller, ‘‘Electrophoretic Infiltration of Woven Carbon Fibre Mats with Sic Powder Suspensions’’; in 7th Conference of the European Ceramic Society, September 9—13, 2001. Edited by A. R. Boccaccini, Trans Tech Publications Ltd., Brugge, 2001. 6 A. R. Boccaccini, I. MacLaren, M. H. Lewis, and C. B. Ponton, ‘‘Electropho￾retic Deposition Infiltration of 2-D Woven Sic Fibre Mats with Mixed Sols of Mullite Composition,’’ J. Eur. Ceram. Soc., 17 [13] 1545–50 (1997). 7 P. A. Trusty, A. R. Boccaccini, E. G. Butler, and C. B. Ponton, ‘‘Novel Tech￾niques for Manufacturing Woven Fiber Reinforced Ceramic Matrix Composites. I. Preform Fabrication,’’ Mater. Manuf. Process., 10 [6] 1215–26 (1995). 8 T. J. Illston, C. B. Ponton, P. M. Marquis, and E. G. Butler, ‘‘Electrophoretic Deposition of Silica/Alumina Colloids for the Manufacture of Cmc’s’’; pp. 1052–9 Ceramic Engineering and Science Proceedings, Proceedings of the 18th Annual Con￾ference on Composites and Advanced Ceramic Materials—B. Part 2 (of 2), January 9–14, 1993, 1994, 15(5). Table I. Suspension and Electrophoretic Infiltration Deposi￾tion Parameters in the Capillaries and in the Suspension In capillaries In suspension Suspension concentration cs (g/mL) 0.04 0.04 Suspension conductivity ss (mS/cm) 2.5 2.5 Electrophoretic mobility mEP mm s (V/cm) 0.8 0.8 Sub-cell coefficient f 1.0 0.52 Current density L (mA/cm2 ) 0.109 0.048 Electric field E (V/cm) 44 10 Deposition/electrophoretic rate (mg/min) 23 12 April 2007 Constant Current EPID of Fiber-Reinforced Ceramic Composites 1069

1070 Journal of the American Ceramic Society--Bao and Nicholson Vol. 90. No. 4 PA R Boccaccini and C B. Ponton, "Processing Ceramic-Matrix Composites us Kooner, w.S. Westby. C. M. A. Watson, and P. M. Farries, "Processing ng Electrophoretic Deposition, "Jomk-/. Min. Met. Mater. Soc., 47[10] 34-7 Nextel"720/Mullite Composition Composite Using Electrophoretic Depositie Kaya, F. Kaya. A.R. Boccaccini, and K.K. Chawla, ""Fabrication and w.S. Westby, S. Kooner, P. M. Farries, P. Boother, and R. A Shatwell, Characterisation Coated Carbon Fibre- Reinforced Alumina Ceramic mat Processing of Nextel 720/Mullite C ix Composites Using Electrophoretic Deposition, Acta Mater. 49[7 1189-97 and FH. e Reinforced Composites, U. Kaya, A.R. Boccaccini, and K. K. Chawla, "Electrophoretic 683.58. General Atomics, San Diego. CA. 199 orming of nickel coated Carbon Fiber -Rein forced borosilicate C Kaya, E. G. Butler, A Selcuk, A. R Boccaccini, and M. H. Lewis. ""Mullite Composites, J. Am. Ceram. Soc.,83 [8] 1885-8(2000). 2H.H perties, "J. Eur. Ceran. Soc., 22[13] 2333-42(2002). N Shape, J. Mater. Sci, 37[19]4145-53(2002). cation of Electrophoretically Infiltrated Silicon Carbide Ca ma and A. R sing of Complex Shape c232p246429-33(1997) and A.R. Bo- ites Using Electrophoretic Deposition, " J. Mater. Sci. Left, 20 [16] ccaccini, Fabrication Technologies for Oxide-Oxide Cera 567-76(2006 26M. N. Rahaman, Ceramic Processing and Sintering. ew p15A.R. Boccaccini, J. Ovenstone, and P. A. Trusty, " Fabrication of Woven letal Fibre Reinforced Glass Matrix Composites, " Appl. Composite Mater 4[31 aN. da, Engineering Electromagnetics, p.333. Springer, New York, 2000 Deposition of Ceramic Particles inside Porous Substrates. 2. Experimental-Moc Boccaccini, C. Kaya, and KK. Chawla, ""Use of Electrophoretic Dep- el, " J Electrochem. Soc., 139[41 1(1992) Processing of Fibre reir 32m R. J. Hunter, Fundations of Colloid Science, Vol. 1. Oxford University Press. lity of Oxide Particles in Polar Organic on of oxic Am Ceram Soc.8791767-70(2004 aya. A.R. Boccaccini, and P. A. Trusty, ""Processing and Characteriza- 8-51(96261 Coagulation of Col- lodal Dispersions, Trans. Faraday Soc., 62. 1638-51( 1966). n of 2-D Woven metal fibre- Reinfo bilayer Silica Matrix Com etween a Spherical oretic Deposition and Pressure Filtration, J. Eur. Ceram. Soc, 19 Particle and a Cylinder, J. Colloid Interface Sci. 231 [1] 199-203(2000) SH. C. Hamaker, "Formation of a Deposit by Electrophoresis. "Trans. Fara- u. I Al-Dawery, and E.G. Butler. ""Microstructural Develop- day Soc. 36, 279-87(1940) ullite Fibre- Reinforced mullite P Sarkar. D. De nd H. rho. ""Synthesis and Microstructural man Infiltration Processing, " Sci. Technol. Adv. Mater, 3[1]35-44(2002). Ceramics by electrophoretic position,"J. Mater.Sci,3国819230204).百

9 A. R. Boccaccini and C. B. Ponton, ‘‘Processing Ceramic–Matrix Composites Using Electrophoretic Deposition,’’ Jom-J. Min. Met. Mater. Soc., 47 [10] 34–7 (1995). 10C. Kaya, F. Kaya, A. R. Boccaccini, and K. K. Chawla, ‘‘Fabrication and Characterisation of Ni-Coated Carbon Fibre-Reinforced Alumina Ceramic Mat￾rix Composites Using Electrophoretic Deposition,’’ Acta Mater., 49 [7] 1189–97 (2001). 11C. Kaya, A. R. Boccaccini, and K. K. Chawla, ‘‘Electrophoretic Deposition Forming of Nickel Coated Carbon Fiber-Reinforced Borosilicate–Glass–Matrix Composites,’’ J. Am. Ceram. Soc., 83 [8] 1885–8 (2000). 12C. Kaya, F. Kaya, and A. R. Boccaccini, ‘‘Electrophoretic Deposition Infil￾tration of 2-D Metal Fibre-Reinforced Cordierite Matrix Composites of Tubular Shape,’’ J. Mater. Sci., 37 [19] 4145–53 (2002). 13C. Kaya and A. R. Boccaccini, ‘‘Colloidal Processing of Complex Shape Stainless Steel Woven Fiber Mat Reinforced Alumina Ceramic Matrix Compos￾ites Using Electrophoretic Deposition,’’ J. Mater. Sci. Lett., 20 [16] 1465–7 (2001). 14C. Kaya, F. Kaya, and A. R. Boccaccini, ‘‘Fabrication of Stainless-Steel￾Fiber-Reinforced Cordierite-Matrix Composites of Tubular Shape Using Electro￾phoretic Deposition,’’ J. Am. Ceram. Soc., 85 [10] 2575–7 (2002). 15A. R. Boccaccini, J. Ovenstone, and P. A. Trusty, ‘‘Fabrication of Woven Metal Fibre Reinforced Glass Matrix Composites,’’ Appl. Composite Mater., 4 [3] 145–55 (1997). 16A. R. Boccaccini, C. Kaya, and K. K. Chawla, ‘‘Use of Electrophoretic Dep￾osition in the Processing of Fibre Reinforced Ceramic and Glass Matrix Com￾posites: A Review,’’ Composites Part a-Appl. Sci. Manuf., 32 [8] 997–1006 (2001). 17Y. Bao and P. S. Nicholson, ‘‘Conductive, Polypyrrole Coating on Mullite/ Alumina Fibers for Electrophoretic Deposition of Oxide Matrices,’’ J. Am. Ceram. Soc., 87 [9] 1767–70 (2004). 18C. Kaya, A. R. Boccaccini, and P. A. Trusty, ‘‘Processing and Characteriza￾tion of 2-D Woven Metal Fibre-Reinforced Multilayer Silica Matrix Composites Using Electrophoretic Deposition and Pressure Filtration,’’ J. Eur. Ceram. Soc., 19 [16] 2859–66 (1999). 19C. Kaya, X. Gu, I. Al-Dawery, and E. G. Butler, ‘‘Microstructural Develop￾ment of Woven Mullite Fibre-Reinforced Mullite Ceramic Matrix Composites by Infiltration Processing,’’ Sci. Technol. Adv. Mater., 3 [1] 35–44 (2002). 20S. Kooner, W. S. Westby, C. M. A. Watson, and P. M. Farries, ‘‘Processing of Nextelt 720/Mullite Composition Composite Using Electrophoretic Deposition,’’ J. Eur. Ceram. Soc., 20 [5] 631–8 (2000). 21W. S. Westby, S. Kooner, P. M. Farries, P. Boother, and R. A. Shatwell, ‘‘Processing of Nextel 720/Mullite Composition Composite Using Electrophoretic Deposition,’’ J. Mater. Sci., 34 [20] 5021–31 (1999). 22T. Ohkawa and F. H. Elsner, Fabrication of Fiber-Reinforced Composites, US Patent No. 5468358. General Atomics, San Diego, CA, 1995. 23C. Kaya, E. G. Butler, A. Selcuk, A. R. Boccaccini, and M. H. Lewis, ‘‘Mullite (Nextelt 720) Fibre-Reinforced Mullite Matrix Composites Exhibiting Favour￾able Thermomechanical Properties,’’ J. Eur. Ceram. Soc., 22 [13] 2333–42 (2002). 24H. H. Streckert, K. P. Norton, D. Katz, and J. O. Freim, ‘‘Microwave Den￾sification of Electrophoretically Infiltrated Silicon Carbide Composite,’’ J. Mater. Sci., 32 [24] 6429–33 (1997). 25E. Stoll, P. Mahr, H. G. Kruger, H. Kern, B. J. C. Thomas, and A. R. Bo￾ccaccini, ‘‘Fabrication Technologies for Oxide–Oxide Ceramic Matrix Composites Based on Electrophoretic Deposition,’’ J. Eur. Ceram. Soc., 26 [9] 1567–76 (2006). 26M. N. Rahaman, Ceramic Processing and Sintering. Marcel Dekker, New York, 1995. 27N. Ida, Engineering Electromagnetics, p. 333. Springer, New York, 2000. 28L. Galor, S. Liubovich, and S. Haber, ‘‘Deep Electrophoretic Penetration and Deposition of Ceramic Particles inside Porous Substrates. 2. Experimental–Mod￾el,’’ J. Electrochem. Soc., 139 [4] 1078–81 (1992). 29R. J. Hunter, Fundations of Colloid Science, Vol. I. Oxford University Press, New York, 1989. 30G. Wang, ‘‘Ionic Stability of Oxide Particles in Polar Organic Media’’; p. 191 in Materials Science and Engineering. McMaster University, Hamilton, 1998. 31R. Hogg, T. W. Healy, and D. W. Fuerstenau, ‘‘Mutual Coagulation of Col￾loidal Dispersions,’’ Trans. Faraday Soc., 62, 1638–51 (1966). 32Y. G. Gu, ‘‘The Electrical Double-Layer Interaction Between a Spherical Particle and a Cylinder,’’ J. Colloid Interface Sci., 231 [1] 199–203 (2000). 33H. C. Hamaker, ‘‘Formation of a Deposit by Electrophoresis,’’ Trans. Fara￾day Soc., 36, 279–87 (1940). 34P. Sarkar, D. De, and H. Rho, ‘‘Synthesis and Microstructural Manipulation of Ceramics by Electrophoretic Deposition,’’ J. Mater. Sci., 39 [3] 819–23 (2004). & 1070 Journal of the American Ceramic Society—Bao and Nicholson Vol. 90, No. 4

Copyright of Journal of the American Ceramic Society is the property of Blackwell Publishing Limited and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder s express written permission. However, users may print, download, or email articles for individual use