Confluence of electronic structure calculations (DFT) and machine learning (ML) for lithium and sodium-ion batteries: a theoretical perspective

The world is rapidly transitioning towards clean energy solutions, and batteries are the key drivers of this transition. With increasing demand for large-scale energy storage systems, the need for cost-effective and sustainable battery storage systems is also increasing. Until now, lithium-ion batteries have completely dominated the commercial rechargeable battery storage space. Due to sodium’s greater affordability and abundance compared to lithium, sodium-ion batteries have drawn interest as a complementary technology to lithium-ion batteries in various applications, like grid storage devices. First-principles studies are often used today to effectively study the key properties of alkali-ion batteries that are difficult to access otherwise, such as the electronic structure effects, ion diffusivity, and quantitative comparison with experiments, to name a few. Understanding the electronic structure of battery materials can help researchers design more efficient and longer-lasting batteries. Recently, machine learning (ML) approaches have emerged as a very attractive tool both for prediction (forward) problems as well as design (or inverse) problems. Dramatic reductions in computational costs, coupled with the rapid development of ML tools in general and deep learning methods in particular, have kindled keen interest. This is so because they can supplement the traditional experimental, theoretical, and computational tools to significantly augment the quest for rapid development and deployment of new products. Furthermore, the integration of electronic structure calculations and ML benefits society by accelerating the development at considerably lower costs for more efficient and sustainable batteries, which can lead to longer-lasting portable devices, cleaner energy storage solutions, and lower environmental impact. This topical review article will focus on how density functional theory (DFT) and ML can facilitate Li-ion and Na-ion battery research via material discovery, rapid screening, and tuning of the electrode properties.