Fig. 2. Equivalent stress-strain curves for samples deformed in continuous torsion testing at several temperatures and strain rates.maximum time intervals spent in the reversed torsion experiments. It can be seen that softening is negligible and almost nil up to 0.5 s, at the test condition. There is then an increase in the softening fraction to 10 % at times of the order of 1 s. Softening then stops and the curve shows, apparently, a small plateau which is followed by an increase in the fraction of softening up to 30 % at 3 s. The curve was extrapolated using an Avrami type equation to full softening taking about 100 s to completion. It can be concluded that no relevant static softening could have occurred during the time intervals spent in the strain reversals, since they were always smaller than 0.3 s.
MetallographySamples were taken from a longitudinal section of the torsion specimens, at a depth of about 150 mm from the surface, for optical and electron scanning microscopy. Preparation of samples for optical microscopy involved grinding down to paper graded 1 000 mesh and polishing with diamond paste sizes 9, 3, 1 and 0.25 mm. Etching was carried out with 6 % nital for a period varying from 20 to 120 s.
Results
Continuous Torsion Testing Stress-Strain CurvesFigsures 2(a) to 2(d) show the stress-strain curves for samples deformed in continuous torsion testing at several temperatures and strain rates. The stresses increased up to a peak stress, s p , and then decreased until a steady state, s ss , was reached. This shape of the stress-strain curve indicates the occurrence of DRX in all tests. As expected, the values of the peak stress and peak strain increased as the temperature decreased or the strain rate increased. Samples tested at 0.01 s Ϫ1 showed oscillating values of the s ss related to grain size growth during recrystallization whereas all other curves displayed no such oscillations.As already mentioned, the temperature of 1 223 K and the strain rate of 0.1 s Ϫ1 were chosen as the base conditions for all tests in the present research. Figure 4(a) highlights the stress-strain curve obtained under these conditions, whereas Fig. 4(b) shows the work hardening rate (Q) versus stress (s) curve. The value of the peak stress can be measured at the point of the Q-s curve at which the value of Q is nil and, in this case, s p is 102 MPa. The value of the peak strain can then be measured from Fig. 4(a), producing e p ϭ0.6. The values of the critical stress and critical strain at which DRX is initiated can also be determined from the Q-s curve, as seen in Fig. 4(b). The value of the critical stress, e c , is obtained at the first point where the slope of the Q-s curve changes from decreasing to increasing values 18) and then the critical strain, e c , can be determined from the stress-strain curve shown in Fig. 4(a). In the present work, these values were s c ϭ95 MPa and e c ϭ0.38, respectively. In these cases, however, there are straining cycles which tend to be above the value of the monotonic steady state stre...