Stochastic current switching in semiconductor superlattices: observation of non-exponential kinetics

We report the experimental measurement of first-passage-time distributions associated with noise-induced current switching in doped, weakly-coupled GaAs/AlAs superlattices, in a regime of nonlinear electronic transport where the static current-voltage ($I - V$) curves exhibit multiple branches and bistability. For applied voltages near the end of each branch, internal shot noise induces switching of measured current to the next branch with a stochastically varying switching time. Switching time distributions are constructed by carrying out up to $10^5$ measurements under identical initial conditions. We have implemented a novel, high bandwidth technique that permits measurement of switching times over very large dynamic range of approximately $10^9$, with measured times ranging from $4$ ns to $10$ s. For relatively small times ($<$ 10$\mu$s), the switching time distributions show exponential tails, as expected for activated escape from an initial metastable state. However, at larger times ($>$ 10 $\mu$s), the distributions exhibit approximate power law tails that extend over several decades of time, with additional fine structure. A rate equation model indicates the possible role of multiple, nearly degenerate metastable states in producing the long tail behavior.