Bond Breaking Kinetics in Mechanically Controlled Break Junction Experiments: A Bayesian Approach

Abstract

Breakjunction experiments allow investigating electronic and spintronic properties at the atomic and molecular scale. These experiments generate by their very nature broad and asymmetric distributions of the observables of interest, and thus a full statistical interpretation is warranted. We show here that understanding the complete distribution is essential for obtaining reliable estimates. We demonstrate this for Au atomic point contacts, where by adopting Bayesian reasoning we can reliably estimate the distance to the transition state, $x{\ddag}$, the associated free energy barrier, ${\Delta}G{\ddag}$, and the curvature $\nu$ of the free energy surface. Obtaining robust estimates requires less experimental effort than with previous methods, fewer assumptions, and thus leads to a significant reassessment of the kinetic parameters in this paradigmatic atomic-scale structure. Our proposed Bayesian reasoning offers a powerful and general approach when interpreting inherently stochastic data that yield broad, asymmetric distributions for which analytical models of the distribution may be developed.