Research Projects

Battery electrolytes contain ions which produce strong local electric fields and induce polarization in the surrounding solvent. This induced polarization contributes significantly to the ion-solvent interactions and influences ion mobility, and it necessitates the use of either polarizable force-fields or ab initio molecular dynamics techniques. Experimentally, low frequency vibrational spectroscopy in the terahertz (THz) region provides powerful means to probe intermolecular interactions, structure and dynamics in these systems. Induced polarization again plays a key role in calculations of these spectral properties as the spectrum is directly related to polarization fluctuations. In the current project, we aim to calculate the THz absorption spectra, apart from ion solvation and ion transport properties, for a variety of battery electrolytes of different types. Accurate polarizable force fields are not readily available for most of the battery electrolyte systems. Such force fields need to be generated and validated. Ab initio molecular dynamics simulations provide accurate and alternative ways to calculate electrolyte properties, but such calculations are computationally very expensive. Ab initio simulation based THz spectral calculations are even more expensive due to requirement to calculate the dipole moments through Wannier centres for multiple parallel trajectories. Statistical convergences for small system sizes and short run lengths are other issues that need to be dealt with when ab initio molecular dynamics simulations are used for battery electrolytes. The current project aims to fill the above gaps by taking a two-fold approach: First, it is planned to develop explicit polarizable force fields within the framework of atomic multipole based AMOEBA model and validate the newly generated force fields through ab initio simulations of smaller systems. Once optimized, the explicit polarizable force fields will be used to perform long simulations of large electrolyte systems with multiple parallel trajectories generated for each system, and the structural, dynamical and THz spectral properties will be calculated. Secondly, ab initio simulations of smaller systems will be used to develop machine learning neural network potentials for selected battery electrolyte systems, including development of proper machine learning models for dipole calculations. Once developed, these neural network potentials will be used to simulate large electrolyte systems for sufficiently long times and the electrolyte properties will be investigated from the perspective of their ion-solvent structure, ion conductivity and vibrational spectroscopy in the THz frequency region. The proposed work will fill important gaps that exist at present in making accurate theoretical predictions of THz absorption spectra of battery electrolytes and connecting them to the underlying interactions, structure and intermolecular dynamical modes involving ions and polar solvents in those electrolyte systems of interest which include solvents of ethers, carbonates, acetals and nitriles apart from water, and salts of Li+, Na+ and also Mg2+and Zn2+ metal ions.