(Invited) Challenges with Quantum Chemistry-Based Screening of Electrochemical Stability of Lithium Battery Electrolytes

Abstract

Development of novel electrolytes and additives for lithium batteries would greatly benefit from high throughput screening of key electrolyte properties such as reduction and oxidation stability and initial decomposition reactions. The computational framework for screening electrolyte first and second reduction and oxidation potentials and initial decomposition reactions is reported. An initial set of ~1000 modified carbonate, phosphate and sulfone-based molecules was screened using density functional theory calculations. The isolated molecules were surrounded by implicit solvent during screening. Analysis of this initial test set has provided insight into the importance of reorganization energy on the oxidation and reduction potentials and the relationship between the second and first reduction and oxidation potentials. However, screening of isolated molecules was found to often miss a number of important electrochemical reactions. For example, H-abstraction reaction during electrolyte oxidation significantly decreased the oxidation stability of carbonates, phosphates and sulfone-based solvents.1 Such reactions were found to be ubiquitous in a number of electrolyte classes. Interestingly, oxidation reactions concerted with the H-abstraction reactions tend to have a significantly higher reorganization energy than the electrolyte oxidation reactions that do not exhibit H-transfer suggesting that such reactions tend to have a significantly lower rate and could often be overlooked. Our results suggest that the minimal system for the examination of electrolyte oxidation stability should consist of a cluster containing two solvent molecules and an anion. While screening such clusters is computationally expensive it results in a much more realistic picture than the screening of isolated solvents and anions and leads to an improved agreement with the experimentally measured electrolyte electrochemical stabilities. Interestingly, the initial oxidation reactions of ethylene carbonate and dimethyl carbonate at the surface of the completely de-lithiated Ni0.5Mn1.5O4 high voltage cathode were found to be similar to reactions occurring in electrolyte clusters, highlighting the potential similarity between electrolyte oxidation on inert electrodes and active LiNi0.5Mn1.5O4 cathode. Analogously, the formation of lithium fluoride (LiF) or solvent decomposition during reduction of semifluorinated solvents and anions was found to shift the electrochemical potential by as much as 1.5-2 V. Thus, inclusion of lithium in the model electrolyte cluster and consideration of the low barrier reactions within the computational methodology is important and could decrease the electrochemical stability window of electrolytes by as much as 3.5 V compared to a procedure that focuses on the isolated compounds surrounded by implicit solvent. Quantum chemistry study of the concentrated LiFSI-based electrolyte reduction presents an interesting example, when examination of the low rate reduction reaction resulted in LiF and Li-F-Li formation and predicted reduction stability of DME-LiFSI electrolyte around 1.6-2.4 V vs. Li/Li+.2 Quantum chemistry predictions were found to be in good agreement with CV measurements for this electrolyte and revealed initial reactions leading to the formation of a stable passivation layer on electrodes.2