Molecular Dynamics Simulation Study of Diffusion Dynamics in [EMIm]⁺Cl⁻+AlCl₃ Ionic Liquid Based on Deep Learning Potential

The [EMIm]⁺Cl⁻+AlCl₃ ion liquid is a promising prototype electrolyte for aluminumion batteries (AIBs). Its ionic transport behavior involves multiple mobile species (Al³⁺、 AlCl₃ 、 [AlCl₄]⁻ and [Al₂Cl₇]⁻), with experimentally unresolved ion migration mechanisms and conversion reactions among these species. This complexity results in heterogeneous ion migration mechanisms and sluggish diffusion kinetics, which cannot be accurately and reliably captured by the conventional first-principles molecular dynamics (FPMD) simulations within the very limited time duration (tens of ps) and relatively small modelling structure (less than 103 atoms). Otherwise, classic molecular dynamics simulations based on various force fields are also scarce for studying and predicting the atomic structure evolution and ion diffusion dynamics of the complex electrolyte system such as ion liquids. In this work, we successfully developed deep neural network interatomic potential (DP-potential) through machine learning techniques, combining first-principles accuracy with classical molecular dynamics efficiency, to systematically investigate various chemical and physical properties for [EMIm]⁺Cl⁻+AlCl₃ ion-liquid at finite temperature. Training and validation of DP potential for [EMIm]⁺Cl⁻+AlCl₃ ion liquid are implemented with a two-stage protocol, including the primary training stage and the refining stage. Before initiating the two training stages, a series of first-principles molecular dynamics (FPMD) simulations are performed for [EMIm]⁺Cl⁻+AlCl₃ ion liquids with different molar ratios (1.0, 1.3, 1.5, 1.7 and 2.0) and equilibrium densities (1.09 g/cm³ ~ 1.56 g/cm³) at finite temperatures (300 K and 400 K), resulting in a highly diverse training datasets spanning a board range of chemical composition and density during the primary training stage for DP potential. Then, the trained DP-potential is employed to conduct long-timescale classic molecular dynamics simulations using LAMMPS program for the [EMIm]⁺Cl⁻+AlCl₃ ion liquids to produce the atomic configurations that either show significant errors in the calculated atomic forces and total energies or exhibit the unusual atomic evolution before crashing. Those highly extrapolated atomic configurations are merged with the initial training datasets to reoptimize the DP potential in the second refining stage. Through this two-stage training approach, we successfully constructed a deep learning neural network interatomic potential with high accuracy, achieving energy prediction errors of 5×10⁻⁴ eV/atom and force prediction errors of 5×10⁻² eV/Å. The reliability of the finally obtained machine learning potential is further validated through a systematic comparison of radial distribution functions (RDF) for some representative atomic pairs such as C-N, C-H, Al-Cl and Cl-H, obtained from both DP-MD and FPMD, demonstrating excellent consistency for the results between two methods. The DP-MD simulations are systematically carried out to investigate vibrational spectrum and Al³⁺ diffusion dynamics as well as possible conversion reactions among molecular or ionic species (Al³⁺、AlCl₃、[AlCl₄]⁻ and [Al₂Cl₇]⁻) in [EMIm]⁺Cl⁻+AlCl₃ ion liquids within 104 atoms at finite temperature. From the calculated vibrational density of states (VDOS), it is revealed that the VDOS of [EMIm]⁺Cl⁻+AlCl₃ ion liquids could be approximated as the simple superposition of vibrational spectra of individual species ([EMIm]⁺, [AlCl₄]⁻, and [Al₂Cl₇]⁻) with the dominance of H related vibrational modes above 90 THz and the Al-Cl modes below 20 THz. At 300 K, DP-MD predicts that Al³⁺ diffusion coefficients remain consistently around 4×10⁻⁷ cm²/s at 300 K regardless of the chemical compositions, and the estimated diffusion activation energy of ~0.20 eV closely matching experimental measurements (0.15 eV). In addition, the calculated ionic conductivity of 27.37 mS/cm for [EMIm]⁺Cl⁻+AlCl₃ at room temperature, and which is only 18.2% deviated from the experimental value (23.15 mS/cm). Notably, two distinct Al³⁺ diffusion mechanisms are identified in [EMIm]⁺Cl⁻+AlCl₃ ion liquid: (1) direct migration processes conducted solely by molecular species including [AlCl₄]⁻ and [Al₂Cl₇]⁻; (2) the migration of the neutral AlCl₃ molecule mediated with two neighboring [AlCl₄]⁻ anions through the conversion reaction between [Al₂Cl₇]⁻ and AlCl₃+[AlCl₄]⁻ moieties. Furthermore, first-principles calculations on the probable dissociation pathways of [Al₂Cl₇]⁻ revealed from DP-MD predict a reaction energy barrier height of 0.49 eV for the AlCl₃ transferring between two [AlCl₄]⁻ anions with an increased reaction probability from 0.00047 events/(ps·Al³⁺) at 1:1.3 molar ratio to 0.00347 events/(ps·Al³⁺) at 1:1.75 molar. Overall, we have successfully proposed a highly efficient and reliable workflow to train and validate the deep neural network interatomic potential for complex electrolyte system such as [EMIm]⁺Cl⁻+AlCl₃ ion liquids, thus providing a more comprehensive investigation of Al³⁺ transport mechanisms in ionic liquid electrolytes for aluminum-ion batteries. In conclusion, this work could further advance the application of machine learning-based potentials in simulating electrolyte systems characterized by complex molecular architectures and sluggish diffusion dynamics.