Availableonlineatwww.sciencedirect.com SCIENCE DIRECT o WEAR ELSEVIER Wear256(2004)233-242 www.elsevier.com/locate/wear Solid-particle erosion and strength degradation of Si3N4/BN fibrous monoliths K C. Gorettaa F Gutierrez-Moraa Nan Chen a J. L. Routbort a, * T.A. Orlova b B.I. Smirnov b, A.R. de Arellano-LOpe Energy Technology Division, Argonne National Laboratory, Argonne, IL 60439-4838, USA loffe Physical-Technical Institute of the Russian Academy of Sciences, St. Petersburg 194021, Russia Departamento de Fisica de la Materia Condensada. Universidad de sevilla, 41080 Sevilla, Spain Received 7 November 2002; received in revised form 4 March 2003; accepted 14 March 2003 Abstract Erosive damage was studied in Si3 N4/Bn fibrous monoliths(FMs) and the individual constituents of the cells and cell boundary monolithic Si3N4 and BN. Unidirectional, 0/90, and +45 FMs were tested Specimens were subjected to impact at 90 by angular Sic particles of average diameter 143 um, traveling at 50-100 m/s. Steady-state erosion rates in the FMs were higher than predicted by a rule of mixtures based on erosion rates of the cell and cell-boundary phases. The relatively rapid FM erosion was attributed to chipping of the Si3 n4 cells caused by radial cracks. Bending strengths were measured before and after erosion testing to steady state at 100 m/s. the ength of monolithic Si3 N4 decreased 22%, the bn was not tested because insufficient material was available. Within experimental error, the strengths of the FMs were unaffected by erosion. Fracture data obtained approximately 1.5 years apart suggested that the FMs w susceptible to environmentally assisted slow crack growth O 2003 Elsevier B V. All rights reserved Keywords: Fibrous monoliths; Erosion; Strength; Silicon nitride; Boron nitride 1. Introduction BN FMs are produced commercially by Advanced Ceramics Research, Tucson, AZ. These and related fMs are being con Powder-derived ceramic fibrous monoliths(FMs) gener- sidered for aerospace applications in which foreign-object ally consist of strong ceramic cells that are surrounded by a damage is possible [11]. Resistance to impact erosion and weaker cell boundary. The cells are typically 100-500 um strength retention following damage are of concern in vide[1-10). FMs are produced by fabrication techniques aerospace and other possible applications for many FM such as extrusion or dip-coating. They are often fabricated We have previously studied erosive damage in Zrsioz as laminates that are laid up from duplex extruded filaments based FMs [12]. The cell boundary in those FMs consisted that consist of a cell phase surrounded by a sheath that forms of large-grained highly porous ZrSiO4. We found that the a continuous cell boundary. In flexure, FMs exhibit grace- FMs eroded more rapidly than would be expected from a ful failure, with energy dissipation arising from substantial simple rule-of-mixtures [13] for composite structures. The sliding of the cells[8]. FMs constitute a lower-cost alterna- anomalously rapid erosion was attributed to wholesale re- tive to conventional continuous-fiber ceramic composites in moval of cells once the supporting cell boundary had been Among ceramic FMs, those consisting of Si3 Na cells and In this study, we have conducted solid-particle erosion a continuous BN cell boundary have achieved the best over tests on Si3N4/BN FMs produced by Advanced Ceramics all mechanical properties and, therefore, have been studied Research. The cell boundary in these FMs is dense, in con most thoroughly [2-8]. Their flexural strengths can exceed trast to that of the ZrSio4 FMs. For comparison,mono- 700 MPa and work-of-fracture values, although typically 3- lithic Si3N4 and BN specimens were also tested. Strengths 6kJ/m, can exceed 10 kJ/m[3-7]. A wide variety of Si3 N4/ in flexure were measured before and after erosion testing The goals of this work were to determine the basic response ponding author. Tel:+1-603-252-5065; fax: +1-630-252-4298. of Si3 Na/BN FMs to erosion and the extent to which erosive E-jmail address: routbortaanl. gow (J.L. Routbort) damage affected strength 0043-1648/S-see front matter 2003 Elsevier B V. All rights reserved doi:10.1016S0043-1648(03)00392-2
Wear 256 (2004) 233–242 Solid-particle erosion and strength degradation of Si3N4/BN fibrous monoliths K.C. Goretta a, F. Gutierrez-Mora a, Nan Chen a, J.L. Routbort a,∗, T.A. Orlova b, B.I. Smirnov b, A.R. de Arellano-López c a Energy Technology Division, Argonne National Laboratory, Argonne, IL 60439-4838, USA b Ioffe Physical-Technical Institute of the Russian Academy of Sciences, St. Petersburg 194021, Russia c Departamento de Fisica de la Materia Condensada, Universidad de Sevilla, 41080 Sevilla, Spain Received 7 November 2002; received in revised form 4 March 2003; accepted 14 March 2003 Abstract Erosive damage was studied in Si3N4/BN fibrous monoliths (FMs) and the individual constituents of the cells and cell boundary, monolithic Si3N4 and BN. Unidirectional, 0/90◦, and ±45◦ FMs were tested. Specimens were subjected to impact at 90◦ by angular SiC particles of average diameter 143m, traveling at 50–100 m/s. Steady-state erosion rates in the FMs were higher than predicted by a rule of mixtures based on erosion rates of the cell and cell-boundary phases. The relatively rapid FM erosion was attributed to chipping of the Si3N4 cells caused by radial cracks. Bending strengths were measured before and after erosion testing to steady state at 100 m/s. The strength of monolithic Si3N4 decreased 22%; the BN was not tested because insufficient material was available. Within experimental error, the strengths of the FMs were unaffected by erosion. Fracture data obtained approximately 1.5 years apart suggested that the FMs were susceptible to environmentally assisted slow crack growth. © 2003 Elsevier B.V. All rights reserved. Keywords: Fibrous monoliths; Erosion; Strength; Silicon nitride; Boron nitride 1. Introduction Powder-derived ceramic fibrous monoliths (FMs) generally consist of strong ceramic cells that are surrounded by a weaker cell boundary. The cells are typically 100–500 m wide [1–10]. FMs are produced by fabrication techniques such as extrusion or dip-coating. They are often fabricated as laminates that are laid up from duplex extruded filaments that consist of a cell phase surrounded by a sheath that forms a continuous cell boundary. In flexure, FMs exhibit graceful failure, with energy dissipation arising from substantial sliding of the cells [8]. FMs constitute a lower-cost alternative to conventional continuous-fiber ceramic composites in some applications. Among ceramic FMs, those consisting of Si3N4 cells and a continuous BN cell boundary have achieved the best overall mechanical properties and, therefore, have been studied most thoroughly [2–8]. Their flexural strengths can exceed 700 MPa and work-of-fracture values, although typically 3– 6 kJ/m2, can exceed 10 kJ/m2 [3–7]. A wide variety of Si3N4/ ∗ Corresponding author. Tel.: +1-603-252-5065; fax: +1-630-252-4298. E-mail address: routbort@anl.gov (J.L. Routbort). BN FMs are produced commercially by Advanced Ceramics Research, Tucson, AZ. These and related FMs are being considered for aerospace applications in which foreign-object damage is possible [11]. Resistance to impact erosion and strength retention following damage are of concern in aerospace and other possible applications for many FMs. We have previously studied erosive damage in ZrSiO4- based FMs [12]. The cell boundary in those FMs consisted of large-grained highly porous ZrSiO4. We found that the FMs eroded more rapidly than would be expected from a simple rule-of-mixtures [13] for composite structures. The anomalously rapid erosion was attributed to wholesale removal of cells once the supporting cell boundary had been degraded [12]. In this study, we have conducted solid-particle erosion tests on Si3N4/BN FMs produced by Advanced Ceramics Research. The cell boundary in these FMs is dense, in contrast to that of the ZrSiO4 FMs. For comparison, monolithic Si3N4 and BN specimens were also tested. Strengths in flexure were measured before and after erosion testing. The goals of this work were to determine the basic response of Si3N4/BN FMs to erosion and the extent to which erosive damage affected strength. 0043-1648/$ – see front matter © 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0043-1648(03)00392-2
K.C. Goretta et al /Wear 256(2004)233-242 2. Experimental procedures 2.1. Specimen preparation and microstructure FM specimens were obtained in three laminated forms unidirectional,o90°,and±45°. All were fabricated by Ad vanced Ceramic Research and were made from Si3 N4/BN green filaments [2 that were produced by melt coextru- sion of a blend of 52 vol. ceramic powder mixture in an ethylene-based copolymer binder [6]. The coex truded filaments contained A82 vol. Si3 N4 core(E-10 H Ube Industries, Tokyo)and a18 vol. Bn sheath(HCP Grade, Advanced Ceramics Corporation, Cleveland, OH) The Si3N4 was a sinterable composition, 92 wt %com- mercial Si3N4 powder, 6 wt% Y203, and 2 wt. Al2O3 Sheets of uniaxially aligned green filaments were wound on a cylindrical mandrel. The filaments were held in place with a spray adhesive that allowed removal upon drying of unidirectional green sheets from the mandrel. The sheets were then stacked to fabricate various laminates [2, 3]. The laminates were warm pressed at a160C to produce solid panels, which were then subjected to a binder pyrolysis 1 mm step that consisted of slow heating in flowing N2 to 600C over a period of 42 h. The Si3N4/BN panels were then hot pressed at 1740° for I h under≈28 MPa pressure, which Fig2. SEM photomicrograph of fractured(a)090°and(b)±45°sc yielded 3mm-thick billets with densities that were >98% mens, showing laminated layers of flattened SiN4 cells surrounded by of theoretical The hot-pressed FM specimens consisted of flattened Two sets of Si3N4 specimens were examined: mono- Si3N4 cells surrounded by a continuous bn cell bound- lithic Si3N4 fabricated by Advanced Ceramics Research ary(Figs. I and 2). The plate-like bn grains were highly for which the composition was the same as that of the cells textured [8, 14] of the FMs, and archival in-situ-reinforced Si N4 fabricated 50 um Si3N4 fibers BN flakes Fig. 1. Composite SEM phe graph showing basic FM structure
234 K.C. Goretta et al. / Wear 256 (2004) 233–242 2. Experimental procedures 2.1. Specimen preparation and microstructure FM specimens were obtained in three laminated forms: unidirectional, 0/90◦, and ±45◦. All were fabricated by Advanced Ceramic Research and were made from Si3N4/BN green filaments [2] that were produced by melt coextrusion of a blend of ≈52 vol.% ceramic powder mixture in an ethylene-based copolymer binder [6]. The coextruded filaments contained ≈82 vol.% Si3N4 core (E-10, Ube Industries, Tokyo) and ≈18 vol.% BN sheath (HCP Grade, Advanced Ceramics Corporation, Cleveland, OH). The Si3N4 was a sinterable composition, 92 wt.% commercial Si3N4 powder, 6 wt.% Y2O3, and 2 wt.% Al2O3 [6]. Sheets of uniaxially aligned green filaments were wound on a cylindrical mandrel. The filaments were held in place with a spray adhesive that allowed removal upon drying of unidirectional green sheets from the mandrel. The sheets were then stacked to fabricate various laminates [2,3]. The laminates were warm pressed at ≈160 ◦C to produce solid panels, which were then subjected to a binder pyrolysis step that consisted of slow heating in flowing N2 to 600 ◦C over a period of ≈42 h. The Si3N4/BN panels were then hot pressed at 1740 ◦C for 1 h under ≈28 MPa pressure, which yielded 3 mm-thick billets with densities that were >98% of theoretical. The hot-pressed FM specimens consisted of flattened Si3N4 cells surrounded by a continuous BN cell boundary (Figs. 1 and 2). The plate-like BN grains were highly textured [8,14]. Fig. 1. Composite SEM photomicrograph showing basic FM structure. Fig. 2. SEM photomicrograph of fractured (a) 0/90◦ and (b) ±45◦ specimens, showing laminated layers of flattened Si3N4 cells surrounded by BN cell boundary. Two sets of Si3N4 specimens were examined: monolithic Si3N4 fabricated by Advanced Ceramics Research, for which the composition was the same as that of the cells of the FMs, and archival in-situ-reinforced Si3N4 fabricated
K.C. Goretta et al/Wear 256(2004)233-242 by AlliedSignal (Morristown, ND)[15] and Dow Chemical (Midland, MD)[16]. BN specimens were also fabricated by Advanced Ceramics Research [17 2.2. Erosion and strength test 0.6 Solid-particle erosion tests were carried out in a slinger- type apparatus that has been described previously [18-201 Tests were conducted in vacuum(500 m Torr), and thus aerodynamic effects were negligible. The feed rate of the erodent was 8gmin, for which interactions between pa d ticles were also negligible The erodent particles were angular SiC abrasives(Norton Co., Worcester, MA, USA) with mean diameter of 143 um [12, 15, 20]. The test velocity() was 50, 70, or 100 m/s and the angle of impact was 90. Specimens were impacted par- Fig. 3. Weight loss by143μ m Sic at )m/s: BN (open ci allel to the hot-pressing direction. Steady-state erosion rate (open diamonds),#4: circles,o90°FM (ER, in mg/g)values were determined from plots of the spec triangles), and ACR imen weight loss versus weight of particles impacting the surface. At least three runs were conducted for each speci- men, and at least four runs were conducted when a transient BN specimen was by far the smallest. Each ER value was was observed in the initial erosion response Following each defined as the slope of linear least-squares fits to the data run, specimens were removed, brushed, cleaned by an air The results for the three types of FMs were similar. The blast, and weighed. The average weight-loss measurements unidirectional FM was tested only at 100 m/s. Insufficient were accurate to 2%. This uncertainty arose due to slightly material was available to allow for more testing. The two incomplete or inconsistent cleaning of the surfaces Si3N4 specimens exhibited similar erosion rates, which were Four-point fracture tests were conducted with an Instron much lower than those of the FMs. The soft BN specimen Model 4505 apparatus(Canton, MA, USA). The loading rate exhibited the highest erosion rate was 1.3 mm/min. The inner load span was 9.5 mm and the Data on erosion rate versus the velocity of impact revealed outer load span was 15 mm [20, 21]. As-received specimens interesting trends(Fig. 4). As might be expected, the ER were polished with 1 um diamond paste prior to testing. Of values for the Si3Na/BN FMs were between those of the the eroded specimens, only those eroded at 100 m/s were hard Si3 N4 and soft BN. Power-law fits to the data of Fig. 4 tested for strength; they were not polished. Four specimens are shown in Table 1. ER was a stronger function of yfor were tested per condition. The strength of only one of the the FMs than for either of the monolithic Si3N4 ceramics or Si3N4 ceramics, the one produced by Advanced Ceramics the BN Research(ACR), was tested. The BN specimens were too The single-impact sites were generally characteristic of small to test after erosion, but as-polished data were avail erosion of brittle solids. Many of the impacts evinced all able from a previous study [22, 23]. Fracture data sets on polished-and-beveled specimens were obtained in the late winter of 2000 and the summer of 2001. In between tests the samples were stored in an open laboratory environment. Eroded and fractured surfaces were examined by scanning electron microscopy(SEM). Single-impact damage sites were also examined by Sem to elucidate the weight-loss mechanism [ 18-21. All specimens were coated with Au-Pd and examined in a JEOL 5400(Peabody, MA)or Hitachi S-4700-ll microscope(Tokyo, Japan) 3. Results 3.. Erosion tests Representative data for weight loss versus dose of impact ate vs SiC erodent velocity for BN(open circles), 0/90o monds),+45. FM(filled diamonds), AlliedSignal Si3 N4 ing SiC particles are shown in Fig. 3. Differences in dose (open which are almost completely obscured by filled triangles), are a consequence of the surface area that was eroded; the and ACR Si3 N4(filled triangles)
K.C. Goretta et al. / Wear 256 (2004) 233–242 235 by AlliedSignal (Morristown, NJ) [15] and Dow Chemical (Midland, MI) [16]. BN specimens were also fabricated by Advanced Ceramics Research [17]. 2.2. Erosion and strength tests Solid-particle erosion tests were carried out in a slingertype apparatus that has been described previously [18–20]. Tests were conducted in vacuum (≈500 mTorr), and thus aerodynamic effects were negligible. The feed rate of the erodent was ≈8 g/min, for which interactions between particles were also negligible. The erodent particles were angular SiC abrasives (Norton Co., Worcester, MA, USA) with mean diameter of 143 m [12,15,20]. The test velocity (V) was 50, 70, or 100 m/s and the angle of impact was 90◦. Specimens were impacted parallel to the hot-pressing direction. Steady-state erosion rate (ER, in mg/g) values were determined from plots of the specimen weight loss versus weight of particles impacting the surface. At least three runs were conducted for each specimen, and at least four runs were conducted when a transient was observed in the initial erosion response. Following each run, specimens were removed, brushed, cleaned by an air blast, and weighed. The average weight-loss measurements were accurate to ±2%. This uncertainty arose due to slightly incomplete or inconsistent cleaning of the surfaces. Four-point fracture tests were conducted with an Instron Model 4505 apparatus (Canton, MA, USA). The loading rate was 1.3 mm/min. The inner load span was 9.5 mm and the outer load span was 15 mm [20,21]. As-received specimens were polished with 1 m diamond paste prior to testing. Of the eroded specimens, only those eroded at 100 m/s were tested for strength; they were not polished. Four specimens were tested per condition. The strength of only one of the Si3N4 ceramics, the one produced by Advanced Ceramics Research (ACR), was tested. The BN specimens were too small to test after erosion, but as-polished data were available from a previous study [22,23]. Fracture data sets on polished-and-beveled specimens were obtained in the late winter of 2000 and the summer of 2001. In between tests, the samples were stored in an open laboratory environment. Eroded and fractured surfaces were examined by scanning electron microscopy (SEM). Single-impact damage sites were also examined by SEM to elucidate the weight-loss mechanism [18–21]. All specimens were coated with Au–Pd and examined in a JEOL 5400 (Peabody, MA) or Hitachi S-4700-II microscope (Tokyo, Japan). 3. Results 3.1. Erosion tests Representative data for weight loss versus dose of impacting SiC particles are shown in Fig. 3. Differences in dose are a consequence of the surface area that was eroded; the Fig. 3. Weight loss vs. dose for specimens eroded by 143 m SiC at 100 m/s: BN (open circles), unidirectional FM (filled circles), 0/90◦ FM (open diamonds), ±45◦ FM (filled diamonds), allied signal Si3N4 (open triangles), and ACR Si3N4 (filled triangles). BN specimen was by far the smallest. Each ER value was defined as the slope of linear least-squares fits to the data. The results for the three types of FMs were similar. The unidirectional FM was tested only at 100 m/s. Insufficient material was available to allow for more testing. The two Si3N4 specimens exhibited similar erosion rates, which were much lower than those of the FMs. The soft BN specimen exhibited the highest erosion rate. Data on erosion rate versus the velocity of impact revealed interesting trends (Fig. 4). As might be expected, the ER values for the Si3N4/BN FMs were between those of the hard Si3N4 and soft BN. Power-law fits to the data of Fig. 4 are shown in Table 1. ER was a stronger function of V for the FMs than for either of the monolithic Si3N4 ceramics or the BN. The single-impact sites were generally characteristic of erosion of brittle solids. Many of the impacts evinced all Fig. 4. Erosion rate vs. SiC erodent velocity for BN (open circles), 0/90◦ FM (open diamonds), ±45◦ FM (filled diamonds), AlliedSignal Si3N4 (open triangles, which are almost completely obscured by filled triangles), and ACR Si3N4 (filled triangles)
K.C. Goretta et al /Wear 256(2004)233-242 Table I The steady-state erosion surfaces all indicated similar Velocity exponent n and linear correlation coefficient R for ER data mechanisms of material removal. Brittle, cleavage-like frac- n ture was dominant(Fig. 6). The eroded surfaces of the 2.0±0.2 BN contained many flake-like features, which are presum- SigNa(ACR) 2.2±0.2 0.9990 ably related to fracture of BNs highly anisotropic crystals 2.1±0.2 0.9997 The Si3 N4 surfaces contained fine debris, similar to that 090° 5.4士04 0.9999 observed at the single-impact sites. The FM surfaces ±45° 4.1±0.5 tained some of the flake-like features observed in the Low-magnification observations of the specimens revealed that the surfaces of the BN and Si3N4 specimens were flat, of the features that lead to material removal: indenting, and that all of the FMs were undulating. For each of the FMs, radial-crack formation, and spalling of the target caused by the undulations scaled with the size of the cells(Fig. 7) propagation of lateral cracks. Important differences were ap- parent among the targets. The bn damage craters were the 3. 2. Strength tests largest(average diameter s200 um) and probably evinced he most evidence of microplasticity. Many of the impacts Results of the four-point flexural tests, for specimens be- into the Si3 N4 caused only scuffing, but little or no material fore and after erosion testing, are shown in Table 2. The data removal. When a clear impact crater was evident, it was rel- for as-polished specimens exhibited expected trends and atively large(average diameter 50-70 um), and inevitably moderate surprises. As expected, the monolithic Si3N4 was he site was strewn with fine debris. Little difference was strongest, the monolithic bn the weakest, and the Fms were observed among the various types of FMs. Despite the com- in between. Among the FMs, the unidirectional one was the paratively high erosion rates of the FMs, the single-impact strongest. All of the strength values were, however, lower damage sites of the FMs(average diameter N50 um)were than those obtained in the year 2000 from the same panels almost always smaller than those that led to material re- [22,23]. The specimens from 2000 were, however, longer, moval in the Si3N4(Fig. 5). Their average width was nearly their outer loading span was 40 mm, and their inner loading one order of magnitude smaller than the Si3N4 cells. As ex- span was 15 mm Solid-particle erosion to steady-state con- pected, the size of the damage site scaled with I ditions for the surfaces appeared to reduce the strengths of a (b) c 50 Fig. 5. SEM photomicrographs of single impact by 143 um SiC erodent:(a)BN, (b) Si3N4, (c) FM in cell region, and(d) FM near cell/cell-boundary
236 K.C. Goretta et al. / Wear 256 (2004) 233–242 Table 1 Velocity exponent n and linear correlation coefficient R for ER data Specimen n R Si3N4 (self-reinforced) 2.0 ± 0.2 0.9999 Si3N4 (ACR) 2.2 ± 0.2 0.9990 BN 2.1 ± 0.2 0.9997 0/90◦ 5.4 ± 0.4 0.9999 ±45◦ 4.1 ± 0.5 0.9534 of the features that lead to material removal: indenting, radial-crack formation, and spalling of the target caused by propagation of lateral cracks. Important differences were apparent among the targets. The BN damage craters were the largest (average diameter ≈200m) and probably evinced the most evidence of microplasticity. Many of the impacts into the Si3N4 caused only scuffing, but little or no material removal. When a clear impact crater was evident, it was relatively large (average diameter ≈50–70m), and inevitably the site was strewn with fine debris. Little difference was observed among the various types of FMs. Despite the comparatively high erosion rates of the FMs, the single-impact damage sites of the FMs (average diameter ≈50m) were almost always smaller than those that led to material removal in the Si3N4 (Fig. 5). Their average width was nearly one order of magnitude smaller than the Si3N4 cells. As expected, the size of the damage site scaled with V. Fig. 5. SEM photomicrographs of single impact by 143 m SiC erodent: (a) BN, (b) Si3N4, (c) FM in cell region, and (d) FM near cell/cell-boundary interface. The steady-state erosion surfaces all indicated similar mechanisms of material removal. Brittle, cleavage-like fracture was dominant (Fig. 6). The eroded surfaces of the BN contained many flake-like features, which are presumably related to fracture of BN’s highly anisotropic crystals. The Si3N4 surfaces contained fine debris, similar to that observed at the single-impact sites. The FM surfaces contained some of the flake-like features observed in the BN. Low-magnification observations of the specimens revealed that the surfaces of the BN and Si3N4 specimens were flat, and that all of the FMs were undulating. For each of the FMs, the undulations scaled with the size of the cells (Fig. 7). 3.2. Strength tests Results of the four-point flexural tests, for specimens before and after erosion testing, are shown in Table 2. The data for as-polished specimens exhibited expected trends and moderate surprises. As expected, the monolithic Si3N4 was strongest, the monolithic BN the weakest, and the FMs were in between. Among the FMs, the unidirectional one was the strongest. All of the strength values were, however, lower than those obtained in the year 2000 from the same panels [22,23]. The specimens from 2000 were, however, longer; their outer loading span was 40 mm, and their inner loading span was 15 mm. Solid-particle erosion to steady-state conditions for the surfaces appeared to reduce the strengths of
K.C. Goretta et al/Wear 256(2004)233-242 237 b 0.2 mm Fig. 7. Low-magnification SEM photomicrograph of eroded surface of unidirectional FM; topology indicates loss of cells all, but the unidirectional and +45 Si3N4/BN specimens Relatively large error bars cast some doubt on each of the comparisons between before and after erosion SEM examination of the Si3 N4 and Si3 N4/BN surfaces revealed that cleavage dominated the fracti cesses. The Si3 N4 surfaces were characteristic of a dense fine-grained ceramic. The Si3N4 contained many elongate grains(Fig. &a). The FMs exhibited larger features, and some Bn was evident on the surfaces(Fig. 8b) 10 Fig. 6. SEM photomicrographs of steady-state surfaces for erosion by 10um 143 um SiC erodent at 100 m/s:(a)BN, (b) Si3N4, and (c)representative Table 2 Four-point flexural strength before(oo)and(oi) after(oer )erosion testing, including data from Refs. [21, 22]a (MPa) do(MPa) Oer(MPa) Change SinA 677±127601±128470±27-2 38士3 Unidirectional476±30194±6193±28-1 10m 0/90 379±86109±5 94±8 175±13100±21139±25+39 Fig. 8. SEM photomicrographs of fracture surfaces of (a)ACR Si3N4 a ai values are from early 2000 and ao are from mid 2001 and(b)0/90] Si3 N4/BN FM
K.C. Goretta et al. / Wear 256 (2004) 233–242 237 Fig. 6. SEM photomicrographs of steady-state surfaces for erosion by 143 m SiC erodent at 100 m/s: (a) BN, (b) Si3N4, and (c) representative FM. Table 2 Four-point flexural strength before (σo) and (σi) after (σer) erosion testing, including data from Refs. [21,22]a Specimen σi (MPa) σo (MPa) σer (MPa) Change (σo: σer) (%) Si3N4 677 ± 127 601 ± 128 470 ± 27 −22 BN 38 ± 3– – – Unidirectional 476 ± 30 194 ± 6 193 ± 28 −1 0/90◦ 379 ± 86 109 ± 5 94 ± 8 −14 ±45◦ 175 ± 13 100 ± 21 139 ± 25 +39 a σi values are from early 2000 and σo are from mid 2001. Fig. 7. Low-magnification SEM photomicrograph of eroded surface of unidirectional FM; topology indicates loss of cells. all, but the unidirectional and ±45◦ Si3N4/BN specimens. Relatively large error bars cast some doubt on each of the comparisons between before and after erosion. SEM examination of the Si3N4 and Si3N4/BN fracture surfaces revealed that cleavage dominated the fracture processes. The Si3N4 surfaces were characteristic of a dense, fine-grained ceramic. The Si3N4 contained many elongated grains (Fig. 8a). The FMs exhibited larger features, and some BN was evident on the surfaces (Fig. 8b). Fig. 8. SEM photomicrographs of fracture surfaces of (a) ACR Si3N4 and (b) 0/90◦ Si3N4/BN FM
K.C. Goretta et al /Wear 256(2004)233-242 4. Discussion The er data for the Bn and Si3 Na specimens were gene ally consistent with prevailing models for erosion of brittle 4.. Erosion materials. In each model The erosion results are addressed most easily by first ER o vn examining the responses of the monoliths and then con- depending on the assumption of impact conditions, n ranges trasting them to those of the FMs. Erosive damage from from 2.3 to 3. 2 [24-30]. Each of the Si3 N4 monoliths fell sharp particles impacting brittle materials has been stud- near the low end of the range ied for decades, and models are effective in predicting a The only significant deviation from model response to ero- monolithic materials response [24-31]. With impact at nor- sion for the monoliths was the fracturing of the SiC erodent mal or near-normal incidence. material loss occurs from when striking the Si3N4 targets. As Srinivasan and Scatter the following sequence of events:(1)indentation creates good have argued, the chief effect of the fractured erodent an elastic-plastic zone beneath the impacting particle (2)a is that multiple impacts may be required to remove target radial-crack perpendicular to the specimen surface is created material; the basic mechanism of material removal remains recoils. the resulting tensile stress induces formation of l The FMs, in contrast, exhibited three responses that did eral cracks approximately parallel to the surface; (4) the lat- not immediately meet with expectations eral cracks propagate to the surface, and chips are removed. 1. The single-impact damage craters were consistently The models are based complete transfer of the kinetic en- smaller for the FMs than for the Si3N4, yet the FM ergy of the erodent to the target. If the erodent particles frac ture significantly, the models can lose some of their utility 2. The calculated velocity exponents for the FMs were much larger than for the monoliths and were larger than pre In relation to this study, clearly the hard Si3N4 targets dicted by any model vere responsible for fracturing the softer SiC erodent [34]3. As will be discussed below the erosion rates for the FMs severe fracturing of this same erodent has also been ob- were higher than would be predicted by an accepted rule served with erosion of Sic targets [19, 351). Negligible, if of mixtures [13 based on erosion rates of the individual any, fracturing of the SiC erodent occurred with the soft BN constituents, BN and si3N4 Thus, in absolute terms, the Er values for the Si3 N4 repre- ent lower limits because some of the energy of impact was These three deviations from expectations are related and dissipated in fracturing the erodent. If a hard abrasive, which are attributable to the complex microstructures of the FMs could have remained intact upon impact, had been used, the The explanation as to why relatively small individual ER values of the Si3N4 would have been higher. SEM did impact craters in pristine surfaces led eventually to compar not reveal an obvious difference in extent of Sic erodent atively rapid erosion rates for the FMs is to be found within fracture among the various velocities of impact. The trend the details of the erosion process and the pertaining models of ER with velocity might then approximate that predicted For the monoliths and FMs, initial material loss(as revealed by any of the applicable models [32] by observations of individual impact sites) was induced The eroded surfaces of all specimens were typical of that largely by formation of lateral cracks [24-30].Although of a brittle ceramic, and the single-impact sites supported the models based on lateral-crack formation agree relatively the conclusion that material was removed by elastic-plastic well with respect to effects of erodent velocity and target indentation and by formation of cracks. In all models for hardness and toughness on materials removal, they disagree erosion of brittle materials, erosion rate is dependent on as to the effect of target elastic modulus E. Wiederhorn and target fracture toughness, Kc, and hardness, H [24-31. In coworkers [25-27 stipulate that lateral-cI representative model, Ritter et al. define the volume, X, of and hence ER, is not a function of the elastic modulus of material removed per impact the target. Ritter et al. [ 30] report that the volume of mate- rial removed to be proportional to the target modulus to the X=φE4kcH-17 m12U7/6 (1) 5/4 power; thus stiffer targets erode faster, all other factors being equal. Evans and coworkers [24, 28, 29] agree, but set where o is a constant, E the target elastic modulus, and U a slightly different exponent. For impact pressure P, the the kinetic energy of the impacting particle [30]. This model lateral cracks that form have length cI implies that hard tough targets are most resistant to erosion bN is softer and much less tough than Si3 N4 and, hence, CL= A F(E/H3412 erodes faster. Based on multiple regression analysis of a KH1/4 large body of erosion data, Wiederhorn and Hockey [27 where A is a constant, F is a geometric factor, and the other argue for the primacy of kc in the relation between X (or terms have been defined above [29]. Longer cracks imply Er), H and Kc, but certainly agree with the prediction that more rapid erosion, with the volume of material removed bn should erode significantly more rapidly than Si3N varying as approximately c[30]
238 K.C. Goretta et al. / Wear 256 (2004) 233–242 4. Discussion 4.1. Erosion The erosion results are addressed most easily by first examining the responses of the monoliths and then contrasting them to those of the FMs. Erosive damage from sharp particles impacting brittle materials has been studied for decades, and models are effective in predicting a monolithic material’s response [24–31]. With impact at normal or near-normal incidence, material loss occurs from the following sequence of events: (1) indentation creates an elastic–plastic zone beneath the impacting particle; (2) a radial-crack perpendicular to the specimen surface is created beneath the elastic–plastic zone; (3) as the erodent particle recoils, the resulting tensile stress induces formation of lateral cracks approximately parallel to the surface; (4) the lateral cracks propagate to the surface, and chips are removed. The models are based complete transfer of the kinetic energy of the erodent to the target. If the erodent particles fracture significantly, the models can lose some of their utility [19,31–33]. In relation to this study, clearly the hard Si3N4 targets were responsible for fracturing the softer SiC erodent [34] (severe fracturing of this same erodent has also been observed with erosion of SiC targets [19,35]). Negligible, if any, fracturing of the SiC erodent occurred with the soft BN. Thus, in absolute terms, the ER values for the Si3N4 represent lower limits because some of the energy of impact was dissipated in fracturing the erodent. If a hard abrasive, which could have remained intact upon impact, had been used, the ER values of the Si3N4 would have been higher. SEM did not reveal an obvious difference in extent of SiC erodent fracture among the various velocities of impact. The trend of ER with velocity might then approximate that predicted by any of the applicable models [32]. The eroded surfaces of all specimens were typical of that of a brittle ceramic, and the single-impact sites supported the conclusion that material was removed by elastic–plastic indentation and by formation of cracks. In all models for erosion of brittle materials, erosion rate is dependent on target fracture toughness, Kc, and hardness, H [24–31]. In a representative model, Ritter et al. define the volume, X, of material removed per impact: X = φE5/4K−1 C H−17/12U7/6, (1) where φ is a constant, E the target elastic modulus, and U the kinetic energy of the impacting particle [30]. This model implies that hard tough targets are most resistant to erosion. BN is softer and much less tough than Si3N4 and, hence, erodes faster. Based on multiple regression analysis of a large body of erosion data, Wiederhorn and Hockey [27] argue for the primacy of KC in the relation between X (or ER), H and KC, but certainly agree with the prediction that BN should erode significantly more rapidly than Si3N4. The ER data for the BN and Si3N4 specimens were generally consistent with prevailing models for erosion of brittle materials. In each model: ER ∝ Vn, (2) depending on the assumption of impact conditions, n ranges from 2.3 to 3.2 [24–30]. Each of the Si3N4 monoliths fell near the low end of the range. The only significant deviation from model response to erosion for the monoliths was the fracturing of the SiC erodent when striking the Si3N4 targets. As Srinivasan and Scattergood have argued, the chief effect of the fractured erodent is that multiple impacts may be required to remove target material; the basic mechanism of material removal remains unchanged [32]. The FMs, in contrast, exhibited three responses that did not immediately meet with expectations: 1. The single-impact damage craters were consistently smaller for the FMs than for the Si3N4, yet the FM steady-state erosion rates were higher for the FMs. 2. The calculated velocity exponents for the FMs were much larger than for the monoliths and were larger than predicted by any model. 3. As will be discussed below, the erosion rates for the FMs were higher than would be predicted by an accepted rule of mixtures [13] based on erosion rates of the individual constituents, BN and Si3N4. These three deviations from expectations are related and are attributable to the complex microstructures of the FMs. The explanation as to why relatively small individual impact craters in pristine surfaces led eventually to comparatively rapid erosion rates for the FMs is to be found within the details of the erosion process and the pertaining models. For the monoliths and FMs, initial material loss (as revealed by observations of individual impact sites) was induced largely by formation of lateral cracks [24–30]. Although the models based on lateral-crack formation agree relatively well with respect to effects of erodent velocity and target hardness and toughness on materials removal, they disagree as to the effect of target elastic modulus E. Wiederhorn and coworkers [25–27] stipulate that lateral-crack formation, and hence ER, is not a function of the elastic modulus of the target. Ritter et al. [30] report that the volume of material removed to be proportional to the target modulus to the 5/4 power; thus stiffer targets erode faster, all other factors being equal. Evans and coworkers [24,28,29] agree, but set a slightly different exponent. For impact pressure P, the lateral cracks that form have length cL cL = A F(E/H)3/4 KcH1/4 1/2 P5/8, (3) where A is a constant, F is a geometric factor, and the other terms have been defined above [29]. Longer cracks imply more rapid erosion, with the volume of material removed varying as approximately c2 L [30].
K.C. Goretta et al/Wear 256(2004)233-242 239 For most individual impacts into an FM, a cell was struck, and the operant hardness and toughness were those of Si3N4 but the relevant elastic modulus was most likely nearer to that of the FM. Measurements with these specimens have determined that the elastic modulus was 311 GPa for Si3N4 and 128 GPa for FMs through the short-transverse direc tion [22, 36]. From Eq (3), taking the ratio(128/311)/8 the expected cL ratio would be a0. 7. This value is approxi- mately what was observed for the relative widths of the im- pact craters. The less-stiff FMs responded initially as Ritter et al. [30] and Evans and coworkers [29] predicted ig. vo, f ar ph stmicro gr pel ls s osive damage in ZISi(Os-based FM If this explanation is correct, and Eq(3)represents the erosive processes accurately, then the FMs would seem to be ideal materials: hard, yet compliant, and therefore resis- In the FMs, a combination of materials loss by both lat- tant to erosion. These explanation are not correct because eral [28] and radial [29] cracks, with a possible contribution Eq. (3)describes only part of the erosion process of FMs. In from subsurface cell damage, led to erosion rates that were conventional monolithic materials impacted at near-normal more rapid than would be predicted by the average size of incidence, radial cracks do not contribute significantly to ingle-impact damage craters. The radial-crack contribution material removal [24-30] For impact at oblique incidence, to material loss is the key difference between the FMs and however, radial cracks can lead to material removal [371 any of the monoliths. Radial cracks of length, cR, form with In the Fms. too. radial-crack formation can induce ma- impact pressure Pis terial removal because of the cellular structure of fms. as F(E/H212/3 shown schematically in Fig 9, because the bonding between CR (4) the cells and the cell boundary is weak, edges of the cells can be removed by radial-crack formation. In addition, be- where F is a geometric factor and the other terms are as cause the BN erodes much more quickly, the cells are in defined in Eq (3)[28]. Radial-crack lengths are significant places poorly supported, and damage to the cells can perhaps relative to lateral-crack lengths. When, along with lateral occur below the surface when the bn has been removed. cracks, they too lead to material loss, the volume of material Although intact SiC erodent particles would be too large to More important to engineering applications is the question penetrate deeply below the surface, debris from shattered erodent could. Evidence for large-scale removal of cells f whether the erosion rates exhibited by the FMs are as rather than piece-by-piece removal from lateral-crack for- one would expect based on the erosion rates of the cell and mation Is pro the morphologies of the eroded sur- cell-boundary phases For discrete composite materials such faces. This SEM evidence was even stronger for erosion of as the Si3N4/BN FMs, erosion rates should be related to the ZrSiO4-based FMs [12]. A representative photomicrograph rates of the individual constituents by from Ref. [12], is shown in Fig 10 in which the FMs con- tained porous, very weak cell boundaries where m refers to the mass fraction of the constituent in () the FM, and the subscripts FM, C, and CB refer to the FM, cell, and cell-boundary specimens, respectively [13, 38]. This model. which has been called the"inverse rule of mixtures is based on parallel erosive processes and holds so long as the erosion rates of the constituents in the FM are independent of each other and the scale of the damage is small relative to the scale of the microstructure. For the Si3 N4/BN FMs, the cells are approximately one order of magnitude wider than an average damage zone Fig. 9. Schematic diagrams of erosive damage in Si3 N4/BN FMs:(a Material loss in conventional brittle materials, in which elastic-plastic zone shaded region) forms beneath indenter. Radial cracks perpendicular posite material will erode in series [38]. In that case, aliea Alternatively, it is possible that the constituents of a com- to surface are driven into target and, when indenter recoils, lateral cracks rule of mixtures could apply urface and material is lost.(b) Material loss in FM, in which soft Bn ERFM= mcERc mCBERcB loss (lightly shaded region) of cell, and(2)exposed cells from loss of The constituents of Si]n4/BN FMs do, in fact, erode inde- ell boundary can be subject to subsurface damage Process(1)is much pendently by lateral-crack formation. However, the weak in- more significant than Process(2) for material loss. terface between the Si3 Na and bn allows for material loss
K.C. Goretta et al. / Wear 256 (2004) 233–242 239 For most individual impacts into an FM, a cell was struck, and the operant hardness and toughness were those of Si3N4, but the relevant elastic modulus was most likely nearer to that of the FM. Measurements with these specimens have determined that the elastic modulus was 311 GPa for Si3N4 and ≈128 GPa for FMs through the short-transverse direction [22,36]. From Eq. (3), taking the ratio (128/311)3/8, the expected cL ratio would be ≈0.7. This value is approximately what was observed for the relative widths of the impact craters. The less-stiff FMs responded initially as Ritter et al. [30] and Evans and coworkers [29] predicted. If this explanation is correct, and Eq. (3) represents the erosive processes accurately, then the FMs would seem to be ideal materials: hard, yet compliant, and therefore resistant to erosion. These explanation are not correct because Eq. (3) describes only part of the erosion process of FMs. In conventional monolithic materials impacted at near-normal incidence, radial cracks do not contribute significantly to material removal [24–30]. For impact at oblique incidence, however, radial cracks can lead to material removal [37]. In the FMs, too, radial-crack formation can induce material removal because of the cellular structure of FMs. As shown schematically in Fig. 9, because the bonding between the cells and the cell boundary is weak, edges of the cells can be removed by radial-crack formation. In addition, because the BN erodes much more quickly, the cells are in places poorly supported, and damage to the cells can perhaps occur below the surface when the BN has been removed. Although intact SiC erodent particles would be too large to penetrate deeply below the surface, debris from shattered erodents could. Evidence for large-scale removal of cells rather than piece-by-piece removal from lateral-crack formation is provided by the morphologies of the eroded surfaces. This SEM evidence was even stronger for erosion of ZrSiO4-based FMs [12]. A representative photomicrograph from Ref. [12], is shown in Fig. 10 in which the FMs contained porous, very weak cell boundaries. Fig. 9. Schematic diagrams of erosive damage in Si3N4/BN FMs: (a) Material loss in conventional brittle materials, in which elastic–plastic zone (shaded region) forms beneath indenter. Radial cracks nearly perpendicular to surface are driven into target and, when indenter recoils, lateral cracks form nearly parallel to surface. These lateral cracks propagate to the surface and material is lost. (b) Material loss in FM, in which soft BN cell-boundary erodes between cells and (1) radial crack produces material loss (lightly shaded region) of cell, and (2) exposed cells from loss of cell boundary can be subject to subsurface damage. Process (1) is much more significant than Process (2) for material loss. Fig. 10. SEM photomicrograph of erosive damage in ZrSiO4-based FM; removal of large sections of cells is apparent. In the FMs, a combination of materials loss by both lateral [28] and radial [29] cracks, with a possible contribution from subsurface cell damage, led to erosion rates that were more rapid than would be predicted by the average size of single-impact damage craters. The radial-crack contribution to material loss is the key difference between the FMs and any of the monoliths. Radial cracks of length, cR, form with impact pressure P is cR = F∗(E/H)1/2 KcH1/4 2/3 P2/3, (4) where F∗ is a geometric factor and the other terms are as defined in Eq. (3) [28]. Radial-crack lengths are significant relative to lateral-crack lengths. When, along with lateral cracks, they too lead to material loss, the volume of material removed can be large. More important to engineering applications is the question of whether the erosion rates exhibited by the FMs are as one would expect based on the erosion rates of the cell and cell-boundary phases. For discrete composite materials such as the Si3N4/BN FMs, erosion rates should be related to the rates of the individual constituents by 1 ERFM = mC ERC + mCB ERCB , (5) where m refers to the mass fraction of the constituent in the FM, and the subscripts FM, C, and CB refer to the FM, cell, and cell-boundary specimens, respectively [13,38]. This model, which has been called the “inverse rule of mixtures”, is based on parallel erosive processes and holds so long as the erosion rates of the constituents in the FM are independent of each other and the scale of the damage is small relative to the scale of the microstructure. For the Si3N4/BN FMs, the cells are approximately one order of magnitude wider than an average damage zone. Alternatively, it is possible that the constituents of a composite material will erode in series [38]. In that case, a linear rule of mixtures could apply: ERFM = mCERC + mCBERCB. (6) The constituents of Si3N4/BN FMs do, in fact, erode independently by lateral-crack formation. However, the weak interface between the Si3N4 and BN allows for material loss
K.C. Goretta et al /Wear 256(2004)233-242 similarity in flaw populations, and thus near equivalence of urface prepara The decreases in strengths of the FMs with testing appro imately 1.5 years apart were dramatic, generally >50%. The added component of shear in the current strength tests may have contributed to the remarkable decreases in strength, but given their magnitudes, slow crack growth is a likely con tributor to the decreases. Thermal expansion has been stud led in unidirectional and cross-ply Si3N4/FMs [40], and the resulting residual stress in the radial direction of the si3n4 cells has been measured [41]. A value of N120-150 MPa was estimated. The residual stress along the cell lengths was not reported. Given the large residual stresses present, and the significant concentrations of glass in both the Si3 N4 cells and BN cell boundary [7], slow crack growth would be ex- Fig. I1. Predicted ER of FMs based on simple rules of mixture. Data pected. Its effects in the FM appear to have been substantial amonds ),+45 FM(filled diamonds), inverse rule Erosion of surfaces would be expected to produce differ of mixtures(thin line), and linear rule of mixtures(bold line)especially ent results in Si3N4 and the FMs. In Si3N4, for the most symbols defined in Fig. 4 part, a smooth surface was replaced by a roughened one. A new and more severe population of flaws was introduced The sem observations suggested that radial cracks did in- by radial cracks, and removal of the BN may allow for un- deed form in the Si3N4. These would be responsible for any dermining of the Si3N4 cells. Predictions of Eqs. (5)and strength reduction. Fracture strength af can be related to (6)are plotted vs. measurements for the BN, Si3N4, and the flaw length cr by FMs in Fig. 11. The FMs eroded more rapidly than pre- dicted by a rule of mixtures, especially at higher velocities of AKC Because all ceramic FMS produced to date rely on en- where A is a constant equal to 0.47 for a theoretical inden- ergy dissipation through a weak cell boundary, whether the tation radial crack, Z the flaw-shape parameter(equal to /2 cell boundary is dense [3-8] or porous [91, it appears that for a semicircular faw), and y the numerical constant that FMs are comparatively susceptible to erosion by sharp par depends on loading and flaw size [30]. The value of A can ticles. Rounder erodent particles tend to induce damage by increase if residual stresses are relieved by the flaw [30, 42] formation of radial, lateral, and Hertzian cone cracks [39] Taking values for Eq. (7) from the analysis of Ritter et al One would presume that FM cells would be susceptible [30], A=0.82, Z=T/2, and y =1.9, we estimate the to edge chipping by all of these cracks, and thus the FMs critical flaw size after erosion for the Si3N4. Kc was taken as themselves would evince relatively rapid erosion rates from 5MPam /2[34]. A value of 52 um was calculated, which impact by round erodent. Tests are required to substantiate is reasonable when compared with the average crater size this hypothesis (Fg.5) It is tempting to apply the same faw analysis to the FMs 4.2. Strength results In the FMs, a relatively smooth surface that contained small arp cracks because of presumed slow crack growth was Nothing out of the ordinary was observed on any of replaced through erosion by a rough and irregular one. How- the fracture surfaces. Cleavage dominated the fracture pro- ever, for the FMs, the flaws introduced by erosion may or cesses. Three independent sets of fracture data were ob- may not have been more severe than those that were re- tained: specimens tested in the year 2000(oi); specimens moved along with the material loss. Furthermore, because from identical billets, but tested in the middle of 2001 the effects of relief of residual stresses cannot be accurately (oo); and specimens tested shortly after erosion testing addressed for the Fms, and no values can be assigned with (oer). Comparisons between a; and oo, and oo and er, are confidence to Kc, no calculations of flaw size were made warranted The results of Table 2 will be addressed qualitatively only All values of oo were lower than the corresponding values Within experimental error, none of the FMs exhibited a of oi. The oo and oi values of the Si3N4 specimens were, strength increase or decrease with erosion. The closest that however, equal within the calculated experimental errors. It any set of specimens came to establishing an unequivocal is likely that the minor differences in strength between Si3N4 trend other than no change was the +45. FM, for which specimens in strength, from 677+127 to 601+128 MPa, are erosion appeared to increase strength. In each of the FMs attributable to specimen configuration. The remarkable sim- previous mechanical-test results have indicated that the Bn larity in the sizes of the error bars suggests a corresponding cell boundary isolates each of the cells quite effectivel
240 K.C. Goretta et al. / Wear 256 (2004) 233–242 Fig. 11. Predicted ER of FMs based on simple rules of mixture. Data are 0/90◦ FM (open diamonds), ±45◦ FM (filled diamonds), inverse rule of mixtures (thin line), and linear rule of mixtures (bold line) especially at higher V, the FMs erode faster than predicted by either rule. Other symbols defined in Fig. 4. by radial cracks, and removal of the BN may allow for undermining of the Si3N4 cells. Predictions of Eqs. (5) and (6) are plotted vs. measurements for the BN, Si3N4, and the FMs in Fig. 11. The FMs eroded more rapidly than predicted by a rule of mixtures, especially at higher velocities of impact. Because all ceramic FMs produced to date rely on energy dissipation through a weak cell boundary, whether the cell boundary is dense [3–8] or porous [9], it appears that FMs are comparatively susceptible to erosion by sharp particles. Rounder erodent particles tend to induce damage by formation of radial, lateral, and Hertzian cone cracks [39]. One would presume that FM cells would be susceptible to edge chipping by all of these cracks, and thus the FMs themselves would evince relatively rapid erosion rates from impact by round erodents. Tests are required to substantiate this hypothesis. 4.2. Strength results Nothing out of the ordinary was observed on any of the fracture surfaces. Cleavage dominated the fracture processes. Three independent sets of fracture data were obtained: specimens tested in the year 2000 (σi); specimens from identical billets, but tested in the middle of 2001 (σo); and specimens tested shortly after erosion testing (σer). Comparisons between σi and σo, and σo and σer, are warranted. All values of σo were lower than the corresponding values of σi. The σo and σi values of the Si3N4 specimens were, however, equal within the calculated experimental errors. It is likely that the minor differences in strength between Si3N4 specimens in strength, from 677±127 to 601±128 MPa, are attributable to specimen configuration. The remarkable similarity in the sizes of the error bars suggests a corresponding similarity in flaw populations, and thus near equivalence of surface preparation. The decreases in strengths of the FMs with testing approximately 1.5 years apart were dramatic, generally >50%. The added component of shear in the current strength tests may have contributed to the remarkable decreases in strength, but given their magnitudes, slow crack growth is a likely contributor to the decreases. Thermal expansion has been studied in unidirectional and cross-ply Si3N4/FMs [40], and the resulting residual stress in the radial direction of the Si3N4 cells has been measured [41]. A value of ≈120–150 MPa was estimated. The residual stress along the cell lengths was not reported. Given the large residual stresses present, and the significant concentrations of glass in both the Si3N4 cells and BN cell boundary [7], slow crack growth would be expected. Its effects in the FM appear to have been substantial. Erosion of surfaces would be expected to produce different results in Si3N4 and the FMs. In Si3N4, for the most part, a smooth surface was replaced by a roughened one. A new and more severe population of flaws was introduced. The SEM observations suggested that radial cracks did indeed form in the Si3N4. These would be responsible for any strength reduction. Fracture strength σf can be related to flaw length cf by σf = AZKc (Yc1/2 f ) , (7) where A is a constant equal to 0.47 for a theoretical indentation radial crack, Z the flaw-shape parameter (equal to π/2 for a semicircular flaw), and Y the numerical constant that depends on loading and flaw size [30]. The value of A can increase if residual stresses are relieved by the flaw [30,42]. Taking values for Eq. (7) from the analysis of Ritter et al. [30], A = 0.82, Z = π/2, and Y = 1.9, we estimate the critical flaw size after erosion for the Si3N4. Kc was taken as 5 MPa m1/2 [34]. A value of 52m was calculated, which is reasonable when compared with the average crater size (Fig. 5). It is tempting to apply the same flaw analysis to the FMs. In the FMs, a relatively smooth surface that contained small sharp cracks because of presumed slow crack growth was replaced through erosion by a rough and irregular one. However, for the FMs, the flaws introduced by erosion may or may not have been more severe than those that were removed along with the material loss. Furthermore, because the effects of relief of residual stresses cannot be accurately addressed for the FMs, and no values can be assigned with confidence to Kc, no calculations of flaw size were made. The results of Table 2 will be addressed qualitatively only. Within experimental error, none of the FMs exhibited a strength increase or decrease with erosion. The closest that any set of specimens came to establishing an unequivocal trend other than no change was the ±45◦ FM, for which erosion appeared to increase strength. In each of the FMs, previous mechanical-test results have indicated that the BN cell boundary isolates each of the cells quite effectively
K.C. Goretta et al/Wear 256(2004)233-242 B3-9]. One might expect little loss of strength with ero- [6]S W. Lee, D.K. Kim, High-temperature characteristics of Si3N4/BN sion because the vast majority of cells should be unaffected fibrous monolithic ceramics, Ceram. Eng. Sci. Proc. 18(4)(1997) The data largely reflect that expectation. Given uncertainties 81-486 with respect to residual stresses and effects of environmen- R W. rice, JW Halloran Infiuence of microstructure and tal exposure, little more in the way of assessment can be nitride fibrous monolithic ceramics, J. Am. Ceram Soc. 82(1999) 2502-2508 [8]JC. McNulty, M.R. Begley, F.w. Zok, In-plane fracture resistance of cross-ply fibrous monoliths, J Am. Ceram Soc. 84(2001)367-375 5. Conclusions [9]BJ. Polzin, T.A. Cruse, R.L. Houston, J.J. Picciolo, D. Singh, K.C. etta, Fabrication and characterization of oxide fibrous monoliths produced by coextrusion, Ceram. Trans. 103(2000)237-244 Si3N4/BN FMs and monolithic specimens of the two [10]S-]. Lee, W.M. Kriven, Toughened oxide composites based on porous constituents were subjected to impact at 90 by angu- lumina-platelet interphases, J Am Ceram Soc. 84(2001)767-774 lar 143 um-diameter SiC particles traveling at 50, 70, or [I]A C. Mulligan, Advanced Ceramics Research, personal commu- 100 m/s Steady-state erosion rates were highest for the BN nication. 2000 and lowest for the Si3 N4. Erosion rates of the FMs were [12]KC. Goretta, E.Gutierrez-Mora, T Tran, J.Katz, J.L. Routbort,TS Orlova, A R. de Arellano-Lopez, Solid-particle erosion of ZrSiO4 higher than predicted by a rule of mixtures based on inde fibrous monoliths. Ceram. Trans. 1 2)139-146 pendent erosion rates of the cell and cell-boundary phases. [13]SK. Hovis, J.E. Talia, R.O. Scattergood, Erosion in multiphase chipping of the rapid erosion was attributed primarily to The relatively ems, Wear 108(1986)139155 cells caused by radial-crack formation upon [14 A.R. de Arellano-Lopez, s. Lopez-Pombero, A. Domingue Impact odriguez, J. L. Routbort, D. Singh, K C. Goretta, Plastic deformation Surface damage induced by erosion at 100 m/s caused a of silicon nitride/boron nitride fibrous monoliths, J. Euro. Ceram. Soe.21(2001)245-250 decrease in bending strength for the Si3 N4 of 22%. The [15]M. Marrero, J.L.Routbort,P.Whalen, C.-w KR Karas FMs each exhibited significant strength loss with storage le erosion of in-situ reinforced SinA, Wear 162-164 over 1.5 years. Environmentally assisted slow crack growth 993)280-284 ppeared to contribute to the strength reduction. within ex 16 M.A. Boling- Risser, K C. Goretta, J.L. Routbort, K.T. Faber, Effect perimental error, the FMs strengths neither decreased nor of microstructure on the high-temperature mechanical behavior of f-reinforced hot-pressed silicon nitride: compressive creep, J. Am. increased after being eroded at 100 m/s Ceram.Soc.83(2000)3065-3069 I7)BK. Kardashev, Yu.A. Burenkov, B I. Smimov, V.V. Shpeizman, V.A. Stepanov, V.M. Chernov, D. Singh, K.C. Goretta, Elasticity and Acknowledgements anelasticity of ceramic samples of boron nitride, Phys. Sol. State 43 (2001)1084-1088( translated from Russian ). We thank T. Tran and J. Katz for experimental assistance [ 18]T H. Kosel, R.O. Scattergood, A P L. Turner, An electron microscopy study of erosive wear, in: K.C. Ludema, W.A. Glaeser, S.K. Rhee both were partially supported by the Division of Educa (Eds ) Proceedings of the International Conference on Wear of tional Programs, Argonne National Laboratory. This work Materials, American Society Mechanical Engineering, New York, was supported by the Defense Advanced Research Projects 1979,pp.192-204 Agency through an Interagency agreement with the US De- [19).L. Routbort, R.O. Scattergood, Solid particle erosion of artment of Energy(DOE); by dOE, under Contract No composites, Key Eng Mater. 71(1992)23-50 220]P. Strzepa, E.J. Zamirowski, J.B. Kupperman, K.C. Goretta W-31-109-Eng-38; by the North Atlantic Treaty Organiza J. L. Routbort, Indentation, erosion, and strength degradation of tion Grant PST CLG.977016; by the Russian Academy of silicon-alloyed pyrolytic carbon, J Mater. Sci. 28(1993)5917-5921 Sciences, by the Ministerio de Educacion y Ciencias of [21]J.L. Routbort, Degradation of structural ceramics by erosion,J Spain, under CICYT Project No. MAT2000-1533-C03-03 Nondestruct. Eval. 15(1996)107-112. [22]KC. Coretta, D. Singh, T.A. Cruse, W.A.Ellingson, J.J. Picciolo, B.J. Polzin, J. L. Routbort, T W. Spohnholtz, F W. Zok, J.C. McNulty. M. He, W.M. Kriven, SJ. Lee, D K. Kim, G.E. Hilmas, AJ. Mercer. References M.R. Begley, A R. de Arellano-Lopez, Development of Advanced Fibrous Monoliths: Final Report For Project of 1998-2000, Report []w.s. Coblenz, Fibrous monolithic ceramic and method for oduction, US Patent 4772524(September 20, 1988) [23]M. Tlustochowitz, D Singh, w.A. Ellingson, K C. Goretta, M. Rigal [2]D J.W. Halloran, G.E. Hilmas, G.A. Brady, S Somas, A. M. Sutaria, Mechanical-property characterization of multidirectional Bard, G. Zywicki, Process for preparing textured ceramic composites, Si3 N4/BN fibrous monoliths, Ceram. Trans. 103(2000)245-25 US Patent 5645 781 (July 8, 1997) [24]AG. Evans, M.E. Gulden, M E Rosenblatt, Impact damage in brittle B3G.A. Danko, G.E. Hilmas, J. Halloran, B. King, Fabrication and materials in the elastic-plastic response regime, Proc. Roy. Soc properties of quasi-isotropic silicon nitride-boron nitride fibl Lond.Ser.A361(1978)343-36 onoliths, Ceram. Eng. Sci. Proc. 18(3)(1997)607-613 [25]S M. Wiederhorn, B.R. Lawn, Strength degradation of 14G. Hilmas U. Abdali, G. Zywicki, J. Halloran, Fibrous with sharp particles. I. Annealed surfaces, J. Am. otoliths a由:6 fracture from powder-processed ceramics, (1979)66-70 1995)263-268. [26A. w. Ruff, S.M. wiederhorn, Erosion 5]D. Kovar, B.H. King, R.W. Trice, J.W. Halloran, Fibrous monolithic ceramics, J. Am. Ceram Soc. 80(1997)2471-2487 Academic Press, New York, 1979,Pp.69-126, d Technet, in C M. Preece(Ed) Treatise on Materials Sciene
K.C. Goretta et al. / Wear 256 (2004) 233–242 241 [3–9]. One might expect little loss of strength with erosion because the vast majority of cells should be unaffected. The data largely reflect that expectation. Given uncertainties with respect to residual stresses and effects of environmental exposure, little more in the way of assessment can be offered. 5. Conclusions Si3N4/BN FMs and monolithic specimens of the two constituents were subjected to impact at 90◦ by angular 143m-diameter SiC particles traveling at 50, 70, or 100 m/s. Steady-state erosion rates were highest for the BN and lowest for the Si3N4. Erosion rates of the FMs were higher than predicted by a rule of mixtures based on independent erosion rates of the cell and cell-boundary phases. The relatively rapid erosion was attributed primarily to chipping of the cells caused by radial-crack formation upon impact. Surface damage induced by erosion at 100 m/s caused a decrease in bending strength for the Si3N4 of ≈22%. The FMs each exhibited significant strength loss with storage over 1.5 years. Environmentally assisted slow crack growth appeared to contribute to the strength reduction. Within experimental error, the FMs strengths neither decreased nor increased after being eroded at 100 m/s. Acknowledgements We thank T. Tran and J. Katz for experimental assistance; both were partially supported by the Division of Educational Programs, Argonne National Laboratory. This work was supported by the Defense Advanced Research Projects Agency through an Interagency agreement with the US Department of Energy (DOE); by DOE, under Contract No. W-31-109-Eng-38; by the North Atlantic Treaty Organization Grant PST.CLG.977016; by the Russian Academy of Sciences; by the Ministerio de Educación y Ciencias of Spain, under CICYT Project No. MAT2000-1533-C03-03. References [1] W.S. Coblenz, Fibrous monolithic ceramic and method for production, US Patent 4 772 524 (September 20, 1988). [2] D. Popovic, J.W. Halloran, G.E. Hilmas, G.A. Brady, S. Somas, A. Bard, G. Zywicki, Process for preparing textured ceramic composites, US Patent 5 645 781 (July 8, 1997). [3] G.A. Danko, G.E. Hilmas, J. Halloran, B. King, Fabrication and properties of quasi-isotropic silicon nitride-boron nitride fibrous monoliths, Ceram. Eng. Sci. Proc. 18 (3) (1997) 607–613. [4] G. Hilmas, A. Brady, U. Abdali, G. Zywicki, J. Halloran, Fibrous monoliths: non-brittle fracture from powder-processed ceramics, Mater. Sci. Eng. A195 (1995) 263–268. [5] D. Kovar, B.H. King, R.W. Trice, J.W. Halloran, Fibrous monolithic ceramics, J. Am. Ceram. Soc. 80 (1997) 2471–2487. [6] S.W. Lee, D.K. Kim, High-temperature characteristics of Si3N4/BN fibrous monolithic ceramics, Ceram. Eng. Sci. Proc. 18 (4) (1997) 481–486. [7] R.W. Trice, J.W. Halloran, Influence of microstructure and temperature on the interfacial fracture energy of silicon nitride/boron nitride fibrous monolithic ceramics, J. Am. Ceram. Soc. 82 (1999) 2502–2508. [8] J.C. McNulty, M.R. Begley, F.W. Zok, In-plane fracture resistance of cross-ply fibrous monoliths, J. Am. Ceram. Soc. 84 (2001) 367–375. [9] B.J. Polzin, T.A. Cruse, R.L. Houston, J.J. Picciolo, D. Singh, K.C. Goretta, Fabrication and characterization of oxide fibrous monoliths produced by coextrusion, Ceram. Trans. 103 (2000) 237–244. [10] S.-J. Lee, W.M. Kriven, Toughened oxide composites based on porous alumina-platelet interphases, J. Am. Ceram. Soc. 84 (2001) 767–774. [11] A.C. Mulligan, Advanced Ceramics Research, personal communication, 2000. [12] K.C. Goretta, F. Gutierrez-Mora, T. Tran, J. Katz, J.L. Routbort, T.S. Orlova, A.R. de Arellano-Lopez, Solid-particle erosion of ZrSiO4 fibrous monoliths, Ceram. Trans. 139 (2002) 139–146. [13] S.K. Hovis, J.E. Talia, R.O. Scattergood, Erosion in multiphase systems, Wear 108 (1986) 139–155. [14] A.R. de Arellano-Lopez, S. Lopez-Pombero, A. DominguezRodriguez, J.L. Routbort, D. Singh, K.C. Goretta, Plastic deformation of silicon nitride/boron nitride fibrous monoliths, J. Euro. Ceram. Soc. 21 (2001) 245–250. [15] M. Marrero, J.L. Routbort, P. Whalen, C.-W. Li, K.R. Karasek, Solid-particle erosion of in-situ reinforced Si3N4, Wear 162–164 (1993) 280–284. [16] M.A. Boling-Risser, K.C. Goretta, J.L. Routbort, K.T. Faber, Effect of microstructure on the high-temperature mechanical behavior of self-reinforced hot-pressed silicon nitride: compressive creep, J. Am. Ceram. Soc. 83 (2000) 3065–3069. [17] B.K. Kardashev, Yu.A. Burenkov, B.I. Smirnov, V.V. Shpeizman, V.A. Stepanov, V.M. Chernov, D. Singh, K.C. Goretta, Elasticity and anelasticity of ceramic samples of boron nitride, Phys. Sol. State 43 (2001) 1084–1088 (translated from Russian). [18] T.H. Kosel, R.O. Scattergood, A.P.L. Turner, An electron microscopy study of erosive wear, in: K.C. Ludema, W.A. Glaeser, S.K. Rhee (Eds.), Proceedings of the International Conference on Wear of Materials, American Society Mechanical Engineering, New York, 1979, pp. 192–204. [19] J.L. Routbort, R.O. Scattergood, Solid particle erosion of ceramics and ceramic composites, Key Eng. Mater. 71 (1992) 23–50. [20] P. Strzepa, E.J. Zamirowski, J.B. Kupperman, K.C. Goretta, J.L. Routbort, Indentation, erosion, and strength degradation of silicon-alloyed pyrolytic carbon, J. Mater. Sci. 28 (1993) 5917–5921. [21] J.L. Routbort, Degradation of structural ceramics by erosion, J. Nondestruct. Eval. 15 (1996) 107–112. [22] K.C. Goretta, D. Singh, T.A. Cruse, W.A. Ellingson, J.J. Picciolo, B.J. Polzin, J.L. Routbort, T.W. Spohnholtz, F.W. Zok, J.C. McNulty, M. He, W.M. Kriven, S.J. Lee, D.K. Kim, G.E. Hilmas, A.J. Mercer, M.R. Begley, A.R. de Arellano-López, Development of Advanced Fibrous Monoliths: Final Report For Project of 1998–2000, Report ANL-01/04, Argonne National Laboratory, May 2001. [23] M. Tlustochowitz, D. Singh, W.A. Ellingson, K.C. Goretta, M. Rigali, M. Sutaria, Mechanical-property characterization of multidirectional Si3N4/BN fibrous monoliths, Ceram. Trans. 103 (2000) 245–254. [24] A.G. Evans, M.E. Gulden, M.E. Rosenblatt, Impact damage in brittle materials in the elastic–plastic response regime, Proc. Roy. Soc. Lond. Ser. A 361 (1978) 343–365. [25] S.M. Wiederhorn, B.R. Lawn, Strength degradation of glass impacted with sharp particles. I. Annealed surfaces, J. Am. Ceram. Soc. 62 (1979) 66–70. [26] A.W. Ruff, S.M. Wiederhorn, Erosion by solid particle impact, in: C.M. Preece (Ed.), Treatise on Materials Science and Technology, Academic Press, New York, 1979, pp. 69–126
242 K.C. Goretta et al /Wear 256(2004)233-242 27S.M. wiederhom, B.J. Hockey, Effect of material parameters on the 36B.. Smirnov, Yu. A. Burenkov, B K. Kardashev, D. Singh, K.C. erosion resistance of brittle materials, J. Mater. Sci. 18(1980)766- Goretta, A.R. de Arellano-Lopez, Elasticity and anelasticity of silicon nitride/boron nitride fibrous monoliths, Phys. Sol. State 43(2001) 28]BR. Lawn, A.G. Evans, D B. Marshall, Elastic/plastic indentation 2094-2098 ramics: the median/radial crack system, J. Am. Ceram 37]S. Srinivasan, R.O. Scattergood, On lateral cracks in glass, J. Mater Dc63(1980)574581 Sci.22(19873463-3469 229].B. Marshall, B R. Lawn, A.G. Evans, Elastic/plastic indentation [38 C.T. Morrison, J.L. Routbort, R.O. Scattergood, R. Warren, Erosion damage in ceramics: the lateral crack system, J. Am. Ceram. S of an aligned alumina-stainless steel composite, Wear 160(1993) 65(1982)561-566 B0JJ E Ritter, P. Strzepa, K Jakus, L. Rosenfeld, K.J. Buckman, Erosion B9]AG. Evans, Impact damage in ceramics, in: R.C. Bradt, DPH amage in glass and alumina, J. Am. Ceram. Soc. 67(1984)769- Hasselman, FF. Lange(Eds ) Fracture Mechanics of Ceramics, voL. 3, Plenum, New York, 1978, pp. 303-331 B1]J.L. Routbort, Degradation of structural ceramics by erosion, J. [40] M.Y. He, D Singh, J.C. McNulty, F.w. Zok, Thermal expansion of Nondestruct. Eval. 15 (1996)107-112 nidirectional and cross-ply fibrous monoliths, Comp. Sci. Technol. 32]S.Srinivasan, R.O. Scattergood, Effect of erodent hardness on erosion 62(2002)967-976. of brittle materials, Wear 128(1988)139-152. [41]D. Singh, K.C. Goretta, J.w. Richardson Jr, A R de Arellano-Lopez, 33]J.L. Routbort, D.A. Helberg, K.C. Goretta, Erosion of whisker Interfacial sliding stress in Si3 N4/BN fibrous monoliths, Scripta forced ceramics, J. Hard Mater. 1(1990)123-135 Mater.46(2002)747-75 34] Chemical and physical properties of Si3N4 and Sic ceramics, in (42] D.B. Marshall, Surface damage, in: F.L. Riley (Ed ) Ceramics: Ceramic Source, voL. 6, American Ceramic Society, Westerville, OH, Implications for Strength Degradation, Erosion and Wear, in Nitrogen 1990,p.351l Ceramics, Nijhoff, The Hague, 1983, pp 635-656 B35A.R de Arellano-Lopez, unpublished results
242 K.C. Goretta et al. / Wear 256 (2004) 233–242 [27] S.M. Wiederhorn, B.J. Hockey, Effect of material parameters on the erosion resistance of brittle materials, J. Mater. Sci. 18 (1980) 766– 780. [28] B.R. Lawn, A.G. Evans, D.B. Marshall, Elastic/plastic indentation damage in ceramics: the median/radial crack system, J. Am. Ceram. Soc. 63 (1980) 574–581. [29] D.B. Marshall, B.R. Lawn, A.G. Evans, Elastic/plastic indentation damage in ceramics: the lateral crack system, J. Am. Ceram. Soc. 65 (1982) 561–566. [30] J.E. Ritter, P. Strzepa, K. Jakus, L. Rosenfeld, K.J. Buckman, Erosion damage in glass and alumina, J. Am. Ceram. Soc. 67 (1984) 769– 774. [31] J.L. Routbort, Degradation of structural ceramics by erosion, J. Nondestruct. Eval. 15 (1996) 107–112. [32] S. Srinivasan, R.O. Scattergood, Effect of erodent hardness on erosion of brittle materials, Wear 128 (1988) 139–152. [33] J.L. Routbort, D.A. Helberg, K.C. Goretta, Erosion of whiskerreinforced ceramics, J. Hard Mater. 1 (1990) 123–135. [34] Chemical and physical properties of Si3N4 and SiC ceramics, in Ceramic Source, vol. 6, American Ceramic Society, Westerville, OH, 1990, p. 351. [35] A.R. de Arellano-Lopez, unpublished results. [36] B.I. Smirnov, Yu.A. Burenkov, B.K. Kardashev, D. Singh, K.C. Goretta, A.R. de Arellano-Lopez, Elasticity and anelasticity of silicon nitride/boron nitride fibrous monoliths, Phys. Sol. State 43 (2001) 2094–2098. [37] S. Srinivasan, R.O. Scattergood, On lateral cracks in glass, J. Mater. Sci. 22 (1987) 3463–3469. [38] C.T. Morrison, J.L. Routbort, R.O. Scattergood, R. Warren, Erosion of an aligned alumina-stainless steel composite, Wear 160 (1993) 345–350. [39] A.G. Evans, Impact damage in ceramics, in: R.C. Bradt, D.P.H. Hasselman, F.F. Lange (Eds.), Fracture Mechanics of Ceramics, vol. 3, Plenum, New York, 1978, pp. 303–331. [40] M.Y. He, D. Singh, J.C. McNulty, F.W. Zok, Thermal expansion of unidirectional and cross-ply fibrous monoliths, Comp. Sci. Technol. 62 (2002) 967–976. [41] D. Singh, K.C. Goretta, J.W. Richardson Jr., A.R. de Arellano-Lopez, Interfacial sliding stress in Si3N4/BN fibrous monoliths, Scripta Mater. 46 (2002) 747–751. [42] D.B. Marshall, Surface damage, in: F.L. Riley (Ed.), Ceramics: Implications for Strength Degradation, Erosion and Wear, in Nitrogen Ceramics, Nijhoff, The Hague, 1983, pp. 635–656