Abstract
Craze fibril stability of polymer glasses can be characterized by measuring the median strain (craze fibril stability) at the onset of craze fibril breakdown in thin films under a constant low strain rate. Monodisperse polystyrenes (PS's) of molecular weight M = 37 000-20 000 000 were used. A strong increase in craze fibril stability was found as M increased from 50 000 to 20 0000, The increase occurred over the same range as the increase in macroscopic fracture toughness of PS. Voids were observed to always nucleate at the bulk-craze interface and never in the craze midrib. Higher strain rates reduced the craze fibril stability. Foreign particle inclusions, e.g., dust, in specimens can significantly decrease the fibril stability from its intrinsic (clean) value. Even in the presence of such particles, however, the M dependence of fibril stability is qualitatively similar to that of crazes in "dust-free" films. The fibril breakdown statistics can be investigated by examining simultaneously the failure events in a large number of independent film specimens. The statistics of craze fibril breakdown are found to follow a Weibull distribution. Two parameters may be extracted by fitting this distribution to the breakdown data: (1) a Weibull scale parameter ε W , which is a measure of craze fibril stability, and (2) a Weibull modulus ρ, which is a measure of variability, that is, the breadth of the distribution of fibril stability (high ρ's produce narrow distributions and vice versa). The Weibull distribution in a weakest-link setting can be used to predict the effect of sample size on the probability of craze fibril breakdown somewhere in the sample. A microscopic statistical model in which craze fibrils fail by random disentanglement of molecular strands at the craze-bulk interface is developed. The experimental observations are in good agreement with the predictions of the model. ©1986 American Chemical Society.