Effects of High and Low Salt Concentrations in Electrolytes at Lithium–Metal Anode Surfaces Using DFT-ReaxFF Hybrid Molecular Dynamics Method

Abstract

Due to creating a passivated solid electrolyte interphase (SEI), high concentration (HC) electrolytes demonstrate peculiar physicochemical properties and outstanding electrochemical performance. However, the structures of such SEI remains far from clear. In this work, a hybrid ab initio and reactive molecular dynamics (HAIR) scheme is employed to investigate the concentration effect of SEI formation by simulating the reductive degradation reactions of lithium bis(fluorosulfonyl)imide (LiFSI) in 1,3 dioxalane (DOL) electrolytes at concentrations of 1 M, 4 M, and 10 M. The efficient HAIR scheme allows the simulations to reach 1 ns to predict electrolytes' deep products at different concentrations. The simulation findings show that the most critical distinction between HC and its low concentration (LC) analogue is that anion decomposition in HC is much more incomplete when only S−F breaking is observed. These insights are important for the future development of advanced electrolytes by rational design of electrolytes.

Frontiers in Theoretical Analysis of Solid Electrolyte Interphase Formation Mechanism

“…Various computational methods have been applied, such as quantum chemical (QC) calculations, [33][34][35][36][37][38][39] first-principles molecular dynamics (FPMD), [40][41][42][43][44][45][46][47][48][49] density functional tight binding (DFTB)-MD, [50,51] and reaction force field (ReaxFF)-MD. [52][53][54][55][56][57] As a result, primary electrochemical reactions have been identified, and the predicted products are consistent with those observed experimentally. However, there are still many unexplained phenomena regarding the SEI, such as the formation of a heterogeneous bilayer structure and morphological variations depending on the electrolyte composition.…”

“…[72][73][74] On the other hand, MD based on the reactive force field (ReaxFF), [75] which can qualitatively describe the forming/ breaking of chemical bonds, is useful for analyzing chemical reactions in rather large systems beyond the scope of FPMD. [52][53][54][55][56][57] For example, an improved method, eReaxFF, successfully simulated not only the ring-opening reaction of EC but also the radical polymerization of its reductive decomposition products. [53] Furthermore, the continuous decomposition process of electrolytes on the Li metal electrode surface has been investigated by ReaxFF-MD, [54][55][56][57] demonstrating the gas generation with separation of Li clusters is the possible origin of unfavorable dendrite growth.…”

“…[52][53][54][55][56][57] For example, an improved method, eReaxFF, successfully simulated not only the ring-opening reaction of EC but also the radical polymerization of its reductive decomposition products. [53] Furthermore, the continuous decomposition process of electrolytes on the Li metal electrode surface has been investigated by ReaxFF-MD, [54][55][56][57] demonstrating the gas generation with separation of Li clusters is the possible origin of unfavorable dendrite growth. [57] The disadvantages of both methods, however, are that the calculation accuracy strongly depends on empirical parameters and that the determination of the parameters for each target system consumes time.…”

“…[47,128] A key finding on the previous literature is that the SEI growth is triggered not only by the electrochemical reactions but also by the chemical reactions. [46][47][48][49][54][55][56][57] The RM method is even effective such longterm SEI growth processes but with more efforts to accurately incorporate elementary chemical reactions.…”

In situ formation of circular and branched oligomers in a localized high concentration electrolyte at the lithium-metal solid electrolyte interphase: a hybrid ab initio and reactive molecular dynamics study

“…This paper uses a hybrid method, a hybrid ab initio and reactive force eld dynamics (HAIR), to extend the simulation to nanoseconds. HAIR can extend the AIMD simulation to 10 to 100 times while maintaining DFT accuracy, [35][36][37][38] because reactive force eld (ReaxFF) 39 can replace the costly AIMD simulation in accelerating mass transfer when the accuracy or ReaxFF can be trained to level up with AIMD. The details of the simulation setup and HAIR method are as follows.…”