Tensile Creep and Rupture of Silicon Nitride

We have characterized the tensile creep, rupture lifetime, and cavitation behavior of a commercial, gas-pressure-sintered silicon nitride in the temperature range 1150° to 1400°C and stress range 70 to 400 MPa. Individual creep curves generally show primary, secondary, and tertiary creep. The majority of the primary creep is not recoverable. The best representation of the data is one where the creep rate depends exponentially on stress, rather than on the traditional power law. This representation also removes the need to break the data into high and low stress regimes. Cavitation of the interstitial silicate phase accompanies creep under all conditions, and accounts for nearly all of the measured strain. These observations are consistent with a model where creep proceeds by the redistribution of silicate phase from cavitating interstitial pockets, accommodated by grain-boundary sliding of silicon nitride.     

Tensile Creep Behavior of a Gas-Pressure-Sintered Silicon Nitride Containing Silicon Carbide

The tensile creep behavior of a gas-pressure-sintered silicon nitride containing silicon carbide was characterized at temperatures between 1375° and 1450°C with applied stresses between 50 and 250 MPa. Individual specimens were tested at fixed temperatures and applied loads. Each specimen was pin-loaded within the hot zone of a split-tube furnace through silicon carbide rods connected outside the furnace to a pneumatic cylinder. The gauge length was measured by laser extensometry, using gauge markers attached to the specimen. Secondary creep rates ranged from 0.54 to 270 Gs⁻¹, and the creep tests lasted from 6.7 to 1005 h. Exponential functions of stress and temperature were fitted to represent the secondary creep rate and the creep lifetime. This material was found to be more creep resistant than two other silicon nitride ceramics that had been characterized earlier by the same method of measurement as viable candidates for high-temperature service.     

Comparison of Tensile and Compressive Creep Behavior in Silicon Nitride

The creep behavior of a commercial grade of Si₃N₄ was studied at 1350° and 1400°C. Stresses ranged from 10 to 200 MPa in tension and from 30 to 300 MPa in compression. In tension, the creep rate increased linearly with stress at low stresses and exponentially at high stresses. By contrast, the creep rate in compression increased linearly with stress over the entire stress range. Although compressive and tensile data exhibited an Arrhenius dependence on temperature, the activation energies for creep in tension, 715.3 ± 22.9 kJ/mol, and compression, 489.2 ± 62.0 kJ/mol, were not the same. These differences in creep behavior suggests that mechanisms of creep in tension and compression are different. Creep in tension is controlled by the formation of cavities. The cavity volume fraction increased linearly with increased tensile creep strain with a slope of unity. A cavitation model of creep, developed for materials that contain a triple-junction network of second phase, rationalizes the observed creep behavior at high and low stresses. In compression, cavitation plays a less important role in the creep process. The volume fraction of cavities in compression was ∼18% of that in tension at 1.8% axial strain and approached zero at strains <1%. The linear dependence of creep rate on applied stress is consistent with a model for compressive creep involving solution–precipitation of Si₃N₄. Although the tensile and compressive creep rates overlapped at the lowest stresses, cavity volume fraction measurements showed that solution–precipitation creep of Si₃N₄ did not contribute substantially to the tensile creep rate. Instead, cavitation creep dominated at high and low stresses.     

Comparison of the Creep and Creep Rupture Performance of Two HIPed Silicon Nitride Ceramics

Measurements of the tensile creep and creep rupture behavior were used to evaluate the long-term mechanical reliability of a commercially available and a developmental hot isostatically pressed (HIPed) silicon nitride. Measurements were conducted at 1260° and 1370°C utilizing button–head tensile specimens. The stress and temperature sensitivities of the secondary creep rates were used to estimate the stress exponent and activation energy associated with the dominant creep mechanism. The stress and temperature dependencies of creep rupture life were determined by continuing individual creep tests to specimen failure. Creep deformation in both materials was associated with cavitation at multigrain junctions. Two-grain cavitation was also observed in the commercial material. Failure in both materials resulted from the evolution of an extensive damage zone. The failure times were uniquely related to the creep rates, suggesting that the zone growth was constrained by the bulk creep response. The fact that the creep and creep rupture behaviors of the developmental silicon nitride were significantly improved compared to those of the commercial material was attributed to the absence of cavitation along two-grain junctions in the developmental material.     

Results of an International Round-Robin for Tensile Creep Rupture of Silicon Nitride

Fourteen laboratories participated in an interlaboratory study to establish the within- and between-laboratory repeatability of tensile creep rupture of silicon nitride. In air at 1375°C at 200 MPa, the times to failure ranged over a factor of 50, and the minimum creep rates ranged over a factor of 20. Despite these large ranges, taken individually, no one laboratory stands out from any other; all produced equally acceptable data. Consumers of silicon nitride tensile creep data must accept this magnitude of variability in reported creep data. The wide variety of specimen shapes and sizes, gripping systems, extensometry techniques, and temperature measurement strategies makes it impossible to assign definitively the root cause of the variability. However, there was a significant specimen size effect. As a group, the small-diameter specimens lasted roughly five times longer and crept three times more slowly than the large-diameter buttonhead specimens. A possible interpretation of the origin of this difference is that the oxidizing conditions affected more of the volume of the small specimens during the test.     

Strength and Creep Behavior of Rare-Earth Disilicate–Silicon Nitride Ceramics

The flexural strength and creep behavior of RE₂Si₂O₇–Si₃N₄ materials were examined. The retention in room-temperature strengths displayed by these ceramics at 1300°C was 80–91%, with no evidence of inelastic deformation preceding failure. The steady-state creep rates, at 1400°C in flexural mode, displayed by the most refractory materials are among the lowest reported for sintered Si₃N₄. The creep behavior was found to be strongly dependent on residual amorphous phase viscosity as well as on the oxidation behavior of these materials. All of the rare-earth oxide sintered materials, with the exception of Sm₂Si₂O₇–Si₃N₄, had lower creep strains than the Y₂Si₂O₇–Si₃N₄ material.     

Creep and Creep Rupture of an Advanced Silicon Nitride Ceramic

Creep and creep rupture behavior of an advanced silicon nitride ceramic were systematically characterized in the temperature range 1150° to 1300°C using uniaxial tensile creep tests. Absence of tertiary creep and the order-of-magnitude breaks in both creep rate and rupture lifetime at certain threshold combinations of stress and temperature were two characteristic features of the creep behavior observed. Thermal annealing was found to have enhanced both subsequent creep resistance and creep rupture life. The stress exponent (n) and the activation energy (Q) defined in the Norton relation were found to be 12.6 and 1645 kJ/mol for the material investigated. Both values appear to fall in the general range of those reported for other but similar types of Si₃N₄ ceramic materials. The stress exponent, m , equivalent to the slope of the Larson–Miller equation was found to be in the range 13 to 14.4, and that defined as p in the Monkman–Grant relation to be 0.91, based on the available experimental data. The values of m , n , and p obtained above approximately support the interrelationship of the three exponents given by p = m/n.     

Creep and Stress Rupture Behavior of an Advanced Silicon Nitride: Part II, Creep Rate Behavior

In Part I of this paper, experimental observations on creep testing of 74 tensile specimens of an advanced silicon nitride were presented. In this part, equations are developed for predicting creep rates in the primary and secondary regimes in the temperature range 1477–1673 K. The resulting model predicts creep strain rates to within a factor of two. The underlying phenomenological basis, which employs an activation energy approach, is discussed. The mechanisms that are likely to be responsible for the transiency of the primary creep regime and for the unique stress and temperature dependencies of the creep rates are explored.     

Interlaboratory Verification of Silicon Nitride Tensile Creep Properties

Five laboratories tested NIST-supplied, pin-loaded, 76-mm-long tensile creep specimens at 1400°C under a 150 MPa load using flag-based, laser extensometry. The laboratories reported failure time and strain and supplied the individual creep curves. Only one of the laboratories produced failure times that were significantly less than the others. It is likely that their reduced failure times resulted from small load calibration and test temperature errors. After steps were taken to ameliorate these problems, three additional tests yielded failure times that agreed with those of the other four laboratories. Although the times to failure from the four laboratories that initially agreed were statistically indistinguishable, their creep curves exhibited subtle differences. These differences probably arose because the laboratories used different gage length definitions. When we recalculated the creep curves to the same gage length definition, the differences between the four laboratories whose times to failure agreed, vanished. Although a number of the specimens exhibited edge chips, creep cracks, and obvious chemical interactions with the flags, the presence of these defects did not reduce the time or strain to failure. Two additional creep tests in our laboratory, using specimens that were grossly misaligned, yielded failure times and strains that were commensurate with those from well-aligned specimens.     

Influence of Grain Size on the Tensile Creep Behavior of Ytterbium-Containing Silicon Nitride

The effect of grain size on the tensile creep of silicon nitride was investigated on two materials, one containing 5% by volume Yb₂O₃, the other containing 5% by volume Yb₂O₃ and 0.5% by mass Al₂O₃. Annealing of the Al₂O₃-free silicon nitride for a longer period during processing increased the grain size by a factor of 2. This increase did not affect the tensile creep rate; the grain size exponent of the creep rate differed little from zero, p =−0.20 ± 1.37 (95% confidence level). This finding supports the more recent theories of tensile creep for which p = 0 or −1 and rejects the more classical theory of solution-precipitation. In compression, a more limited data set showed p =−1.89 ± 1.72 (95% confidence level). In contrast to the Al₂O₃-free material, a longer term anneal of the Al₂O₃-containing material during processing did not increase its grain size. Despite this, the longer-annealed Al₂O₃-containing material crept 10 to 100 times slower than the short-annealed material. The enhancement of creep resistance may be a consequence of SiAlON formation during the additional annealing, which reduces the Al content in the amorphous phase and increases its viscosity. Such changes in chemical composition of the grain boundaries are more effective in controlling tensile creep rate than is the grain size.     

Creep Behavior of a Sintered Silicon Nitride

A commercial sintered silicon nitride has been crept in bending and compression at temperatures of 1100°C to 1400°C. In the as-sintered condition the material contains an amorphous intergranular phase. This phase undergoes partial devitrification as a result of high temperature exposure. Preannealing the material to a stable microstructure has very little effect on the creep properties. Deformation behavior compares well with that predicted from a model for creep due to viscous flow of a non-Newtonian grain boundary phase. In bending, the model predicts an initial constant strain rate at low strains as the intergranular phase is squeezed out from between grains under compression. Samples crept in compression are not expected to have this same initial constant strain rate regime. The model also predicts a strong initial strain rate dependence (in bending) on the initial thickness of the amorphous grain boundary layer. Experimentally this strain rate is not affected by partial grain boundary crystallization, suggesting that partial devitrification does not alter the intergranular film thickness or viscosity. This is supported by transmission electron microscopy, which has shown that crystallization of the intergranular phase occurs largely in the pockets between grains, leaving amorphous films between grains.     

A New Model for Tensile Creep of Silicon Nitride

The tensile creep rate of most commercial grades of Si₃N₄ increases strongly with stress. Although the usual power-law functions can represent the creep data, the data often show curvature and systematic variations of slope with temperature and stress. In this article, we present a new approach to understanding the creep of ceramics, such as Si₃N₄, where a deformable second phase bonds a deformation-resistant major phase. A review of experimental data suggests that the rate of formation and growth of cavities in the second phase controls creep in these materials. The critical step for deformation is the redistribution of the second phase away from the cavitation site to the surrounding volume. The effective viscosity of the second phase and the density of active cavities determine the creep rate. Assuming that the hydrostatic stresses in pockets of the second phase are normally distributed leads to a model that accurately describes the tensile creep rate of grades of Si₃N₄. In this model, the creep rate increases exponentially with the applied stress, is independent of Si₃N₄ grain size, is inversely proportional to the effective viscosity of the deformable phase, and is proportional to the cube of the volume fraction of the deformable phase.     

Elevated-Temperature Mechanical Properties of Silicon Nitride/Boron Nitride Fibrous Monolithic Ceramics

A unique, all-ceramic material capable of nonbrittle fracture via crack deflection and delamination has been mechanically characterized from 25° through 1400°C. This material, fibrous monoliths, was comprised of unidirectionally aligned 250 μm diameter silicon nitride cells surrounded by 10 to 20 μm thick boron nitride cell boundaries. The average flexure strengths of fibrous monoliths were 510 and 290 MPa for specimens tested at room temperature and 1300°C, respectively. Crack deflection in the BN cell boundaries was observed at all temperatures. Characteristic flexural responses were observed at temperatures between 25° and 1400°C. Changes in the flexural response at different temperatures were attributed to changes in the physical properties of either the silicon nitride cells or boron nitride cell boundary.     

Cavitation Contributes Substantially to Tensile Creep in Silicon Nitride

During tensile creep of a hot isostatically pressed (HIPed) silicon nitride, the volume fraction of cavities increases linearly with strain; these cavities produce nearly all of the measured strain. In contrast, compressive creep in the same stress and temperature range produces very little cavitation. A stress exponent that increases with stress (ε∞σ ⁿ , 2 < n < 7) characterizes the tensile creep response, while the compressive creep response exhibits a stress dependence of unity. Furthermore, under the same stress and temperature, the material creeps nearly 100 times faster in tension than in compression. Transmission electron microscopy (TEM) indicates that the cavities formed during tensile creep occur in pockets of residual crystalline silicate phase located at silicon nitride multigrain junctions. Small-angle X-ray scattering (SAXS) from crept material quantifies the size distribution of cavities observed in TEM and demonstrates that cavity addition, rather than cavity growth, dominates the cavitation process. These observations are in accord with a model for creep based on the deformation of granular materials in which the microstructure must dilate for individual grains to slide past one another. During tensile creep the silicon nitride grains remain rigid; cavitation in the multigrain junctions allows the silicate to flow from cavities to surrounding silicate pockets, allowing the dilatation of the microstructure and deformation of the material. Silicon nitride grain boundary sliding accommodates this expansion and leads to extension of the specimen. In compression, where cavitation is suppressed, deformation occurs by solution—reprecipitation of silicon nitride.     

Tensile Creep in an in Situ Reinforced Silicon Nitride

The tensile creep of an in situ reinforced silicon nitride is described in terms of the rheological behavior of the thin intergranular film present in this liquid-phase sintered silicon nitride. The high stress exponents and apparent activation energies (at constant stress) can be explained assuming non-Newtonian flow behavior of the film during grain boundary sliding. Time-to-failure is related to the minimum creep rate, even for samples which fail by slow crack growth. In addition, the primary creep region and the relaxation effects observed on unloading are described in terms of grain boundary sliding modified by the presence of a grain boundary phase with a lower elastic modulus than silicon nitride.     

Stress Rupture in Tension and Flexure of a SiC-Whisker-Reinforced Silicon Nitride Composite at Elevated Temperatures

Stress rupture of a 20 vol% SiC whisker-reinforced Si₃N₄ composite processed by gas pressure sintering was investigated by both tension and flexure methods. The stress exponents for the stress rupture decrease with increasing temperature. The fracture surfaces of both tensile and flexural stress rupture at 1000°C consist of mirror, mist, and hackle regions. The size of the mirror region increases with decreasing stress. Crack propagation is a mixture of intergranular and transgranular modes at 1000°C. Both tensile and flexural fracture surfaces under constant stress at 1200°C were characterized by a rough zone and a mirror zone; the size of the rough zone increased with decreasing stress. Creep crack growth occurred at 1200°C, which is a process of cavity nucleation, growth, and interlinkage in front of a crack. The transition of fracture mechanisms with temperature is discussed.     

Microstructure and Creep Behavior of Silicon Nitride and SiAlONs

Microstructural evolution of silicon nitride (Si₃N₄) and SiAlON materials and its influence on creep resistance is reviewed. Grain size, grain morphology, and the ratio of α- to β-phase grains play a part in resistance to creep. The glassy, intergranular phase typically has the strongest influence on creep. Creep data are usually obtained using uniaxial tensile or compressive tests, where creep in tension is controlled by cavitation and grain boundary sliding controls creep in compression. The impression creep methodology is also reviewed. An additional creep mechanism, dilation of the SiAlON grain structure, was found to be active in impression creep.     

Fracture Behavior of Multilayer Silicon Nitride/Boron Nitride Ceramics

The fracture behavior of multilayer Si₃N₄/BN ceramics in bending has been studied. The materials were prepared by a process of tape casting, coating, laminating, and hot pressing. The Si₃N₄ layers were separated by thin, weak BN interlayers. Crack patterns in bending bars were examined with a scanning electron microscope. The weak layers deflected cracks in bending and thus prevented catastrophic failure. In one well-aligned multilayer ceramic A, a main crack propagated through the specimen although along a zigzag path. A second multilayer ceramic B was made to simulate a wood grain structure. Its failure was dominated by shear cracking along the weak BN layers. Besides crack deflection, interlock bridging between toothlike layers in the wake of the main crack appeared also to contribute to toughening.     

Creep and Stress Rupture Behavior of an Advanced Silicon Nitride: Part I, Experimental Observations

Cylindrical buttuohead specimens of an advanced silicon nitride were tested in uniaxial tension at temperatures between 1422 and 1673 K. In the range 1477 to 1673 K, creep deformation was reliably measured using high-temperature contact probe extensometry. Extensive scanning and transmission electron microscopy has revealed the formation of lenticular cavities at two-grain junctions at all temperatures (1422–1673 K) and extensive triple-junction cavitation occurring at the higher temperatures (1644–1673 K). Cavitation is believed to be part of the net creep process. The stress rupture data show stratification of the Monkman–Grant lines with respect to temperature. Failure strain increased with increase in rupture time or temperature, or decrease in stress. Fractography showed that final failure occurred by subcritical crack growth in all specimens.     

Creep Behavior of a SiC-Whisker-Reinforced Alumina

Creep tests in four-point flexure loading configuration in air employing applied stresses of 37 to 300 MPa at temperatures of 1200°, 1300°, and 1400°C were performed on 20-vol%-SiC-whisker-reinforced alumina and unreinforced single-phase polycrystalline alumina. The creep rate of polycrystalline alumina was significantly reduced through the addition of SiC whiskers, although strain to failure was lower. Transmission and scanning electron microscopy results suggest that substantial increase in the creep resistance in flexure of alumina composites originates from the retardation of grain-boundary sliding by the SiC whiskers.