Statistical analysis of current-induced switching in metallic atomic-size contacts

Current-driven conductance switching in atomic-size contacts has been attributed to reversible and vibrationmediated atomic rearrangements. Here, we present a comprehensive statistical analysis on the switching properties of atomic gold, copper, and aluminium contacts fabricated by mechanically controllable break junctions. The comparative analysis shows that various bi- and multilevel switching patterns can exist with volatile as well as nonvolatile features, and with distinct material-dependent preferences among the three metals. In addition, we apply different current-biasing protocols to identify intrinsic and material-dependent switching properties under variable boundary conditions. To explore the stability of such memory states, we reveal the stochastic nature of the underlying switching mechanism and suggest a simple qualitative approach to estimate the creation and failure rate for such switching events. The approach is inspired by atomic-scale electromigration models accounting for the observed switching voltages and conductance jumps.