《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_brittleness29

CERAMICS INTERNATIONAL ELSEVIER Ceramics International 29(2003)777-784 www.elsevier.com/locate/ceramint Fracture toughness of ceramics and ceramic composites George A Gogotsi Institute for Problems of strength, 2 Timiryazevskaya Street, Kiev 01014, Ukrain Received 22 July 2002: received in revised form 9 September 2002: accepted 10 November 2002 Abstract The fracture toughness of zirconia, alumina, and silicon nitride ceramics, zirconia and alumina single crystals, silicon carbide as well as silicon nitride ceramic particulate composites, silicon nitride laminated composites, and other ceramics materials were studied by a single edge v-notched beam (SEVNB) method Manual and mechanical procedures for V-notch polishing-out and three-and four-point flexure tests were developed. Load-deflection diagrams for V-notched specimens contributed to better understanding of the deformation behavior of ceramics at room temperature and 1300-1400C sEnb (single edge notched beam) and sEPB (single edge precracked beam) as well as micro-Raman spectroscopy data were used to analyze the sevPb results. The ratio between SevNb and senb results is about 0.6 for elastic materials, over 0.9 for inelastic ones, and about 1.0 for laminated ceramic composite. The polishing-out of a V-notch does not lead to the tetragonal-monoclinic phase transformation in zirconia C 2003 Elsevier Ltd and Techna S r l. All rights reserved Keywords: B Composites; B Spectroscopy: C. Fracture; C. Toughness; Ceramics; SEVNB-method 1. ntroduction steps in this field. Leading specialists from different countries engaged in those investigations performed Ceramics and ceramic composites are promising statistical analyses to compare the test methods sum materials having rather high strength characteristics but marized in Table 1. They were helpful for creating the luite low crack resistance properties at the same time. basis of more reliable determination of the fracture This is one of the major factors hindering the wide-scale toughness characteristics of ceramics. For a long time, application of these materials in various fields of human no particular preference was given to any of the meth activities. The crack resistance is critical not only for ods considered because each of them had appreciable ceramic products operating under extreme mechanical shortcomings, which restricted their application [2, 3] and thermal loads but also for structural components Recently, much attention has been focused on the whose brittle fracture is intolerable even under arbitrary SEVNB method [4], which is a further development of a loads simple and widely used the single edge notched beam For many years, the performance of ceramics has (SENB) method suitable only for rough fracture tough been evaluated on the basis of full-scale tests. However, ness evaluations. The test data obtained by the SEVnB their fracture toughness characteristics have not always method were found to be comparable to those obtained been the object of scientific interest. Wide-scale fracture by the single edge precracked beam (SEPB) method toughness investigations were started only in the late usually considered as providing the real values of cri- 1980s International prestandard studies(Round Robin tical stress intensity factor(Klc) but this test is often on Fracture Toughness-RRFT [ID aimed at the assess- difficult to realize in practice [5]. To evaluate the poten ment of the accuracy and reliability of the data obtained tials of the SEvnb method as a standard one for testing by commonly accepted test methods were important ceramics, its analysis was performed [6]. We verified the efficiency of our procedures and demonstrated the applicability of the SEVNB method to different ceramic E-mail address: ggogotsi @ipp. adam. kiev. ua(GA. Gogotsi). materials 0272-8842/03/$30.00@ 2003 Elsevier Ltd and Techna S.r.l. All rights reserved. doi:10.l016/S0272-8842(02)00230-4

Fracture toughness of ceramics and ceramic composites George A. Gogotsi Institute for Problems of Strength, 2 Timiryazevskaya Street, Kiev 01014, Ukraine Received 22 July 2002; received in revised form 9 September 2002; accepted 10 November 2002 Abstract The fracture toughness of zirconia, alumina, and silicon nitride ceramics, zirconia and alumina single crystals, silicon carbide as well as silicon nitride ceramic particulate composites, silicon nitride laminated composites, and other ceramics materials were studied by a single edge V-notched beam (SEVNB) method. Manual and mechanical procedures for V-notch polishing-out and three- and four-point flexure tests were developed. Load–deflection diagrams for V-notched specimens contributed to better understanding of the deformation behavior of ceramics at room temperature and 1300–1400
C. SENB (single edge notched beam) and SEPB (single edge precracked beam) as well as micro-Raman spectroscopy data were used to analyze the SEVPB results. The ratio between SEVNB and SENB results is about 0.6 for elastic materials, over 0.9 for inelastic ones, and about 1.0 for laminated ceramic composite. The polishing-out of a V-notch does not lead to the tetragonal-monoclinic phase transformation in zirconia ceramics. # 2003 Elsevier Ltd and Techna S.r.l. All rights reserved. Keywords: B. Composites; B. Spectroscopy; C. Fracture; C. Toughness; Ceramics; SEVNB-method 1. Introduction Ceramics and ceramic composites are promising materials having rather high strength characteristics but quite low crack resistance properties at the same time. This is one of the major factors hindering the wide-scale application of these materials in various fields of human activities. The crack resistance is critical not only for ceramic products operating under extreme mechanical and thermal loads but also for structural components whose brittle fracture is intolerable even under arbitrary loads. For many years, the performance of ceramics has been evaluated on the basis of full-scale tests. However, their fracture toughness characteristics have not always been the object of scientific interest. Wide-scale fracture toughness investigations were started only in the late 1980s. International prestandard studies (Round Robin on Fracture Toughness–RRFT [1]) aimed at the assess￾ment of the accuracy and reliability of the data obtained by commonly accepted test methods were important steps in this field. Leading specialists from different countries engaged in those investigations performed statistical analyses to compare the test methods sum￾marized in Table 1. They were helpful for creating the basis of more reliable determination of the fracture toughness characteristics of ceramics. For a long time, no particular preference was given to any of the meth￾ods considered because each of them had appreciable shortcomings, which restricted their application [2,3]. Recently, much attention has been focused on the SEVNB method [4], which is a further development of a simple and widely used the single edge notched beam (SENB) method suitable only for rough fracture tough￾ness evaluations. The test data obtained by the SEVNB method were found to be comparable to those obtained by the single edge precracked beam (SEPB) method usually considered as providing the real values of cri￾tical stress intensity factor (KIc) but this test is often difficult to realize in practice [5]. To evaluate the poten￾tials of the SEVNB method as a standard one for testing ceramics, its analysis was performed [6]. We verified the efficiency of our procedures and demonstrated the applicability of the SEVNB method to different ceramic materials. 0272-8842/03/$30.00 # 2003 Elsevier Ltd and Techna S.r.l. All rights reserved. doi:10.1016/S0272-8842(02)00230-4 Ceramics International 29 (2003) 777–784 www.elsevier.com/locate/ceramint E-mail address: ggogotsi@ipp.adam.kiev.ua (G.A. Gogotsi)

G.A. Gogotsi/Ceramics International 29(2003)777-784 2. Experimental 2.2. Test specimens 2. Materials Rectangular beams (3 mmx4 mm cross-section) v- notched in two stages were used as test specimens [7 Isostatically pressed silicon nitride ceramics (Si3N4) The package of 2-6 specimens were glued onto a cera- [7], sintered (Y-PSZ)[8] and sintered under pressure (Y- mic plate and then V-notched. Two additional speci PSZa) partially-stabilized zirconia ceramics [9] were the mens were fixed on both sides of the package for its major materials used in the process of verifying the protection from the edge effects due to notching. A experimental procedures. Gas-pressure sintered silicon preliminary notch 0.8 mm deep(spec simens for lou nitride(GPSSN) as well as sintered alumina(Al2O3-998) point flexure)or 1.6 mm deep(specimens for three-point and silicon carbide (ssic)[6] analyzed in the rrFt97 flexure) was cut in the package by a 150-300 um thick also became the objects of the investigation. In our diamond disk. At the next stage, the V-notch was experiments we also used zirconia and alumina single manually (convenient for comparatively"soft'cera- posites with compressed SiaN4 and tensile-stressed polished-out after filling the preliminary notch wih S) crystals as well as silicon nitride based laminated com- mics)or mechanically (necessary for"hard"ceramics) Si3N4 n wt. TiN layers [10]. This composite, layers diamond paste of grain sizes ranging from 2 to 7 um of which were prepared by rolling, was produced by hot and a razor blade 150-300 um thick. In the case of pressing at 1850C for 20 min under pressure 150 MPa. manual polishing-out, the walls of the preliminary notch For high-temperature investigations, reaction-bonded (1.6 mm deep) served as the razor blade guides. The and sintered Si3N4 30% SiC 3% Mgo [11] and purpose of the test determined the number of notches hot-pressed (in graphite molds)Sic 50%ZrB 2 (one or three) cut in a standard 45-mm long specimen 10% B4C[12] ceramic particulate composites were also (Fig. 1). The criterion for the proper V-notch sharpness used. Some data on these ceramics are given in Table 2. is its width, S, equal to two root radii p. The notch Conventional ceramics fracture toughness test methods Specimen Method and fracture Specimen Method and fracture surface of specimen surface of specimen SEPB (surface crack cracked beam) SENB i hgea (single edg (indentation IF (chevron fracture) notched beam) Fig. 1. SEVNB specimens: a, b, three-point fiexure; c, four-point flexure

2. Experimental 2.1. Materials Isostatically pressed silicon nitride ceramics (Si3N4) [7], sintered (Y-PSZ) [8] and sintered under pressure (Y￾PSZa) partially-stabilized zirconia ceramics [9] were the major materials used in the process of verifying the experimental procedures. Gas-pressure sintered silicon nitride (GPSSN) as well as sintered alumina (Al2O3-998) and silicon carbide (SSiC) [6] analyzed in the RRFT’97 also became the objects of the investigation. In our experiments we also used zirconia and alumina single crystals as well as silicon nitride based laminated com￾posites with compressed Si3N4 and tensile-stressed Si3N4 + n wt.% TiN layers [10]. This composite, layers of which were prepared by rolling, was produced by hot pressing at 1850
C for 20 min under pressure 150 MPa. For high-temperature investigations, reaction-bonded and sintered Si3N4 + 30% SiC + 3% MgO [11] and hot-pressed (in graphite molds) SiC + 50%ZrB2 + 10% B4C [12] ceramic particulate composites were also used. Some data on these ceramics are given in Table 2. 2.2. Test specimens Rectangular beams (3 mm4 mm cross-section) V￾notched in two stages were used as test specimens [7]. The package of 2–6 specimens were glued onto a cera￾mic plate and then V-notched. Two additional speci￾mens were fixed on both sides of the package for its protection from the edge effects due to notching. A preliminary notch 0.8 mm deep (specimens for four￾point flexure) or 1.6 mm deep (specimens for three-point flexure) was cut in the package by a 150–300 mm thick diamond disk. At the next stage, the V-notch was manually (convenient for comparatively ‘‘soft’’ cera￾mics) or mechanically (necessary for ‘‘hard’’ ceramics) polished-out after filling the preliminary notch with a diamond paste of grain sizes ranging from 2 to 7 mm and a razor blade 150–300 mm thick. In the case of manual polishing-out, the walls of the preliminary notch (1.6 mm deep) served as the razor blade guides. The purpose of the test determined the number of notches (one or three) cut in a standard 45-mm long specimen (Fig. 1). The criterion for the proper V-notch sharpness is its width, S, equal to two root radii
. The notch Table 1 Conventional ceramics fracture toughness test methods Specimen Method and fracture surface of specimen Specimen Method and fracture surface of specimen SEPB (single edge precracked beam) SCF (surface crack in flexure) SENB (single edge notched beam) IS (indentation strength) CNB (chevron notched beam) IF (indentation fracture) Fig. 1. SEVNB specimens: a, b, three-point flexure; c, four-point flexure. 778 G.A. Gogotsi / Ceramics International 29 (2003) 777–784

G.A. Gogotsi/ Ceramics International 29(2003)777-784 length and radius were measured with optical(Olimpus For the recording of the load-deflection curves, a hi BX5IM)and scanning electron microscopes (x500 or sensitivity LVDT-based deflectometer was suspended better) on a specimen(Fig. 2)and was in no way connected to the testing arrangement [13]. The LVDT was located 23. Procedures outside the heating chamber in case of high-temperature The specimens were tested in three-point (16 mm span In all the experiments, the speed of the testing between the bearing rollers) or four-point flexure(20-40 machine crosshead was constant and equal to 0.5 mm, machine and fitted with a hard load cell, a system for Fracture toughness, Kle values were calculated in precision displacement of a loading rod, and loading accordance with ASTM standard [14](three-point flex- supports. The latter are equipped with an attachment ure tests) and Din one [15](four-point flexure tests) for the rotation of the bearing rollers during specimen Additional procedures for evaluating Klc and other loading. Three-point flexure required the precise align- mechanical characteristics are described elsewhere [16] ment of a specimen on the bearing rollers, with the axis a micro-Raman imaging m of the central bearing roller and the radius of the v- JSA)with a 514.5 nm excitation line of an argon laser notch root being in the same plane as the applied load .(100 objective and a- l-um diameter spot) was used In high-temperature tests, we used a loading block to examine the surfaces of fractured Y-PSZ specimens which is similar to that used in room temperature tests. by a procedure described in Ref [171 3. Results and discussion The results of comparative three-and four-point flex ure tests of monolithic ceramics and particular ceramic composites are summarized in Table 3, where the data obtained within the RRFT97 program are also cited The variation of Ki values as a function of notch root Fig. 2. Loading scheme for three-point flexure(a) and deflectometer radius was studied for silicon nitride and zirconia( Fig 3) Ispended on a specimen 3x4x45 mm'(b) Load-deflection diagrams for V-notched specimens are Table 2 Characteristics of some material Material Density (g/cm Strength at 20C(MPa) Hardness(GPa) Method of manufacture 14.1 GPSS 224 Y-PSZ 05 425 IP sintered Material of RRTF97 Comparative fracture toughness tests(MPa m/2) Three-point flexure(a/wa point flexure(a/Wa0. 2...0.3 Our results RRFT97 results 5.5±0.07(5 5.35±0.16(5) GPSS 5.3±0.04(5) 5.2±0.18(5) 5.36±0.34(129) Si3 N4+30%SiC+ 3% Mgo 27±0.14(4) 240±0.16(5) SSiC 266±0.20(4) 2.61±0.18(56) SiC+ 50%oRb,+10% BC 3.59±0.12(3) 3.51±0.15(3) A2O3998 3.5±0.05(5) 3.6±0.06(5) 3.57±0.22(135) 5.7±0.17(5) 59±0.19(5) ± Standard deviation. The number of specimen tested (in parentheses)

length and radius were measured with optical (Olimpus BX51M) and scanning electron microscopes (500 or better). 2.3. Procedures The specimens were tested in three-point (16 mm span between the bearing rollers) or four-point flexure (20–40 mm spans between the bearing rollers) on a home-made Ceramtest block, mounted on a universal testing machine and fitted with a hard load cell, a system for precision displacement of a loading rod, and loading supports. The latter are equipped with an attachment for the rotation of the bearing rollers during specimen loading. Three-point flexure required the precise align￾ment of a specimen on the bearing rollers, with the axis of the central bearing roller and the radius of the V￾notch root being in the same plane as the applied load. In high-temperature tests, we used a loading block which is similar to that used in room temperature tests. For the recording of the load–deflection curves, a high￾sensitivity LVDT-based deflectometer was suspended on a specimen (Fig. 2) and was in no way connected to the testing arrangement [13]. The LVDT was located outside the heating chamber in case of high-temperature tests. In all the experiments, the speed of the testing machine crosshead was constant and equal to 0.5 mm/ min. Load cell and deflectometer readings were recorded with a coordinate potentiometer. Fracture toughness, KIc values were calculated in accordance with ASTM standard [14] (three-point flex￾ure tests) and DIN one [15] (four-point flexure tests). Additional procedures for evaluating KIc and other mechanical characteristics are described elsewhere [16]. A micro-Raman imaging microscope (Renishaw, USA) with a 514.5 nm excitation line of an argon laser (100 objective and a1-mm diameter spot) was used to examine the surfaces of fractured Y-PSZ specimens by a procedure described in Ref. [17]. 3. Results and discussion The results of comparative three- and four-point flex￾ure tests of monolithic ceramics and particular ceramic composites are summarized in Table 3, where the data obtained within the RRFT’97 program are also cited. The variation of KIc values as a function of notch root radius was studied for silicon nitride and zirconia (Fig. 3). Load–deflection diagrams for V-notched specimens are Fig. 2. Loading scheme for three-point flexure (a) and deflectometer suspended on a specimen 3445 mm3 (b). Table 3 Comparative fracture toughness tests (MPa m1/2) Test method Three-point flexure (a/W0.5) Four-point flexure (a/W0.2...0.3) Our results RRFT’97 results Si3N4 5.50.07 (5)a 5.350.16 (5) – GPSSN 5.30.04 (5) 5.20.18 (5) 5.360.34 (129) Si3N4+30%SiC+3% MgO 2.270.14 (4) 2.400.16 (5) – SSiC 2.45 (1) 2.660.20 (4) 2.610.18 (56) SiC+50%ZrB2+10% B4C 3.590.12 (3) 3.510.15 (3) – Al2O3-998 3.50.05 (5) 3.60.06 (5) 3.570.22 (135) Y-PSZ 5.70.17 (5) 5.90.19 (5) –  Standard deviation. a The number of specimen tested (in parentheses). Table 2 Characteristics of some materials Material Density (g/cm3 ) Strength at 20
C (MPa) Hardness (GPa) Method of manufacture Si3N4 3.14 700 14.1 HIP GPSSN 3.23 >920 13.5 Gas-pressure sintereda SSiC 3.15 – 22.4 Sintereda Al2O3-998 3.86 350 19.3 Sintereda Y-PSZ 6.05 425 12.1 IP + sintered a Material of RRTF’97. G.A. Gogotsi / Ceramics International 29 (2003) 777–784 779

G.A. Gogotsi/ Ceramics international 29(2003)777-784 三 Y-PSZ v月9 Y-PSZa , P__SEPB-method V-notch root radius (um) V-notch root radius(um) Fig 3. Effect of V-notch root radii on the Kls values for Si,N4(a)and Y-PSz (b)ceramics 60 z40 4 mm x 5mm cr 3mm x 4mm cross-section pecten 024681012141618 6 Deflection(um) Deflection(um) Fig 4. Load-deflection diagrams for notched Si3 N4+ 30% SiC+ 3% Mgo (a) and SiC+ 50% ZrB2+10% BC(b)specimens tested at room tem- perature(1, 3)and at 1400C(2, 4) presented in Figs 4 and 5. The results of micro-Raman racy of our test procedures. The comparison of the data analysis are given in Fig. 6. The comparative data on presented in Table 3 also points to the fact that essential SEVNB and senB results are summarized in Tables 4-6. differences between three- and four-point flexure results High-temperature test results are cited in Table 7 are absent. a similar conclusion was also made else- It is useful to start the analysis with emphasis on the where [18] for ceramic matrix composites. Conse Kle values obtained in three- and four-point flexure quently, both test methods might be considered (Table 3). The data presented in this table demonstrate identical. Moreover, four-point flexure can be more good agreement between our results and the average easily applied in practice because it does not require a results of RRFT97 [6], which confirms sufficient accu- precise placement of specimens on the bearing rollers, which is difficult to achieve without a trained operator On the other hand, three-point flexure tests can utilize small-size specimens, which is advantageous for materi- p1=20m als science research t The analysis of results(Fig. 3)shows that a decrease decrease in the Kle values for Si3 N4 and Y-PSZ ceramics (similar relation was also observed for other materials [2D. It is interesting to note that the fracture toughness- Si3N4(Fig. 3a) Deflection 8, um the vicinity of a small notch root radius as compared to Fig. 5. Load-deflection diagrams for notched SiC+50%- that of the Y-PSZ(Fig. 3b). Such an effect is probably TiB+10%BC specimens with V-notch root radiuses PI and p2 are determined by different sensitivity of these materials to qual (1 and 2)and differed (3)in values on them opposite side stress concentrations because of differences in their

presented in Figs. 4 and 5. The results of micro-Raman analysis are given in Fig. 6. The comparative data on SEVNB and SENB results are summarized in Tables 4–6. High-temperature test results are cited in Table 7. It is useful to start the analysis with emphasis on the KIc values obtained in three- and four-point flexure (Table 3). The data presented in this table demonstrate good agreement between our results and the average results of RRFT’97 [6], which confirms sufficient accu￾racy of our test procedures. The comparison of the data presented in Table 3 also points to the fact that essential differences between three- and four-point flexure results are absent. A similar conclusion was also made else￾where [18] for ceramic matrix composites. Conse￾quently, both test methods might be considered identical. Moreover, four-point flexure can be more easily applied in practice because it does not require a precise placement of specimens on the bearing rollers, which is difficult to achieve without a trained operator. On the other hand, three-point flexure tests can utilize small-size specimens, which is advantageous for materi￾als science research. The analysis of results (Fig. 3) shows that a decrease in the V-notch radius of a specimen leads to an essential decrease in the KIc values for Si3N4 and Y-PSZ ceramics (similar relation was also observed for other materials [2]). It is interesting to note that the fracture toughness￾notch root radius curve for Si3N4 (Fig. 3a) flattens in the vicinity of a small notch root radius as compared to that of the Y-PSZ (Fig. 3b). Such an effect is probably determined by different sensitivity of these materials to stress concentrations because of differences in their Fig. 3. Effect of V-notch root radii on the KIc values for Si3N4 (a) and Y-PSZ (b) ceramics. Fig. 4. Load–deflection diagrams for notched Si3N4+30% SiC+3% MgO (a) and SiC+50% ZrB2+10% B4C (b) specimens tested at room tem￾perature (1, 3) and at 1400
C (2, 4). Fig. 5. Load–deflection diagrams for notched SiC+50%- TiB2+10%B4C specimens with V-notch root radiuses
1 and
2 are equal (1and 2) and differed (3) in values on them opposite sides. 780 G.A. Gogotsi / Ceramics International 29 (2003) 777–784

G.A. Gogotsi/ Ceramics international 29(2003)777-784 grain sizes( 4 um for Si3 N4 and >I um for Y-PSZ) as contradict the results of sintered SiC whisker-reinforced it was mentioned elsewhere [19]. It should be empha- Si3N4[18] tests. A similar tendency is also typical of Sic sized that only for notch root radii less than 5-7 um [7] ceramics [191 e Kle values for Y-PSZ ceramics agree with SEPb data The examination of the specimens fractured in the (similar behavior is also typical of fine-grained alumina tests revealed a fracture crack that propagated from the [20]. A similar conclusion follows from [6], where an points where "additional"stress concentrators were attempt was made to relate the V-notch radius values present. This observation also confirms the assumption required for the correct determination of Kle values to that the fracture of a loaded ceramic specimen starts the averaged values of this parameter obtained in the from a small crack ahead of a machined notch root [19] RRFT97 studies. As follows from Fig 3a, the decrease In this context, it is interesting to mention that K in Klc values for Si, N4 ceramics occurs with a decrease magnitudes are influenced by the sharpness of a notch in notch radii down to about 30 um, which does not root rather than by its shape [7] W 250 150 l00 Raman shift(cm") Raman shift(cm") Fig. 6. Raman spectra of Y-PSZ specimens fractured by the SEVNB method (a)and by indentation(b): nonfractured specimen surface (1), dia- mond saw surface (2), fracture surface(3), razor blade surface(4), indentation edge(5), and indentation center(6) Table 4 Kic values for ceramics obtained by SEVNB and SENB methods(MPa m/2) Test method Index p Brittleness measure, X SEVNB SENB 5.14±0.29 1.18±0 SiC+ 50%ZrB,+ 10%B, C 3.52±0.08 6.24士 0.564 5.17±0.06 9.12±0.29 Three 0.567 SSiC 4.42±0.23 Si3 N4+ 30%SiC+ 3%Mgo 2.49±0.16 0.920 Mg- PSZ (TSgrade) 944±0.1 10.2±0.27 0.925 s& coo Three 0.25 The width Table 5 KIe values for single crystals obtained by SEVNB and SENB methods(MPa m) Single crystals Peculiarity Test method Index p modulus(GPa) measure, X SENB SEVN Zirconia Partially stabilized (3%Y203 9.33±0.95 0.33±2.171 Alumina Specimen axis 45 to optical axis of crystal 403 2.31±0.34 0.94 Specimen axis 90 to optical axis of cryst 410 3.19±0.53 85±0.50 1.12

grain sizes (4 mm for Si3N4 and >1 mm for Y-PSZ) as it was mentioned elsewhere [19]. It should be empha￾sized that only for notch root radii less than 5–7 mm [7] the KIc values for Y-PSZ ceramics agree with SEPB data (similar behavior is also typical of fine-grained alumina [20]. A similar conclusion follows from [6], where an attempt was made to relate the V-notch radius values required for the correct determination of KIc values to the averaged values of this parameter obtained in the RRFT’97 studies. As follows from Fig. 3a, the decrease in KIc values for Si3N4 ceramics occurs with a decrease in notch radii down to about 30 mm, which does not contradict the results of sintered SiC whisker-reinforced Si3N4 [18] tests. A similar tendency is also typical of SiC ceramics [19]. The examination of the specimens fractured in the tests revealed a fracture crack that propagated from the points where ‘‘additional’’ stress concentrators were present. This observation also confirms the assumption that the fracture of a loaded ceramic specimen starts from a small crack ahead of a machined notch root [19]. In this context, it is interesting to mention that KIc magnitudes are influenced by the sharpness of a notch root rather than by its shape [7]. Fig. 6. Raman spectra of Y-PSZ specimens fractured by the SEVNB method (a) and by indentation (b): nonfractured specimen surface (1), dia￾mond saw surface (2), fracture surface (3), razor blade surface (4), indentation edge (5), and indentation center (6). Table 4 KIc values for ceramics obtained by SEVNB and SENB methods (MPa m1/2) Material Test method Flexure (points) Index ’ Brittleness measure,  SEVNB SENBa Y-PSZ 5.140.29 9.540.47 Four 0.538 1 Soda limit glass 0.660.07 1.180.09 Four 0.562 1 SiC+50%ZrB2+10%B4C 3.520.08 6.240.37 Three 0.564 1 Si3N4 5.170.06 9.120.29 Three 0.567 1 SSiC 2.610.18 4.420.23 Four 0.590 1 Si3N4+30%SiC+3%MgO 2.270.14 2.490.16 Three 0.920 0.88 Mg-PSZ (TS-grade) 9.440.12 10.20.27 Four 0.925 0.41 La8.8Ca0.2CoO3 2.2 2.2 Three 10.25 a The width of notch is 0.2–0.3 mm. Table 5 KIc values for single crystals obtained by SEVNB and SENB methods (MPa m1/2) Single crystals Peculiarity Elastic modulus (GPa) Test method Brittleness measure,  Index ’ SENB SEVNB Zirconia Partially stabilized (3% Y2O3) 245 9.330.95 10.332.17 1 0.9 Alumina Specimen axis 45
to optical axis of crystal 403 2.310.34 2.450.29 10.94 Specimen axis 90
to optical axis of crystal 410 3.190.53 2.850.50 1 1.12 G.A. Gogotsi / Ceramics International 29 (2003) 777–784 781