Explainable Band Gap Prediction in ABX₃ Perovskites Using Composition-Based Features

DOI

10.2991/978-94-6239-750-7_54How to use a DOI?

Keywords

Band gap prediction; Deep learning; ElemNet; LightGBM; NAdam optimizer; Perovskite materials; Random Forest; Residual network; XGBoost

Abstract

Perovskite materials are widely used in technologies like LEDs, solar cells, and photodetectors, requires to accurate prediction of band gap for better performance, here convolutional experiments and simulations can take a long time and use a lot of resources. This work introduces an ElemNet inspired residual network optimised with NAdam (ElemNet-NAdam) that estimates band gaps from elemental composition, enabling fast screening without structural inputs. Using stable perovskite formulas, the model shown better performance against a 1D convolutional neural network and machine learning regressors like Random Forest, XGBoost, and LightGBM. ElemNet-NAdam had an overall performance with MSE 0.6926, MAE 0.5679, RMSE 0.8322, and R² 0.7580, showing accurate predictions and converged steadily. It also used SHAP to explain how elements were present in ABX compensation. These results show that composition only deep learning can find nonlinear connections between the makeup of elements and their electronic response. This will speed up the discovery and design of perovskite compounds. The framework easily works with interpretability to help chemically guided optimisation of A, B, and X site choices.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Gayatri Mantena
AU  - Kritesh Kumar Gupta
PY  - 2026
DA  - 2026/08/31
TI  - Explainable Band Gap Prediction in ABX₃ Perovskites Using Composition-Based Features
BT  - Proceedings of the International Conference on Advanced Design, Manufacturing, and Sustainable Energy Systems (ICADMSES 2026)
PB  - Atlantis Press
SP  - 752
EP  - 772
SN  - 2589-4943
UR  - https://doi.org/10.2991/978-94-6239-750-7_54
DO  - 10.2991/978-94-6239-750-7_54
ID  - Mantena2026
ER  -