A FAIR Approach to Constructing a Computational Perovskite Database

In the last decade, ABX3 perovskites have revolutionized the fields of photovoltaics and photocatalysis due to their exceptional electronic, optical, and defect properties. Halide perovskites have shown immense potential in applications such as LEDs, lasers, and sensors for both UV and IR spectra. Given the massive compositional space spanned by ABX3 perovskites in terms of cation/anion choices and alloying, high-throughput density functional theory (DFT) computations have become a pivotal tool for exploring the perovskite chemical space, helping generate extensive property datasets that feed into advanced machine learning (ML) models for predictive analysis [1-3]. Despite the wealth of studies and data, there remains a significant gap in how these datasets are stored, shared, and utilized in perpetuity for research and education. Ensuring data is Findable, Accessible, Interoperable, and Reusable (FAIR) [4] is critical for advancing research and facilitating collaboration within the scientific community.<br/><br/>Our work introduces a comprehensive workflow that addresses this challenge by utilizing nanoHUB’s Sim2Ls framework [5,6] to systematically parse DFT calculations for perovskites and store them in a universally indexed database. This database is designed to be easily queried via a Python-based API, simplifying data access and manipulation for researchers. Moreover, by integrating compositional and crystal graph-based machine learning models, our workflow enables rapid and accurate predictions of fundamental material properties. This capability accelerates the discovery and development of new materials and guides exploration into uncharted chemical spaces. Our comprehensive online tool walks users through visualization and understanding of the ever-growing dataset and enables on-demand prediction and design of novel perovskites with multiple desired properties. The approach presented here is a significant step in organizing high-throughput computational data. By ensuring that this data is systematically stored and readily available, we promote a more collaborative research environment and support ongoing advancements in studying perovskite materials.<br/><br/><b>References</b><br/><br/>1. Mannodi-Kanakkithodi et al., "Data-Driven Design of Novel Halide Perovskite Alloys", Energy Environ. Sci. 15, 1930–1949 (2022). https://pubs.rsc.org/en/content/articlehtml/2022/ee/d1ee02971a<br/>2. J. Yang et al., “High-throughput computations and machine learning for halide perovskite discovery”, MRS Bulletin. (2022). https://doi.org/10.1557/s43577-022-00414-2<br/>3. Yang, Jiaqi, Panayotis Manganaris, and Arun Mannodi-Kanakkithodi. "Discovering novel halide perovskite alloys using multi-fidelity machine learning and genetic algorithm." <i>The Journal of Chemical Physics</i> 160.6 (2024). https://pubs.aip.org/aip/jcp/article/160/6/064114/3263011<br/>4. L. C. Brinson, L. M. Bartolo, B. Blaiszik, D. Elbert, I. Foster, A. Strachan, and P. W. Voorhees, “Community action on fair data will fuel a revolution in materials research,” MRS Bulletin, 49, 12-16 (2024).<br/>5. Rushik Desai, Alejandro Strachan, Arun Kumar Mannodi Kanakkithodi (2024), "Perovskite VASP-Data Extractor," https://nanohub.org/resources/perovsvasp. (DOI: 10.21981/M4G4-1C36).