E驅≈3S Journal of the European Ceramic Society 21(2001)251-261 elsevier. com/locate/jeurc Impact response and mechanical behavior of 3-D ceramic matrix composites Hsien-Kuang Liu", Chuan-Cheng Huang Department of Mechanical Engineering, Feng-Chia University, 100 Wenhua Road, 407 Taichung, Taiwan Received 29 September 1999: received in revised form 24 May 2000: accepted 4 June 2000 Abstract This paper examines the impact response, compressive strength, and flexural strength of three-dimensional carbon fiber rein- forced ceramics. The composite was fabricated by combination of the pressure infiltration method and sol-gel processing, using the mixture of silica sol and alumina particles. The effects of sol viscosity on impact response and flexural strength, and infiltration pressure on compressive strength are investigated. Impact tests were carried out using an Izod impact testing machine. It is found that impact energy of the specimens increases linearly with sol viscosity. The sol viscosity affects impact response by two factors, interfacial strength and volume fraction of silica. Higher sol viscosity leads to weaker interface and more residual silica, resulting in higher impact energy. Flexural strength decreases with sol viscosity according to an exponential decay function because higher viscosity leads to weak interface as well as lower flexural strength. Compressive strength increases with infiltration pressure, which follows a parabolic function. Higher infiltration pressure enhances infiltration of the mixture, resulting in dense matrix and strong interface. As a result, the composite shows a higher compressive strength due to less buckling of longitudinal fibers strongly con- fined by neighboring dense matrix and transverse fiber bundles. The compressive stress-strain history has been examined and related to how the specimens respond to the compression. c 2001 Elsevier Science Ltd. All rights reserved. Keywords: Alumino-silicate matrix; Carbon fibre; Composites; Mechanical properties; Sol-gel methods 1. Introduction process compared to that of a slurry. 4 The latter demonstrated that the slurry during consolidation must To fabricate ceramic matrix composites by combina- have a sufficiently high viscosity to prevent sedimenta ion of pressure infiltration method and sol-gel proces- tion and lower packing density. However, in our studies sing takes advantages of both methods. -3 The for the short fiber composite it was proved that high advantages include variety of reinforcement, low densi- viscosity of the mixture of sol and ceramic particles fication temperature, low shrinkage, and reduced drying leads to the reaction of sol with ceramic particles, and stresses. Authors have fabricated three-dimensional (3 herefore more sedimentation and low packing density D )green ceramic matrix composites' and short fiber Therefore, one of the goals of this work is to study the ceramic matrix composites by combination of both influence of sol viscosity on microstructure and mechan methods, using the mixture of sol and ceramic particles. ical properties of 3-D ceramic matrix composites In the former study, interparticle forces and particle size Recently, major progress has been made in the devel- distribution are the major factors that influence green pment of 3-D ceramic matrix composites. The driving density and drying stresses, while in the latter study, forces behind the development of these materials have sedimentation, infiltration rate, and viscosity of mixture been the need for high strength and high energy absorp- of sol and ceramic particles are the major factors that tion under multidirectional loadings, and conformity for influence mechanical properties of composites. Further- shape forming. The preform used in the composite cre- more, the mixture of sol and ceramic particles was found ates 3-D networks of reinforcing fibers that eliminate to have different behavior during the consolidation weak planes and prevent the material from planar type of failure. such as delamination. In the absence of weak planes, the impact energy could be dissipated in a more E-mail address: hkliu(@fcu.edu. tw(H.-K. Liu) localized area 6 Whether or not a 3-D ceramic matri 0955-2219/01/S. see front matter C 2001 Elsevier Science Ltd. All rights reserved PII:S0955-2219(00)00181-3
Impact response and mechanical behavior of 3-D ceramic matrix composites Hsien-Kuang Liu *, Chuan-Cheng Huang Department of Mechanical Engineering, Feng-Chia University, 100 Wenhwa Road, 407 Taichung, Taiwan Received 29 September 1999; received in revised form 24 May 2000; accepted 4 June 2000 Abstract This paper examines the impact response, compressive strength, and ¯exural strength of three-dimensional carbon ®ber reinforced ceramics. The composite was fabricated by combination of the pressure in®ltration method and sol±gel processing, using the mixture of silica sol and alumina particles. The eects of sol viscosity on impact response and ¯exural strength, and in®ltration pressure on compressive strength are investigated. Impact tests were carried out using an Izod impact testing machine. It is found that impact energy of the specimens increases linearly with sol viscosity. The sol viscosity aects impact response by two factors, interfacial strength and volume fraction of silica. Higher sol viscosity leads to weaker interface and more residual silica, resulting in higher impact energy. Flexural strength decreases with sol viscosity according to an exponential decay function because higher viscosity leads to weak interface as well as lower ¯exural strength. Compressive strength increases with in®ltration pressure, which follows a parabolic function. Higher in®ltration pressure enhances in®ltration of the mixture, resulting in dense matrix and strong interface. As a result, the composite shows a higher compressive strength due to less buckling of longitudinal ®bers strongly con- ®ned by neighboring dense matrix and transverse ®ber bundles. The compressive stress-strain history has been examined and related to how the specimens respond to the compression. # 2001 Elsevier Science Ltd. All rights reserved. Keywords: Alumino-silicate matrix; Carbon ®bre; Composites; Mechanical properties; Sol±gel methods 1. Introduction To fabricate ceramic matrix composites by combination of pressure in®ltration method and sol±gel processing takes advantages of both methods.1ÿ3 The advantages include variety of reinforcement, low densi- ®cation temperature, low shrinkage, and reduced drying stresses. Authors have fabricated three-dimensional (3- D) green ceramic matrix composites1 and short ®ber ceramic matrix composites2 by combination of both methods, using the mixture of sol and ceramic particles. In the former study, interparticle forces and particle size distribution are the major factors that in¯uence green density and drying stresses; while in the latter study, sedimentation, in®ltration rate, and viscosity of mixture of sol and ceramic particles are the major factors that in¯uence mechanical properties of composites. Furthermore, the mixture of sol and ceramic particles was found to have dierent behavior during the consolidation process3 compared to that of a slurry.4 The latter demonstrated that the slurry during consolidation must have a suciently high viscosity to prevent sedimentation and lower packing density. However, in our studies for the short ®ber composite it was proved that high viscosity of the mixture of sol and ceramic particles leads to the reaction of sol with ceramic particles, and therefore more sedimentation and low packing density. Therefore, one of the goals of this work is to study the in¯uence of sol viscosity on microstructure and mechanical properties of 3-D ceramic matrix composites. Recently, major progress has been made in the development of 3-D ceramic matrix composites. The driving forces behind the development of these materials have been the need for high strength and high energy absorption under multidirectional loadings, and conformity for shape forming.5 The preform used in the composite creates 3-D networks of reinforcing ®bers that eliminate weak planes and prevent the material from planar type of failure, such as delamination. In the absence of weak planes, the impact energy could be dissipated in a more localized area.6 Whether or not a 3-D ceramic matrix 0955-2219/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved. PII: S0955-2219(00)00181-3 Journal of the European Ceramic Society 21 (2001) 251±261 www.elsevier.com/locate/jeurceramsoc * Corresponding author. E-mail address: hkliu@fcu.edu.tw (H.-K. Liu)
H.K. Liu, C -C. Huang/Journal of the European Ceramic Society 21(2001)251-261 composite is more impact resistant depends not only on pressure infiltration and sol-gel methods. The effect of the constituents used but, more important, on how the sol viscosity and infiltrate n pressure on mechanical network is constructed and how the impact load is properties of composites is studied. The correlation applied. It is therefore the goal of this research to char- between mechanical properties and damage modes of acterize the impact behavior of the materials. composites is investigated A review of literature indicates that most previous research was dedicated to the impact behavior of uni directional or 2-D ceramic composites, but very little 2. Experimental procedure as been done on that of 3-D ceramic composites. The major reason is the difficulty in preparing specimens 2. 1. Materials uitable for impact tests. For 3-D ceramic composites under Charpy impact testing, several interesting con The silica sol was prepared by the following recipe in clusions were proposed. First, the matrix is divided by which tetraethyl-orthosilicate (TEOS), ethanol, deio- the 3-D network structure and increases dynamic tough- nized water and 7 wt. of HNO3 were mixed and stir ness of the composite. Second, the compliant fiber net- red at a volume ratio of 1: 1: 1. 6: 0.06 to obtain 175 ml work raises the resistance to crack propagation. Third, sol. In order to study the effect of sol viscosity on the coating of fiber improves impact resistance. Fourth, mechanical properties of 3-D ceramic matrix compo- fiber pull-out increases impact resistance. Damaged by sites, four kinds of viscosity are selected. The sol with delamination, 2-D C/C composite has higher impact viscosity 3.78 cP was obtained according to the proce rupture work than that of 3-D C/C composite For cross- dure above, while sols with viscosities 4.17 and 8.5 cP ply ceramic composites under drop-weight impact, high are prepared by evaporating 250 and 350 ml sols, temperature environment and applied tensile loading respectively, to 175 ml. The lowest viscosity 3.02 cP was make the damage of the composite more severe. This prepared by changing volume ratio of TEOS from 1.0 to indicates that different testing conditions significantly 0. 25. 17 g of alumina particles(a-Al2O3, average dia- influence the impact response of the materials meter d=0.3 um, purity =99.99%, p=3.965 g/cm To deeply understand impact response of the materi- ProChem Co, USA)were added to the sol to prepare ls, static tests are necessary. For ceramic matrix com the infiltrate. The infiltrate was further mixed both in an posites under compression, o dilatational fracture within ultrasonic bath and electromagnetic blender in turns the matrix dominates composite failure at low confine until the particles in the infiltrate were dispersed un ment pressures, while fiber kinking caused by high con- formly and well wetted. The carbon fiber used in the 3- finement pressure indicates shear dominated mechanisms. D preform is Toho T-300 fiber. The properties of fiber The processing and mechanical properties of 3-D car- are as follows: tensile strength 3920 MPa, tensile mod bon/SiC and carbon/Si N4 at high temperature under ulus 235 GPa, diameter(d)=7 um, p=1.77 g/cm3, Sic were reviewed. A brocesses of cross-weave carbon/ 12000 filaments in a fiber tow. The 3-D preform is a vacuum and fracture xural strengths of 3-D carbon/ three-axis orthogonal structure. Fiber volume fraction SiC at room and high temperatures are much lower of the 3-D preform is 38.2%. A one-shuttle weaving than those unidirectional carbon/SiC composites. The scheme is adopted to fabricate the 3-D fabric. 6 high-temperature strength is higher than that at room In order to predict volume fraction of silica in the temperature for 3-D carbon/SiC, while the trend is silica/alumina matrix, solid content in the silica sol was reverse for 3-D carbon/ Si3N4. The fracture process of measured. A drying and sintering experiment was con cross-weave carbon/SiC does not involve cracking by a ducted on the silica sol with four kinds of viscosity. The single dominant crack but occurs by the development of silica sol was dried at 80C to obtain dried gel. There- multiple transverse fractures of groups of four to eight after the dried gel was sintered at 1300C for 2 h to yield fibers followed by longitudinal cracking at the interface. silica. The volume of silica sol was controlled as 175 cm3 The cracks were temporary arrested by the internal(ml), and the weight of dried gel and silica were mea- voids until specimens fail and there is extensive fiber sured. Therefore volume of silica can be obtained with its debonding and pull-out. Although ceramic matrix com- density of 2.2 g/cm. These data are shown in Table posites reinforced by carbon fibers are vulnerable to degradation in an oxidizing environment at relatively 2. 2. Fabrication low temp hey provide valuable information for he material used in oxygen free environment as well as A pressure infiltration apparatus was used to perform for other material systems the processing of 3-D ceramic matrix composites. The Based on the knowledge we have gained, the prime apparatus consists of a cylinder with an inner diameter objective of this work is aimed at evaluating impact of 50 mm, a base cavity for collecting the liquid, a response, flexural strength, and compressive strength of plunger, and a filter assembly. The filter assembly con- 3-D woven ceramic matrix composites fabricated by sists of nitrate cellulose membrane filter paper with pore
composite is more impact resistant depends not only on the constituents used but, more important, on how the network is constructed and how the impact load is applied. It is therefore the goal of this research to characterize the impact behavior of the materials. A review of literature indicates that most previous research was dedicated to the impact behavior of unidirectional or 2-D ceramic composites,7ÿ9 but very little has been done on that of 3-D ceramic composites. The major reason is the diculty in preparing specimens suitable for impact tests. For 3-D ceramic composites under Charpy impact testing,9 several interesting conclusions were proposed. First, the matrix is divided by the 3-D network structure and increases dynamic toughness of the composite. Second, the compliant ®ber network raises the resistance to crack propagation. Third, the coating of ®ber improves impact resistance. Fourth, ®ber pull-out increases impact resistance. Damaged by delamination, 2-D C/C composite has higher impact rupture work than that of 3-D C/C composite. For crossply ceramic composites under drop-weight impact,7 high temperature environment and applied tensile loading make the damage of the composite more severe. This indicates that dierent testing conditions signi®cantly in¯uence the impact response of the materials. To deeply understand impact response of the materials, static tests are necessary. For ceramic matrix composites under compression,10 dilatational fracture within the matrix dominates composite failure at low con®nement pressures, while ®ber kinking caused by high con- ®nement pressure indicates shear dominated mechanisms. The processing and mechanical properties of 3-D carbon/SiC and carbon/Si3N4 at high temperature under vacuum11 and fracture processes of cross-weave carbon/ SiC were reviewed.12 Flexural strengths of 3-D carbon/ SiC at room and high temperatures are much lower than those unidirectional carbon/SiC composites. The high-temperature strength is higher than that at room temperature for 3-D carbon/SiC, while the trend is reverse for 3-D carbon/Si3N4. The fracture process of cross-weave carbon/SiC does not involve cracking by a single dominant crack but occurs by the development of multiple transverse fractures of groups of four to eight ®bers followed by longitudinal cracking at the interface. The cracks were temporary arrested by the internal voids until specimens fail and there is extensive ®ber debonding and pull-out. Although ceramic matrix composites reinforced by carbon ®bers are vulnerable to degradation in an oxidizing environment at relatively low temperatures, they provide valuable information for the material used in oxygen free environment as well as for other material systems. Based on the knowledge we have gained, the prime objective of this work is aimed at evaluating impact response, ¯exural strength, and compressive strength of 3-D woven ceramic matrix composites fabricated by pressure in®ltration and sol±gel methods. The eect of sol viscosity and in®ltration pressure on mechanical properties of composites is studied. The correlation between mechanical properties and damage modes of composites is investigated. 2. Experimental procedure 2.1. Materials The silica sol was prepared by the following recipe in which tetraethyl-orthosilicate (TEOS), ethanol, deionized water and 7 wt.% of HNO3 were mixed and stirred at a volume ratio of 1:1:1.6:0.06 to obtain 175 ml sol. In order to study the eect of sol viscosity on mechanical properties of 3-D ceramic matrix composites, four kinds of viscosity are selected. The sol with viscosity 3.78 cP was obtained according to the procedure above, while sols with viscosities 4.17 and 8.5 cP are prepared by evaporating 250 and 350 ml sols, respectively, to 175 ml. The lowest viscosity 3.02 cP was prepared by changing volume ratio of TEOS from 1.0 to 0.25. 17 g of alumina particles (a-Al2O3, average diameter d 0:3 mm, purity=99.99%, =3.965 g/cm3 , ProChem Co., USA) were added to the sol to prepare the in®ltrate. The in®ltrate was further mixed both in an ultrasonic bath and electromagnetic blender in turns until the particles in the in®ltrate were dispersed uniformly and well wetted. The carbon ®ber used in the 3- D preform is Toho T-300 ®ber. The properties of ®ber are as follows: tensile strength 3920 MPa, tensile modulus 235 GPa, diameter (d)=7 mm, =1.77 g/cm3 , 12000 ®laments in a ®ber tow. The 3-D preform is a three-axis orthogonal structure. Fiber volume fraction of the 3-D preform is 38.2%. A one-shuttle weaving scheme is adopted to fabricate the 3-D fabric.6 In order to predict volume fraction of silica in the silica/alumina matrix, solid content in the silica sol was measured. A drying and sintering experiment was conducted on the silica sol with four kinds of viscosity. The silica sol was dried at 80C to obtain dried gel. Thereafter the dried gel was sintered at 1300C for 2 h to yield silica. The volume of silica sol was controlled as 175 cm3 (ml), and the weight of dried gel and silica were measured. Therefore volume of silica can be obtained with its density of 2.2 g/cm3 . These data are shown in Table 1. 2.2. Fabrication A pressure in®ltration apparatus was used to perform the processing of 3-D ceramic matrix composites. The apparatus consists of a cylinder with an inner diameter of 50 mm, a base cavity for collecting the liquid, a plunger, and a ®lter assembly. The ®lter assembly consists of nitrate cellulose membrane ®lter paper with pore 252 H.-K. Liu, C.-C. Huang / Journal of the European Ceramic Society 21 (2001) 251±261
H.K. Liu, C -C Huang /Journal of the European Ceramic Society 21(2001)251-261 Table I Summary of mechanical properties of 3-D ceramic matrix Specimen Sol Infiltration Green Impact energy Impact Compression Compression Compression vIscosIty pressure composite energy/area strain (cP)a density (% (%) (MPa) 1.25 MPa 25MPa74.1 19. 8702870 401 455.59 0.775 1059 0.50MPa271.9 021.00MPa76.8 enti-Poise. unit of sol viscosity Mega Pascal, unit of infiltration pressure size of 0.I um, a stainless steel wire cloth, and a perfo- mm in width(b), and 7 mm in thickness (h1). The speci rated solid steel disk. The filter assembly was placed on mens were clamped perpendicularly at one end, and left the top of the base cavity. The stainless steel wire cloth 30 mm free for the impact. The mass of the impactor, a was used to keep the filter paper from rupturing Cylin- pendulum, was 4.54 kg. After the pendulum impacted der, filter assembly, and base were tightly sealed by he specimen at top 10 mm, a pointer can read the thread and O-ring. The 3-D fabric was cut as a rectangle impact energy on the machine based on the difference of with dimension 3.8 x25x0.85 cm, and then inserted the height of the pendulum before and after the impact into a rectangular hole within a circular acrylic disk Four different specimens(IA, IB, IC, and ID)were with 5.0 cm in diameter and 0.85 cm thick. The fabric impacted, which were fabricated by sol viscosity of 3.02, and disk were put in the cylinder and attached to the fil- 3.78, 4.17, and 8.50 cP, respectively. First letter"I ter assembly closely. The infiltrate was then poured into the specimen number means impact the cylinder and an mts testing machine set at constant pressure mode was used to provide the infiltration pres- 2.3.2. Compressive test sure on the plunger. Four kinds of infiltration pressure The dimension of specimens for compression is were used: 0.5, 0.75, 1.0 and 1.25 MPa. The infiltration 37. 5x24x7 mm(L x bx h). The compression loading is procedure stops when the plunger reaches the upper sur- applied in the length direction, and both ends are grip- face of the fabric Using pressure infiltration, the thick 3- ped by the fixture for 15 the through-thickness D fabric can be efficiently infiltrated and consolidated direction. The test is conducted by Instron 4468 testing via silica sol/alumina particle route in a single step. machine and crosshead speed is set at 0. 1 mm/s. Com After infiltration, the green composite is dried in an pressive strength o and compressive modulus E are cal oven at 60C and humidity 95% for 24 h. In order to culated by the following formula raise solid content, the dried composite is soaked into silica sol under vacuum for 3 h and then dried for 24 h P (1) Densification of the composite was conducted by hot pressing furnace(FCPHP-R-5, FRET-20, High Multi 5000, Multi-purpose High Temperature Furnace, E PL Japan). Conditions for hot pressing are temperature 1600C and pressure 10 MPa for 1f h by flowing nitro- gen gas. After hot pressing, mechanical properties and where P is maximum load, P/8 is the slope on the com microstructure of composites were evaluated pressive load-displacement curve at final stage. Four different specimens PA, PB, PC, and PD were com 23. Characterization pressed, which were fabricated by infiltration pressure of 0.50, 0.75, 1.0, and 1. 25 MPa, respectively 2.3.1. Impact test The impact tests were conducted using an Izod test 2.3.3. Flexural test machine(Model: TMI-43-01, Test Machine Co, USA) The dimension of specimens for three-point bending The dimension of specimens is 40 mm in length(L), 12 test is 40x 12x7 mm(Lx). The span is 30 mm
size of 0.1 mm, a stainless steel wire cloth, and a perforated solid steel disk. The ®lter assembly was placed on the top of the base cavity. The stainless steel wire cloth was used to keep the ®lter paper from rupturing. Cylinder, ®lter assembly, and base were tightly sealed by thread and O-ring. The 3-D fabric was cut as a rectangle with dimension 3.82.50.85 cm, and then inserted into a rectangular hole within a circular acrylic disk with 5.0 cm in diameter and 0.85 cm thick. The fabric and disk were put in the cylinder and attached to the ®lter assembly closely. The in®ltrate was then poured into the cylinder and an MTS testing machine set at constant pressure mode was used to provide the in®ltration pressure on the plunger. Four kinds of in®ltration pressure were used: 0.5, 0.75, 1.0 and 1.25 MPa. The in®ltration procedure stops when the plunger reaches the upper surface of the fabric. Using pressure in®ltration, the thick 3- D fabric can be eciently in®ltrated and consolidated via silica sol/alumina particle route in a single step. After in®ltration, the green composite is dried in an oven at 60C and humidity 95% for 24 h. In order to raise solid content, the dried composite is soaked into silica sol under vacuum for 3 h and then dried for 24 h. Densi®cation of the composite was conducted by a hot pressing furnace (FCPHP-R-5, FRET-20, High Multi 5000, Multi-purpose High Temperature Furnace, Japan). Conditions for hot pressing are temperature 1600C and pressure 10 MPa for 1 1 2 h by ¯owing nitrogen gas. After hot pressing, mechanical properties and microstructure of composites were evaluated. 2.3. Characterization 2.3.1. Impact test The impact tests were conducted using an Izod test machine (Model: TMI-43-01, Test Machine Co., USA). The dimension of specimens is 40 mm in length (L), 12 mm in width (b), and 7 mm in thickness (h). The specimens were clamped perpendicularly at one end, and left 30 mm free for the impact. The mass of the impactor, a pendulum, was 4.54 kg. After the pendulum impacted the specimen at top 10 mm, a pointer can read the impact energy on the machine based on the dierence of the height of the pendulum before and after the impact. Four dierent specimens (IA, IB, IC, and ID) were impacted, which were fabricated by sol viscosity of 3.02, 3.78, 4.17, and 8.50 cP, respectively. First letter ``I'' in the specimen number means impact. 2.3.2. Compressive test The dimension of specimens for compression is 37.5247 mm (L b h). The compression loading is applied in the length direction, and both ends are gripped by the ®xture for 15 mm in the through-thickness direction. The test is conducted by Instron 4468 testing machine and crosshead speed is set at 0.1 mm/s. Compressive strength s and compressive modulus E are calculated by the following formula P bh 1 E P L bh 2 where P is maximum load, P= is the slope on the compressive load±displacement curve at ®nal stage. Four dierent specimens PA, PB, PC, and PD were compressed, which were fabricated by in®ltration pressure of 0.50, 0.75, 1.0, and 1.25 MPa, respectively. 2.3.3. Flexural test The dimension of specimens for three-point bending test is 40127 mm (L b h). The span is 30 mm. Table 1 Summary of mechanical properties of 3-D ceramic matrix composites Specimen type Sol viscosity (cP)a In®ltration pressure Green composite density (%) Flexural strength (MPa)b Flexural modulus (GPa) Impact energy (J) Impact energy/area (J/m2 ) Compression strain (%) Compression strength (MPa) Compression modulus (MPa) BA 3.02 1.25 MPa 78.1 23.92 1.483 BB 3.78 1.25 MPa 74.1 19.82 0.840 BC 4.17 1.25 MPa 73.4 19.03 0.793 BD 8.50 1.25 MPa 70.0 18.27 0.694 IA 3.02 1.25 MPa 78.1 0.598 8813.81 IB 3.78 1.25 MPa 74.1 0.721 9455.59 IC 4.17 1.25 MPa 73.4 0.733 9502.57 ID 8.50 1.25 MPa 70.0 0.775 10590.16 PA 3.02 0.50 MPa2 71.9 3.51 24.97 1.392 PB 3.02 0.75 MPa 73.7 2.35 27.32 1.496 PC 3.02 1.00 MPa 76.8 2.43 31.07 3.042 PD 3.02 1.25 MPa 78.1 3.53 37.69 4.065 a cp, centi-Poise, unit of sol viscosity. b MPa, Mega Pascal, unit of in®ltration pressure. H.-K. Liu, C.-C. Huang / Journal of the European Ceramic Society 21 (2001) 251±261 253
H.K. Liu, C-C. Huang /Journal of the European Ceramic Society 21(2001)251-261 The test is conducted by Instron 4468 testing machine and crosshead speed is set at 0.008 mm/s. Flexural strength o and flexural modulus E are calculated by the 40 following formula e 3 PL E=84bh3 where p is maximum load, P/8 is the initial slope on the load-displacement curve. Four different specimens BA BB, BC, and Bd were tested, which were fabricated by sol viscosity of 3.02, 3.78, 4.17, and 8.50 cP, respec Infiltration pressure(MPa) tively Fig. 1. Compressive strength as a function of infiltration pressure for 3-D ceramic matrix composites. 3. Results and discussion Table I summarizes the effect of infiltration pressure on compressive strength, and sol viscosity on flexural strength and impact response of 3-D ceramic matrix composites (CMC). Green composite density of each pecimen is also listed. The correlation among two pro- cessing parameters and three mechanical properties are discussed as follows 3.l.C As shown in Fig. 1, compressive strength of PA, PB, and PD specimens increases with infiltration pres- sure according to a parabolic function. Compressive stress-strain curves for four specimens are shown in Fig 0.00 2. In the curve for the PD specimen fabricated by the highest infiltration pressure of 1. 25 MPa, the peak value and slope near the peak value lead to its highest com- pressive strength and modulus, respectively. Based on Fig. 2. Stress-strain curves for compression of PA, PB, PC, and PD our observation, the damage of a compressed 3-D CMC is complicated because the dominated damage mode is unclear. The damage may include three interactive expands perpendicular to the direction of compression modes: expansion mode, shear mode, and buckling loading due to Poissons effect and leads to debonding mode. Although the expansion mode was concluded as Both debonding and compressive loading cause buck ne dominated damage mode for most brittle materi- ling of the fiber bundle. As shown in Fig. 2, the gradual ls, o the 3-D fiber network may resist the expansion slope in the curve for PA specimen indicates its lowest and play an important role in compression of the 3-d modulus of 1. 392 GPa among four compressed speci- MC. Therefore the effect of infiltration pressure on mens. Low infiltration pressure prevents infiltration of impressive strength and the related damage modes are alumina particles into fiber bundles. As a result, low investigated y her porosity The lowest infiltration pressure, 0.5 MPa, for PA obtained, resulting in lower modulus and weak inter pecimens leads to its lowest compressive strength of face. Therefore, under compression the crack propa 24.97 MPa. Serious buckling of the fiber bundle in the gates through weak interface and causes debonding compressed PA specimen is shown in Fig. 3a. At the Besides, transverse fiber bundles become compliant due onset of compression, the matrix crack initiates from to poor infiltration. Without strong confinement force surface ceramic layer and propagates into the fiber/ from transverse fiber bundles, serious buckling of the matrix interface. As the crack extends, the matrix longitudinal fiber bundle occurs
The test is conducted by Instron 4468 testing machine and crosshead speed is set at 0.008 mm/s. Flexural strength and ¯exural modulus E are calculated by the following formula 3 2 PL bh2 3 E P L3 4bh3 4 where p is maximum load, P= is the initial slope on the load±displacement curve. Four dierent specimens BA, BB, BC, and BD were tested, which were fabricated by sol viscosity of 3.02, 3.78, 4.17, and 8.50 cP, respectively. 3. Results and discussion Table 1 summarizes the eect of in®ltration pressure on compressive strength, and sol viscosity on ¯exural strength and impact response of 3-D ceramic matrix composites (CMC). Green composite density of each specimen is also listed. The correlation among two processing parameters and three mechanical properties are discussed as follows. 3.1. Compressive strength and damage As shown in Fig. 1, compressive strength of PA, PB, PC, and PD specimens increases with in®ltration pressure according to a parabolic function. Compressive stress±strain curves for four specimens are shown in Fig. 2. In the curve for the PD specimen fabricated by the highest in®ltration pressure of 1.25 MPa, the peak value and slope near the peak value lead to its highest compressive strength and modulus, respectively. Based on our observation, the damage of a compressed 3-D CMC is complicated because the dominated damage mode is unclear. The damage may include three interactive modes: expansion mode, shear mode, and buckling mode. Although the expansion mode was concluded as the dominated damage mode for most brittle materials,10 the 3-D ®ber network may resist the expansion and play an important role in compression of the 3-D CMC. Therefore the eect of in®ltration pressure on compressive strength and the related damage modes are investigated. The lowest in®ltration pressure, 0.5 MPa, for PA specimens leads to its lowest compressive strength of 24.97 MPa. Serious buckling of the ®ber bundle in the compressed PA specimen is shown in Fig. 3a. At the onset of compression, the matrix crack initiates from surface ceramic layer and propagates into the ®ber/ matrix interface. As the crack extends, the matrix expands perpendicular to the direction of compression loading due to Poisson's eect and leads to debonding. Both debonding and compressive loading cause buckling of the ®ber bundle. As shown in Fig. 2, the gradual slope in the curve for PA specimen indicates its lowest modulus of 1.392 GPa among four compressed specimens. Low in®ltration pressure prevents in®ltration of alumina particles into ®ber bundles. As a result, low green density (Table 1) as well as higher porosity is obtained, resulting in lower modulus and weak interface. Therefore, under compression the crack propagates through weak interface and causes debonding. Besides, transverse ®ber bundles become compliant due to poor in®ltration. Without strong con®nement force from transverse ®ber bundles, serious buckling of the longitudinal ®ber bundle occurs. Fig. 1. Compressive strength as a function of in®ltration pressure for 3-D ceramic matrix composites. Fig. 2. Stress-strain curves for compression of PA, PB, PC, and PD specimens. 254 H.-K. Liu, C.-C. Huang / Journal of the European Ceramic Society 21 (2001) 251±261
H.K. Liu, C -C Huang /Journal of the European Ceramic Society 21(2001)251-261 compression Fig 3.(a) Serious buckling of the fiber bundle in the PA specimen compression, (b) intermediate buckling of the fiber bundle in the PB Fig 4.(a)Moderate buckling of the fiber bundle in the PC specimen The infiltration pressure for PB specimens is 0.75 in compression,(b)slight buckling of the fiber bundle in the PD spe. MPa, which is higher than that of PA specimens and results in higher compressive strength of 27.32 MPa and The curve implies the microstructure evolution in the higher modulus of 1.496 GPa. Higher infiltration pres- compressed 3-D ceramic composite. At the beginning of sure results in better infiltration of sol/particle mixture, compression, shear force causes matrix microcracks leading to better packing density of the particles. This inside the fiber bundle. Later, compression loading enhances interfacial strength and increases compressive tends to close the microcracks and leads to gradual strength. As shown in Fig. 3b, intermediate buckling of increase of modulus of the composite indicated by the two fiber bundles in the specimen is observed, which is slope of stress-strain curve before strain of 2.8%. As the less serious than that in the PA specimen. The buckling strain is larger than 2.8%, the matrix begins to expand also causes cracks in inter-bundle regions among long- due to Poissons effect, but at the same time is con- itudinal and transverse fiber bundles, and deformation strained by the transverse fiber bundles, leading to rapid of transverse fiber bundles originally in a rectangular increase of the slope(modulus). Finally, transverse fiber bundles can not provide sufficient confinement force Similarly, the higher infiltration pressure of 1.0 MPa and therefore compression causes failure of the compo- for PC specimens than for PB specimens leads to higher site by the buckling mode. As shown in Fig. 4b, the compressive strength of 31.07 MPa and modulus of longitudinal fiber bundle only slightly buckles because 3.042 GPa. As shown in Fig. 4a, compression causes better confinement force is provided through stronger moderate buckling of the fiber bundle and leads to fiber interface, denser matrix, and 3-D fiber network breakage at the convex side. Few cracks occur in the transverse fiber bundle due to buckling of longitudinal 3. 2. Flexural strength and damage bundle For PD specimens, the infiltration pressure is 1.25 As shown in Fig. 5, flexural strength of BA, BB, BC, MPa, which is the highest among four compressed spe- and BD specimens decreases with silica sol viscosity cimens and results in the highest compressive strength according to an exponential decay function. Flexural of 37.69 MPa and modulus of 4.065 GPa. This result is stress strain curves for BA, BB, BC, and BD specimens due to better matrix strength caused by higher compo- are depicted in Fig. 6. Four curves indicate similar trend site green density. As shown in Fig. 2, the slope for PD except for load history after maximum load, affected by promptly increases when the strain is larger than 2.8%. sol viscosity. a typical flexural damage configuration is
The in®ltration pressure for PB specimens is 0.75 MPa, which is higher than that of PA specimens and results in higher compressive strength of 27.32 MPa and higher modulus of 1.496 GPa. Higher in®ltration pressure results in better in®ltration of sol/particle mixture, leading to better packing density of the particles. This enhances interfacial strength and increases compressive strength. As shown in Fig. 3b, intermediate buckling of two ®ber bundles in the specimen is observed, which is less serious than that in the PA specimen. The buckling also causes cracks in inter-bundle regions among longitudinal and transverse ®ber bundles, and deformation of transverse ®ber bundles originally in a rectangular shape. Similarly, the higher in®ltration pressure of 1.0 MPa for PC specimens than for PB specimens leads to higher compressive strength of 31.07 MPa and modulus of 3.042 GPa. As shown in Fig. 4a, compression causes moderate buckling of the ®ber bundle and leads to ®ber breakage at the convex side. Few cracks occur in the transverse ®ber bundle due to buckling of longitudinal bundle. For PD specimens, the in®ltration pressure is 1.25 MPa, which is the highest among four compressed specimens and results in the highest compressive strength of 37.69 MPa and modulus of 4.065 GPa. This result is due to better matrix strength caused by higher composite green density. As shown in Fig. 2, the slope for PD promptly increases when the strain is larger than 2.8%. The curve implies the microstructure evolution in the compressed 3-D ceramic composite. At the beginning of compression, shear force causes matrix microcracks inside the ®ber bundle. Later, compression loading tends to close the microcracks and leads to gradual increase of modulus of the composite indicated by the slope of stress-strain curve before strain of 2.8%. As the strain is larger than 2.8%, the matrix begins to expand due to Poisson's eect, but at the same time is constrained by the transverse ®ber bundles, leading to rapid increase of the slope (modulus). Finally, transverse ®ber bundles can not provide sucient con®nement force, and therefore compression causes failure of the composite by the buckling mode. As shown in Fig. 4b, the longitudinal ®ber bundle only slightly buckles because better con®nement force is provided through stronger interface, denser matrix, and 3-D ®ber network. 3.2. Flexural strength and damage As shown in Fig. 5, ¯exural strength of BA, BB, BC, and BD specimens decreases with silica sol viscosity according to an exponential decay function. Flexural stress strain curves for BA, BB, BC, and BD specimens are depicted in Fig. 6. Four curves indicate similar trend except for load history after maximum load, aected by sol viscosity. A typical ¯exural damage con®guration is Fig. 3. (a) Serious buckling of the ®ber bundle in the PA specimen in compression, (b) intermediate buckling of the ®ber bundle in the PB specimen in compression. Fig. 4. (a) Moderate buckling of the ®ber bundle in the PC specimen in compression, (b) slight buckling of the ®ber bundle in the PD specimen in compression. H.-K. Liu, C.-C. Huang / Journal of the European Ceramic Society 21 (2001) 251±261 255
H.K. Liu, C -C Huang /Journal of the European Ceramic Society 21(2001)251-261 2086 Fig. 7. Macroscopic failure of specimen under three-point bending. Sol viscosity(cP) Fig. 5. Flexural strength as a function of sol viscosity for 3-D ceramic infiltration for BA specimens, and yield highest compo- site green density as shown in Table 1. This should insure better wetting and contact at the interface and therefore leads to better interfacial strength. Under bending, the crack is difficult to propagate through bet- ter interfac fibers and yields higher flexural strength. As shown in Fig. 8a, better interface leads to bundle pull-out, adhe- sion of matrix on the fibers and fiber fracture. Those damage mo strength of the BA specimen among four bended speci mens The higher viscosity of silica sol for BB than for BA pecimens leads to lower flexural strength of 19. 8 MPa The higher viscosity prevents the infiltration of the matrix materials and results in weaker interface as well as lower flexural strength. As shown in Fig. &b, fiber pull-out is found in the BB specimen, contributing to Displacement(mm) lower flexural strength. Similarly, higher viscosity of Fig. 6. Flexural load-displacement curves A. BB. BC. and BD silica sol for BC specimens than for BB specimens leads to lower flexural strength of 19.03 MPa. sem observa- tion of failure of the BC specimen in Fig. 9a shows extensive debonding and pull-out of fibers in y and z shown in Fig. 7. At the onset of flexural test, linear directions elastic behavior on initial stage of the stress-strain curve The highest viscosity of silica sol 8.50 cP for BD spe ndicates that the ceramic matrix undertakes most of the cimens among four bended specimens results in the loading. As the loading increases, matrix cracks occur lowest strength of 18.27 MPa. It is found in the load- and propagate to the interface. The nature of interface displacement curve(Fig. 6) that the curve rebounds was affected by sol viscosity At maximum load, a lot of after it drops sharply from the maximum load. This fibers break followed by fiber pull-out, resulting in the trend is different from other curves probably due to reduction of flexural loading Due to the constraint of 3- more contribution of 3-D fiber network. The sharp drop D fiber network, the crack can only detour around of the curve from the maximum load infers extensive fibers; therefore, stress-strain curve gradually descends debonding in three directions (x, y, and z) because the and the composite fails in a dissipative manner. The highest sol viscosity would result in the weakest inter- effect of sol viscosity on flexural strength and damage face. Thereafter, the curve ascends because the crack mode of the four specimens is then discussed deflects around 3-D fiber network. Extensive fiber pull- For Ba specimens the viscosity of silica sol is 3.02 cP, out(Fig. 9b)contributes to the local rise of the curve which is the lowest among four bended specimens, until the composite fails leading to the highest flexural strength of 23.92 MPa Low flexural strengths of four bended specimens sug The lowest viscosity of silica sol results in the good gest that the fiber degrade at the processing temperature
shown in Fig. 7. At the onset of ¯exural test, linear elastic behavior on initial stage of the stress±strain curve indicates that the ceramic matrix undertakes most of the loading. As the loading increases, matrix cracks occur and propagate to the interface. The nature of interface was aected by sol viscosity. At maximum load, a lot of ®bers break followed by ®ber pull-out, resulting in the reduction of ¯exural loading. Due to the constraint of 3- D ®ber network, the crack can only detour around ®bers; therefore, stress±strain curve gradually descends and the composite fails in a dissipative manner. The eect of sol viscosity on ¯exural strength and damage mode of the four specimens is then discussed. For BA specimens the viscosity of silica sol is 3.02 cP, which is the lowest among four bended specimens, leading to the highest ¯exural strength of 23.92 MPa. The lowest viscosity of silica sol results in the good in®ltration for BA specimens, and yield highest composite green density as shown in Table 1. This should insure better wetting and contact at the interface and therefore leads to better interfacial strength. Under bending, the crack is dicult to propagate through better interface; instead it tends to break strong carbon ®bers and yields higher ¯exural strength. As shown in Fig. 8a, better interface leads to bundle pull-out, adhesion of matrix on the ®bers, and ®ber fracture. Those damage modes correspond to the highest ¯exural strength of the BA specimen among four bended specimens. The higher viscosity of silica sol for BB than for BA specimens leads to lower ¯exural strength of 19.8 MPa. The higher viscosity prevents the in®ltration of the matrix materials and results in weaker interface as well as lower ¯exural strength. As shown in Fig. 8b, ®ber pull-out is found in the BB specimen, contributing to lower ¯exural strength. Similarly, higher viscosity of silica sol for BC specimens than for BB specimens leads to lower ¯exural strength of 19.03 MPa. SEM observation of failure of the BC specimen in Fig. 9a shows extensive debonding and pull-out of ®bers in y and z directions. The highest viscosity of silica sol 8.50 cP for BD specimens among four bended specimens results in the lowest strength of 18.27 MPa. It is found in the load± displacement curve (Fig. 6) that the curve rebounds after it drops sharply from the maximum load. This trend is dierent from other curves probably due to more contribution of 3-D ®ber network. The sharp drop of the curve from the maximum load infers extensive debonding in three directions (x, y, and z) because the highest sol viscosity would result in the weakest interface. Thereafter, the curve ascends because the crack de¯ects around 3-D ®ber network. Extensive ®ber pullout (Fig. 9b) contributes to the local rise of the curve until the composite fails. Low ¯exural strengths of four bended specimens suggest that the ®ber degrade at the processing temperature Fig. 6. Flexural load-displacement curves for BA, BB, BC, and BD specimens. Fig. 5. Flexural strength as a function of sol viscosity for 3-D ceramic matrix composites. Fig. 7. Macroscopic failure of specimen under three-point bending. 256 H.-K. Liu, C.-C. Huang / Journal of the European Ceramic Society 21 (2001) 251±261
H.K. Liu, C -C Huang /Journal of the European Ceramic Society 21(2001)251-261 15KU1.88KXi8,gN1336 fiber pull-out, breakage, and undulation, (b) failure mode of the BD Fig.8.(a)Failure mode of the BA specimen under bending showing specimen under bending showing extensive fiber pull-out. bundle pull-out, (b) failure mode of the BB specimen in bending showing extensive fiber pull-out face and penetrate the composite instead of detouring through interface. White surface matrix layer still atta- and less fibers oriented in the loading direction, com- ches well with the fiber preform, suggesting good inter- pared to unidirectional composites. I facial bonding. Eventually the specimen broke into two pieces. Figure 1 lb shows smooth broken surface parallel 3.3. Impact behavior to the impact loading direction in the IA specimens This leads to lower impact energy for the lack of local The impact energy of four impacted specimens IA, IB, damage mode for energy absorption. IC, and Id as a function of silica sol viscosity is shown For IB specimens, the medium-low viscosity of silica in Fig. 10. It is found that the impact energy increases sol leads to impact energy of 9.46 kJ/m. As shown in linearly with sol viscosity. The correlation between Fig. 12a, the specimen also broke into two pieces, and impact energy and impact damage modes is discussed white surface matrix layer detaches seriously from the follows fiber preform on left-hand side. Further, separation of low viscosity of silica sol enhances the infiltration of the IB specimen(Fig. 12b) shows a rough crack surface mixture of silica sol and alumina particles according to indicating a damage mode of crack deflection. This local Darcy's law 3 A typical impact damage configuration damage mode contributes to higher impact energy of IB of IA specimens is shown in Fig. lla. At onset of than that of IA impact, the impact loading simultaneously causes fiber For IC specimens, the medium-high viscosity of breakage and matrix cracking at the impact site. The sol leads to higher impact energy of 9.5 kJ/m. The impact causes the crack to propagate across the inter- damage configuration and failure modes are shown
and less ®bers oriented in the loading direction, compared to unidirectional composites.11 3.3. Impact behavior The impact energy of four impacted specimens IA, IB, IC, and ID as a function of silica sol viscosity is shown in Fig. 10. It is found that the impact energy increases linearly with sol viscosity. The correlation between impact energy and impact damage modes is discussed as follows. The lowest viscosity of silica sol for IA specimens results in the lowest impact energy of 8.81 kJ/m2 . The low viscosity of silica sol enhances the in®ltration of the mixture of silica sol and alumina particles according to Darcy's law.13 A typical impact damage con®guration of IA specimens is shown in Fig. 11a. At onset of impact, the impact loading simultaneously causes ®ber breakage and matrix cracking at the impact site. The impact causes the crack to propagate across the interface and penetrate the composite instead of detouring through interface. White surface matrix layer still attaches well with the ®ber preform, suggesting good interfacial bonding. Eventually the specimen broke into two pieces. Figure 11b shows smooth broken surface parallel to the impact loading direction in the IA specimens. This leads to lower impact energy for the lack of local damage mode for energy absorption. For IB specimens, the medium-low viscosity of silica sol leads to impact energy of 9.46 kJ/m2 . As shown in Fig. 12a, the specimen also broke into two pieces, and white surface matrix layer detaches seriously from the ®ber preform on left-hand side. Further, separation of ®ber bundles suggests that interfacial bonding be weaker than that of IA specimens. SEM observation on IB specimen (Fig. 12b) shows a rough crack surface indicating a damage mode of crack de¯ection. This local damage mode contributes to higher impact energy of IB than that of IA. For IC specimens, the medium-high viscosity of silica sol leads to higher impact energy of 9.5 kJ/m2 . The damage con®guration and failure modes are shown in Fig. 8. (a) Failure mode of the BA specimen under bending showing bundle pull-out, (b) failure mode of the BB specimen in bending showing extensive ®ber pull-out. Fig. 9. (a) Failure mode of the BC specimen under bending showing ®ber pull-out, breakage, and undulation, (b) failure mode of the BD specimen under bending showing extensive ®ber pull-out. H.-K. Liu, C.-C. Huang / Journal of the European Ceramic Society 21 (2001) 251±261 257
H.K. Liu, C -C Huang/Journal of the European Ceramic Society 21(2001)251-261 13a and b, respectively. The specimen broke into modes of IC specimens show more debonding for higher pieces, and weak interfacial bonding was observed as energy absorption indicated by separation of fibers in Fig. 13a. Failure The highest viscosity of silica sol for ID specimens results in the highest impact energy of 10.5 kJ/ value is approximately the same as critical rate of release of strain energy (10.3 kJ/m ). GIDm, for 3-D Al2O3/SiO2 from Ref. 9. The highest viscosity prevents good infiltration and results in lowest packing density of alumina particles in the 3-D preform, indicated by the lowest green density shown in Table 1. This leads to weak interface As shown in Fig. 14a, the specimen does mpletely separate. Up g causes crack to partially penetrate the composite. As the Impactor proceeds, matrix cracks propagate through weak interface and this dissipates most of the impact energy before complete breakage of the composite Therefore there are some unbroken fibers remained to connect two broken pieces of specimen. Fig 14b shows extensive fiber pull-out indicated by holes in the matrix and crack deflection indicated by rather rough fracture Fig10. Impact energy as a function of silica sol viscosity for 3-D surface. More damage modes contribute to better ceramic matrix composites energy absorption Fig. 11. Impact damage for the nowing (a) complete Fig. 12. Impact damage for the IB specimen from (a) complete breakage of the composite and Irface layer, and(b) breakage of the composite and separation of surface layer, and(b) remained holes after fiber pull-out
Figs. 13a and b, respectively. The specimen broke into two pieces, and weak interfacial bonding was observed as indicated by separation of ®bers in Fig. 13a. Failure modes of IC specimens show more debonding for higher energy absorption. The highest viscosity of silica sol for ID specimens results in the highest impact energy of 10.5 kJ/m2 . This value is approximately the same as critical rate of release of strain energy (10.3 kJ/m2 ), GIDm, for 3-D Al2O3/SiO2 from Ref. 9. The highest viscosity prevents good in®ltration and results in lowest packing density of alumina particles in the 3-D preform, indicated by the lowest green density shown in Table 1. This leads to weak interface. As shown in Fig. 14a, the specimen does not completely separate. Upon impact, the loading causes crack to partially penetrate the composite. As the impactor proceeds, matrix cracks propagate through weak interface and this dissipates most of the impact energy before complete breakage of the composite. Therefore, there are some unbroken ®bers remained to connect two broken pieces of specimen. Fig. 14b shows extensive ®ber pull-out indicated by holes in the matrix, and crack de¯ection indicated by rather rough fracture surface. More damage modes contribute to better energy absorption. Fig. 10. Impact energy as a function of silica sol viscosity for 3-D ceramic matrix composites. Fig. 11. Impact damage for the IA specimen showing (a) complete breakage of the composite and adhesion of surface layer, and (b) smooth fracture surface. Fig. 12. Impact damage for the IB specimen from (a) complete breakage of the composite and separation of surface layer, and (b) remained holes after ®ber pull-out. 258 H.-K. Liu, C.-C. Huang / Journal of the European Ceramic Society 21 (2001) 251±261
H.K. Liu, C -C Huang /Journal of the European Ceramic Society 21(2001)251-261 (a) 今 Fig 14. Impact damage for the ID specimen from(a) bridging of the Fig 13. Impact damage for the IC specimen from(a) complete break. osite by unbroken fibers, and(b) extensive fiber pull-out age of the composite, and(b)extensive fiber bending and pull-out 3.4. Volume fraction of silica modify the matrix composition. A schematic diagram shown in Fig 15 can be used to summarize the influence The impact damage modes are also influenced by the of silica sol viscosity on matrix composition of 3-D matrix composition which is relative to volume fraction ceramic matrix composites. Fabricated by the lowest sol of silica. To predict the volume fraction of silica in viscosity, more alumina particles can be infiltrated into lica/alumina matrix for the 3-D green composite, the intra-bundle pores among fibers of IA specimens under packing density of alumina particles must be assumed. lower viscous force. When the sol viscosity increases, the The packing density can be either 52.3% via simple particles become more and more difficult to infiltrate the cubic packing or 60% via hexagonal packing, depend- pore, as can be seen in the change of green density in ing on sol viscosity For the sol with lower viscosity, the Table I from IB to ID. After hot-pressing, silica sol can alumina particles can flow easily and packing individu- react with alumina particles and produce mullite matrix, lly during infiltration process; therefore, the particles depending on stoichiometric proportion of alumina to tend to pack densely as a hexagonal packing. The silica. 4 Therefore, it is possible to simultaneously detect empirical data for the volume fraction of silica are silica, alumina, and mullite in the matrix in which their shown in Table 2. For the sol with viscosity of 3.02 and fracture toughness ranks from high to low. 5 For IA 3.78 cP, volume fraction of silica would be 0.65 and specimens, a very small amount of silica(Table 1)may 2.1%, respectively. In contrast, alumina particles in the react with alumina to produce mullite as a thin surface higher viscosity sol would lead to simple cubic packing, layer around alumina grains. Thus, the composition of resulting in volume fraction of silica as 2.68%(4.17 cP) IA would be a large amount of alumina plus little mul- nd28%(8.50cP) lite. For IB specimens, more silica and less alumina tend Based on prediction of volume fraction of silica, the to yield some mullite, however, the residual silica can reaction between silica and alumina can be involved to toughen the matrix and overwhelm the reduction of
3.4. Volume fraction of silica The impact damage modes are also in¯uenced by the matrix composition which is relative to volume fraction of silica. To predict the volume fraction of silica in silica/alumina matrix for the 3-D green composite, the packing density of alumina particles must be assumed. The packing density can be either 52.3% via simple cubic packing or 60% via hexagonal packing, depending on sol viscosity. For the sol with lower viscosity, the alumina particles can ¯ow easily and packing individually during in®ltration process; therefore, the particles tend to pack densely as a hexagonal packing. The empirical data for the volume fraction of silica are shown in Table 2. For the sol with viscosity of 3.02 and 3.78 cP, volume fraction of silica would be 0.65 and 2.1%, respectively. In contrast, alumina particles in the higher viscosity sol would lead to simple cubic packing, resulting in volume fraction of silica as 2.68% (4.17 cP) and 2.8% (8.50 cP). Based on prediction of volume fraction of silica, the reaction between silica and alumina can be involved to modify the matrix composition. A schematic diagram shown in Fig. 15 can be used to summarize the in¯uence of silica sol viscosity on matrix composition of 3-D ceramic matrix composites. Fabricated by the lowest sol viscosity, more alumina particles can be in®ltrated into intra-bundle pores among ®bers of IA specimens under lower viscous force. When the sol viscosity increases, the particles become more and more dicult to in®ltrate the pore, as can be seen in the change of green density in Table 1 from IB to ID. After hot-pressing, silica sol can react with alumina particles and produce mullite matrix, depending on stoichiometric proportion of alumina to silica.14 Therefore, it is possible to simultaneously detect silica, alumina, and mullite in the matrix in which their fracture toughness ranks from high to low.15 For IA specimens, a very small amount of silica (Table 1) may react with alumina to produce mullite as a thin surface layer around alumina grains. Thus, the composition of IA would be a large amount of alumina plus little mullite. For IB specimens, more silica and less alumina tend to yield some mullite, however, the residual silica can toughen the matrix and overwhelm the reduction of Fig. 13. Impact damage for the IC specimen from (a) complete breakage of the composite, and (b) extensive ®ber bending and pull-out. Fig. 14. Impact damage for the ID specimen from (a) bridging of the broken composite by unbroken ®bers, and (b) extensive ®ber pull-out. H.-K. Liu, C.-C. Huang / Journal of the European Ceramic Society 21 (2001) 251±261 259
H.K. Liu, C -C Huang /Journal of the European Ceramic Society 21(2001)251-261 Table 2 Predicted volume fraction of silica in silica/alumina matrix for the 3-D green composite Sol viscosity Volume of Weight of Weight of Volume of Packing density of alumina (%) Volume fraction of silica (%) (cP) silica sol(cm) dried gel(g) silica(g) silica(cm) Cubic Hexagonal Cubic Hexagonal .72 52.3 3.78 13.6 11.62 52.3 13.54 12.12 51 52.3 0.6 mixture, but also the matrix composition. The ease of the flow of the mixture under lower viscosity led to denser particle packing both in inter-yarn and intra yarn pores, and this resulted in better interfacial bond- ing after hot-pressing, and lower impact energy. Fur ther, less residual silica and some mullite fabricated by lower sol viscosity caused lower impact energy of com posites due to more brittle nature of mullite than silica With sol viscosity from low to high, the corresponding failure modes of specimens IA, IB, IC, and ID are straight cracks, deflected cracks, debonding, and hot-pressing debonding and pull-out. With more failure modes and energy dissipative mode such as pull-out, ID specimens possess highest impact energy It was found that the compressive strength of the composites increased with infiltration pressure accord ng to a pa function. Higher int assisted the infiltration of the mixture of sol and parti cles into intra-bundle pores, leading to dense matrix with less porosity. It was found that fiber buckling is unavoidable for undulated fibers in compressed 3-D composites. The dense matrix is expected to provide stronger interface and lateral confined force. This redu- Fig. 15. The schematic diagram for mechanisms that influences ces buckling of fibers and contributes to better com impact response of IA, IB, IC, and ID specimens pressive strength. The flexural strength of the composites was found to decrease with sol viscosity according to an exponential decay function. The low sol viscosity leads toughness caused by mullite. Eventually, IB has higher to dense matrix and strong interface, and these result in impact energy than IA. For IC specimens, the compo- high flexural strength because high flexural strength is sition is similar to that of IB except for more residual usually accompanied by strong interface. Furthermore, silica, and this leads to slightly higl her impact energy or specimens under the same infiltration pressure and Ithough there is a great amount of silica in ID speci- sol viscosity the flexural strength was lower than the mens, only few amount of mullite can be produced due compressive strength. This is due to the nature of brittle to a small amount of alumina particles. Therefore, a materials that compression loading inhibits the crack large amount of residual silica leads to the highest propagation, while tensile loading associated with the impact energy of ID specimens flexural test promotes crack propagation It was further found that the improved toughness of he 3-D composites was achieved at the expense of the 4. Conclusions strength. Although higher compressive and flexural trengths were not obtained in this paper, a contribution The effect of processing parameters on impact beha- was made on the influence of processing parameters on vior, compressive strength, and flexural strength of 3-D both strengths and impact performance. Further, the ceramic matrix composites has been examined. The silica findings of correlation between the two strengths and sol viscosity affects impact performance of composites, impact response should provide valuable information not only because it influences the flow of sol/particle for the community of ceramic matrix composites
toughness caused by mullite. Eventually, IB has higher impact energy than IA. For IC specimens, the composition is similar to that of IB except for more residual silica, and this leads to slightly higher impact energy. Although there is a great amount of silica in ID specimens, only few amount of mullite can be produced due to a small amount of alumina particles. Therefore, a large amount of residual silica leads to the highest impact energy of ID specimens. 4. Conclusions The eect of processing parameters on impact behavior, compressive strength, and ¯exural strength of 3-D ceramic matrix composites has been examined. The silica sol viscosity aects impact performance of composites, not only because it in¯uences the ¯ow of sol/particle mixture, but also the matrix composition. The ease of the ¯ow of the mixture under lower viscosity led to denser particle packing both in inter-yarn and intrayarn pores, and this resulted in better interfacial bonding after hot-pressing, and lower impact energy. Further, less residual silica and some mullite fabricated by lower sol viscosity caused lower impact energy of composites due to more brittle nature of mullite than silica. With sol viscosity from low to high, the corresponding failure modes of specimens IA, IB, IC, and ID are straight cracks, de¯ected cracks, debonding, and debonding and pull-out. With more failure modes and energy dissipative mode such as pull-out, ID specimens possess highest impact energy. It was found that the compressive strength of the composites increased with in®ltration pressure according to a parabolic function. Higher in®ltration pressure assisted the in®ltration of the mixture of sol and particles into intra-bundle pores, leading to dense matrix with less porosity. It was found that ®ber buckling is unavoidable for undulated ®bers in compressed 3-D composites. The dense matrix is expected to provide stronger interface and lateral con®ned force. This reduces buckling of ®bers and contributes to better compressive strength. The ¯exural strength of the composites was found to decrease with sol viscosity according to an exponential decay function. The low sol viscosity leads to dense matrix and strong interface, and these result in high ¯exural strength because high ¯exural strength is usually accompanied by strong interface. Furthermore, for specimens under the same in®ltration pressure and sol viscosity the ¯exural strength was lower than the compressive strength. This is due to the nature of brittle materials that compression loading inhibits the crack propagation, while tensile loading associated with the ¯exural test promotes crack propagation. It was further found that the improved toughness of the 3-D composites was achieved at the expense of the strength. Although higher compressive and ¯exural strengths were not obtained in this paper, a contribution was made on the in¯uence of processing parameters on both strengths and impact performance. Further, the ®ndings of correlation between the two strengths and impact response should provide valuable information for the community of ceramic matrix composites. Table 2 Predicted volume fraction of silica in silica/alumina matrix for the 3-D green composite Sol viscosity (cP) Volume of silica sol (cm3 ) Weight of dried gel (g) Weight of silica (g) Volume of silica (cm3 ) Packing density of alumina (%) Volume fraction of silica (%) Cubic Hexagonal Cubic Hexagonal 3.02 175 4.2 3.78 1.72 52.3 0.6 0.88 0.65 3.78 175 13.6 12.36 5.62 52.3 0.6 2.8 2.1 4.17 175 13.44 11.62 5.28 52.3 0.6 2.68 1.97 8.50 175 13.54 12.12 5.51 52.3 0.6 2.8 2.06 Fig. 15. The schematic diagram for mechanisms that in¯uences impact response of IA, IB, IC, and ID specimens. 260 H.-K. Liu, C.-C. Huang / Journal of the European Ceramic Society 21 (2001) 251±261