Journal of the European Ceramic Society 15(1995)191-199 Printed in Great Brita Fractography of Fatigued and Fractured Regions in a Silicon Carbide Whisker Reinforced Alumina Composite Ashok Kr. Ray, Swapan Kr. Das, Prabir Kr. Roy National Metallurgical Laboratory, Jamshedpur 831007, Bihar, India S Banerjee Research and Developmerit Centre for Iron and Steeh SAIL, Ranchi 834002, Bihar, India (Received 18 May 1994; revised version received 19 July 1994; accepted 1 August 1994) Abstract whe h applications, these ceramic uld often encounter fatigue cracked and fast fractured regions in four and cyclic loading which produce crack extension point bend specimens prepared from 25 wt% silicon Therefore, the fractographic features of the fatigue carbide whisker reinforced alumina composites were failed samples need to be examined to identify the examined by Scanning Electron Microscopy. In the likely micromechanism of crack advance under fatigue cracked region, the alumina matrix failed monotonic and cyclic loading in this composite mainly in a transgranular mode and the whiskers Recently, Dauskardt et al. have made an exten failed mainly with a fat fracture surface but with- sive fractography of fatigue failed regions in a 15 out pullout. On the other hand, in the fast fractured vol% SiC whisker-reinforced alumina composite region, the whiskers failed predominantly by pullout The identification of the fractographic features at and the alumina matrix failed in a mixed mode with the low, medium and high Stress Intensity Range about half in transgranular and the other half in (AK) fatigue region as well as in the fast fracture intergranular fracture. Thus, to improve the fracture region can give us a clue to the likely mechanisms toughness of the material, the grain boundar of fracture in our material strength of alumina and the matrix whisker inter facial bonding should be improved. To increase the resistance to fatigue, the fracture strength of the 2 Experimental Procedure alumina grains should be improved by using finer a-alumina particles and the fatigue strength of theThe four point bend specimens were sliced, pre- whisker has to be increased by improving the uni- pared, surface finished and randomised from a 15 formity in distribution of B-Sic whiskers during hot mm X 50 mm x 12 5 mm preformed billet of 25 wt% pressing silicon carbide whisker reinforced alumina com posite material(Fig. 1(a). The details of fabrication of the billet are given elsewhere. 1 Introduction The alumina powder used was of a-Lype. The Sic whiskers had been produced by a carburize Fatigue crack growth rate(FCGr), 2and fracture tion process at 1600 C where the sources of silicon toughness-3of 25 wt% silicon carbide whisker and carbon were rice husk ash and rice husk reinforced alumina ceramic composite have been hydrocarbons. studied and reported. -Such studies are impor- The particle size of the alumina powder was less tant since these materials have potential applica- than 1 um. The average whisker diameter was tion in the production of structural components 0-1-025 um as revealed in the lt plane of the used at elevated temperatures, in high efficiency montage of the ceramic composite investigated heat engines and heat recovery systems and for here( fig. 1 (b)). The length of the whiskers varied making cutting tools to rnachine special materials. between 10 and 30 um
Journal of Ihe European Ceramic Sociery 15 (1995) 191-199 Elsevier Science Limited Printed in Great Britain 0955-2219/95/$9.50 Fractography of Fatigued and Fractured Regions in a Silicon Carbide Whisker Reinforced Alumina Composite Ashok Kr. Ray, Svvapan Kr. Das, Prabir Kr. Roy National Metallurgical Laboratory, Jamshedpur 83 1007, Bihar, India & S. Banerjee Research and Development Centre for Iron and Steel-SAIL, Ranchi 834002, Bihar, India (Received 18 May 1994; revised version received 19 July 1994; accepted 1 August 1994) Abstract Fatigue cracked and fast fractured regions in four point bend specimens prepared from 2.5 wt% silicon carbide whisker reinforced alumina composites were examined by Scanning Electron Microscopy. In the fatigue cracked region, the alumina matrix failed mainly in a transgranular mode and the whiskers failed mainly with a flat fracture surface but without pullout. On the other hand, in the fast fractured region, the whiskers failed predominantly by pullout and the alumina matrix j&led in a mixed mode with about half in transgranular and the other half in intergranular fracture. Titus, to improve the fracture toughness of the material, the grain boundary strength of alumina and the matrix whisker interfacial bonding should be improved. To increase the resistance to fatigue, the fracture strength of the alumina grains should be improved by using finer a-alumina particles and #the fatigue strength of the whisker has to be increased by improving the uniformity in distribution of P-Sic whiskers during hot pressing. 1 Introduction Fatigue crack growth rate (FCGR)‘,2 and fracture toughness1-3 of 25 wt% silicon carbide whisker reinforced alumina ceramic’ composite have been studied and reported.‘-3 Such studies are important since these materials have potential application in the production of structural components used at elevated temperatures, in high efficiency heat engines and heat recovery systems and for making cutting tools to machine special materials. 191 When used in such applications, these ceramic components would often encounter monotonic and cyclic loading which produce crack extension. Therefore, the fractographic features of the fatigue failed samples need to be examined to identify the likely micromechanism of crack advance under monotonic and cyclic loading in this composite. Recently, Dauskardt et ~1.~ have made an extensive fractography of fatigue failed regions in a 15 ~01% Sic whisker-reinforced alumina composite. The identification of the fractographic features at the low, medium and high Stress Intensity Range (AK) fatigue region as well as in the fast fracture region can give us a clue to the likely mechanisms of fracture in our material. 2 Experimental Procedure The four point bend specimens were sliced, prepared, surface finished and randomised from a 15 mm X 50 mm X 12.5 mm preformed billet of 25 wt% silicon carbide whisker reinforced alumina composite material (Fig. l(a)). The details of fabrication of the billet are given elsewhere.3 The alumina powder used was of a-type.’ The Sic whiskers had been produced by a carburization process at 1600°C where the sources of silicon and carbon were rice husk ash and rice husk hydrocarbons.3s5 The particle size of the alumina powder was less than 1 pm. The average whisker diameter was 0.1-0.25 pm as revealed in the LT plane of the montage of the ceramic composite investigated here (Fig. l(b)). The length of the whiskers varied between 10 and 30 pm.3
K. Ray, S. K. Das, P. K. Roy, S. Ba Fig. 1(a). Portion of the billet prepared from the 25 wt% silicon carbide-alumina composite, All dimensions are in mm Fig. 2. Fracture surface of the composite showing the size and shape of the alumina matrix Fig. 1(b). Montage of the microstructures shows distribution of SiCw along the three plane The montage of the ceramic composite revealed the 3D-distribution pattern of the whiskers in the C The SEM (Scanning Electron Microscope) longitudinal (L), long transverse(LT)and short amination of the fractured surface of the speci- transverse(ST)planes. In the L plane, the whiskers mens which failed due to premature crack exten- appeared to be randomly oriented as the hot press- sion during precracking in the conventional bridge ing direction is perpendicular to this L plane. dur fixture, showed that the alumina grain size after ing hot pressing, whiskers which are not normal to fabrication varied bctwccn 1 and 6 um. This the l plane gct furthcr inclined thus producing the confirms that during hot pressing, the alumina random orientation of the whiskers. Since maxi underwent substantial grain coarsening. In areas mum material flow occurred along the Lt planes where either whiskers were absent or there was during hot pressing, the whiskers tend to get evidence of pores(Fig. 2 ), the grain size was as high oriented parallel to the L plane In the St plane, as 6 um. However, in other areas, the majority of there was a mixture of random orientation as we the alumina grain size was in the range of 24 um. as normal alignment of the whisker
A. K. Ray, S. K. Das, P. K. Roy, S. Banerjee Fig. l(a). Portion of the billet prepared from the 25 wt% silicon carbide-alumina composite. All dimensions are in mm. Fig. l(b). Montage of the microstructures shows distribution of Sic, along the three planes. The SEM (Scanning Electron Microscope) examination of the fractured surface of the specimens which failed due to premature crack extension during precracking in the conventional bridge fixture,1,3 showed that the alumina grain size after fabrication varied between 1 and 6 pm. This confirms that during hot pressing, the alumina underwent substantial grain coarsening. In areas where either whiskers were absent or there was evidence of pores (Fig. 2), the grain size was as high as 6 pm. However, in other areas, the majority of the alumina grain size was in the range of 24 pm. Fig. 2. Fracture surface of the composite showing the size and shape of the alumina matrix. The montage of the ceramic composite revealed the 3D-distribution pattern of the whiskers in the longitudinal (L), long transverse (LT) and short transverse (ST) planes. In the L plane, the whiskers appeared to be randomly oriented as the hot pressing direction is perpendicular to this L plane. During hot pressing, whiskers which are not normal to the L plane get further inclined thus producing the random orientation of the whiskers.’ Since maximum material flow occurred along the LT planes during hot pressing, the whiskers tend to get oriented parallel to the L plane. In the ST plane, there was a mixture of random orientation as well as normal alignment of the whiskers
fractography of fatigued and fractured regions 193 Wei had concluded in their work6 that whisker orient during processing of hot R=0.1.f=1Hz pressed Sic-whisker reinforced alumina leads to anisotropy in both fracture toughness and fracture x- with v notch 8 strength of the composites. In other words, their fracture strengths are limited by the nonunifor mity of the distribution of the whiskers, i.e by the ability to disperse the sic whiskers. They also n=15.5 found that the dispersion of the whiskers improved by using finer alumina powder and hence an increase in the fracture strength of the compos- ite was observed. Nevertheless, they have clearly stress Intensity Rang,△K(MPa「) observed6, , that similar to our composite under Fig. 4. Fatigue crack growth data of 25 wt% SiC reinforced investigation(Fig. 1(b)), the whiskers were prefer- Al,O3 composite entially aligned perpendicular to the hot pressing distribution of whiskers sug- four point bend specimens. On the remaining half, gested that a great deal of rearrangement of notches were inserted exactly at the centre of the whiskers and powder occurred in the initial stage 3 mm x 50 mm surface of the specimens with a of densification of the composites and/or the carborundum wheel to a depth of 0. 1 mm. There- matrix material underwent considerable deformation after, the specimens were subjected to fatigue loading in a servohydraulic test machine MTS-880 The dimensions of the billet and the orientation (100 kn capacity) to precrack the specimen in an of the four point bend specimens(3 mm x 4 mm articulated bridge fixture, 2 till the ratio of crack X 50 mm)in the billet is given in Fig. 1(a). The length to width was 005 on the (4 mm X 50 mm) 4 mm X 50 mm faces, were normal to the hot-press- surface of the specimen. Precracking was con ing direction so that both the direction of crack ducted at a force of 4-5 kN, load ratio, R=0-1 propagation( Fig 3)and the crack plane were par- and frequency, f= 20 Hz. After this, the fatigue allel to the LT plane(Fig. l(a)) and normal to the crack growth rate in the specimen was determined hot-pressing direction. The 3 mm X 50 mm faces at increasing values of Stress Intensity Range AK of the specimen were parallel to the st plane (Fig. 4). under normal four point bend loading A Vickers indentation produced at 0-8 kN load(see Fig 3)until the ratio of crack length to width at the centre of the 3 mm x 50 mm face of the was about 0.45-0.5. on the 4 mm x 50 mm sur- sample acted as the crack starter on half of the face of the specimen. Thereafter, the test to deter- mine the fatigue crack growth rate(FCGr) was P/2 terminated and the specimen was subjected to External span monotonic loading to determine the fracture toughness of the material as per ASTM STP 410. For both FCGR as well as fracture toug ness(Kic tests, a l kn load range of the MTS 880 test machine was used. The test frequency was I Hz and the loading rate was 0.25N s". Tests were conducted in laboratory atmosphere and at ambient temperature. The fracture toughness testing produced the fast fracture. the details of fracture toughness and FCgr tests are given elsewhere. 1,2 P/2 The fracture surfaces were coated with a thin film of gold(thickness 0-02 um) prior to SEM (Scanning Electron Microscopic examination, in a JEOL JSM 840A microscope) The three regions, namely low AK correspond ing to 0-8-1-8 MPa vm; high AK corresponding to 28-3 MPa vm; and, the fast fracture region corre- ALL DIMENSIONS ARE IN MM sponding to a Kic value of 5.9 MPa vm Fig. 3. Indented and precrad.a sf pe four corners of cation of 30 X under SEM(Fig. 5)-using the imen for four point first, identified on the fracture surface at a bend loading. Cracks are located a indentation details of the crack growth rate data generated on
Fractography of fatigued and fractured regions 193 Becher and Wei had concluded in their work6 that whisker orientation during processing of hot pressed Sic-whisker reinforced alumina leads to anisotropy in both fracture toughness and fracture strength of the composites. In other words, their fracture strengths are limited by the nonuniformity of the distribution of the whiskers, i.e by the ability to disperse the SIC whiskers. They also found that the dispersion of the whiskers improved by using finer alumina powder and hence an increase in the fracture strength of the composite was observed. Nevertheless, they have clearly observed6,’ that similar to our composite under investigation (Fig. l(b)), the whiskers were preferentially aligned perpendicular to the hot pressing axis. This type of distribution of whiskers suggested that a great deal of rearrangement of whiskers and powder occurred in the initial stage of densification of the composites and/or the matrix material underwent considerable deformation or creep during hot pressing. The dimensions of thle billet and the orientation of the four point bend #specimens (3 mm X 4 mm X 50 mm) in the billet is given in Fig. l(a). The 4 mm X 50 mm faces, were normal to the hot-pressing direction so that both the direction of crack propagation (Fig. 3) and the crack plane were parallel to the LT plane (Fig. l(a)) and normal to the hot-pressing direction. 3 The 3 mm X 50 mm faces of the specimen were parallel to the ST plane. A Vickers indentation produced at 0.8 kN load at the centre of the 3 mm X 50 mm face of the sample acted as the crack starter on half of the ip” Extcrn;D span 1 p’2 ,&---25 -4 * 2. L-501 ALL DIMENSIONS ARE IN MM. Fig. 3. Indented and precracked specimen for four point bend loading. Cracks are located at the four comers of indentation. 3 $5 ; 2 R= 0.1, f q 1Hz z P k- with V notch & % precrack pi ,g l with xi indentation & a precrack 5 : [ n I 16.1 g nz15.5 x 8 ,g 3 4 5 Stress Intensity Range, AK (MP afi) Fig. 4. Fatigue crack growth data of 25 wt% Sic reinforced A1,03 composite. four point bend specimens. On the remaining half, notches were inserted exactly at the centre of the 3 mm X 50 mm surface of the specimens with a Carborundum wheel to a depth of 0.1 mm. Thereafter, the specimens were subjected to fatigue loading in a servohydraulic test machine MTS-880 (100 kN capacity) to precrack the specimen in an articulated bridge fixture,2 till the ratio of crack length to width was 0.05 on the (4 mm X 50 mm) surface of the specimen. Precracking was conducted at a force of 4-5 kN, load ratio, R = 0.1 and frequency, f = 20 Hz. After this, the fatigue crack growth rate in the specimen was determined at increasing values of Stress Intensity Range AK (Fig. 4) under normal four point bend loading (see Fig. 3) until the ratio of crack length to width was about 0.45-0.5, on the 4 mm X 50 mm surface of the specimen. Thereafter, the test to determine the fatigue crack growth rate (FCGR) was terminated and the specimen was subjected to monotonic loading to determine the fracture toughness of the material’ as per ASTM STP 410.* For both FCGR as well as fracture toughness (K,,) tests, a 1 kN load range of the MTS- 880 test machine was used. The test frequency was I Hz and the loading rate was 0.25 N ss’. Tests were conducted in laboratory atmosphere and at ambient temperature. The fracture toughness testing produced the fast fracture. The details of fracture toughness and FCGR tests are given elsewhere.‘,2 The fracture surfaces were coated with a thin film of gold (thickness 0.02 pm) prior to SEM (Scanning Electron Microscopic examination, in a JEOL JSM 840A microscope). The three regions, namely low AK corresponding to 0.8-l .8 MPa &; high AK corresponding to 2.8-3 MPa 6; and, the fast fracture region corresponding to a K,, value of 5.9 MPa G, were, at first, identified on the fracture surface at a magnification of 30 X under SEM (Fig. 5) - using the details of the crack growth rate data generated on
A.K. Ray, S. K. Das, P. K. Roy, S. Banerjee re surface. Right-hand side--low AK region left-hand side-fast fracture Fig. 6. At low AK (0-8-1.8 MPa Vm)region, the majority of region(FF) the whiskers failed with a square fracture without evidence the specimen as a guideline the difference between the fatigue and fast fracture could, however, be discerned through observation even with the naked eye. The distinction between the low and the high AK fatigue regions was made from the crack length versus the number of cycles data generated during the fatigue loading of the specimen Each region was, at first, carefully scanned at low magnification, to identify the general and uni- formly distributed features. Thereafter, it was examined at two different magnifications 4500 X and 7500 X, in order to identify the fractographic features and the mechanism of fracture in the low AK and the fast fracture regions Fig. 7. At high AK(28-3 MPa vm) region the whiskers 3 Fractographic Observations failed in a mixed mode, i.e. both with a square fracture and also by pullo A summary of the results of the fractographic observations is reported in Table 1 vm), the whisker failed in two different ways: that Table 1 states that, at low AK(08-1. 8 MPa is about 60% of the whiskers failed with a flat Nm), the whiskers failed predominantly by shear- fracture and the balance failed by pullout(Fig. 7) ing with a fat or square fracture without any visi- On the other hand, in the fast fracture region, the ble evidence of necking; this was typical of fatigue whiskers failed predominantly by pullout(Fig 8) fracture(Fig. 6). However, at high AK(2- 8-3 MPa Table 1 also reports that the matrix alumina grains failed predominantly through transgranular fracture( Fig 9)at low AK(0-8-1-8 MPav m).In gions in a 25 wt% silicon carbide whisker reinforced alumina failed in a mixed mode wherein about 45% of the Monotonic alumina grains failed by intergranular and the bal- Mechanisi 2345 ance by the transgranular mode as shown in Fig MPaym MPaym MPavm 10 During monotonic loading, the alumina grains Whisker Pullout ( in the fast fracture region also failed in a mixed mode with a slightly increased percentage of inter- granular (55%)and the balance by transgranular S intergranular (% mode-as shown in Fig. 11 It is noteworthy that in the low AK region, the Crack o River pattern 5PbP A crack had frequently branched(Fig. 12)and deflected(Fig. 13)while it propagated. River pattern
194 A. K. Ray, S. K. Das, P. K. Roy, S. Banerjee Fig. 5. Entire fracture surface. Right-hand side-low AK region; middle-high AK region; left-hand side - fast fracture region (FF). the specimen as a guideline. The difference between the fatigue and fast fracture could, however, be discerned through observation even with the naked eye. The distinction between the low and the high AK fatigue regions was made from the crack length versus the number of cycles data generated during the fatigue loading of the specimen. Each region was, at first, carefully scanned at low magnification, to identify the general and uniformly distributed features. Thereafter, it was examined at two different magnifications 4500 X and 7500 X, in order to identify the fractographic features and the mechanism of fracture in the low AK and the fast fracture regions. 3 Fractographic Observations A summary of the results of the fractographic observations is reported in Table 1. Table 1 states that, at low AK (0.8-1.8 MPa &), the whiskers failed predominantly by shearing with a flat or square fracture without any visible evidence of necking; this was typical of fatigue fracture (Fig. 6). However, at high AK (2.8-3 MPa Table 1. Fractographic features in the fatigue and fracture regions in a 25 wt% silicon carbide whisker reinforced alumina Mechanism of failure Fatigue Monotonic Low AK High AK Fracture 0.8-1.8 2.8-3.0 5.96 MPad6 MPavSi MPa fi Whisker PntloQt T/&) 5 40 85 vs. shear (%) 95 60 15 Matrix Transgranular (x) 95 55 45 vs. intergranular (%) 5 45 55 o Branching P A A o Deflection P A A Crack Q River pattern P A A P = present; A = absent. Fig. 6. At low AK (0.8-1.8 MPa 6) region, the majority of the whiskers failed with a square fracture without evidence of large scale pullout. Fig. 7. At high AK (2.8-3 MPa 6) region the whiskers failed in a mixed mode, i.e. both with a square fracture and also by pullout. &), the whisker failed in two different ways: that is about 60% of the whiskers failed with a flat fracture and the balance failed by pullout (Fig. 7). On the other hand, in the fast fracture region, the whiskers failed predominantly by pullout (Fig. 8). Table 1 also reports that the matrix alumina grains failed predominantly through transgranular fracture (Fig. 9) at low AK (0.8-l .8 MPa&). In the high AK (2.8-3 MPa&) region, the matrix failed in a mixed mode wherein about 45% of the alumina grains failed by intergranular and the balance by the transgranular mode, as shown in Fig. 10. During monotonic loading, the alumina grains in the fast fracture region also failed in a mixed mode with a slightly increased percentage of intergranular (-55%) and the balance by transgranular mode-as shown in Fig. 11. It is noteworthy that in the low AK region, the crack had frequently branched (Fig. 12) and deflected (Fig. 13) while it propagated. River pattern
fractography of fatigued and fractured regions Fig.8. In the fast fracture region(Kc =5.9 MPavm) Fig. ll. In the fast fracture region(monotonic loading),the hikers failed predominantly by pullout mechanism alumina grains failed in a mixed mode, i.e. intergranular (-5S%o)and transgranular(-45) Fig.9. At low AK region, the alumina matrix failed prede nantly through transgranular fracture Fig 12. Crack branching in the low AK region Fig. 10. At high AK region, the alumina grains failed in a Fig. 13. Crack deflection in the low AK region mixed mode, i. e. intergranular (-45%)and transgranular (-55% 4 Discussion markings with steps(Fig. 14) were also present in Twenty-five weight percent silicon carbide whisker the cleavage facets. However, in the fast fracture reinforced alumina composite is susceptible to a region, the branching and deflection of the crack fatigue crack growth phenomenon(Fig. 4), were virtually absent which is simila that in the case of metallic
Fractography of fatigued and fractured regions Fig. 8. In the fast fracture region (K,, = 59 MPa&), whiskers failed predominantly by pullout mechanism. Fig. 11. In the fast fracture region (monotonic loading), the alumina grains failed in a mixed mode, i.e. intergranular (-55%) and transgranular (-45%). Fig. 9. At low AK region, the alumina matrix failed predominantly through transgranular fracture. Fig. 12. Crack branching in the low AK region. Fig. 10. At high AK region, the alumina grains failed in a mixed mode, i.e. intergranular (-45%) and transgranular (-5 5%). markings with steps (Fig.14) were also present in Twenty-five weight percent silicon carbide whisker the cleavage facets. However, in the fast fracture reinforced alumina composite is susceptible to a region, the branching and deflection of the crack fatigue crack growth phenomenon*,2 (Fig. 4), were virtually absent. which is similar to that in the case of metallic Fig. 13. Crack deflection in the low AK region. 4 Discussion
A. K Ray, S. K. Das, P. K. Roy, S. Banerjee Thus, it is likely that the whiskers continue to bridge the crack even after some of the surround ing alumina grains had failed by transgranular cleavage. Therefore, unde circumstances the whisker could fail due to tensile fracture or One might rule out the possibility of tensile fracture if one had examined the relative values of strength and y modulus ed in Refs 9 and 10. The fracture strength of the whisker is about 6 5-7 times higher than that of alumina Accordingly, the tensile fracture of the whisker was unlikely if the traction forces ahead of the crack tip were to be distributed in proportion to the relative cross sections of the whisker and al Fig. 14. River pattern marking with steps in the low AK region(YZ-modulation SEM image) mina in the plane of the crack. Even if the whiskers were to carry the major part of the trac- metals, but with a higher crack growth exponent tion forces ahead of the crack tip, the high trac- (n =15 5). Figure 4 shows that the FCgr data of tion forces would produce failure by pullout this material fits the usual Paris equation da/dN= rather than by tensile fracture, as was observed in A(△K) the fast fracture region where Kic values and con Since the manner of failure of the whiskers and sequently the traction forces, were high the a-alumina grains were entirely different in the Since the possibility of large scale failure by low AK fatigue and the fast fracture regions, the pullout or tensile fracture schemes of crack growth under these two condi- whiskers were most likely to fail by fatigue. At tions are discussed separately, to focus on their low AK, the whiskers which had bridged the crack contradistinctions after the surrounding alumina grains had failed by cleavage, would continue to experience fatigue 4.1 Low Ak fatigue region loading, then failed with a square fracture which During fatigue loading at the low AK regions, is typical of fatigue. The fatigue failure of the sili- where AK=0-8-1-8 MPa vm the alumina grains con carbide whisker under such circumstances, of the matrix failed mainly by transgranular cleav- could be significantly influenced by the stacking ge with frequent crack branching and crack defl- fault such as is reported Ref. 1l in Fig. 15, since ection. Also, the cleavage facets revealed river such fault could be responsible for nucleating the pattern and steps. At low AK, the Kc values and fatigue crack. However, this aspect needs careful consequently the traction force ahead of the crack study tip, were not adequate to promote intergranular Since the cleavage of the a-alumina grains prob- fracture ibly occurred before the whiskers (located mainly At low AK, the amount of crack extension is at the grain boundaries) had failed, it was very low. So even though the energy dissipated to improve the fatigue resistance of the a-alum per cycle is low, the number of cycles and the ina grains. This apart, the dislocation pile-up cumulative energy required for a given amount of crack extension is quite significant and can thus account for the higher surface area required fo the various features encountered like crack branching, river pattern and steps t low AK fatigue, the whisker could fail in three different modes ● pullout, ● tensile fracture e fatigue Table I shows that only a few of the whiskers 0-IR failed due to pullout at low AK fatigue. This natu- rally indicated that the maximum traction forces generated ahead of the crack tip during the low Fig. 15. TEM of B-Sic whisker showing stacking fault by AK fatigue was not adequate to produce a pullout lark field and bright field techniques
196 A. K. Ray, S. K. Das, P. K. Roy, S. Banerjee Fig. 14. River pattern marking with steps in the low AK region (YZ-modulation SEM image). metals, but with a higher crack growth exponent (n = 15.5). Figure 4 s h ows that the FCGR data of this material fits the usual Paris equation da/dN = A(AK)“. Since the manner of failure of the whiskers and the a-alumina grains were entirely different in the low AK fatigue and the fast fracture regions, the schemes of crack growth under these two conditions are discussed separately, to focus on their contradistinctions. 4.1 Low AK fatigue region During fatigue loading at the low AK regions, where AK = 0.8-1.8 MPa drn the alumina grains of the matrix failed mainly by transgranular cleavage with frequent crack branching and crack deflection. Also, the cleavage facets revealed river pattern and steps. At low AK, the K, values and consequently the traction force ahead of the crack tip, were not adequate to promote intergranular fracture. At low AK, the amount of crack extension is very low. So even though the energy dissipated per cycle is low, the number of cycles and the cumulative energy required for a given amount of crack extension is quite significant and can thus account for the higher surface area required for the various features encountered like crack branching, river pattern and steps. At low AK fatigue, the whisker could fail in three different modes 0 pullout, 0 tensile fracture, 0 fatigue. Table 1 shows that only a few of the whiskers failed due to pullout at low AK fatigue. This naturally indicated that the maximum traction forces generated ahead of the crack tip during the low AK fatigue was not adequate to produce a pullout. Thus, it is likely that the whiskers continue to bridge the crack even after some of the surrounding alumina grains had failed by transgranular cleavage. Therefore, under these circumstances, the whisker could fail due to tensile fracture or fatigue. One might rule out the possibility of tensile fracture if one had examined the relative values of strength and Young’s modulus reported in Refs 9 and 10. The fracture strength of the whisker is about 6.5-7 times higher than that of alumina. Accordingly, the tensile fracture of the whisker was unlikely if the traction forces ahead of the crack tip were to be distributed in proportion to the relative cross sections of the whisker and alumina in the plane of the crack. Even if the whiskers were to carry the major part of the traction forces ahead of the crack tip, the high traction forces would produce failure by pullout rather than by tensile fracture, as was observed in the fast fracture region where K,, values and consequently the traction forces, were high. Since the possibility of large scale failure by pullout or tensile fracture was discounted, whiskers were most likely to fail by fatigue. At low AK, the whiskers which had bridged the crack after the surrounding alumina grains had failed by cleavage, would continue to experience fatigue loading, then failed with a square fracture which is typical of fatigue. The fatigue failure of the silicon carbide whisker under such circumstances, could be significantly influenced by the stacking fault such as is reported Ref. 11 in Fig. 15, since such fault could be responsible for nucleating the fatigue crack. However, this aspect needs careful study. Since the cleavage of the a-alumina grains probably occurred before the whiskers (located mainly at the grain boundaries) had failed, it was necessary to improve the fatigue resistance of the a-alumina grains. This apart, the dislocation pile-up Fig. 15. TEM of P-Sic whisker showing stacking fault by dark field and bright field techniques
fractography of fatigued and fractured regions model would obviously indicate that the cleav- nantly by transgranular cleavage and the advanc failure of a-alu prevented ing crack frequently branched and tilted and through the refinement of alumina grains in the twisted in its course of deflection. Since the aggregate structure. The size of alumina grains whiskers were predominantly located at the grain varied between 1 and 6 um. In this context, one boundaries and the crack propagation along the should note that one could obtain higher resis- grain boundary at low AK fatigue was negligible, tance to low AK fatigue, if ultra-fine or nano- the possibility that the crack would encounter size alumina grains, such as that produced by the these whiskers as it advances would be rather low. solgel technique, were to be used to fabricate this Only a few of those whiskers which had bridged the crack with an orientation normal to the A schematic view of crack propagation mecha- advancing crack plane, could fail by pullout. The nism is representated in Fig. 16(a). It illustrates other whiskers were likely to fail by directly expe- that at low AK fatigue the matrix failed predomi- riencing fatigue loading. This could explain why SILICO WHISKER CRACK REDOMINANTLY INTERGRANULAR WHISKER PULLOUT MONOTONIC LOADING PREDOMINANTLY TRANSGRANULAR WHISKER SHEAR Low△ K FATIGUE Fig. 16. Schematic of the crack propagation during monotonic and cyclic loading
Fractography of fatigued and fractured regions 197 model12 would obviously indicate that the cleavage failure of a-alumina could be prevented through the refinement of alumina grains in the aggregate structure. Tlhe size of alumina grains varied between 1 and 6 pm. In this context, one should note that one could obtain higher resistance to low AK fatigue, if ultra-fine or nanosize alumina grains, such as that produced by the solgel technique, were to be used to fabricate this composite. A schematic view of crack propagation mechanism is representated in Fig. 16(a). It illustrates that at low AK fatigue, the matrix failed predominantly by transgranular cleavage and the advancing crack frequently branched and tilted and twisted in its course of deflection. Since the whiskers were predominantly located at the grain boundaries and the crack propagation along the grain boundary at low AK fatigue was negligible, the possibility that the crack would encounter these whiskers as it advances would be rather low. Only a few of those whiskers which had bridged the crack with an orientation normal to the advancing crack plane, could fail by pullout. The other whiskers were likely to fail by directly experiencing fatigue loading. This could explain why WHISKER’ PULLOUT CRACK PREDOMINANTLY INTERGRANULAR MONOTONIC LOADING PREDOMIYANTLY TRANSGRANULAR WHISKER SHEAR LOW AK FATIGUE Fig. M, Schematic of the crack propagation during monotonic and cyclic loading
A.K. Ray, S. K. Das, P. K. Roy, S. Banerjee at low AK fatigue, the majority of the whiskers Since the ceramic material is often expected to failed with a square fracture and without pullout. survive both monotonic and fatigue loading, one could make the following general observations 4.2 Fast fracture region based on the scheme of crack growth under the During the fast fracture under monotonic loading two conditions as discussed above such as during the fracture toughness testing, To prevent fracture failure of this ceramic mate- some of the a-alumina grains of the matrix failed rial under monotonic loading, its microstructural by intergranular fracture; the others failed by features would have to be strengthened to avoid transgranular cleavage. The value of Kic during intergranular fracture of a-alumina al ind the pull- such monotonic loading was 5.9 MPa Vm. At such out of the whisker. This requires that the grain values of Kc, the traction force generated ahead of boundary strength of a-alumina and the the crack tip was adequate to produce the inter- whisker--matrix interfacial strength should be granular failure in those grains which were increased Similarly, to retard crack extension dur- favourably oriented with respect to the direction ing low AK fatigue, intergranular cleavage fracture of the traction force. Debonding and pullout of of a-alumina grains and fatigue failure resistance those whiskers at the fractured grain boundaries of the silicon carbide whiskers should be would therefore occur It is difficult to summarise from the fracto- powder, ,and by improving the uniformity of graphic features as to which of these, i.e. trans- distribution of whiskers in the matrix during hot granular cleavage or intergranular fracture, pressing occurred, at first, under monotonic loading After some of the grains had failed through the intergranular mode, the other neighbouring grains 5 Conclusions hich were less favourably oriented or which were strongly anchored by the whiskers across the grain The micromechanism of fracture of a 25 wt% sili- boundaries, failed through transgranular cleavage. con carbide whisker reinforced alumina composite The pullout of the whiskers could be avoided when loaded in fatigue was quite different from if the whisker a-alumina interfacial strength was that when loaded monotonically. Consequently improved or if finer and more numerous whiskers the fractographic features in the two instances of and also finer a-alumina particles were used in loading were significantly different fabricating the composite. Similarly, the intergran When loaded in fatigue, the alumina matrix ular fracture could be avoided by strengthening failed in a transgranular mode and the whiskers the boundaries failed by producing a fat fracture but without As illustrated schematically in Fig. 16(b), the their pullout. fracture due to monotonic loading showed a sub- When loaded monotonically, the whiskers pre- stantial amount of intergranular failure of the doninantly failed by pullout and the alumina matrix and pullout of the whiskers. Crack deflec- matrix failed in a mixed mode with about 55% in tion, branching and the characteristic river intergranular and the balance in transgranular pattern marking with steps, were absent in this mode. Whereas the grain boundary strength of region a-alumina and the whisker-matrix interfacial The matrix failure was predominantly inter- bonding should be increased to produce increased granular. Nevertheless, the whiskers were located resistance to failure under monotonic loading. The mainly at the grain boundaries. Therefore, the fatigue failure resistance of the composite could probability of the advancing crack front interact- be improved by improving the cleavage strength of ing with the whiskers was extremely high. The the a-alumina grains, i. e. by using finer a-alumina hikers tending to bridge the advancing crack particles and by increasing the fatigue strength of quite frequently would therefore get debonded the whiskers by improving the uniformity in the and pulled out-which is schematically illustrated distribution of B-SiC whiskers during hot pressing nd shown in Fig. 16(b). This justifies the fact that during the monotonic fracture, failure of whiskers by the pullout mechanism was frequently observed Acknowledgements in the fractograph. During fatigue loading at high AK, mechanism The authors gratefully acknowledge the help and of fracture was an intermediate situation between advice of Mr N. K. das and professor o. n the two extreme cases: that is between the case of Mohanty in the SEM work. The authors are thankful monotonic fast fracture and that of the low ak to professor p. ramachandrarao director fatigue fracture National Metallurgical Laboratory, for allowing us
198 A. K. Ray, S. K. Das, P. K. Roy, S. Banerjee at low AK fatigue, the majority of the whiskers failed with a square fracture and without pullout. 4.2 Fast fracture region During the fast fracture under monotonic loading such as during the fracture toughness testing, some of the o-alumina grains of the matrix failed by intergranular fracture; the others failed by transgranular cleavage. The value of K,, during such monotonic loading was 5.9 MPa fi. At such values of Kc, the traction force generated ahead of the crack tip was adequate to produce the intergranular failure in those grains which were favourably oriented with respect to the direction of the traction force. Debonding and pullout of those whiskers at the fractured grain boundaries would therefore occur. Since the ceramic material is often expected to survive both monotonic and fatigue loading, one could make the following general observations based on the scheme of crack growth under the two conditions as discussed above. It is difficult to summarise from the fractographic features as to which o&these, i.e. transgranular cleavage or intergranular fracture, occurred, at first, under monotonic loading. To prevent fracture failure of this ceramic material under monotonic loading, its microstructural features would have to be strengthened to avoid intergranular fracture of o-alumina and the pullout of the whisker. This requires that the grain boundary strength of a-alumina and the whisker-matrix interfacial strength should be increased. Similarly, to retard crack extension during low AK fatigue, intergranular cleavage fracture of a-alumina grains and fatigue failure resistance of the silicon carbide whiskers should be improved, possibly by using finer a-alumina powder6,’ and by improving the uniformity of distribution of whiskers in the matrix during hot pressing.6,7 After some of the grains had failed through the intergranular mode, the other neighbouring grains which were less favourably oriented or which were strongly anchored by the whiskers across the grain boundaries, failed through transgranular cleavage. 5 Conclusions The pullout of the whiskers could be avoided if the whisker o-alumina interfacial strength was improved or if finer and more numerous whiskers and also finer a-alumina particles were used in fabricating the composite. Similarly, the intergranular fracture could be avoided by strengthening the boundaries. The micromechanism of fracture of a 25 wt% silicon carbide whisker reinforced alumina composite when loaded in fatigue was quite different from that when loaded monotonically. Consequently, the fractographic features in the two instances of loading were significantly different. As illustrated schematically in Fig. 16(b), the fracture due to monotonic loading showed a substantial amount of intergranular failure of the matrix and pullout of the whiskers. Crack deflection, branching and the characteristic river pattern marking with steps, were absent in this region. When loaded in fatigue, the alumina matrix failed in a transgranular mode and the whiskers failed by producing a flat fracture but without their pullout. The matrix failure was predominantly intergranular. Nevertheless, the whiskers were located mainly at the grain boundaries. Therefore, the probability of the advancing crack front interacting with the whiskers was extremely high. The whiskers tending to bridge the advancing crack quite frequently would therefore get debonded and pulled out-which is schematically illustrated and shown in Fig. 16(b). This justifies the fact that during the monotonic fracture, failure of whiskers by the pullout mechanism was frequently observed in the fractograph. When loaded monotonically, the whiskers predominantly failed by pullout and the alumina matrix failed in a mixed mode with about 55% in intergranular and the balance in transgranular mode. Whereas the grain boundary strength of a-alumina and the whisker-matrix interfacial bonding should be increased to produce increased resistance to failure under monotonic loading. The fatigue failure resistance of the composite could be improved by improving the cleavage strength of the a-alumina grains, i.e. by using finer a-alumina particles and by increasing the fatigue strength of the whiskers by improving the uniformity in the distribution of P-Sic whiskers during hot pressing. Acknowledgements During fatigue loading at high AK, mechanism The authors gratefully acknowledge the help and of fracture was an intermediate situation between advice of Mr N. K. Das and Professor 0. N. the two extreme cases: that is between the case of Mohanty in the SEM work. The authors are thankful monotonic fast fracture and that of the low AK to Professor P. Ramachandrarao Directorfatigue fracture. National Metallurgical Laboratory, for allowing us
fractography of fatigued and fractured regions to publish this work and to NIST(National Insti- 5. Cook, J. L&Rhodes,J.R whiskers ute of Standards and Technology) NBS-USA for Presented at the 88th annual upplying us the ceramic specimens Ceramic Society, Chicago, IL (Engineer- ing Ceramics Division, Paper No. 18-C-86) References pment o whisker--reinforced ceramics. J. Am. Ceram. Soc. -. 1. Ray, A.K., Fuller, E. 64(2)(1985)298-304. growth rate and fractur 8. Brown, W.F., Jr& rawley, J. E, American Society for Test silicon carbide whisk ag and Materials STP 410. ASTM, Philadelphia, PA, 1966 with residual porosity 9. Ray, A.K., Mohanty, G.& Ghose, A, Efects of cata (1994) lysts and temperature on silicon carbide whiskers forma- 2. Ray, A. K.& banerjee, s, Precracking of ceramic speci tion from rice husk.J. Mater. Sci. Lett., 10( 1991)227-9 men and determination of fracture crack growth rate of 10. Fisher, E. S, Manghnani, M. H.& Routbort, J. L 25 wt% Sic reinforced AlO3 composite. Accepted for Study of the clastic propcrties of Al,O, and Si3N4 mate publication in J. Am. Ceram Soc.(1995) rial composites with SiC, whisker reinforcement. High 3. Krause. R, F. jr fuller, E.r. Jr. fracture resistance Performance Composites for the 1990s, ed S.K. Das, C behaviour of silicon carbide whisker-reinforced alumina P. Ballard and F. Marikar. The Minerals, Metals and opposites with different porosites J. Am. Ceram. Soc. Materials Society, 1991 3(3)(1990)55966 11. Reinforcing Tumurrowy Technolugy Ceram Indust. April 4. Dauskardt R. H. James, m.R.. porter,J. R.& Ritchie (1992). R Cyclic fatigue crack growth in a Sic whisker 12. Kingery. w.D., Bowen, H. K.& Uhlmann, D.R., Intro- reinforced alumina ceramic composite. J. Am. Ceram ion to Ceramics, 2nd ed, John Wiley and Sons, Inc Soc,755)(1992)759-71 York,1976,p.795
Fractography of fatigued and fractured regions 199 to publish this work and to NIST (National Institute of Standards and Technology) NBS-USA for supplying us the ceramic: specimens. References Ray, A. K., Fuller, E. R.. & Banerjee, S., Fatigue crack growth rate and fracture toughness of 25 weight percent silicon carbide whisker reinforced alumina composite with residual porosity. Submitted to J. Eur. Ceram. Sot. (1994). Ray, A. K. & Banerjee, S., Precracking of ceramic specimen and determination of fracture crack growth rate of 25 wt% Sic reinforced .41,0, composite. Accepted for publication in J. Am. Ceram. Sot. (1995). Krause, R. F. Jr & Fuller, E. R. Jr, Fracture resistance behaviour of silicon carbide whisker-reinforced alumina composites with different porosites. J. Am. Ceram. Sot., 73(3) (1990) 559-66. Dauskardt, R. H., James, M. R., Porter, J. R. & Ritchie, R. O., Cyclic fatigue crack growth in a Sic whiskerreinforced alumina ceramic composite. J. Am. Ceram. Sot., 75(5) (1992) 759-71. 5. 6. 7. 8. 9. 10. 11. 12. Cook, J. L. & Rhodes, J. R., Silicon carbide whiskers. Presented at the 88th Annual Meeting of the American Ceramic Society, Chicago, IL, April 28, 1986 (Engineering Ceramics Division, Paper No. 18-C-86). Becher, P. F. & Wei, G. C., Toughening behaviour in Sic-whisker-reinforced alumina. J. Am. Ceram. Sot., 67( 12) (1984) 267-9. Wei, G. C. & Becher, P. F., Development of SiCwhisker-reinforced ceramics, J. Am. Ceram. Sot.-Bull., 64(2) (1985) 298-304. Brown, W. F., Jr & Srawley, J. E., American Society for Testing and Materials STP--410. ASTM, Philadelphia, PA, 1966. Ray, A. K., Mohanty, G. & Ghose, A., Effects of catalysts and temperature on silicon carbide whiskers formation from rice husk. J. Mater. Sci. Lett., 10 (1991) 227-9. Fisher, E. S., Manghnani, M. H. & Routbort, J. L., Study of the elastic properties of A&O, and Si,N, material composites with Sic, whisker reinforcement. High Performance Composites for the 199Os, ed. S. K. Das, C. P. Ballard and F. Marikar, The Minerals, Metals and Materials Society, 199 1. Reinforcing Tomorrows Technology. Ceram. Indust. April (1992). Kingery, W. D., Bowen, H. K. & Uhlmann, D. R., Zntroduction to Ceramics, 2nd ed., John Wiley and Sons, Inc. New York, 1976, p. 795
