A “double-torsion” test jig for operation in vacuum or a selected gas atmosphere at temperatures up to 1500°C is described. The jig has been used to determine stress intensity (K_I)-crack velocity (v) relationships for single-phase β' Si-Al-O-N ceramics which have been hot-pressed with different “impurity” densification aids and which have received “post-sintering” heat treatments.

The K_I-v data have been calculated using three testing procedures; (a) for high crack velocities (10⁻² to 10⁻⁴ ms⁻¹) from the rate of load relaxation, (b) for intermediate velocities from the “constant displacement rate” technique and, (c) for low velocities (10⁻⁶ to 10⁻¹⁰ ms⁻¹) from direct measurements of crack velocity on sequential application of a constant load. A procedure for a “machine-relaxation” correction for technique (a), evaluation of the deviation in the “constant K_I” assumption with crack length and the sensitivity to specimen thickness, is described.

The data can be fitted to a relationship of the form v = constant K_Iⁿ over a wide range of crack velocity with excellent continuity in data obtained from the different techniques and good agreement with isolated values from “notched-beam” tests.

Marked variation in the stress-intensity exponent n which occurs for ceramics with different sintering aids may be correlated with differing susceptibility to subcritical crack growth by means of interlinkage of creep cavities and, in turn, on grain boundary structure. A progressive change in the v-K_I relationship with heat treatment in “noncavitating” ceramics is discussed in terms of current theories for diffusive intergranular crack growth.

The consistency of these data, sensitivity to change in grain-boundary structure, and a correlation with creep behavior indicate that double-torsion tests on small ceramic specimens at elevated temperatures may be used with confidence in determining K_I-v relationships, either for failure prediction or in distinguishing between different slow-crack growth mechanisms.