(389cl) Artificial Neural Network model for capturing the effect of local atomic environment on diffusion processes in metals and alloys

The local chemical environment can greatly influence rate processes. In most systems, the types of local environments can be very large numbers, which parecludes the use of molecular dynamics-based approaches to study the effect of all possible environments. A systematic artificial neural network (ANN) modelling approach is used to capture this environment effect is introduced. We focus on surface diffusion in metals and their alloys, namely the 𝐴𝑔𝑥−𝐴𝑢𝑦−𝑃𝑡𝑧−𝑃𝑑1−𝑥−𝑦−𝑧 system. Here the rate constants can vary over several orders of magnitude depending on the environment. The ANN model developed can estimate the activation barrier of diffusion. Any composition of the alloy is allowed. The hopping adatom can be either Ag/Au/Pt or Pd. The ANN model is trained using activation barriers calculated for different atomic environments with binary alloys. First, local environment with 96 different sites including various adatom and substrate sites are explored for the creation of a database of activation barriers. The database is next analysed with help of decision trees to identify the top-17 most important sites. These sites form the input layer of the ANN model. Good predictions are obtained with the ANN model for the ternary and quaternary alloys. The ANN model can be useful for modelling morphological changes in nanomaterials involving metal alloys.

Research Interests

My primary research interest lies in developing advanced methodologies that integrate AI and machine learning with multi-scale simulations to address complex material simulation problems across diverse length and timescales. This unique approach allows for innovative solutions in predicting and understanding atomic-scale behaviors and processes in various materials.

Further, I wish to gain experience in fine-tuning Generative AI, LLMs, and deep learning frameworks (PyTorch, TensorFlow) for scientific applications.

My Ph.D. research leveraged machine learning algorithms, density functional theory (DFT), kinetic Monte Carlo (KMC), and molecular dynamics (MD) simulations to predict materials' properties at atomic and molecular scales. During my tenure at Applied Materials India, I further honed my skills in VASP-based simulations for thin-film deposition processes and machine learning-enhanced screening of material properties.

I was invited as speaker at SC'19, Denver, USA by Lenovo Data Center Group, for being a winner of the Lenovo AI Challenge, 2019. Presented my research at Supercomputing Conference (SC’19) , CHISA 2024.