The creep properties of niobium-stabilised stainless steels of carbon contents in the range 0.01 to 0.05% carbon can be accounted for by the general recovery theory of creep. The high stress dependencies of recovery and creep rate can be adequately explained through an internal friction stress or impedance term, retarding recovery. Measurement of this friction term by dislocation density and stress relaxation techniques provides the correct stress dependencies when applied to the modified recovery theory.
t. IntroductionThe recovery theory of creep is based on the knowledge that metals harden with strain, and soften with time on heating [1,2]. During steady state creep, it was postulated that there will be an exact balance between strain-hardening and recovery, such that the steady state creep rate is given by McLean et al [4][5][6][7] who have shown that the above relation is a general one and holds for pure metals and single phase alloys. More recent work on these materials has shown that the theory can be extended into the primary and tertiary stages of creep with equal success [8][9][10][11].The theories of McLean et al [4][5][6][7] although very attractive due to their inherent simplicity, are not generally applicable to high-strength particle-hardened material. In general they suffer from an inability to predict the high stress dependence of the recovery and creep rates found in such materials, although McLean [7] has shown in principle the way in which this discrepancy arises. In addition, it has been found difficult to test the recovery theory on materials containing second phase particles, as it is often not easy and sometimes impossible [12] McLean's theories [4][5][6][7] assume that the dislocations formed during creep exist in a threedimensional network, and that the growth of such a network, essentially by climb of dislocations, constitutes the recovery process. If the average mesh size of the network is x, the driving force for recovery is inversely proportional to x [13] and the growth rate of the average network size is given by dx M. Todt x where M = mobility factor and TD = line tension of the dislocation. This treatment is analogous to that applied to normal grain growth, and the influence of particles in retarding the growth rate [14] can be applied in a similar manner. Lagneborg [15] has introduced an impedance factor, Z, into equation 2 such that Z depends both on the number and size of the impeding particles, and on the nature of the particle-dislocation interaction. A growing network will thus experience a retarding force