Understanding conductance fluctuations in single-molecule break junctions: a statistical approach

Single-molecule break junction experiments suffer from a lack of understanding of molecular orientation at the electrodes. One emerging approach to tackle this challenge is change point detection (CPD), a tool commonly used in econometrics to identify points of structural break in data. Thus far, however, applications of CPD to single-molecule break junctions assume that only one structural break exists in conductance versus displacement data – limiting their ability to study the subtle but meaningful conductance fluctuations within the molecular plateau. Modifying the Pruned Exact Linear Time method of change point detection, we develop a CPD algorithm, PELT-R, which detects a variable number of change points and seamlessly differentiates blank traces from those with molecular character. Across multiple experimental and synthetic single-molecule datasets, we demonstrate that PELT-R enables the study of trace history and indicates that the molecular plateau cannot be characterized by a single conductance value as most contemporary approaches suggest. The CPD algorithm developed in this work is broadly applicable to many molecular systems and serves as a starting point for investigating the atomic reorganizations responsible for observed conductance changes.