(203f) Chemical Identification in Multicomponent Electrolytes Using Voltammetry, Physics-Based Modeling, and Bayesian Inference | AIChE

(203f) Chemical Identification in Multicomponent Electrolytes Using Voltammetry, Physics-Based Modeling, and Bayesian Inference

Authors 

Brushett, F., Massachusetts Institute of Technology
The prospects of a renewably powered grid continue to grow with the decreasing cost of sustainable energy technologies [1,2]. However, the intermittency of variable resources inhibits their deployment, necessitating energy management systems capable of operating across a range of time scales, such as rechargeable batteries (e.g., lithium-ion batteries, redox flow batteries). These batteries are complex devices whose performance and durability are dictated by the interplay of many constituent components (e.g., electrodes, membranes/separators), with the liquid electrolyte being especially important for long-term operation. Understanding and mitigating electrolyte decomposition within electrochemical systems is key to extending lifetime [3-5], and studies that probe electrolyte degradation are typically conducted using post mortem analysis to retroactively elucidate the decay pathway(s) [4-6]. While this approach is largely successful, it relies on a suite of ex situ techniques (e.g., nuclear magnetic resonance, mass spectrometry, and UV-Vis spectroscopy) that are time-consuming and expensive. Further, preparatory steps necessary for these ex situ analyses (e.g., dilution) can alter electrolyte composition, potentially providing incomplete information about electrolyte behavior. To this end, machine learning algorithms are emerging that label electroactive compounds from in situ electroanalytical techniques such as voltammetry [7,8].While noteworthy, existing protocols can be improved; some methods identify compounds less accurately when the training and testing data are obtained under different electrolyte or experimental conditions [7], while others do not evaluate all possible species combinations in multicomponent solutions [8].

Building upon this prior work, we have developed a versatile protocol that uses physical modeling and binary hypothesis testing to identify redox-active compounds in multicomponent electrolytes. The protocol references a compound library that catalogues physical descriptors (e.g., redox potential, diffusion coefficients) and compares the library to experimental data using Bayesian inference to identify the redox-active species present. The compound identities are then reported to the user, the knowledge of which can be used to perform more targeted ex situ experiments by narrowing the feasible set of candidate compounds. This process, in turn, can simplify and streamline electrolyte analysis and may consequently accelerate the development of electrochemical technologies. In this presentation, we will describe protocol development and validation using a model set of phenothiazine derivatives across two distinct voltammetric techniques. We will also discuss how this approach may be used to monitor the complex phenomena underlying electrolyte behavior in practical electrochemical systems.

Acknowledgments: This work was supported as part of the National Science Foundation (NSF) under Award Number 1805566. Any opinion, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the NSF. We also gratefully acknowledge the MIT Supercloud and Lincoln Laboratory Supercomputing Center for providing HPC resources that have contributed to the research results reported within this work. We finally thank Professor Susan Odom, Dr. Aman Kaur, and the Odom Research Group at the University of Kentucky for synthesizing, purifying, and shipping phenothiazines.

References:

(1) Bolinger, M.; Seel, J. Utility-Scale Solar: Empirical Trends in Project Technology, Cost, Performance, and PPA Pricing in the United States - 2018 Edition; 2018.

(2) Lazard. Lazard’s Levelised Cost of Energy Analysis - Version 13.0; 2019; Vol. 13.

(3) Day, R. P.; Xia, J.; Petibon, R.; Rucska, J.; Wang, H.; Wright, A. T. B.; Dahn, J. R. Differential Thermal Analysis of Li-Ion Cells as an Effective Probe of Liquid Electrolyte Evolution during Aging. Journal of The Electrochemical Society 2015, 162 (14), A2577–A2581. https://doi.org/10.1149/2.0181514jes.

(4) Goulet, M.-A.; Aziz, M. J. Flow Battery Molecular Reactant Stability Determined by Symmetric Cell Cycling Methods. Journal of The Electrochemical Society 2018, 165 (7), A1466–A1477. https://doi.org/10.1149/2.0891807jes.

(5) Narayana, K. A.; Casselman, M. D.; Elliott, C. F.; Ergun, S.; Parkin, S. R.; Risko, C.; Odom, S. A. N-Substituted Phenothiazine Derivatives: How the Stability of the Neutral and Radical Cation Forms Affects Overcharge Performance in Lithium-Ion Batteries. ChemPhysChem 2015, 16 (6), 1179–1189. https://doi.org/10.1002/cphc.201402674.

(6) Kwabi, D. G.; Lin, K.; Ji, Y.; Kerr, E. F.; Goulet, M.-A.; De Porcellinis, D.; Tabor, D. P.; Pollack, D. A.; Aspuru-Guzik, A.; Gordon, R. G.; Aziz, M. J. Alkaline Quinone Flow Battery with Long Lifetime at pH 12. Joule 2018, 2, 1894–1906. https://doi.org/10.1016/j.joule.2018.07.005.

(7) Ye, J.-J.; Lin, C.-H.; Huang, X.-J. Analyzing the anodic stripping square wave voltammetry of heavy metal ions via machine learning: Information beyond a single voltammetric peak. Journal of Electroanalytical Chemistry 2020, 113934. https://doi.org/10.1016/j.jelechem.2020.113934.

(8) Dean, S. N.; Shriver-Lake, L. C.; Stenger, D. A.; Erickson, J. S.; Golden, J. P.; Trammell, S. A. Machine Learning Techniques for Chemical Identification Using Cyclic Square Wave Voltammetry. Sensors (Switzerland) 2019, 19, 2392. https://doi.org/10.3390/s19102392.