Oral: Machine Learning Analysis of Phase Stability of Cerium-Magnesium Based Alloys forHydrogen Storage Applications

The unique magnetic and electronic properties of cerium-based compounds combined

with their potential applications in high-performance magnets, catalysts etc., motivate

the need for accurate predictive models of material properties. In this work, we use

Machine Learning (ML) techniques to predict the structural stability of Magnesium-

based alloys for Hydrogen Storage Application, using structural characteristics,

including lattice symmetry, bond lengths, and coordination environments. We aim to

develop ML models based on Center-Environment (CE) features to predict the enthalpy

of formation in chemical doping studies. Our goal is to achieve high prediction accuracy

that could potentially be comparable to experimental results of chemical doping in

phase stabilization.

We aim to understand the pivotal influence of lattice symmetry, valency, and bonding

and coordination numbers in the enthalpy of formation, which could provide insights into

the underlying mechanisms governing the phase stability in cerium-based systems. By

employing machine learning to extend the predictive capabilities of first principles

methods, we hope to develop a robust framework for evaluating and designing new

cerium-based magnetic and hydrogen storage materials and their range of stability in

chemical doping studies. This approach seeks to advance the understanding of

chemical and physical properties of cerium-magnesium based alloys and offer a

pathway for the targeted development of novel materials for technological applications,

including gap magnets, hydrogen storage and other lanthanide-based systems.