Structure and transport properties of LiTFSI-based deep eutectic electrolytes from machine-learned interatomic...

Area of Science:

• Materials Science
• Electrochemistry
• Computational Chemistry

Background:

• Deep eutectic solvents (DES) are gaining traction as cost-effective alternatives to ionic liquids for electrochemical applications.
• Their unique properties make them suitable for use as liquid electrolytes.
• Understanding ion transport in DES is crucial for optimizing their performance.

Purpose of the Study:

• To investigate the structural and dynamic properties of deep eutectic electrolytes based on lithium bis(trifluoromethanesulfonyl)imide (LiTFSI).
• To compare the accuracy of local equivariant neural network potentials with classical force fields for DES simulations.
• To elucidate the mechanisms of lithium-ion transport in different LiTFSI-based DES.

Main Methods:

• Utilized molecular dynamics (MD) simulations with a local equivariant neural network interatomic potential model.
• Performed large-scale MD simulations with first-principles accuracy.
• Analyzed ion-ion interactions, Li+-amide interactions, cationic transport numbers, and ionic conductivity.

Main Results:

• Classical force fields inaccurately characterize ion-ion interactions in DES.
• Observed close contacts between lithium ions bridged by amide oxygen atoms.
• Identified distinct Li+ transport mechanisms (structural, vehicular, solvent-exchange) in LiTFSI:urea, LiTFSI:N-methylacetamide, and LiTFSI:acetamide systems.

Conclusions:

• Equivariant neural network potentials enable accurate, large-scale MD simulations of DES.
• Li+ transport mechanisms in DES are complex and depend on the specific amide co-former.
• Findings provide insights into designing advanced electrolytes for electrochemical devices.