Using Voltammetry Augmented with Physics-Based Modeling and Bayesian Hypothesis Testing to Estimate Electrolyte Composition

Abstract

Voltammetry is a foundational electrochemical technique that can qualitatively and quantitatively probe electroactive species in electrolytes and as such has been used in numerous fields of study. Recently, automation has been introduced into voltammetric analyses to extend their capabilities (e.g., Bayesian parameter estimation, compound identification <i>via</i> machine learning); however, opportunities exist to enable more versatile methods across a wider range of electrolyte and experimental conditions. Here, we present a protocol that uses experimental voltammetry, physics-driven models, binary hypothesis testing, and Bayesian inference to enable robust labeling of electroactive species in multicomponent electrolytes across multiple techniques. We first describe the development of this protocol, and we subsequently validate the methodology in a case study involving five <i>N</i>-functionalized phenothiazine derivatives. In this analysis, the protocol correctly labeled an electrolyte containing 10H-phenothiazine and 10-methylphenothiazine from both cyclic voltammograms and cyclic square wave voltammograms, demonstrating its ability to identify electroactive constituents of a multicomponent solution. Finally, we identify areas of further improvement (e.g., achieving greater detection accuracy) and future applications to potentially enhance <i>in situ</i> or <i>operando</i> diagnostic workflows.