Pergamon 09567151(9400343-2 Printed in Great Britain HIGH TEMPERATURE DEFORMATION OF AN ALUMINA COMPOSITE REINFORCED WITH SILICON CARBIDE WHISKERS KENONG XIA and TERENCE G. LANGDON Department of Mechanical and Manufacturing Engineering, University of Melbourne, Parkville, Victoria, Australia 3052 and"Departments of Materials Science and Mechanical Engineering, University of Southern California, Los Angeles, CA90089-1453, U.S.A (Received I February 1994; in revised form 20 July 1994) Abstract-Four-point bending creep tests we rried out in air on an alumina matrix composite reinforced with 9.3 vol. of silicon carbide whiskers. Typical three-stage creep was observed. In the temperature range of 1673-1823 K, the composite exhibited an average stress exponent of 3.8. The activation energy for creep was estimated as -820-830 kJ mol-. microstructure of the composite wa haracterized before and after deformation, dislocation networks and other configurations were observed It is concluded that the deformation mechanism consists of intragranular dislocation movement controlled by the lattice diffusion of oxygen ions 1 INTRODUCTION characteristics of an Al,O3 matrix composite ein forced with 9.3 vol. of sic whiskers Ceramic materials are becoming increasingly attrac tive for a wide range of engineering applications because they are generally harder, stronger and 2 EXPERIMENTAL MATERIAL AND PROCEDURES lighter than metals. More importantly, they maintain their high strength at very high temperatures and they The test material, designated Al, O9.3 vol% are more resistant to severe environments. Neverthe- SiC(w), consisted of an alumina matrix (a-Al,O,) less, the potential applications of ceramic materials and 9.3 vol. of silicon carbide whisker reinfor are often limited by their inherent brittleness, suscep- ment (equivalent to 7. 5 wt%). The composite was ibility to sudden catastrophic failure and low ther- fabricated by hot pressing a mixture of high purity mal shock resistance. Recent investigations have Al, O, powder and Sic whiskers without additives demonstrated that it is possible to toughen ceramic The material was produced as hot-pressed discs hav- matrices by adding various composite components: ing a thickness of 3 mm, and rectangular bars we for example, the toughness of composites such as cut from these discs with cross-sections of 2x 3 mm Al,O-SiC [1-3], Si, N SiC [4, 5], Al,O-Zro2[6], and lengths of -50 mm mullite-SiC [7] and SiC-TiC [8] are generally higher Prior to testing, the as-received specimens were han the monolithic matrices. characterized by scanning and transmission electron The creep behavior of ceramics becomes important microscopy in order to determine the grain size of when considering their use in structural applications the matrix, the distribution of the whiskers and the at elevated temperatures. In general, only limited nature of the interfaces between the matrix and the experimental data are available to characterize the whiskers. The grain boundaries of the matrix were creep behavior of composites, and this deficiency revealed by polishing on a series of diamond paste tends to inhibit the establishment of design criteria and then etching in phosphoric acid at 453K for nd the improvement of processing methods >30 min. The whiskers were revealed by etching the There are numerous reports of the high tempera- unpolished surfaces in phosphoric acid at 553K ture creep behavior of monolithic alumina and these for 5-10 min. The samples were coated with gold data are tabulated (9) and analyzed [10] elsewhere. and then examined in a Cambridge Stereoscan S There are also several reports of the creep behavior IV-10 scanning electron microscope operating with of alumina composites [11-18] but there has been no an accelerating voltage of 10 kV. The sample stage systematic investigation of the creep properties and was set at the horizontal position deformation mechanisms occurring in a well For observations of the whisker/ matrix interface, terized alumina composite. Accordingly, the samples were prepared for transmission electron investigation was conducted to determine the microscopy. A slice of 300-500 um thickness was 1421
Pergamon Acta metall, mater. Vol. 43, No. 4, pp. 1421 1427, 1995 Copyright ~ 1995 Elsevier Science Ltd 0956-7151(94)00343-2 Printed in Great Britain. All rights reserved 0956-7151/95 $9.50 + 0.00 HIGH TEMPERATURE DEFORMATION OF AN ALUMINA COMPOSITE REINFORCED WITH SILICON CARBIDE WHISKERS KENONG XIA l and TERENCE G. LANGDON 2 ~Department of Mechanical and Manufacturing Engineering, University of Melbourne, Parkville, Victoria, Australia 3052 and 2Departments of Materials Science and Mechanical Engineering, University of Southern California, Los Angeles, CA 90089-1453, U.S.A. (Received 1 February 1994; in revised Jorm 20 July 1994) Abstract--Four-point bending creep tests were carried out in air on an alumina matrix composite reinforced with 9.3 vol.% of silicon carbide whiskers. Typical three-stage creep was observed. In the temperature range of 1673-1823 K, the composite exhibited an average stress exponent of 3.8. The activation energy for creep was estimated as ~820--830 kJ mol-~. Microstructure of the composite was characterized before and after deformation. Dislocation networks and other configurations were observed in samples deformed to large strains. It is concluded that the deformation mechanism consists of intragranular dislocation movement controlled by the lattice diffusion of oxygen ions. 1. INTRODUCTION Ceramic materials are becoming increasingly attractive for a wide range of engineering applications because they are generally harder, stronger and lighter than metals. More importantly, they maintain their high strength at very high temperatures and they are more resistant to severe environments. Nevertheless, the potential applications of ceramic materials are often limited by their inherent brittleness, susceptibility to sudden catastrophic failure and low thermal shock resistance. Recent investigations have demonstrated that it is possible to toughen ceramic matrices by adding various composite components; for example, the toughness of composites such as A1203-SiC [1-3], Si3N4-SiC [4, 5], A1203-ZrO 2 [6], muilite-SiC [7] and SiC-TiC [8] are generally higher than the monolithic matrices. The creep behavior of ceramics becomes important when considering their use in structural applications at elevated temperatures. In general, only limited experimental data are available to characterize the creep behavior of composites, and this deficiency tends to inhibit the establishment of design criteria and the improvement of processing methods. There are numerous reports of the high temperature creep behavior of monolithic alumina and these data are tabulated [9] and analyzed [10] elsewhere. There are also several reports of the creep behavior of alumina composites [11-18] but there has been no systematic investigation of the creep properties and deformation mechanisms occurring in a well characterized alumina composite. Accordingly, the present investigation was conducted to determine the creep characteristics of an A1203 matrix composite reinforced with 9.3 vol.% of SiC whiskers. 2. EXPERIMENTAL MATERIAL AND PROCEDURES The test material, designated A1203-9.3vol.% SiC(w), consisted of an alumina matrix (~-A1203) and 9.3 vol.% of silicon carbide whisker reinforcement (equivalent to 7.5 wt%). The composite was fabricated by hot pressing a mixture of high purity A1203 powder and SiC whiskers without additives. The material was produced as hot-pressed discs having a thickness of 3 mm, and rectangular bars were cut from these discs with cross-sections of 2 x 3 mm and lengths of ~ J0 ram. Prior to testing, the as-received specimens were characterized by scanning and transmission electron microscopy in order to determine the grain size of the matrix, the distribution of the whiskers and the nature of the interfaces between the matrix and the whiskers. The grain boundaries of the matrix were revealed by polishing on a series of diamond pastes and then etching in phosphoric acid at ~453 K for > 30 min. The whiskers were revealed by etching the unpolished surfaces in phosphoric acid at ~553 K for 5 10min. The samples were coated with gold and then examined in a Cambridge Stereoscan S IV-10 scanning electron microscope operating with an accelerating voltage of 10 kV. The sample stage was set at the horizontal position. For observations of the whisker/matrix interface, samples were prepared for transmission electron microscopy. A slice of ~ 300-500 pm thickness was 1421
1422 XIA and LANGDON: DEFORMATION OF AN ALUMINA COMPOSITE cut from the bulk specimen using a diamond saw the slice was reduced to a disc of 3 mm dia using an ultrasonic cutter, and the disc was ground on diamond to a thickness of w 50 um and ion-milled to perforation. The TEM samples were coated with carbon to improve conductivity and avoid charging effects, and they were examined with a Philips EM420 electron microscope operating at 120 Creep tests were tures using a 4-point bending rig. Alumina was used for the base support, the support for the lower pivots, and the upper loading ram. The four loading pivots, of 3. 2 mm dia, were made of optically-polished sap- phire (single crystal alumina) so that they were stronger than the matrix material of the test speci- (a) ens (inner span) and lower two pivots(outer span)were 6.4 and 19 mm, respectively. All specimens were tested with the bars placed so that the dimension of 3 mm was vertical; the creep load was therefore applied in the hot pressing direction. Tests were performed in air under conditions of constant load and with the temperature controlled to +2 K. Each specimen was maintained at the testing temperature for at least 30 min prior to pplying the load. The vertical displacement of the loading ram was measured with a linear variable differential transducer and recorded with a strip-chart The outer fiber stress, o, was calculated from the applied load, P, through the expression 19 Fig. 1. Distribution of whiskers in Al, O,9.3 vol. SiC(w) at(a) low and(b)high magnifications: the hot-pressing where L is the outer span(19 mm), a is the inner span and in the plane of observation, respectively. The (6.4 mm), w is the width of the specimen (2 mm), h is higher magnifications in Figs 1(b)and 2(b)show that the height of the specimen(3 mm)and n is the stress the whiskers are reasonably sparsely spaced and exponent of creep. The flexure displacement, y, was randomly oriented but with a tendency to lie prefer- relationship [ 19] direction. Observations by TEM indicated that the whiskers were located primarily in the grain bound- ()(+25)y aries or at triple points, as shown in Fig. 3, although (2)a small percentage of the whiskers appeared to be isolated within the grains. There was no evidence for As documented in the Appendix, an iterative pro- any residual porosity at the interfaces between the cedure was adopted to determine n using equations matrix and the whiskers 1)and(2) It was difficult to clearly reveal the grain bound For examination of microstructures after defor- aries in the alumina matrix because of the presence of ation, TEM samples were prepared from the outer whiskers. However, it was estimated by etching and regions of the tensile parts of selected test specimens. use of SEM, and confirmed by TEM, that the aver- age grain size was within the range of 1-2 um 3. EXPERIMENTAL RESULTS Extensive examination by TEM showed the pres- ence of only a small number of dislocations in the 3. 1.Characterization of the material after fabrication as-fabricated material A careful examination, at both low and high mag- nifications, showed that the distribution of whiskers 3.2. Creep results in the matrix was essentially uniform. The results Figure 4 shows a typical plot of strain, f, vs time are shown in Figs I and 2, where the hot pressing 4, for an absolute testing temperature T, of 1773K direction is either perpendicular to or lying vertical with the various curves corresponding to different
1422 XIA and LANGDON: DEFORMATION OF AN ALUMINA COMPOSITE cut from the bulk specimen using a diamond saw, the slice was reduced to a disc of 3 mm dia using an ultrasonic cutter, and the disc was ground on diamond to a thickness of ~50 #m and ion-milled to perforation. The TEM samples were coated with carbon to improve conductivity and avoid charging effects, and they were examined with a Philips EM420 electron microscope operating at 120 kV. Creep tests were conducted at elevated temperatures using a 4-point bending rig. Alumina was used for the base support, the support for the lower pivots, and the upper loading ram. The four loading pivots, of 3.2 mm dia, were made of optically-polished sapphire (single crystal alumina) so that they were stronger than the matrix material of the test specimens; the spacings between the upper two pivots (inner span) and lower two pivots (outer span) were 6.4 and 19mm, respectively. All specimens were tested with the bars placed so that the dimension of 3 mm was vertical; the creep load was therefore applied in the hot pressing direction. Tests were performed in air under conditions of constant load and with the temperature controlled to +2 K. Each specimen was maintained at the testing temperature for at least 30min prior to applying the load. The vertical displacement of the loading ram was measured with a linear variable differential transducer and recorded with a strip-chart recorder. The outer fiber stress, tr, was calculated from the applied load, P, through the expression [19] =( L-a ~f 2n + l tr \ wh2 //~ n )(P) (1) where L is the outer span (19 mm), a is the inner span (6.4 mm), w is the width of the specimen (2 ram), h is the height of the specimen (3 mm) and n is the stress exponent of creep. The flexure displacement, y, was converted to the outer fiber strain, E, through the relationship [19] E = L q-a (-n-+ 1) y" (2) As documented in the Appendix, an iterative procedure was adopted to determine n using equations (1) and (2). For examination of microstructures after deformation, TEM samples were prepared from the outer regions of the tensile parts of selected test specimens. 3. EXPERIMENTAL RESULTS 3.1. Characterization of the material after fabrication A careful examination, at both low and high magnifications, showed that the distribution of whiskers in the matrix was essentially uniform. The results are shown in Figs 1 and 2, where the hot pressing direction is either perpendicular to or lying vertical Fig. 1. Distribution of whiskers in A1203-9.3 vol.% SiC(w) at (a) low and (b) high magnifications: the hot-pressing direction is perpendicular to the plane of observation. and in the plane of observation, respectively. The higher magnifications in Figs 1 (b) and 2(b) show that the whiskers are reasonably sparsely spaced and randomly oriented but with a tendency to lie preferentially in the plane perpendicular to the hot-pressing direction. Observations by TEM indicated that the whiskers were located primarily in the grain boundaries or at triple points, as shown in Fig. 3, although a small percentage of the whiskers appeared to be isolated within the grains. There was no evidence for any residual porosity at the interfaces between the matrix and the whiskers. It was difficult to clearly reveal the grain boundaries in the alumina matrix because of the presence of whiskers. However, it was estimated by etching and use of SEM, and confirmed by TEM, that the average grain size was within the range of ~ 1-2/tm. Extensive examination by TEM showed the presence of only a small number of dislocations in the as-fabricated material. 3.2. Creep results Figure 4 shows a typical plot of strain, ,, vs time, t, for an absolute testing temperature, T, of 1773 K with the various curves corresponding to different
XIA and LANGDON: DEFORMAtION OF AN ALUMINA COMPOSITE 1423 AzO·93w寡siC|叫 a Fig 4. Creep strain vs time at 1773 K for a range of applied a stresses; for specimens bent to failure, the creep curves levels of the applied stress, o. These creep curves exhibit the normal three stages with a brief primary stage, a lengthy and well-defined steady-state con dition, and then a very brief tertiary stage of acceler ating creep before final catastrophic fracture(marked with x in Fig. 4). Using these creep data, Fig. 5 shows the instantaneous strain rate, t, plotted against strain for the same five specimens. This plot confirms the presence of a well-defined steady-state stage t It lso apparent that this material is capable of exhibit ing reasonable ductility with strains to failure of Fig. 2. Distribution of whiskers in Al,O3 9.3 voL. SiC(w) >10% under some experimental conditions at (a)low and (b) high magnifications: the hot-pressing Creep tests were conducted at temperatures from direction lies vertical in the plane of observatic 1673 to 1823 K, and Fig. 6 shows a logarithmic plot of the steady-state strain rate vs the applied stress. At each temperature, the datum points fall along straight lines, and the four lines are give an average value for the stress exponent, n, of ~3.8. Taking n=3.8, Fig. 7 shows a semi-logarithmic plot of the temper Fig 3. Whiskers located in the boundaries and at tril tFor the two specimens tested at the lower stresses in strains of 8%. Tests conducted stress levels mperatures revealed no significant further decrease in strain rate at strains ab 10 it is therefore reasor very close to the Instanta rain rate vs strain at 1773 K
XIA and LANGDON: DEFORMATION OF AN ALUMINA COMPOSITE 1423 161 I AI203 - 9.3 vol I~, SiC (w) 14 I- .r ----T 7--T~--- • 17 -- 12j- ~" 0 39.8 -'-" u 32.8 | .-/ o 25.8 r~L ,,;" a 18.7 '~F ~ ~11.8 0# ,, ~ , t (hr} Fig. 4. Creep strain vs time at 1773 K for a range of applied stresses; for specimens bent to failure, the creep curves terminate with x. Fig. 2. Distribution of whiskers in AI203 9.3 vol.% SiC(w) at (a) low and (b) high magnifications: the hot-pressing direction lies vertical in the plane of observation. levels of the applied stress, a. These creep curves exhibit the normal three stages with a brief primary stage, a lengthy and well-defined steady-state condition, and then a very brief tertiary stage of accelerating creep before final catastrophic fracture (marked with x in Fig. 4). Using these creep data, Fig. 5 shows the instantaneous strain rate, ~, plotted against strain for the same five specimens. This plot confirms the presence of a well-defined steady-state stage.t It is also apparent that this material is capable of exhibiting reasonable ductility with strains to failure of > 10% under some experimental conditions. Creep tests were conducted at temperatures from 1673 to 1823 K, and Fig. 6 shows a logarithmic plot of the steady-state strain rate vs the applied stress. At each temperature, the datum points fall along straight lines, and the four lines are reasonably parallel and give an average value for the stress exponent, n, of ~3.8. Taking n = 3.8, Fig. 7 shows a semi-logarithmic plot of the temperature compensated stress [20], Fig. 3. Whiskers located in the grain boundaries and at triple points. #For the two specimens tested at the lower stresses in Fig. 5, the tests were terminated without failure at strains of ~8%. Tests conducted at low stress levels at other temperatures revealed no significant further decrease in strain rate at strains above ~6-8%, and it is therefore reasonable to conclude that the slowest strain rates recorded in Fig. 5 are very close to the steady-state values. 10"3 t I I ~ I I I AI203 - 9.3 vo~% SiC (.) T=1773 K o'(MPa) o 39.8 ,F 432.8 10 -4 ~._ ,~ 025.8 t c ~"c-c--C°-'°~ / °18.7 / -- I0" i0-~ Fig. 5. Instantaneous strain rate vs strain at 1773 K
1424 XIA and LANGDON: DEFORMATION OF AN ALUMINA COMPOSITE Fig. 8. a dislocation network bound to siC whiskers after deforming to a strain of 414% at 1773 K and 25.8 MPa: the foil plane is perpendicular to the hot-pressing direction Fig. 6. Steady-state strain rate vs stress. o"G-T, vs the reciprocal of the absolute tempera- activity. In general, higher dislocation densities were ture, 1/T, for strain rates of 1.0 x 10-3and 1.0 x observed in specimens deformed to larger strains 10-6s-l, respectively, where G is the shear modulus Contrary to some earlier reports [17, 18], there was of monolithic alumina; the value of G was estimated very little evidence of the development of internal for each testing temperature from the relationship cavities. An example of dislocation activity is shown in Go-(aG /aT)T (3)Fig 8 for a specimen tested to failure at a strain of 14% at 1773 K and under an applied stress of here Go is the extrapolated shear modulus of 25.8 MPa: the foil plane is perpendicular to the hot aG/oT is the variation of G per degree Kelvin pressing direction and the TEM sample was taken (23.4 MPa-[21D. From Fig. 7, the activation from the outer portion of the tensile side of the bend energy for creep, e, is nated as c830+30 specimen. Figure 8 shows a dislocation network nned by whiskers, and Fig. 9 shows, for the same and <820+ 10 kJ mol-1 for the two strain rates, specimen, an array of dislocations associated with the end of a whisk Inspection of several specimens after deformation evealed the occurrence of extensive dislocation 4.1. Deformation mechanism in the composite The creep behavior of monolithic alumina has been well documented and thus provides a reference when Al 03-9. 3 wolx Sicl) considering the creep properties of alumina matrix 101 oL0x10 Fig. 7. Temperature compensated stress vs the reciprocal of Fig. 9. An array of dislocations associated he end o he absolute temperature for strain rates of 1.0 d a whisker after deforming at 1773 K and 25. 8 MPa; the foil 1.0 x 10-65", respectively plane is perpendicular e hot-pressing directio
1424 XIA and LANGDON: DEFORMATION OF AN ALUMINA COMPOSITE i0"3 10"~ ~w~ I0"~ i0"~ I i , i 'lH] AI203 - 9.3 volg SiC (w T(K) t,1823 01773 o 1723 '~ 1673 1 10 I I L L llltlll 0.5 10 ~ 10 z 58 ~(MPa) Fig. 6. Steady-state strain rate vs stress. Fig. 8. A dislocation network bound to SiC whiskers after deforming to a strain of ~ 14% at 1773 K and 25.8 MPa: the foil plane is perpendicular to the hot-pressing direction. a'/G'-IT~ vs the reciprocal of the absolute temperature, l/T, for strain rates of 1.0 x 10 -5 and 1.0 x 10 -6 S -l, respectively, where G is the shear modulus of monolithic alumina; the value of G was estimated for each testing temperature from the relationship G = Go - (aG/OT)T (3) where Go is the extrapolated shear modulus of alumina at absolute zero (1.71 x l05 MPa [21]) and dG/OT is the variation of G per degree Kelvin (23.4MPaK -] [21]). From Fig. 7, the activation energy for creep, Q, is estimated as ~830 + 30 and ~820+ lOkJmo1-1 for the two strain rates, respectively. 3.3. Microstructures after deformation Inspection of several specimens after deformation revealed the occurrence of extensive dislocation r (g) 7xi0-II 1823 1773 17~3 1673 AI2C) 3 - 9.3 vol % SiC(w) [ (s -I ) o1.0x10 "5 10 "11 O 1.0xl0-6 / / 10-13 10-14 7xl0 "15 I I 5.4 5.6 5.8 6.0 ~ xlO 4 (K "l} Fig. 7. Temperature compensated stress vs the reciprocal of the absolute temperature for strain rates of 1.0 x 10 -5 and 1.0 x 10 -6 s -~, respectively. activity. In general, higher dislocation densities were observed in specimens deformed to larger strains. Contrary to some earlier reports [17, 18], there was very little evidence of the development of internal cavities. An example of dislocation activity is shown in Fig. 8 for a specimen tested to failure at a strain of ~ 14% at 1773 K and under an applied stress of 25.8 MPa: the foil plane is perpendicular to the hot pressing direction and the TEM sample was taken from the outer portion of the tensile side of the bend specimen. Figure 8 shows a dislocation network pinned by whiskers, and Fig. 9 shows, for the same specimen, an array of dislocations associated with the end of a whisker. 4. DISCUSSION 4.1. Deformation mechanism in the composite The creep behavior of monolithic alumina has been well documented and thus provides a reference when considering the creep properties of alumina matrix Fig. 9. An array of dislocations associated with the end of a whisker after deforming at 1773 K and 25.8 MPa: the foil plane is perpendicular to the hot-pressing direction
XIA and LANGDON: DEFORMATION OF AN ALUMINA COMPOSITE composites. For monolithic alumina, there are two SiC whiskers and this provides an opportunity for a distinct types of creep behavior with either a stress direct comparison with the present results exponent of unity when the stress is low and the Under conditions of high temperature creep, it is grain size is small or a higher stress exponent, gener- now established that the steady-state creep rate, ly close to 3, when the stress is high and the grain t, may be expressed by a relationship of the form size is large [10]. At high temperatures, similar to [10] those used in the present experiments, the grain size marking the transition between the two types of ADGb/b behavior is 20-30 um. Therefore, it may be antici- kr(a八G pated that in the present experiments, with a grain size of 1-2um, the alumina matrix should display where D is the appropriate diffusion coefficient,b a stress exponent of I in the absence of any silicon is the Burgers vector(4.75 x 10-10 m for A120,),k is Boltzmanns constant, d is the grain size, p is carbide whiskers. Reference to the deformation map the exponent of the inverse grain size and A is a for diffusion creep in unreinforced alumina at elev ated temperatures /22] indicates that the diffusional dimensionless constanyer-law creep, where n >3, creep process is Coble creep controlled by aluminum ion diffusion along the grain boundaries with there is no dependence on grain size so that activation energy of -419 kJ mol[23 Therefore, independent sets of creep data may be However, the composite used in this investigati compared by plotting the temperature compensated exhibits a higher stress exponent of 3. 8 and the strain rate, kT/ DGb, vs the normalized stress hated for the activation energy for creep a/g 820-830kJ mol-')is fairly close to the value In the present experiments, the measured acti- for oxygen diffusion in the alumina lattice vation energy is reasonably close to the value for (-600-800 kJ mol[24-28). Thus, the presence of the diffusion of the oxygen anions in the alumina diffusion creep is classified into Nabarro-Herring or D9=2×10exp(-63500R7m2s-1(5) Coble creep. In both cases, the diffusion creep is where R is the gas constant accompanied by Lifshitz grain boundary sliding [29]. Figure 10 shows the present data plotted in this In the present experiments, it is concluded that the normalized form together with the data reported silicon carbide whiskers, which are located preferen- earlier by Lin and Becher[17 for alumina reinforced tially at the grain boundaries, interfere with the with 10 vol% of SiC whiskers ability of the grains to move with respect to each All of the present datum points fall close to a single other. As a result, Lifshitz grain boundary sliding is line in this normalized plot. The results of Lin and pressed and the rate of diflusion creep is sig- Becher [17] exhibit a stress exponent of x 4 but, at nificantly lowered so that intragranular dislocation any selected stress level, the measured creep rate is processes become important. within the grains. First, the measured stress exponent of -3.8 is typical of an intragranular dislocation process, and the measured activation energ ent with a lattice diffusion process within the alumina matrix controlled by the slower-moving oxy gen anions. Second, there is microstructural evidence for dislocation activity within the grains, especially he samples deformed to strains above -10% 4.2. Comparison with other creep data on Al,O, with Although there are several reports of creep behav or in Al,O, strengthened with SiC whiskers, it is Lin and Becher(1991 difficult to make a direct comparison with most of these data because of significant differences in the However, Lin and Becher [17] reported creep data stress for the present data and for ain rate y volume percentages of the whisker reinforcement. Fig 10. Temperature for an alt lumina composite containing 10 vol % of
XIA and LANGDON: DEFORMATION OF AN ALUMINA COMPOSITE 1425 composites. For monolithic alumina, there are two distinct types of creep behavior with either a stress exponent of unity when the stress is low and the grain size is small or a higher stress exponent, generally close to 3, when the stress is high and the grain size is large [10]. At high temperatures, similar to those used in the present experiments, the grain size marking the transition between the two types of behavior is ~ 20-30 #m. Therefore, it may be anticipated that in the present experiments, with a grain size of ~ 1-2 #m, the alumina matrix should display a stress exponent of 1 in the absence of any silicon carbide whiskers. Reference to the deformation map for diffusion creep in unreinforced alumina at elevated temperatures [22] indicates that the diffusional creep process is Coble creep controlled by aluminum ion diffusion along the grain boundaries with an activation energy of ~419 kJ mol ~ [23]. However, the composite used in this investigation exhibits a higher stress exponent of ~3.8 and the value estimated for the activation energy for creep (~820-830kJmol -~) is fairly close to the value for oxygen diffusion in the alumina lattice (~ 600-800 kJ mol t [24-28]). Thus, the presence of whiskers markedly affects the creep behavior of the alumina matrix. Depending upon the path for diffusion, whether through the lattice or along the grain boundaries, diffusion creep is classified into Nabarro Herring or Coble creep. In both cases, the diffusion creep is accompanied by Lifshitz grain boundary sliding [29]. In the present experiments, it is concluded that the silicon carbide whiskers, which are located preferentially at the grain boundaries, interfere with the ability of the grains to move with respect to each other. As a result, Lifshitz grain boundary sliding is suppressed and the rate of diffusion creep is significantly lowered so that intragranular dislocation processes become important. There are two experimental observations supporting the occurrence of extensive dislocation activity within the grains. First, the measured stress exponent of ~3.8 is typical of an intragranular dislocation process, and the measured activation energy is consistent with a lattice diffusion process within the alumina matrix controlled by the slower-moving oxygen anions. Second, there is microstructural evidence for dislocation activity within the grains, especially in the samples deformed to strains above ~ 10%. 4.2. Cornpar&on with other creep data on Ale0 J with SiC whiskers Although there are several reports of creep behavior in A1203 strengthened with SiC whiskers, it is difficult to make a direct comparison with most of these data because of significant differences in the volume percentages of the whisker reinforcement. However, Lin and Becher [17] reported creep data for an alumina composite containing 10vol.% of SiC whiskers and this provides an opportunity for a direct comparison with the present results. Under conditions of high temperature creep, it is now established that the steady-state creep rate, ~, may be expressed by a relationship of the form [10] ADGb g - (4) ,, where D is the appropriate diffusion coefficient, b is the Burgers vector (4.75 × 10-ram for A1203), k is Boltzmann's constant, d is the grain size, p is the exponent of the inverse grain size and A is a dimensionless constant. In intragranular power-law creep, where n >/3, there is no dependence on grain size so that p = 0. Therefore, independent sets of creep data may be compared by plotting the temperature compensated strain rate, ~kT/DGb, vs the normalized stress, tr/G. In the present experiments, the measured activation energy is reasonably close to the value for the diffusion of the oxygen anions in the alumina lattice. Thus, the diffusion coefficient was taken as the value for lattice diffusion of oxygen, D °2- t , given by [24] D °2 =2x lO-lexp(-635,000/RT)m2s J (5) where R is the gas constant. Figure 10 shows the present data plotted in this normalized form together with the data reported earlier by Lin and Becher [17] for alumina reinforced with 10 vol.% of SiC whiskers. All of the present datum points fall close to a single line in this normalized plot. The results of Lin and Becher [17] exhibit a stress exponent of ~4 but, at any selected stress level, the measured creep rate is 10-5 i i J iiiii] AI203-9.3vol % SiC w A 1~-3 Z' , lo -~ 0 1773 z~' / [] 1723 [~ el 10-8 tJ AI20 g -10 vol % SIC(, / OT:1573 K,d=SFm Lin and Becher (1991) 10"~0. 5 1/.4' ''''"~)_3 5x10- 3 o-/G Fig. 10. Temperature compensated strain rate vs normalized stress for the present data and for the results of Lin and Becher [17] on A1203-10 vol.% SiC(w)
1426 XIA and LANGDON: DEFORMATION OF AN ALUMINA COMPOSITE lower than in the present experiments by about three fiber stress of 61.5 MPa(Fig. 7)whereas the exper- orders of magnitude. iments of Lin and Becher [17], conducted at the lower The procedure employed by Lin and Becher [17] temperature of 1573 K, covered estimated outer fiber was similar to that used in the present experiments. stresses in the range from 100 to 234 MPa. It is he samples were obtained by hot-pressing, tests probable that these higher stress levels account for were conducted in air using a four-point bending the markedly different creep behavior in the two sets rig and the stresses and strains were calculated of experiments using equations (1)and(2). The average grain size was x8 um in the as-fabricated samples, but this larger size should have no effect on the measured 5. SUMMARY AND CONCLUSIONS train rates if an intragranular deformation process (Creep tests were conducted in air on an is dominant alumina composite reinforced with 9.3 vol. of A significant difference lies in the overall creep silicon carbide whiskers behavior of the two sets of samples. Lin and Becher (2)The creep curves exhibited a well-defined [17 tested their specimens at a temperature of 1573 K steady-state region with strains to failure in excess of and obtained total strains to failure of 10% under whereas the present experiments were conducted at conditions temperatures in the range from 1673 to 1823 K and (3)The creep data gave a stress exponent of yielded strains to failure which were consistently + 3.8 and an activation energy for creep >5% and even up to >10%. Further, Lin and + mol cher[17] reported the development of very exten-(4)The microstructure after deformation revealed sive internal cavitation, primarily in the form of small extensive dislocation activity and with very little polyhedral and penny-shaped cavities lying along the evidence for the development of internal cavitation grain boundaries and cracking along boundaries (5)It is concluded that the presence of many where no whiskers were present, whereas very little Sic whiskers in the grain boundaries inhibits the cavitation was visible in the present experiments after occurrence of Lifshitz grain boundary sliding and failure therefore diffusion creep is suppressed. Instead, the The occurrence of extensive cavitation in the expe composite deforms by an intragranular dislocatic ments of Lin and Becher [17] may account for the process apparent discrepancy in the creep data as recorded in anions controlled by lattice diffusion of the oxygen Fig. 10. In four-point bending experiments where equations(1)and (2)necessitates that the neutral axis the Advanced Composite Materl au to Dr J F. Rhodes of samples deform by power-law creep, the use of Acknowledgements-We are gr mains along the central line of the specimen with ing the material used in this investigation. This work was the tensile and compressive behavior obeying the supported in part by the U.S. Army Research Office under Grant No. DAALO3-91-G-0230 same constitutive law. When the stress exponent, n, is greater than s2, the development of microcracking and cavitation shifts the neutral axis away from this REFERENCES central line and this will lead to a significant overes- timation of the outer fiber stress and a consequent L. P. F. Becher W. H underestimation of the equivalent tensile creep rate Warwick, in Fracture Mechanics of Ceramics, Vol. 7 [30, 31]. In view of the large amounts of cavitation Hasselman and F, F. Lange), p. 61. Plenum Press reported by Lin and Becher [17), it is possible that New York(1986) heir estimates of the outer fiber stresses and strains 2. K. Niihara, A. Nadahira, T. Uchiyama and T. Hirai, may be in error. R. C. Bradt, A. G. Evans, D. P. H. Hasselman Finally, it is necessary to address the apparent and F. F. Lange), p 103. Plenum Press, New York dichotomy between the experiments Becher [17] where cavitation was extensive and the L. Vaughn and M. K. Ferber, Am. present experiments where cavity development wa very limited. In the work of de Arellano-Lopez 4.P D. Shalek, JJ. Petrovic, G F. Hurley and F D Gac, et al. [18] on alumina composites containing up 5. R Lundberg, L. Kahlman, R Pompe and RCarlsson to 30 vol. of Sic whiskers, it was reported that, pon the fabrication pro sed to 6. M. Ruhle, N. Clausson and A. H. Heuer. J. Am. ceran Soc.69,195(1986 duce the composites, there is a critical stress below 7. s. C. Samanta and S. Musikant, Ceram. Engng Sci hich the development of cavitation damage is essen- tially negligible. Although insufficient information 8. G. C. Wei and P F Becher, J. Am. Ceram. Soc. 67, 571 is currently available to determine the magnitude of the critical stress under any selected experimental conditions 1(1983) nevertheless noted that the present 10. W.R. Cannon and T.G. Langdon, J Mater. Sci. 23, experiments were performed up to a maximum outer 1(1988)
1426 XIA and LANGDON: DEFORMATION OF AN ALUMINA COMPOSITE lower than in the present experiments by about three orders of magnitude. The procedure employed by Lin and Becher [17] was similar to that used in the present experiments. The samples were obtained by hot-pressing, tests were conducted in air using a four-point bending rig and the stresses and strains were calculated using equations (1) and (2). The average grain size was ~8/~m in the as-fabricated samples, but this larger size should have no effect on the measured strain rates if an intragranular deformation process is dominant. A significant difference lies in the overall creep behavior of the two sets of samples. Lin and Becher [17] tested their specimens at a temperature of 1573 K and obtained total strains to failure of ~5% and even up to >10%. Further, Lin and Becher [17] reported the development of very extensive internal cavitation, primarily in the form of small polyhedral and penny-shaped cavities lying along the grain boundaries and cracking along boundaries where no whiskers were present, whereas very little cavitation was visible in the present experiments after failure. The occurrence of extensive cavitation in the experiments of Lin and Becher [17] may account for the apparent discrepancy in the creep data as recorded in Fig. 10. In four-point bending experiments where samples deform by power-law creep, the use of equations (1) and (2) necessitates that the neutral axis remains along the central line of the specimen with the tensile and compressive behavior obeying the same constitutive law. When the stress exponent, n, is greater than ~ 2, the development of microcracking and cavitation shifts the neutral axis away from this central line and this will lead to a significant overestimation of the outer fiber stress and a consequent underestimation of the equivalent tensile creep rate [30, 31]. In view of the large amounts of cavitation reported by Lin and Becher [17], it is possible that their estimates of the outer fiber stresses and strains may be in error. Finally, it is necessary to address the apparent dichotomy between the experiments of Lin and Becher [17] where cavitation was extensive and the present experiments where cavity development was very limited. In the work of de Arellano-L6pez et al. [18] on alumina composites containing up to 30 vol.% of SiC whiskers, it was reported that, depending upon the fabrication procedure used to produce the composites, there is a critical stress below which the development of cavitation damage is essentially negligible. Although insufficient information is currently available to determine the magnitude of the critical stress under any selected experimental conditions, it is nevertheless noted that the present experiments were performed up to a maximum outer fiber stress of 61.5 MPa (Fig. 7) whereas the experiments of Lin and Becher [17], conducted at the lower temperature of 1573 K, covered estimated outer fiber stresses in the range from ~ 100 to ~234 MPa. It is probable that these higher stress levels account for the markedly different creep behavior in the two sets of experiments. 5. SUMMARY AND CONCLUSIONS (1) Creep tests were conducted in air on an alumina composite reinforced with 9.3vo!.% of silicon carbide whiskers. (2) The creep curves exhibited a well-defined steady-state region with strains to failure in excess of 5% and up to > 10% under some experimental conditions. (3) The creep data gave a stress exponent of ~3.8 and an activation energy for creep of 820-830 kJ mol- i. (4) The microstructure after deformation revealed extensive dislocation activity and with very little evidence for the development of internal cavitation. (5) It is concluded that the presence of many SiC whiskers in the grain boundaries inhibits the occurrence of Lifshitz grain boundary sliding and therefore diffusion creep is suppressed. Instead, the composite deforms by an intragranular dislocation process controlled by lattice diffusion of the oxygen anions. Acknowledgements--We are grateful to Dr J. F. Rhodes of the Advanced Composite Materials Corporation for supplying the material used in this investigation. This work was supported in part by the U.S. Army Research Office under Grant No. DAAL03-91-G-0230. REFERENCES 1. P. F. Becher, T. N. Tiegs, J. C. Ogle and W. H. Warwick, in Fracture Mechanics of Ceramics, Vol. 7 (edited by R. C. Bradt, A. G. Evans, D. P. H. Hasselman and F. F. Lange), p. 61. Plenum Press, New York (1986). 2. K. Niihara, A. Nadahira, T. Uchiyama and T. Hirai, in Fracture Mechanics of Ceramics, Vol. 7 (edited by R. C. Bradt, A. G. Evans, D. P. H. Hasselman and F. F. Lange), p. 103. Plenum Press, New York (1986). 3. J. Homeny, W. L. Vaughn and M. K. Ferber, Am. Ceram. Soc. Bull. 67, 333 (1987). 4. P. D. Shalek, J. J. Petrovic, G. F. Hurley and F. D. Gac, Am. Ceram. Soc. Bull. 65, 351 (1986). 5. R. Lundberg, L. Kahlman, R. Pompe and R. Carlsson, Am. Ceram. Soc. Bull. 66, 330 (1987). 6. M. Riihle, N. Clausson and A. H. Heuer, 3. Am. Ceram. Soc. 69, 195 (1986). 7. S. C. Samanta and S. Musikant, Ceram. Engng Sci. Proc. 6, 663 (1985). 8. G. C. Wei and P. F. Becher, J. Am. Ceram. Soc. 67, 571 (1984). 9. W. R. Cannon and T. G. Langdon, J. Mater. Sci. 18, I (1983). 10. W. R. Cannon and T. G. Langdon, J. Mater. Sci. 23, I (1988)
XIA and LANGDON: DEFORMATION OF AN ALUMINA COMPOSITE 1427 11. A. H. Chokshi and J. R. Porter, J. Am. Ceram. Soc. 26. D. J. Reed and B. J. wuensch, J. Am. Ceram. Soc. 63, 88(1980) 12. P. Lipetzky, S. R. Nutt and P. F. Becher, Mater. Res. 27. K P.R. Reddy and A. R. Cooper, J. Am. Ceram. Soc Soc. Symp. Proc. 120, 271(1988) 65,634(1982). 13. S. R. Nutt, P, Lipetzky and P. F Becher Mater. Sci. 28. Y Oishi, K. Ando and N. Suga, J, Am. Ceram Soc. 66, I4 Dominguez- Rodriguez ok.CF GoretCaumhe a.A 30. M. Chuang. ).oMmarter Sci 2l, 65 ( 1986)963) Routbort, J. Am. Ceram Soc. 73, 1297(1990) 31. T.J. Chuang and s. M. Wiederhorn, J. Am. Ceram 15. H.-T. Lin and P, F. Becher, J. Am. Ceram, Soc. 73, Soc.71,595(1988) 16. P. Lipetzky, S.R. Nutt, D. A. Koester and R. F. Davis, 17. H-T. Lin and P, F. Becher, J. Am. Ceram. Soc. 74. APPENDIX 886(1991) 18.A Method of Using Equations(1)and(2) An iterative procedure was adopted in making use 19. G. W. Hollenberg, G. R. Terwilliger an J. Am. Ceram. Soc. 54, 196(1971) equations(D)and(2)to determine the values of the stress F. A. Mohamed and T. G. Langdon, Physica was assumed initially solidi(a)33,375(1976 used to calculate g and e D. H. Chung and G. Simmons, J, appl. Ph calculated values of strain rate and stress logarithmically to give a first order estimate for n. 22. T. G. Langdon, Cerar 1980) Second, this estimate of n was used to recalculate g and 23.R. M. Cannon and r Deformation of E from equations(I)and (2), and the logarithmic plotting of Ceramic Materials(edit strain rate vs stress was repeated to provide a second order Tressler), p. 61. Plenum Press, New York(1975) estimate for n 33, These various steps were subsequently repeated until there was no significant change in the va 25. Y. Oishi, K. Ando and Y, Kubota, J. Che. Phys. 73, values of n documented in this report represent these final 1410(1980
XIA and LANGDON: DEFORMATION OF AN ALUMINA COMPOSITE 1427 11. A. H. Chokshi and J. R. Porter, J. Am. Ceram. Soc. 68, C144 (1985). 12. P. Lipetzky, S. R. Nutt and P. F. Becher, Mater. Res. Soc. Symp. Proc. 120, 271 (1988). 13. S. R. Nutt, P. Lipetzky and P. F. Becher, Mater. Sci. Engng A126, 165 (1990). 14. A. R. de Arellano-L6pez, F. L. Cumbrera, A. Dominguez-Rodriguez, K. C. Goretta and J. L. Routbort, J. Am. Ceram. Soc. 73, 1297 (1990). 15. H.-T. Lin and P. F. Becher, J. Am. Ceram. Soc. 73, 1378 (1990). 16. P. Lipetzky, S. R. Nutt, D. A. Koester and R. F. Davis, J. Am. Ceram. Soc. 74, 1240 (1991). 17. H.-T. Lin and P. F. Becher, J. Am. Ceram. Soc. 74, 1886 (1991). 18. A. R. de Arellano-L6pez, A. Dominguez-Rodriguez, K. C. Goretta and J. L. Routbort, J. Am. Ceram. Soc. 76, 1425 (1993). 19. G. W. Hollenberg, G. R. TerwiUiger and R. S. Gordon, J. Am, Ceram. Soc. 54, 196 (1971). 20. F. A. Mohamed and T. G. Langdon, Physica Status Solidi (a) 33, 375 (1976). 21. D. H. Chung and G. Simmons, 3. appl. Phys. 39, 5316 (1968). 22. T. G. Langdon, Ceram. Int. 6, 11 (1980). 23. R. M. Cannon and R. L. Coble, in Deformation of Ceramic Materials (edited by R. C. Bradt and R. E. Tressler), p. 61. Plenum Press, New York (1975). 24. Y. Oishi and W. D. Kingery, J. Chem. Phys. 33, 480 (1960). 25. Y. Oishi, K. Ando and Y. Kubota, J. Chem. Phys. 73, 1410 (1980). 26. D. J. Reed and B. J. Wuensch, J. Am. Ceram. Soc. 63, 88 (1980). 27. K. P. R. Reddy and A. R. Cooper, J. Am. Ceram. Soc. 65, 634(1982). 28. Y. Oishi, K. Ando and N. Suga, J. Am. Ceram. Soc. 66, C130 (t983). 29. I. M. Lifshitz, Soviet Phys. JETP 17, 909 (1963). 30. T.-J. Chuang, J. Mater. Sci. 21, 165 (1986). 31. T.-J. Chuang and S. M. Wiederhorn, J. Am. Ceram. Soc. 71, 595 (1988). APPENDIX Method of Using Equations (1) and (2) An iterative procedure was adopted in making use of equations (I) and (2) to determine the values of the stress exponent, n. First, a value of n was assumed initially and this value was used to calculate o and E using equations (1) and (2). The calculated values of strain rate and stress were then plotted logarithmically to give a first order estimate for n. Second, this estimate of n was used to recalculate a and E from equations (1) and (2), and the logarithmic plotting of strain rate vs stress was repeated to provide a second order estimate for n. These various steps were subsequently repeated until there was no significant change in the value estimated for n. The values of n documented in this report represent these final values
