Accelerated discovery of perovskite materials guided by machine learning techniques

Surjeet Kumar, Subhajit Dutta, Russlan Jaafreh, Nirpendra Singh, Abhishek Sharan, Kotiba Hamad, Daeho Yoon

    Research output: Contribution to journalArticlepeer-review

    Abstract

    In this work, machine learning (ML) techniques are used to discover perovskite structures. ML models are built using the bandgap, as a proxy to represent the efficiency of the solar cell materials. The dataset containing well-known perovskite materials along with bandgaps listed in the open-source database, is used in the learning process. 10-fold cross-validation results show that the random forest (RF) algorithm has better performance as compared to other models. A prediction pool of ∼ 240 K compounds with 7 different prototype structures is created, many of these compounds have never been explored. The RF model is then used to predict the bandgap of new perovskite materials. By screening materials based on formability 6855 new candidates are obtained. For the validation of results, DFT calculation is performed and compared with ML-predicted properties for a new compound. © 2023 Elsevier B.V.
    Original languageAmerican English
    JournalMaterials Letters
    Issue number135311
    StatePublished - 2023

    Keywords

    • Bandgap
    • Feature generation
    • Machine learning
    • Magpie
    • Perovskite-structured material
    • Energy gap
    • Perovskite solar cells
    • Machine learning models
    • Machine learning techniques
    • Machine-learning
    • Perovskite structures
    • Solar cell materials
    • Structured materials
    • Work machines
    • Perovskite

    Fingerprint

    Dive into the research topics of 'Accelerated discovery of perovskite materials guided by machine learning techniques'. Together they form a unique fingerprint.

    Cite this