28 February 2018 in The Journal of Chemical Physics
We present a new automated method for structural classification of the traces obtained in break junction experiments. Using recurrent neural networks trained on the traces of minimal cross-sectional area in molecular dynamics simulations, we successfully separate the traces into two classes: point contact or nanowire. This is done without any assumptions about the expected features of each class. The trained neural network is applied to experimental break junction conductance traces, and it separates the classes as well as the previously used experimental methods. The effect of using partial conductance traces is explored, and we show that the method performs equally well using full or partial traces (as long as the trace just prior to breaking is included). When only the initial part of the trace is included, the results are still better than random chance. Finally, we show that the neural network classification method can be used to classify experimental conductance traces without using simulated results for training, but instead training the network on a few representative experimental traces. This offers a tool to recognize some characteristic motifs of the traces, which can be hard to find by simple data selection algorithms.
7 February 2017 in The Journal of Chemical Physics
We review data analysis techniques that can be used to study temporal correlations among conductance traces in break junction measurements. We show that temporal histograms are a simple but efficient tool to check the temporal homogeneity of the conductance traces, or to follow spontaneous or triggered temporal variations, like structural modifications in trained contacts, or the emergence of single-molecule signatures after molecule dosing. To statistically analyze the presence and the decay time of temporal correlations, we introduce shifted correlation plots. Finally, we demonstrate that correlations between the opening and subsequent closing traces may indicate structural memory effects in atomic-sized metallic and molecular junctions. Applying these methods on measured and simulated gold metallic contacts as a test system, we show that the surface diffusion induced flattening of the broken junctions helps to produce statistically independent conductance traces at room temperature, whereas at low temperature repeating tendencies are observed as long as the contacts are not closed to sufficiently high conductance setpoints. Applying opening-closing correlation analysis on Pt–CO–Pt single-molecule junctions, we demonstrate pronounced contact memory effects and recovery of the molecule for junctions breaking before atomic chains are formed. However, if chains are pulled the random relaxation of the chain and molecule after rupture prevents opening-closing correlations.