Applications of density functional theory and machine learning in nanomaterials: A review

Abstract

The development and creation of nanomaterials carry enormous prospects in advancing technology in electronics, energy storage and medicine. The high degree of complexity and diversity in nanomaterials presents a real challenge in their theoretical and experimental studies. Density Functional Theory (DFT) is emerging as a powerful computational tool to model, understand, and predict material properties at a quantum mechanical level for nanomaterials. This review highlights the considerable use of DFT in elucidating the electronic, structural, and catalytic attributes of various nanomaterials. Also, this review considers developments between DFT and machine learning (ML)-based techniques that have paved the way for accelerated discoveries and design of novel nanomaterials. In fact, the ML algorithm has built models based on data from DFT, which predicts with high accuracy the properties of materials at reduced computational costs to expand vast areas of emerging chemistries. Major advances in this new hybrid approach include the development of ML models to predict band gaps, adsorption energies, and reaction mechanisms. The review discusses open topics regarding the future efforts to integrate DFT and ML focusing on model interpretability, data quality and broadened applicability to increasingly complex systems. The review concludes by discussing key advancements, such as those of machine learning interatomic potentials, graph-based models for structure property mapping and generative AI for materials design.