E驅≈3S Journal of the European Ceramic Society 20(2000)537-544 Chemical vapor deposition of zro2 and C/zrO2 on mullite fibers for interfaces in mullite/ aluminosilicate fiber-reinforced composites K. Nubianb B Saruhana,*B Kankaa M. Schmuckera H. Schneidera G. Wahlb a German Aerospace Center, Institute for Materials Research, 51147 Koln, Germany bUniversity of Braunschweig, Institute for Surface Technology, 38108 Braunschweig Germ Accepted 10 August 1999 Abstract For the realization of crack deflection and fiber pull-out in aluminosilicate fiber-reinforced dense mullite-matrix composites suitable fiber /matrix- interfaces are an important requirement in order to obtain sufficiently weak bondings between fibers and matrices. Two types of chemical vapor deposited(CVD)fiber/matrix-interfaces have been studied in the present work porous Zro and C/zrO2-double layers. In the latter case, carbon was burned out to form a gap during the processing of composites(fugitive coating). Porous ZrO2 coatings were produced by an optimized CVD-process with Zr-acetylacetonate as a precursor. The con stancy of the layer thickness depended on the deposition temperature. It was found that at a temperature of approximately 300C and a pressure of 5 hPa, suitably uniform layers with thickness ranging between 100 and 300 nm were achieved. The coatings contained approximately 15 wt% carbon which produced, after annealing in air, a porous structure. The deposition kinetics can be described by a first order reaction. The carbon layer in C/ZrO -double layers was produced by using propane. The thickness of carbon layer was 10 and 100 nm, respectively. Aluminosilicate fiber/mullite matrix composite prepegs were fabricated by infiltration of coated and unidirectionally oriented fiber(0 )with a slurry, containing a pre-mullite powder, calcined at 1.C. Uniaxial hot- pressing of dried prepegs was carried out at 1250C for 15 min, at 20 MPa. Prepegs with ZrO2 fiber/matrix- interfaces were hot pressed in air, while the samples with C/ZrO -interfaces were processed in flowing argon. After hot-pressing, samples with C/zrO2- interfaces were heat-treated in air (1000C) in order to bl arn out the C-layer (fugitive coating). These composites yielded a con- trolled fracture with a high deflection rate and a favorable fracture strength of about 200 MPa, due to crack-defiection and fiber tolerant than those having C/ZrO, double layer systems. C 2000 Elsevier Science Ltd. All rights reserved e, they are less damage pull-out. Composites with ZrO2-interfaces, on the contrary yielded no crack deflection or pull-out. Therefo Keywords: Aluminosilicate fibres; Composites; Interfaces; Mullite matrix; ZRO thus lowering the shear strength and constituting a pre- ferred path for the diversion of matrix-originating Thermal protection systems consisting of oxide-based cracks. An example of this approach is given by Si-O-N fiber-reinforced composites can contribute significantly in the reduction of no and emission in the combustion coating on SiC-fibers. The required porosity was gen erated with latex polystyrene mixed with Si3 N4 and chambers of aircraft engines and stationary gas turbine SiOx-powders. Push-out tests showed that only those m乙mm(③ mics, less cooling, thus, less fuel consum necessary. Nevertheless, stror matrix-interfaces due chanical perature. Another method, describing a successful pseud ss of ceramic composites, porous fiber coating system by means of CVD, used a making them less damage tolerant and, therefore, less The carbon source was methane, and the Sic source reliable for the application was either methyltrichlorosilane or tetrachloride. After For achievement of advantageous failure in the com heat-treatment at a temperature range between 800 and posites, one approach is to apply porous interphases, 1200C under reduced pressure(1-100 Torr), crystalline 4 Corresponding author. Tel. +49-2203-601-3228: fax: +49-2203- SiC with a resulting porosity of 12% was obtained. 696480 The only effective way in coating of fiber surfaces in E-mail address: bilge saruhan(@ drde(B Saruhan) fiber tows or fabrics with thin ceramic layers is to 0955-221999/S- see front matter o 2000 Elsevier Science Ltd. All rights reserved PII:S0955-2219(99)00251-4
Chemical vapor deposition of ZrO2 and C/ZrO2 on mullite ®bers for interfaces in mullite/aluminosilicate ®ber-reinforced composites K. Nubianb, B. Saruhana,*, B. Kankaa , M. SchmuÈckera , H. Schneidera , G. Wahlb a German Aerospace Center, Institute for Materials Research, 51147 KoÈln, Germany bUniversity of Braunschweig, Institute for Surface Technology, 38108 Braunschweig, Germany Accepted 10 August 1999 Abstract For the realization of crack de¯ection and ®ber pull-out in aluminosilicate ®ber-reinforced dense mullite-matrix composites, suitable ®ber/matrix-interfaces are an important requirement in order to obtain suciently weak bondings between ®bers and matrices. Two types of chemical vapor deposited (CVD) ®ber/matrix-interfaces have been studied in the present work porous ZrO2 and C/ZrO2-double layers. In the latter case, carbon was burned out to form a gap during the processing of composites (fugitive coating). Porous ZrO2 coatings were produced by an optimized CVD-process with Zr-acetylacetonate as a precursor. The constancy of the layer thickness depended on the deposition temperature. It was found that at a temperature of approximately 300C and a pressure of 5 hPa, suitably uniform layers with thickness ranging between 100 and 300 nm were achieved. The coatings contained approximately 15 wt% carbon which produced, after annealing in air, a porous structure. The deposition kinetics can be described by a ®rst order reaction. The carbon layer in C/ZrO2-double layers was produced by using propane. The thickness of carbon layer was 10 and 100 nm, respectively. Aluminosilicate ®ber/mullite matrix composite prepegs were fabricated by in®ltration of coated and unidirectionally oriented ®ber (0) with a slurry, containing a pre-mullite powder, calcined at 1100C. Uniaxial hotpressing of dried prepegs was carried out at <1250C for 15 min, at 20 MPa. Prepegs with ZrO2 ®ber/matrix-interfaces were hotpressed in air, while the samples with C/ZrO2-interfaces were processed in ¯owing argon. After hot-pressing, samples with C/ZrO2- interfaces were heat-treated in air (1000C) in order to burn out the C-layer (fugitive coating). These composites yielded a controlled fracture with a high de¯ection rate and a favorable fracture strength of about 200 MPa, due to crack-de¯ection and ®ber pull-out. Composites with ZrO2-interfaces, on the contrary yielded no crack de¯ection or pull-out. Therefore, they are less damage tolerant than those having C/ZrO2 double layer systems. # 2000 Elsevier Science Ltd. All rights reserved. Keywords: Aluminosilicate ®bres; Composites; Interfaces; Mullite matrix; ZRO2 1. Introduction Thermal protection systems consisting of oxide-based ®ber-reinforced composites can contribute signi®cantly in the reduction of NOx and emission in the combustion chambers of aircraft engines and stationary gas turbine engines. Relying on good thermal properties of the ceramics, less cooling, thus, less fuel consumption will be necessary. Nevertheless, strong bonding at the ®ber/ matrix-interfaces due to the chemical and mechanical interactions causes brittleness of ceramic composites, making them less damage tolerant and, therefore, less reliable for the application. For achievement of advantageous failure in the composites, one approach is to apply porous interphases, thus lowering the shear strength and constituting a preferred path for the diversion of matrix-originating cracks. An example of this approach is given by Si±O±N coating on SiC-®bers.1 The required porosity was generated with latex polystyrene mixed with Si3N4 and SiO2-powders. Push-out tests showed that only those coatings which were retreated with a silica layer displayed a reasonably low (25 MPa) friction-stress at room temperature. Another method, describing a successful pseudo porous ®ber coating system by means of CVD, used a 0.5±50 mm thick layer of carbon-enriched SiC-coating.2 The carbon source was methane, and the SiC source was either methyltrichlorosilane or tetrachloride. After heat-treatment at a temperature range between 800 and 1200C under reduced pressure (1±100 Torr), crystalline SiC with a resulting porosity of 12% was obtained. The only eective way in coating of ®ber surfaces in ®ber tows or fabrics with thin ceramic layers is to 0955-2219/99/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved. PII: S0955-2219(99)00251-4 Journal of the European Ceramic Society 20 (2000) 537±544 * Corresponding author. Tel.: +49-2203-601-3228; fax: +49-2203- 696480. E-mail address: bilge.saruhan@dlr.de (B. Saruhan)
employ processes which have a large throwing power. All piping was made of stainless steel and heated to 500 The CVD process is one of such processes. Deposition K. Typical deposition conditions were: deposition tem of SiC or BN coatings on fibers with CVD process has perature 300-500oC, pressure 500 Pa. 4 been proven to be successful. 3 This paper describes The deposition equipment is shown in Fig 3. In this methods for coating of Nextel M 720-mullite/alumina- equipment, two different evaporators were installed, in fibers with a porous ZrO2-layer and with a C/ZrO2- order to produce multi-component coatings if necessary double layer resulting in a fugitive coating, sealed with a The reaction gas was transported into a quartz tube in a porous oxide layer resistance heated furnace. The internal diameter of the quartz tube was 10.5 cm and the aluminosilicate fiber fabrics, cut into rectangles of 50x 50 mm, were installed 2. Experimental in this quartz tube. The fabrics were arranged at a position perpendicular to the gas flow. Maximum 4 2. Materials samples could be placed in the resistance furnace. The deposited quantity was measured by the mass differ Precursors with high vapor pressures at moderate ence before (1) and after the deposition (m2) temperatures are preferred for CVD-coatings. Candi The molar deposition rate was calcu date precursors are chlorides, alkoxides and B-dikato- lated by assuming that ZrO, is deposited: nates. Zirconium acetylacetonate [Zr(acac)4 is a B- dikatonate and has a vapor pressure of approx 700 Pa (1) at 200 C, which is high enough to be used in CVD processes. The CVd processing of Zr(acac)4 to produce ZrO2-coatings on flat substrates was described in the where M is the molar mass of Zro literature. 4-7 These studies reported that some carbon The carbon precoating of the Nextel fibers was carried was entrapped in the coating system, because of the out under a pressure of 12 hPa at a deposition tem incomplete decomposition on the substrate surface. This perature of 950%C for 45 min. The carrier gas was pro characteristic of the precursor may be benefited for the pane and its flow rate was set to 167 sccm synthesis of porous ZrO] -layers, without application of another substance for achievement of porosity. The 2.3. Processing of composite Zr(acac)4 powder was of 98% purity and provided by FLUKA(Germany) The pre-mullite(Siral) powder was calcined at about The woven fiber fabrics(8 harness Atlas) were of 1100C prior to preparation of the slurry. X-ray dif- mullite/alumina [ Nextel(720)] and were provided by 3M fraction data of this calcined powder yielded a weakly (Minnesota, USA). The fabrics had the form shown in crystalline y-Al2O3 phase, accompanied by some Sio Fig. I. Woven fiber tows were consisted of fibers with a rich amorphous phase Formation of mullite under given diameter of 12 um. The chemical composition was 85 hot-pressing conditions occurred after I minute holding wt%Al,O3 and 15 wt%SiO, with a density of 3.4 time at 1250C. Just at the hot-pressing temperature, cm3. Each fiber tow contains approximately 400 single without any holding time, no mullite formation was observed. For a full transformation to mullite, it was The matrix material for the composites was produced necessary to hold for 15 min at 1250C under 20 MPa by hot-pressing pre- mullite powders (Siral, CONDEA, Prolonged holding times yielded no improvement in the Germany). This powder has a purity of 99.99% and is densification and mullitization. In order to avoid fiber amorphous in the as-received form damage, the optimum holding time was limited to 15 min Slurry was prepared by mixing of the calcined pow 2. 2. CVD-process and experimental equipmen ders with binder and disperser in an aqueous media Although, for coating, the fiber fabrics were used, for Decomposition temperature of Zr(acac)4 is rather the preparation of the prepegs, the fiber tows were low, hence evaporation rate becomes unstable, if it is pulled out of the fabric to fabricate ID-unidirectional heated for longer times at temperatures above 160C. fiber composites. As-coated fiber tows were immersed in This behavior requires special evaporation equipment, the slurry and formed to rectangular shaped plates on which is shown schematically in Fig. 2. The powdery plaster of Paris moulds. After drying the prepegs in air, precursor is transported through a rotating plate with the stacked prepegs were uniaxially cold-pressed under holes into the evaporator which is heated to 200C and approximately 2 MPa pressure. Hot-pressing of the there evaporates spontaneously. The carrier gas with ZrO2-coated fiber composites was carried out in air at the precursor then is transported through tubes to the 1250 C for 15 min under 20 MPa uniaxial pressure CI urnace. The precursor concentration in the gas was ZrO2-coated fiber composites were also hot-pressed adjusted to 0. 1 mol% at a gas flow of 200 sccm argon. under the same conditions, however in flowing argon
employ processes which have a large throwing power. The CVD process is one of such processes. Deposition of SiC or BN coatings on ®bers with CVD process has been proven to be successful.3 This paper describes methods for coating of NextelTM 720-mullite/alumina- ®bers with a porous ZrO2-layer and with a C/ZrO2- double layer resulting in a fugitive coating, sealed with a porous oxide layer. 2. Experimental 2.1. Materials Precursors with high vapor pressures at moderate temperatures are preferred for CVD-coatings. Candidate precursors are chlorides, alkoxides and b-dikatonates. Zirconium acetylacetonate [Zr(acac)4] is a bdikatonate and has a vapor pressure of approx. 700 Pa at 200C, which is high enough to be used in CVDprocesses. The CVD processing of Zr(acac)4 to produce ZrO2-coatings on ¯at substrates was described in the literature.4±7 These studies reported that some carbon was entrapped in the coating system, because of the incomplete decomposition on the substrate surface. This characteristic of the precursor may be bene®ted for the synthesis of porous ZrO2-layers, without application of another substance for achievement of porosity. The Zr(acac)4 powder was of 98% purity and provided by FLUKA (Germany). The woven ®ber fabrics (8 harness Atlas) were of mullite/alumina [Nextel(720)] and were provided by 3M (Minnesota, USA). The fabrics had the form shown in Fig. 1. Woven ®ber tows were consisted of ®bers with a diameter of 12 mm. The chemical composition was 85 wt% Al2O3 and 15 wt% SiO2 with a density of 3.4 g/ cm3 . Each ®ber tow contains approximately 400 single ®bers. The matrix material for the composites was produced by hot-pressing pre-mullite powders (Siral, CONDEA, Germany). This powder has a purity of 99.99% and is amorphous in the as-received form. 2.2. CVD-process and experimental equipment Decomposition temperature of Zr(acac)4 is rather low, hence evaporation rate becomes unstable, if it is heated for longer times at temperatures above 160C.4 This behavior requires special evaporation equipment, which is shown schematically in Fig. 2. The powdery precursor is transported through a rotating plate with holes into the evaporator which is heated to 200C and there evaporates spontaneously. The carrier gas with the precursor then is transported through tubes to the furnace. The precursor concentration in the gas was adjusted to 0.1 mol% at a gas ¯ow of 200 sccm argon. All piping was made of stainless steel and heated to 500 K. Typical deposition conditions were: deposition temperature 300±500C, pressure 500 Pa.4 The deposition equipment is shown in Fig. 3. In this equipment, two dierent evaporators were installed, in order to produce multi-component coatings if necessary. The reaction gas was transported into a quartz tube in a resistance heated furnace. The internal diameter of the quartz tube was 10.5 cm and the aluminosilicate ®ber fabrics, cut into rectangles of 5050 mm2 , were installed in this quartz tube. The fabrics were arranged at a position perpendicular to the gas ¯ow. Maximum 4 samples could be placed in the resistance furnace. The deposited quantity was measured by the mass dierence before (m1) and after the deposition (m2) m m2 ÿ m1. The molar deposition rate was calculated by assuming that ZrO2 is deposited: n : m m AM 1 where M is the molar mass of ZrO2. The carbon precoating of the Nextel ®bers was carried out under a pressure of 12 hPa at a deposition temperature of 950C for 45 min. The carrier gas was propane and its ¯ow rate was set to 167 sccm. 2.3. Processing of composites The pre-mullite (Siral) powder was calcined at about 1100C prior to preparation of the slurry. X-ray diffraction data of this calcined powder yielded a weakly crystalline g-Al2O3 phase, accompanied by some SiO2- rich amorphous phase. Formation of mullite under given hot-pressing conditions occurred after 1 minute holding time at 1250C. Just at the hot-pressing temperature, without any holding time, no mullite formation was observed. For a full transformation to mullite, it was necessary to hold for 15 min at 1250C under 20 MPa. Prolonged holding times yielded no improvement in the densi®cation and mullitization. In order to avoid ®ber damage, the optimum holding time was limited to 15 min. Slurry was prepared by mixing of the calcined powders with binder and disperser in an aqueous media. Although, for coating, the ®ber fabrics were used, for the preparation of the prepegs, the ®ber tows were pulled out of the fabric to fabricate 1D-unidirectional ®ber composites. As-coated ®ber tows were immersed in the slurry and formed to rectangular shaped plates on plaster of Paris moulds. After drying the prepegs in air, the stacked prepegs were uniaxially cold-pressed under approximately 2 MPa pressure. Hot-pressing of the ZrO2-coated ®ber composites was carried out in air at 1250C for 15 min under 20 MPa uniaxial pressure. C/ ZrO2-coated ®ber composites were also hot-pressed under the same conditions, however in ¯owing argon 538 K. Nubian et al. / Journal of the European Ceramic Society 20 (2000) 537±544
K Nubian et al. / Journal of the European Ceramic Society 20(2000)537-544 ÷150m composition 85wt% filament density 3. 4 g/cm3 number of filaments / bundle 400 of a bundl 0,57 specific surface of bundle 9,8·10cm2g substrate siz 50 mm x 50 mm Fig 1. Schematic view of the Nextel 720 fiber fabri which was introduced into the system above about 400C 3. Results and discussion [Fig. 4(a)and (b)]. For both cases, the pressure was applied at about 1100 C, as soon as the first shrinkage 3. 1. Deposition of zro2 occurs. Hot-pressing was carried out mould-free between Sic-punches in order to achieve a homogeneous tem- Fig. 5 shows the measured deposition rate on the perature distribution on the sample plate fabric and on the Al2O3 wafer vs the reciprocal absolute temperature. The temperature dependence of the 2.4. Characterization of composites deposition rate can be described by an Arrhenius plot Phase analysis of the matrix and the coating were im exp( T<370°C carried out by X-ray powder diffraction at room tem- perature(SIEMENS, D5000, Germany) using Ni-fil tered CuKg radiation. Difiraction patterns were Thus, the activation energy E= 24+-3 kJ/mol of recorded in step scan mode(3s/0.05, 20)in the 10-800 20 ZrO2 deposition on the fabric was calculated. The cal range. Microstructural investigations were carried out culated activation energy E for the fiber fabric is much with optical and electron microscopy studies were done smaller than the values given in literature for ZrO2- with LEO field emission scanning electron microscopy deposition on flat substrates(E=80+-7 kJ/mol).4A and with PHILLIPS EM 430 transmission electron plane Al2O3-wafer was coated as a reference under the microscopy (TEM), 300 kV accelerating voltage. Sam- same conditions. The activation energy was E=66+-5 ple preparation performed by dimple grinding and kJ/mol which is in good accordance with literature.4 sequent ion beam milling. Samples were coated with The deposition rate related to the fiber surface on the carbon to avoid charging effect fabric is much lower than on the AlO3-wafer. This is
which was introduced into the system above about 400C [Fig. 4(a) and (b)]. For both cases, the pressure was applied at about 1100C, as soon as the ®rst shrinkage occurs. Hot-pressing was carried out mould-free between SiC-punches in order to achieve a homogeneous temperature distribution on the sample plate. 2.4. Characterization of composites Phase analysis of the matrix and the coating were carried out by X-ray powder diraction at room temperature (SIEMENS, D5000, Germany) using Ni-®ltered CuKa radiation. Diraction patterns were recorded in step scan mode (3s/0.05, 2) in the 10±80 2 range. Microstructural investigations were carried out with optical and electron microscopy studies were done with LEO ®eld emission scanning electron microscopy and with PHILLIPS EM 430 transmission electron microscopy (TEM), 300 kV accelerating voltage. Sample preparation performed by dimple grinding and subsequent ion beam milling. Samples were coated with carbon to avoid charging eect. 3. Results and discussion 3.1. Deposition of ZrO2 Fig. 5 shows the measured deposition rate on the fabric and on the Al2O3 wafer vs the reciprocal absolute temperature. The temperature dependence of the deposition rate can be described by an Arrhenius plot n : m exp ÿ E RTT < 370 C 2 Thus, the activation energy E 24 ÿ3 kJ/mol of ZrO2 deposition on the fabric was calculated. The calculated activation energy E for the ®ber fabric is much smaller than the values given in literature for ZrO2- deposition on ¯at substrates (E 80 ÿ7 kJ/mol). 4 A plane Al2O3-wafer was coated as a reference under the same conditions. The activation energy was E=66+ÿ5 kJ/mol which is in good accordance with literature.4 The deposition rate related to the ®ber surface on the fabric is much lower than on the Al2O3-wafer. This is Fig. 1. Schematic view of the Nextel 720 ®ber fabric. K. Nubian et al. / Journal of the European Ceramic Society 20 (2000) 537±544 539
K Nubian et al. Journal of the European Ceramic Society 20(2000)537-544 because the deposition rate on the fibers is not only deter- The scanning electron microscopy observations of the mined by the chemical reaction, but also by the diffusion fibers coated at 310 and 370C demonstrates that the coat- of the reactive gas into the fiber bundle. The deposition ings deposited at 310.C were more uniform and complete conditions are given above. At temperatures T>370 C than those deposited at 370 C. The surface of the Zro the deposition rate of ZrO2 decreases according to Fig. 5 coating, obtained at 310C were smooth and between the because of strong powder formation in the gas phase fibers, no pasting was observed. The 370%C, deposited coat- which was observed on all walls ing surfaces were rough and incomplete. Only a few fibers were coated, while most fibers were free of any coating. In order to determine the distribution of the layer thickness on the single fibers inside the fabric, one fiber engine Slurry Preparation rotating axis Slurry infiltration plates with holes Uniaxial cold-pressing of prepregs(app. 2 MPa) precursor Drying carrier gas ZrO-interface at 1250C under ZrO-interface at 1250C under 20 MPa in flowing argon MPa in to the reactor 250°C,15min 1000°C heated to200°C Fig. 4.(a) Flow chart showing the details in fabrication of composites and(b) schematic presentation of the hot-pressing process during Fig. 2. Evaporation equipment used for CVD-coating experiments. evaporator 2 ubstrate hold th substrates controller 960 V1 Fig 3. Schematic drawing of the CVD-equipment
because the deposition rate on the ®bers is not only determined by the chemical reaction, but also by the diusion of the reactive gas into the ®ber bundle. The deposition conditions are given above. At temperatures T>370C the deposition rate of ZrO2 decreases according to Fig. 5 because of strong powder formation in the gas phase which was observed on all walls. The scanning electron microscopy observations of the ®bers coated at 310 and 370C demonstrates that the coatings deposited at 310C were more uniform and complete than those deposited at 370C. The surface of the ZrO2- coating, obtained at 310C were smooth and between the ®bers, no pasting was observed. The 370C, deposited coating surfaces were rough and incomplete. Only a few ®bers were coated, while most ®bers were free of any coating. In order to determine the distribution of the layer thickness on the single ®bers inside the fabric, one ®ber Fig. 2. Evaporation equipment used for CVD-coating experiments. Fig. 3. Schematic drawing of the CVD-equipment. Fig. 4. (a) Flow chart showing the details in fabrication of composites and (b) schematic presentation of the hot-pressing process during fabrication of composites. 540 K. Nubian et al. / Journal of the European Ceramic Society 20 (2000) 537±544
bundle was removed from the fabric and embedded in medium if a first order reaction for the ZrO2 deposition resin(G-1 Epoxy, GATAN Inc ) After breaking the is assumed. 8 embedded bundle the thickness of the coatings on the single fibers could be measured by SEM. In each sample j=-g dx dc approx 80 single fibers taken at random in the bundle (3) were measured. The resulting thickness distributions are shown in Fig. 6. It shows that the thickness on the fibers e is the porosity of the fiber bundle produced by the is more constant at lower temperatures. This can be fibers, D is the gas diffusion coefficient of the precursor expected because the depletion in the gas phase crea in the carrier gas, c is the molar concentration of the ses with the temperature. The thickness distribution can precursor in the gas phase, q is the tortuosity, j is the be explained by Fick's law for the diffusion in a porous current diffusion density related to the cross-section of 1,E04 380°c370°c360°c350°C340°c330°c320°c310°c 0EE 1E-05 deposition rate related to the fiber surface 1,50E03 1.55E03 1,60E03 1,70E03 175E03 deposition tempeture 1/T[1/K ◆ Nextel720■A2O3 Fig. 5. Deposition rate of ZrO2 from Zrfacac) vs reciprocal deposition temperature on the Nextel 720 fiber fabric and on Al2O]wafer(P=500 Pa, ow=200 sccm argon) 100% 56cE品 g巽50% layer thickness [ur 士T=310°C一T=330°C一T=340%"T=370°C Fig. 6. Cumulated distribution of layer thickness in a single fiber bundle (80 fibers measured for each bundle)
bundle was removed from the fabric and embedded in resin (G-1 Epoxy, GATAN Inc.). After breaking the embedded bundle the thickness of the coatings on the single ®bers could be measured by SEM. In each sample approx. 80 single ®bers taken at random in the bundle were measured. The resulting thickness distributions are shown in Fig. 6. It shows that the thickness on the ®bers is more constant at lower temperatures. This can be expected because the depletion in the gas phase decreases with the temperature. The thickness distribution can be explained by Fick's law for the diusion in a porous medium if a ®rst order reaction for the ZrO2 deposition is assumed.8 j ÿ " q D dc dx 3 " is the porosity of the ®ber bundle produced by the ®bers, D is the gas diusion coecient of the precursor in the carrier gas, c is the molar concentration of the precursor in the gas phase, q is the tortuosity, j is the current diusion density related to the cross-section of Fig. 5. Deposition rate of ZrO2 from Zr(acac)4 vs reciprocal deposition temperature on the Nextel 720 ®ber fabric and on Al2O3-wafer (P=500 Pa, ¯ow=200 sccm argon). Fig. 6. Cumulated distribution of layer thickness in a single ®ber bundle (80 ®bers measured for each bundle). K. Nubian et al. / Journal of the European Ceramic Society 20 (2000) 537±544 541
K Nubian et al. / Journal of the European Ceramic Society 20(2000)537-544 the porous medium, x is the distance to bundle surface. isolated ZrO, grains(Fig. 7). Furthermore, the pores do In order to improve the constancy of the layer thickness not present the open channel-type pore formation typi according to Fig. 6 the deposition temperature should cal for volatilization processes. Therefore, we believe be as low as possible. Experiments show, however, that that crystallization processes as well as burn-out can be no deposition is possible at temperatures below 300 c held responsible because the reaction velocity is too slow. So deposition In spite of their relatively high interface porosity temperatures at about 310C was found optimum composites with ZrO2 fiber/matrix- interfaces show no X-ray diffraction data of the fibers CVD-coated at fiber pull-out [Fig 8(a)] and display stress-strain curves 370C showed that the coatings just after the CVD- corresponding to brittle fracture behavior. Scanning process were amorphous ZrO2. After a heat-treatment electron microscopy investigations on the fracture sur at 1250'C or during hot-pressing of composites at faces reveal that the fiber/matrix bonding, despite the 1250.C, the coatings were converted to monoclinic presence of a porous Zro2 layer, was strong. Especially ZrO2 On determination of mass changes of the coated in hot-pressing direction, the matrix at the contact points fibers after heat-treatment at 700oC in oxygen, a weight to the fibers is highly densified and is in intense contact to loss of approximately 15 wt% was found, indicating the the interfaces. The area around the fibers, perpendicular presence of residual carbon in the as-coated form, due to the hot-pressing direction on the contrary are less den to the incomplete decomposition of the Zr(acac)4. This value is in good agreement with the literature data of einbeck e 3. 2. ZrO, and C/ZrO interfaces in mullite/ aluminosilicate fiber-reinforced composites ZrOrinterfaces Transmission electron microscopic investigations of the composites with ZrO2 fiber /matrix- interface yielded porous Zro2 layers with a thickness of about 200-500 nm(Fig. 7). In a first approach, we assumed that pore formation was mainly caused by burn-out process of the residual carbon which was in the order of 15%. more detailed inspection of the microstructure in ZrO2-layer showed that the porous Zro2 after hot-pressing can not be 8m150kU492E39434979 due only to burn-out. This was derived from the observa tion that the interphase displays an open porosity which is homogeneously distributed between well-rounded and Hot-Press Direction cnse an ss Banc Dense area] on microscopic image of ZrO2 fiber/matrix drawing of interfacial conditions after hot-pressing in unidir terraces in mullite ma omposite. The interfaces consist of porous aluminosilicate fiber-reinforced/porous Zro2 interphase and ZrO2 layers with a th of about 500 nm matrIx composites
the porous medium, x is the distance to bundle surface. In order to improve the constancy of the layer thickness according to Fig. 6 the deposition temperature should be as low as possible. Experiments show, however, that no deposition is possible at temperatures below 300C because the reaction velocity is too slow. So deposition temperatures at about 310C was found optimum. X-ray diraction data of the ®bers CVD-coated at 370C showed that the coatings just after the CVDprocess were amorphous ZrO2. After a heat-treatment at 1250C or during hot-pressing of composites at 1250C, the coatings were converted to monoclinic ZrO2. On determination of mass changes of the coated ®bers after heat-treatment at 700C in oxygen, a weight loss of approximately 15 wt% was found, indicating the presence of residual carbon in the as-coated form, due to the incomplete decomposition of the Zr(acac)4. This value is in good agreement with the literature data of Brenn¯eck et al. 6 3.2. ZrO2 and C/ZrO2-interfaces in mullite/ aluminosilicate ®ber-reinforced composites ZrO2-interfaces Transmission electron microscopic investigations of the composites with ZrO2 ®ber/matrix-interface yielded porous ZrO2 layers with a thickness of about 200±500 nm (Fig. 7). In a ®rst approach, we assumed that pore formation was mainly caused by burn-out process of the residual carbon which was in the order of 15%. More detailed inspection of the microstructure in ZrO2-layer showed that the porous ZrO2 after hot-pressing can not be due only to burn-out. This was derived from the observation that the interphase displays an open porosity which is homogeneously distributed between well-rounded and isolated ZrO2 grains (Fig. 7). Furthermore, the pores do not present the open channel-type pore formation typical for volatilization processes. Therefore, we believe that crystallization processes as well as burn-out can be held responsible. In spite of their relatively high interface porosity, the composites with ZrO2 ®ber/matrix-interfaces show no ®ber pull-out [Fig. 8(a)] and display stress±strain curves corresponding to brittle fracture behavior. Scanning electron microscopy investigations on the fracture surfaces reveal that the ®ber/matrix bonding, despite the presence of a porous ZrO2 layer, was strong. Especially in hot-pressing direction, the matrix at the contact points to the ®bers is highly densi®ed and is in intense contact to the interfaces. The area around the ®bers, perpendicular to the hot-pressing direction on the contrary are less denFig. 7. Transmission electron microscopic image of ZrO2 ®ber/matrix interfaces in mullite matrix composite. The interfaces consist of porous ZrO2 layers with a thickness of about 500 nm. Fig. 8. (a) Scanning electron microscopic image and (b) schematic drawing of interfacial conditions after hot-pressing in unidirectionally aluminosilicate ®ber-reinforced/porous ZrO2 interphase and mullite matrix composites. 542 K. Nubian et al. / Journal of the European Ceramic Society 20 (2000) 537±544
K Nubian et al. / Journal of the European Ceramic Society 20(2000)537-544 sified and the porous matrix is only partly in contact to the fiber interfaces. Obviously, there exist radial stress gradients around the fibers with high stress concentra- tion in hot-pressing direction [Fig. 8(b)]. We believe that the processing conditions(e.g phase combination of the matrix material, temperature and the way how the pres- sure was applied, phase transformation occurring in the matrix and the interphase material, etc ) may have great nfluence on the possible formation of strong bonding at the interface. The slurry infiltrated pre-composite con- tains as a matrix amorphous SiO2-rich phase which soft ens at relatively lower temperatures. During hot- pressing, the amorphous phase in the matrix facilitates sintering, versus a viscous flow of ?-Al2O3-particles before transforming to mullite at 1250C. The sintering experiments show that mullite formation does occur at temperatures >1250C only. This means that the amor- phous phase is present throughout the whole processing ne and may lead to strong bonding with the interface 3.3. C/ZrOr-interfaces A further approach in order to achieve damage toler ant composites, weakening of fiber/matrix-bonding and elimination of the process-related stresses between fibers and matrix has been established by using the fugitive coating concept. These fiber/matrix-interfaces consist of C/ZrO2 double layers and are deposited by CVD. Since Fig 9. Scanning electron image of interface in CIZror-coated mullite hot-pressing was performed in argon atmosphere, the atrix composites (a) and 3-point-bending stress-str urve of this composite carbon layer acted as a buffer zone and that way pre- vented exaggerated interfacial fiber/matrix bonding and Iso the formation of compressive stresses at the matrix/ sipating crack deflection process to take place. There fiber-interfaces in hot-pressing direction. TEM investi- fore, the gap thickness should be at the same order as gations on as-hot-pressed samples confirmed the exis- surface roughness of fiber and matrix. tence of thin(10-100 nm)amorphous carbon layers surrounding fibers. After burn-out of the carbon layers in air thin gaps between fiber and zro2-layer are forme formed 4 Conclusions [Fig 9(a)]. Scanning electron microscopy observations and stress-strain curves of the C/ZrO2-interface comp The ZrO2 deposition was described by a first order sites demonstrate that an extensive fiber pull-out, thus reaction. Deposition temperatures about 300C was damage tolerant behavior is achieved [Fig 9(b)]. Crack found to be optimum for ZrO2-coating on fiber fabrics, deflection occurs at the gap, while the ZrO2-layer compared to higher temperatures used with dense remains attached to the matrix. Apparently, the thick wafers. Use of Zr-acetylacetonate resulted in porous ness of the fugitive coating plays a major role for the coatings after hot-pressing the oxided oxide fiber-rein damage tolerant behavior of the composite. a broad forced composites. C/ZrO2-double coating are achieved gap surrounding the fiber can not work, since the by successive CVD-coating with propane and Zr-acet required load-transfer between matrix and fibers does ylacetonate not occur On the other hand, if the gap is too narrow, Composites are fabricated by infiltration of coated many local contact points between the matrix and fiberfiber yarns with pre-mullite slurry and consequently develop, leading to a strong fiber/matrix bonding and hot-pressing the infiltrated prepegs. Porous ZrOz-coat an associated brittle fracture behavior of the composite. ing at interface of mullite/aluminosilicate fiber-rein ur present studies have shown that carbon layers with forced composites displayed no fiber-pull-out. C/ZrO2, thickness ranging between about 10 and 100 nm are in turn, resulted in damage tolerant fracture of the suitable for fugitive coatings. Obviously the ideal gap composites. The mechanical and microstructural obser- thickness is that which enables the fiber to remain vations of the composites at RT and at 1200C after 2 h uncompressed, at the same time, allows energy dis- heat-treatment showed that a thickness of 10 nm for the
si®ed and the porous matrix is only partly in contact to the ®ber interfaces. Obviously, there exist radial stress gradients around the ®bers with high stress concentration in hot-pressing direction [Fig. 8(b)]. We believe that the processing conditions (e.g. phase combination of the matrix material, temperature and the way how the pressure was applied, phase transformation occurring in the matrix and the interphase material, etc.) may have great in¯uence on the possible formation of strong bonding at the interface. The slurry in®ltrated pre-composite contains as a matrix amorphous SiO2-rich phase which softens at relatively lower temperatures. During hotpressing, the amorphous phase in the matrix facilitates sintering, versus a viscous ¯ow of g-Al2O3-particles before transforming to mullite at 1250C. The sintering experiments show that mullite formation does occur at temperatures 51250C only. This means that the amorphous phase is present throughout the whole processing line and may lead to strong bonding with the interface. 3.3. C/ZrO2-interfaces A further approach in order to achieve damage tolerant composites, weakening of ®ber/matrix-bonding and elimination of the process-related stresses between ®bers and matrix has been established by using the fugitive coating concept. These ®ber/matrix-interfaces consist of C/ZrO2 double layers and are deposited by CVD. Since hot-pressing was performed in argon atmosphere, the carbon layer acted as a buer zone and that way prevented exaggerated interfacial ®ber/matrix bonding and also the formation of compressive stresses at the matrix/ ®ber-interfaces in hot-pressing direction. TEM investigations on as-hot-pressed samples con®rmed the existence of thin (10±100 nm) amorphous carbon layers surrounding ®bers. After burn-out of the carbon layers in air thin gaps between ®ber and ZrO2-layer are formed [Fig. 9(a)]. Scanning electron microscopy observations and stress±strain curves of the C/ZrO2-interface composites demonstrate that an extensive ®ber pull-out, thus damage tolerant behavior is achieved [Fig. 9(b)]. Crack de¯ection occurs at the gap, while the ZrO2-layer remains attached to the matrix. Apparently, the thickness of the fugitive coating plays a major role for the damage tolerant behavior of the composite. A broad gap surrounding the ®ber can not work, since the required load-transfer between matrix and ®bers does not occur. On the other hand, if the gap is too narrow, many local contact points between the matrix and ®ber develop, leading to a strong ®ber/matrix bonding and an associated brittle fracture behavior of the composite. Our present studies have shown that carbon layers with thickness ranging between about 10 and 100 nm are suitable for fugitive coatings. Obviously the ideal gap thickness is that which enables the ®ber to remain uncompressed, at the same time, allows energy dissipating crack de¯ection process to take place. Therefore, the gap thickness should be at the same order as surface roughness of ®ber and matrix. 4. Conclusions The ZrO2 deposition was described by a ®rst order reaction. Deposition temperatures about 300C was found to be optimum for ZrO2-coating on ®ber fabrics, compared to higher temperatures used with dense wafers. Use of Zr-acetylacetonate resulted in porous coatings after hot-pressing the oxided/oxide ®ber-reinforced composites. C/ZrO2-double coating are achieved by successive CVD-coating with propane and Zr-acetylacetonate. Composites are fabricated by in®ltration of coated ®ber yarns with pre-mullite slurry and consequently hot-pressing the in®ltrated prepegs. Porous ZrO2-coating at interface of mullite/aluminosilicate ®ber-reinforced composites displayed no ®ber-pull-out. C/ZrO2, in turn, resulted in damage tolerant fracture of the composites. The mechanical and microstructural observations of the composites at RT and at 1200C after 2 h heat-treatment showed that a thickness of 10 nm for the Fig. 9. Scanning electron image of interface in C/ZrO2-coated mullite ®ber mullite matrix composites (a) and 3-point-bending stress-strain curve of this composite. K. Nubian et al. / Journal of the European Ceramic Society 20 (2000) 537±544 543
K Nubian et al. /Journal of the European Ceramic Society 20(2000)537-544 gap was insufficient for providing an effective long-term 2. Scheffier, w. Multilayer fiber coating. US Patent No. 5,445 106 damage-tolerance. While a gap thickness of 100 nm can be regarded as maximum, thus gap ranging between 50 3. Stolle. R. and Wahl. G. Direct transfer of kinetic data from a and 100 nm can be recommended for successful results microbalance equipment into a tube reactor for CVD-BN on Sic Acknowledgement 4. Pulver. M. and Wahl. G. Proc. of the 14th Int. Conf and 1lth European Conf on CVD, 1997, 960-967 5. Fukuda, R, Nagata, S, Negishi, A, Kasuga. Y, and Okuo. T, Ms. Gudrun Paul is thanked for her careful work in pre- Comm. Eur. Communities, Report, 1991 aration of the difficult TEM samples of the composites 6. Brennfleck, K, Fitzer, E and Schoch, G, In Proc. of Sth Europ Conf. CID, ed J.O. Carlsson and J. Lindstrom, Uppsala, 1985 Referen 7. Balog, M.. Schieber. M., Michman, M. and Patai, S. Thin Solid Films,1977,47,109-120. 1. Ogbuki, L. U.J. T, A porous oxidation-resistant fiber coating 8. Carman, P. C, Flow of Gases through Pore dia. butter for CMC interphase Ceram. Eng. Sci. Proc., 1995. 6(4), 497-505 worths Scientific Publications. London. 1956
gap was insucient for providing an eective long-term damage-tolerance. While a gap thickness of 100 nm can be regarded as maximum, thus gap ranging between 50 and 100 nm can be recommended for successful results. Acknowledgements Ms. Gudrun Paul is thanked for her careful work in preparation of the dicult TEM samples of the composites. References 1. Ogbuki, L. U. J. T., A porous oxidation-resistant ®ber coating for CMC interphase. Ceram. Eng. Sci. Proc., 1995, 6(4), 497±505. 2. Scheer, W., Multilayer ®ber coating, US Patent No. 5,445,106, 1996. 3. Stolle, R. and Wahl, G., Direct transfer of kinetic data from a microbalance equipment into a tube reactor for CVD-BN on SiC fabrics. Advanced Materials, Chem. Vap. Deposition, 2000, in press. 4. Pulver, M. and Wahl, G. Proc. of the 14th Int. Conf. and 11th European Conf. on CVD, l997, 960±967. 5. Fukuda, R., Nagata, S., Negishi, A., Kasuga, Y., and Okuo. T., Comm. Eur. Communities, Report, 1991. 6. Brenn¯eck, K., Fitzer, E. and Schoch, G., In Proc. of 5th Europ. Conf. CVD, ed. J. O. Carlsson and J. LindstroÈm, Uppsala, 1985, pp. 63±70. 7. Balog, M., Schieber, M., Michman, M. and Patai, S., Thin Solid Films, 1977, 47, 109±120. 8. Carman, P. C., Flow of Gases through Porous Media. Butterworths Scienti®c Publications, London, 1956. 544 K. Nubian et al. / Journal of the European Ceramic Society 20 (2000) 537±544