Energy-GNoME: A Living Database of Selected Materials for Energy Applications

Paolo De Angelis,Giovanni Trezza,Giulio Barletta,Pietro Asinari,Eliodoro Chiavazzo

2024-11-15

Abstract:Artificial Intelligence (AI) in materials science is driving significant advancements in the discovery of advanced materials for energy applications. The recent GNoME protocol identifies over 380,000 novel stable crystals. From this, we identify over 33,000 materials with potential as energy materials forming the Energy-GNoME database. Leveraging Machine Learning (ML) and Deep Learning (DL) tools, our protocol mitigates cross-domain data bias using feature spaces to identify potential candidates for thermoelectric materials, novel battery cathodes, and novel perovskites. Classifiers with both structural and compositional features identify domains of applicability, where we expect enhanced accuracy of the regressors. Such regressors are trained to predict key materials properties like, thermoelectric figure of merit (zT), band gap (Eg), and cathode voltage ($\Delta V_c$). This method significantly narrows the pool of potential candidates, serving as an efficient guide for experimental and computational chemistry investigations and accelerating the discovery of materials suited for electricity generation, energy storage and conversion.

Materials Science,Other Condensed Matter,Machine Learning

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to efficiently screen out new materials with potential value in materials science, especially for the field of energy applications. Specifically, the goal of the paper is to identify candidates that may be suitable for thermoelectric materials, new battery cathode materials, and new perovskite materials from a large number of candidate materials through machine learning (ML) and deep learning (DL) techniques. The solution of these problems is crucial for accelerating the development of energy conversion, storage, and generation technologies.

The paper mentions that although first - principles methods such as density functional theory (DFT) can accurately predict the physical and chemical properties of materials, traditional material screening strategies have limitations in practical applications due to the high - dimensional composition space and high computational costs. In addition, researchers usually rely on intuition and limited heuristic methods to identify potential materials, and these methods often lack extrapolation ability when dealing with unseen samples. Therefore, the paper proposes a new method that combines high - throughput computational methods and machine - learning techniques to overcome these challenges.

Specifically, the paper utilizes the Graph Networks for Materials Exploration (GNoME) database, which contains more than 380,000 newly discovered stable crystals. Through a machine - learning model, the authors of the paper carried out a preliminary screening of these materials, aiming to identify materials suitable for further numerical or experimental research, especially in the three fields of thermoelectric materials, perovskite materials, and battery cathode materials. This method can not only significantly narrow the range of potential candidate materials but also improve the reliability and accuracy of the screening process, thereby providing effective guidance for experimental and computational chemistry research and accelerating the discovery of materials suitable for power generation, energy storage, and conversion.