K.C. Gorenta et al /Materials Science and Engineering A 412 (2005)146-152 The discrepancy between the values of 5-7 and 25 MPa 3.4. Creep is believed to be associated mainly with variations in cross- section along the cell length, the result of intrusion of adja- High-temperature compressive creep of unidirectional cent transverse cells during consolidation by pressing( Fig. 2b). Si3 N4/BN has been studied in detail [32] and that of cross-ply The intrusions also led to formation of cusps at many of laminates preliminarily; monolithic BN and Si3N4 have the triple junctions between longitudinal and transverse cells. also been studied [33]. Tests were conducted in inert gas at These features allowed premature disengagement of the cells 1200-1500oC The BN cell-boundary material exhibited micro- during pullout, which led to an overestimation in the con- scopic plasticity, but did not truly creep. Its high-temperature tact area when the measured pullout lengths were used. The fracture strengths were substantially lower than the creep result was a low inferred sliding stress from the load/cMod stresses that the cells themselves could sustain The Fms approach, which is based on the response following complete exhibited good creep resistance in the direction of the cells. The cell fracture, the inferred sliding stress was 23-33 MPa, com- measured creep rates were slightly faster than those of the host parable to the one measured by pushout on a unidirectional Si3 N4 because the bn carried almost no load; a simple rule of material(25 MPa). This correlation indicates that the latter mixtures based on load-bearing area of Si3N4 was obtained approach for determining the bridging parameters was more Data from steady-state creep tests at 1400-1500oC were fit to reliable than the one based on pullout measurements alone. a standard equatio Nevertheless, the two approaches provided corroborating and de complementary information about the form of the bridging drAole-g/RT Bridging was further studied by measuring directly cell pull- where de/dt is the strain rate, Aa constant, o the stress, the out lengths. Two measurements were made for each cell, one stress exponent is unity, o the apparent activation energy, R on each of the two mating faces (a total of >400 measure- the gas constant, and T is absolute temperature. Q was obtained ments; Fig. 6). Pullout lengths were also determined from SEM directly by changing temperature at constant load on a single photomicrographs of cross-sections. The mean pullout length sample, and by comparing strain rates for multiple specimens was found to be 110 um, in broad agreement with the range at the same stress and various temperatures. The calculated inferred from the bridging model and the bending analysis activation energy for the two approaches were 570+150 and (70-90um) 625+40 kJ/mol, respectively. These two values are equal When used together, the two approaches to determine the within experimental error and are equal to that often reported bridging law also provide valuable insights into the factors that for monolithic Si3N4 [32, and references therein] ontrol the efficacy of the bridging, especially those related to the geometry of the cells. That is, the periodic variations in cell 3.5. Erosion and wear resistance cross-section and the cusps that occur at the triple points between the longitudinal and transverse cells limit the extent of debond The solid-particle erosion studies were conducted in vacuum ing and sliding through the bn to a distance comparable to the A stream of angular Sic particles 143 um in average diameter cell width. Elimination of those cusps by better filling space in impacted the targets at normal incidence and 100 m/s. The soft, the green state should increase pullout lengths and hence work- monolithic BN eroded very rapidly. The Si3N4/BN FMs eroded of-fracture values [18, 31] much more rapidly than did monolithic Si3 N4, and, in fact, faster of rule of mixture erosion of a Si3 N4+ BN composite. Rapid loss of the Bn cell boundary and subsequent large-scale removal of the unsupported Si3N4 cells were deemed to be responsible for the rapid erosion 4. Similar rapid erosion of an FM relative to a dens has been reported for ZrSio4-based FMs [35] Erosive damage did not, on the basis of stress, lower the average RT flexural strength of the FMs, but did lower that of he Si3 Na significantly (by 22%)[34] Friction and wear tests have recently been performed between sintered Si3N4(SN220, Kyocera, Kagoshima, Japan) pins and Si3N4(SN220, Kyocera)and Si3N4/BN FM flats, with and without oil lubrication(Mobil 10w30), in a pin-on-d tribometer(CSEM-THT, Neuchatel, Switzerland ). The surfaces of the Si3 N4 flats were initially rough to allow better comparison with wear of the FMs. The oil-lubricated tests were performed at temperatures to 120 C to 10N load in order to create a severe boundary-lubricated sliding regime and to assess the per Fig. 6. SEM photomicrograph of surface of fractured section of 0/90 cross-pl formance of test materials under such severe conditions Tests Si3N4/BN FM; the cell pullout lengths are short were performed up to 5000 m sliding distance to determine the150 K.C. Goretta et al. / Materials Science and Engineering A 412 (2005) 146–152 The discrepancy between the values of 5–7 and ≈25 MPa is believed to be associated mainly with variations in cross￾section along the cell length, the result of intrusion of adja￾cent transverse cells during consolidation by pressing (Fig. 2b). The intrusions also led to formation of cusps at many of the triple junctions between longitudinal and transverse cells. These features allowed premature disengagement of the cells during pullout, which led to an overestimation in the con￾tact area when the measured pullout lengths were used. The result was a low inferred sliding stress. From the load/CMOD approach, which is based on the response following complete cell fracture, the inferred sliding stress was ≈23–33 MPa, com￾parable to the one measured by pushout on a unidirectional material (≈25 MPa). This correlation indicates that the latter approach for determining the bridging parameters was more reliable than the one based on pullout measurements alone. Nevertheless, the two approaches provided corroborating and complementary information about the form of the bridging law. Bridging was further studied by measuring directly cell pull￾out lengths. Two measurements were made for each cell, one on each of the two mating faces (a total of >400 measure￾ments; Fig. 6). Pullout lengths were also determined from SEM photomicrographs of cross-sections. The mean pullout length was found to be ≈110
m, in broad agreement with the range inferred from the bridging model and the bending analysis (70–90
m). When used together, the two approaches to determine the bridging law also provide valuable insights into the factors that control the efficacy of the bridging, especially those related to the geometry of the cells. That is, the periodic variations in cell cross-section and the cusps that occur at the triple points between the longitudinal and transverse cells limit the extent of debond￾ing and sliding through the BN to a distance comparable to the cell width. Elimination of those cusps by better filling space in the green state should increase pullout lengths and hence work￾of-fracture values [18,31]. Fig. 6. SEM photomicrograph of surface of fractured section of 0/90◦ cross-ply Si3N4/BN FM; the cell pullout lengths are short. 3.4. Creep High-temperature compressive creep of unidirectional Si3N4/BN has been studied in detail [32] and that of cross-ply laminates preliminarily; monolithic BN and Si3N4 have also been studied [33]. Tests were conducted in inert gas at 1200–1500 ◦C. The BN cell-boundary material exhibited micro￾scopic plasticity, but did not truly creep. Its high-temperature fracture strengths were substantially lower than the creep stresses that the cells themselves could sustain. The FMs exhibited good creep resistance in the direction of the cells. The measured creep rates were slightly faster than those of the host Si3N4 because the BN carried almost no load; a simple rule of mixtures based on load-bearing area of Si3N4 was obtained. Data from steady-state creep tests at 1400–1500 ◦C were fit to a standard equation: dε dt = Aσ1 e−Q/RT , (4) where dε/dt is the strain rate, A a constant, σ the stress, the stress exponent is unity, Q the apparent activation energy, R the gas constant, and T is absolute temperature. Q was obtained directly by changing temperature at constant load on a single sample, and by comparing strain rates for multiple specimens at the same stress and various temperatures. The calculated activation energy for the two approaches were 570 ± 150 and 625 ± 40 kJ/mol, respectively. These two values are equal within experimental error and are equal to that often reported for monolithic Si3N4 [32, and references therein]. 3.5. Erosion and wear resistance The solid-particle erosion studies were conducted in vacuum. A stream of angular SiC particles 143
m in average diameter impacted the targets at normal incidence and 100 m/s. The soft, monolithic BN eroded very rapidly. The Si3N4/BN FMs eroded much more rapidly than did monolithic Si3N4, and, in fact, faster than would be predicted by any sort of rule of mixtures for erosion of a Si3N4 + BN composite. Rapid loss of the BN cell boundary and subsequent large-scale removal of the unsupported Si3N4 cells were deemed to be responsible for the rapid erosion [34]. Similar rapid erosion of an FM relative to a dense monolith has been reported for ZrSiO4-based FMs [35]. Erosive damage did not, on the basis of stress, lower the average RT flexural strength of the FMs, but did lower that of the Si3N4 significantly (by 22%) [34]. Friction and wear tests have recently been performed between sintered Si3N4 (SN220, Kyocera, Kagoshima, Japan) pins and Si3N4 (SN220, Kyocera) and Si3N4/BN FM flats, with and without oil lubrication (Mobil 10W30), in a pin-on-disk tribometer (CSEM-THT, Neuchatel, Switzerland). The surfaces of the Si3N4 flats were initially rough to allow better comparison with wear of the FMs. The oil-lubricated tests were performed at temperatures to 120 ◦C to 10 N load in order to create a severe boundary-lubricated sliding regime and to assess the per￾formance of test materials under such severe conditions. Tests were performed up to 5000 m sliding distance to determine the