Data-Driven Analysis of High-Throughput Experiments on Liquid Battery Electrolyte Formulations: Unraveling the Impact of Composition on Conductivity

Abstract

An in-house, unique, custom-developed high-throughput experimentation facility, used for discovery of novel and optimization of existing electrolyte formulations for diverse cell chemistries and targeted applications, follows a high-throughput formulation-characterization-performance-elucidation-optimization-evaluation chain based on a set of previously established requirements. Here, we propose a scalable data-driven workflow to predict ionic conductivities of non-aqueous battery electrolytes based on linear and Gaussian regression, considering a dataset acquired from one-of-a-kind high-throughput electrolyte formulation to high-throughput conductivity measurement sequence. Deeper insight into various compositional effects is gained from a generalized Arrhenius analysis, in which conductivities, activation energies and deviations from Arrhenius behavior are determined separately. Each observable displays a specific dependence on the electrolyte salt concentration. The conductivity is fully insensitive to the addition of electrolyte additives for otherwise constant molar composition. We also discuss and interpret qualitative trends predicted by the data-driven model in light of physical features such as viscosity or ion association effects.

Robotic cell assembly to accelerate battery research

“…The overall workflow of the herein presented automatic coin cell assembly robot focuses on the production process after electrode coating and electrolyte formulation. Though the system is principally amendable to manufacturing cells with different electrolyte mixtures 19,20 and electrodes, the herein presented study uses the same electrolyte (1M LiPF6 in 3:7 EC:EMC by weight formulated by Elyte, Germany) and electrodes throughout.…”

One‐Shot Active Learning for Globally Optimal Battery Electrolyte Conductivity**

“…The dataset DS1 used herein is the same underlying the study presented by Flores et al [13] using the formulation and characterization setup reported by Krishnamoorthy et al [14] The herein presented one-shot active learning approach is model free, meaning that we do not utilize any physics or chemistry knowledge except correct pose of the input (formulation) and output (conductivity) and a compartmentalization of the problem by temperature.…”

“…The studies by Dave et al [11,12] consider a wide range of electrolyte formulations but within a narrow range of temperatures. Utilizing an existing dataset [13,14] spanning a wide range of formulations and temperatures, we aim to perform as few as possible additional experiments to discover formulations with maximum conductivity for a wide range of temperatures. This is performed in a workflow called one-shot active learning.…”

“…The existing dataset of lithium hexafluorophosphate (LiPF 6 ) in ethylene carbonate (EC), ethyl methyl carbonate (EMC) and propylene carbonate (PC) was totaling 80 electrolyte formulations with measured conductivities at À 30 to 60 °C. [13,14] The suggestion of new formulations was fully exploitative, [20] i. e., requested formulations were selected solely based on their predicted conductivity at a respective temperature with complete neglect of model uncertainty. Active learning in fully exploitative mode has been shown to significantly increase the so-called "enhancement factor" by Rohr et al [20] The enhancement factor describes the increase in probability of finding an optimum given a fixed budget of experiments.…”

Learning the laws of lithium-ion transport in electrolytes using symbolic regression

“…29 In essence, SR simultaneously learns both adjustable parameters and the functional form relating electrolyte conductivity with its formulation. We make use of a HTE setup 30 to collect thousands of conductivity measurements of LiPF 6 -based electrolytes with ethylene carbonate (EC), propylene carbonate (PC) and ethyl methyl carbonate (EMC) as solvents at different temperatures. With a simple SR approach, we train multiple candidate expressions and show that a particular expression emerges as a clear candidate, complying with strict and often competing criteria of accuracy, simplicity and consistency.…”