An Orbital-based Geometric Deep Learning Framework for Periodic Materials

Minhyuk Kang, Beom Seok Kang, Sergei Kliavinek, Maurice D. Hanisch, William A. Goddard III, Anima Anandkumar

Keywords: orbital-based learning, geometric deep learning, crystal graph neural networks, electronic structure, band gap prediction, delta learning, periodic materials

TL;DR: We introduce the first orbital-based geometric deep learning model for periodic materials and show that semi-empirical orbital features enable accurate and data-efficient band gap prediction across multiple levels of theory.

Abstract: Orbital-based deep learning has achieved notable success in molecular systems by integrating quantum mechanical information into geometric deep learning. In this study, we introduce OrbNet-Crystal, which extends the OrbNet-Equi framework from molecular systems to periodic crystal graphs. Our approach derives orbital features from lower-level semi-empirical quantum mechanical calculations, converts reciprocal-space orbital features into real-space orbital features via inverse Fourier transforms, and embeds them into SE(3)-equivariant graph neural networks for periodic materials. We evaluate OrbNet-Crystal on the Computational 2D Materials Database for band gap prediction at multiple electronic-structure levels of theory. Using semi-empirical orbital features, OrbNet-Crystal achieves strong accuracy across all targets, both for direct prediction and delta-learning across multiple levels of theory. Furthermore, we show that transfer learning substantially improves performance for data-scarce, higher-level theory targets. Overall, this work establishes the first orbital-based deep learning paradigm for crystalline systems and demonstrates the potential of orbital features for transferable, multi-fidelity learning in solid-state materials discovery.

Submission Track: Paper Track (Tiny Paper)

Submission Category: AI-Guided Design

Submission Number: 48