ARTIFICIAL INTELLIGENCE AND DENSITY FUNCTIONAL THEORY DRIVEN DESIGN OF ADVANCED FUNCTIONAL MATERIALS FOR SUSTAINABLE ENERGY CONVERSION AND STORAGE

Abstract

The global transition to a low-carbon energy system depends on materials that can harvest, convert and store energy with high efficiency, long lifetime and low cost. Conventional discovery based on empirical trial and error is too slow to meet this demand, because the chemical and structural search space of inorganic, hybrid and molecular materials is astronomically large. Density functional theory (DFT) has become the standard quantum-mechanical tool for predicting the stability, electronic structure, ionic transport and surface reactivity of candidate materials, yet its computational cost limits the number of systems that can be examined. Artificial intelligence (AI), particularly machine learning (ML) with graph neural networks, machine-learned interatomic potentials and generative models, can learn from DFT data and extend its reach by several orders of magnitude. This article reviews how the combination of DFT and AI is reshaping the design of functional materials for batteries, solid-state electrolytes, perovskite photovoltaics, photocatalysts and electrocatalysts. We summarize the theoretical foundations and known limitations of DFT, the main families of ML models and databases, and an integrated closed-loop workflow in which DFT, ML, active learning and experiment reinforce one another. Representative case studies are analyzed, and open challenges, namely functional accuracy, data quality, out-of-distribution generalization, synthesizability, interpretability and experimental validation, are discussed together with future directions.