Subcritical crack growth models for SiC fiber in air and steam: Low temperature data and possible effects of residual stress

Abstract

Failure times (tf) and brittle creep strain rates (ἑ) for static fatigue of Hi-Nicalon™-S fiber tows were measured at 500 and 600 °C. These measurements added to those done previously at 700–1100 °C. Subcritical crack growth (SCG) based models developed to predict tf and ἑ from the 700–1100 °C data were modified to include the additional data. Parameters for two SCG laws were determined for air and Si(OH)4 saturated steam environments, and an additional parameter for effective room-temperature residual stress (σRo) along the crack path was introduced. Tow failure was modelled as sequential filament failure by SCG from weakest to strongest in a Weibull distribution. The increase in stress on intact filaments as they oxidized and as other filaments failed caused brittle creep strain. SCG-based models for both surface and interior cracks were considered. Orthogonal direction regression (ODR) was used to find the best parameter fits to data. Faster numerical methods to find the global parameter-space minima and avoid local minima were developed. New model parameters were determined using the entire 500–1100 °C tf and ἑ data set, done at initial applied stresses from 260 to 1526 MPa. In air, measured tf and ἑ at 500 and 600 °C follow the trends previously observed for higher temperatures. However, beneath 700 °C in steam, tf are longer and ἑ are much slower than expected. They are similar to values measured for air, suggesting an SCG mechanism change between 700° and 600 °C in steam. An atomistic SCG law-based model with activation energy (Q) of 310 kJ/mol and σRo of ∼200 MPa compression is suggested to be appropriate for static fatigue in air. An atomistic SCG law-based model with Q of 360 kJ/mol and σRo of ∼0 MPa best fit data measured in steam. However, differences in data fit quality between models were small, so firm conclusions about the correct model, residual stress effects, and associated crack paths could not be made. As stress approaches zero, the model converges to tow failure times that are the times for complete fiber oxidation. As failure times approach zero and temperatures decrease, the model converges to tow strengths predicted by fiber bundle theory at room temperature. The sources of error in data fit, the merit of different SCG models, and the possible roles of residual stress in SCG for Hi-Nicalon™-S fibers are discussed.