Electrochemical engineering of nanostructured materials
Motivation: Two-dimensional (2D) monolayers and heterostructures are attractive anode materials for next-generation rechargeable batteries, particularly for alkali metal-ion batteries (AMIBs), owing to their excellent Li+, Na+, and K+ adsorption, diffusion, and high storage capacity. The low-dimensionality of few-atom-thick 2D monolayers and heterostructures offers the opportunity to leverage atomistic tuning approaches such as strain, doping, functional group addition, vacancies, and defects to engineer their electrochemical properties for battery anode applications.
Research: We are investigating the electrochemical properties including metal-ion adsorption, diffusion, charge storage, and open circuit voltage, of novel 2D monolayers and heterostructures as electrodes for metal-ion batteries using first-principles Density Functional Theory (DFT) calculations and Molecular Dynamics (MD) simulations. This research involves multiscale hierarchical modelling methodologies covering all relevant physical domains, time, and length scales for metal-ion batteries.
Thermal engineering of nanostructured materials
Motivation: Predicting and tuning the thermal conductivity of 2D materials hold potential for optimizing the battery performance. By achieving high thermal conductivity in 2D materials, efficient heat dissipation can be facilitated, and energy losses could be minimized. Molecular Dynamics (MD) simulations is the commonly used approach for estimating the thermal conductivity of 2D materials. However, the accuracy of MD simulations is primarily dependent on the interatomic potentials (I.P) utilised. Therefore, there is a need to develop effective interatomic potentials for 2D materials and their heterostructures
Research: We predict the anisotropic in-plane and cross-plane thermal conductivities of 2D TMD monolayers and heterostructures through an innovative hierarchical approach combining Density Functional Theory (DFT), MD and machine learning (ML)-derived interatomic potentials. Leveraged by state-of-the-art ML algorithms, these hierarchical modelling methodologies integrate simulations from atomistic models of the nanostructured electrodes with DFT and MD into reduced-order thermo-electrochemical models of the macroscale battery architecture.
The following are the steps used to generate an interatomic potential using machine learning to estimate thermal conductivity:
- Input structures: The first step is to gather reference data for the 2D material or heterostructure of interest. This phase involves performing ab initio molecular dynamics (AIMD) simulations.
- Training dataset: The training dataset consists of information regarding the atomic positions, energies, and forces of the material for the purpose of training the ML model.
- ML model: The ML model is trained using the provided training dataset to accurately predict the energy, forces, and stresses of the material as a function of atomic positions.
- ML model tuning: The ML model is evaluated by computing an error function, which is the difference between the output value generated by the ML model and the actual value from DFT.
- Prediction: The predicted potential after proper tuning is then used to calculate the phonon properties of the material from which the thermal conductivity (K) is derived.
- Validation: To ensure the ML-predicted interatomic potential is accurate, the predicted K is compared with experimental data or results from other computational methods.
Experimental-Numerical Feedback System
ATOMS Laboratory is conducting multidisciplinary collaborative research with the Laboratory for Strategic Materials to develop an innovative modelling-experimental framework for material discovery combining thermo-electrochemical hierarchical modelling of batteries with state-of-the-art material synthesis processes and characterization techniques.
