Pulse duration memory of weakly pinned charge-density waves

An explanation is proposed for the pulse duration memory (PDM) which certain weakly pinned charge density waves (CDW's) exhibit in their oscillatory response to repetitive pulses of electric field. A steady state is reached in which the phase of the current oscillation at the end of each pulse is practically independent of the pulse duration. Present models of the phenomenon follow Fukuyama, Lee, and Rice (FLR) [Phys. Rev. B 17, 535 (1978); 19, 3970 (1979)] in regarding the CDW as an elastic body, and account for the current Ic carried by the CDW always rising as the pulse ends, as in the first observation of PDM, in K0.3MoO₃. In numerical studies these models have shown PDM only when the pinning by impurities is strong. In experiments, however, PDM is exhibited by weakly-pinned CDW's, notably in NbSe₃, where it differs from that in the models in that Ic is near or past maximum as the pulse ends. In this paper the FLR models are reviewed, and it is concluded that their failure to account for PDM when the pinning is weak, and for certain other features of the experiments, is the result of the CDW being assumed to behave elastically. Alternatives, in which the amplitude of the CDW collapses to allow phase slip, are then considered. A model is developed in which dislocations in the structure of the CDW, generated at Frank-Read sources, allow phase slip between the moving CDW and a layer which remains stationary. With appropriate choices of parameters, the model accounts for the forms of PDM observed in weakly pinned systems, and for other features inexplicable in FLR terms. Evidence of the presence of stationary layers is examined, and it is shown that their disappearance at low temperatures would account for a switching phenomenon, and absence of PDM, seen in some specimens of NbSe₃. The relation of the model to the phase organization displayed by the FLR models is also briefly discussed.