The creep of commercially pure polycrystalline metals under constant stress has four stages: a virtually instantaneous extension, decelerating Andrade β creep, almost steady-state Andrade κ creep, and an acceleration towards failure. Little is known about the first stage, and the fourth stage has been extensively reviewed elsewhere. The limited experimental evidence on the physical mechanism of the second stage is reviewed and a critical discussion is given of various theories of this stage. The dependence of strain rate on stress in the third, steady-state, period seems to fall into two regimes, a power law with an exponent of about 4–5, and a rather closely exponential law. The limits of the parameters within which a simple theory of the exponential dependence can be expected to be valid are discussed, and found to be compatible with experiments. Theories of the power-law dependence are discussed, and, appear to be unconvincing. The theoretical models do not relate closely to the metallographic and other physical observations. In view of the weakness of theory, experiments which may indicate the physical processes dominant in steady-state creep are reviewed. It is usually not clear whether they pertain to the power-law or the exponential regime. While the theories all assume that most of the deformation occurs homogeneously within the grains, most experimental observations point strongly to a large deformation at or close to the grain boundaries. However, a detailed study of dislocation processes in a single grain of polycrystalline foil strained in the electron microscope shows that most of the observed strain can be accounted for by the motion of single dislocations through the subgrain structure. There is no clear reconciliation of these two sets of observations. Grain-boundary sliding cannot occur without intragranular deformation. One or other process may dominate the overall deformation; the geometrically dominant process may not be the rate-controlling one.
Read full abstract