Data-driven discovery of novel 2D materials by deep generative models

Abstract

Efficient algorithms to generate candidate crystal structures with good stability properties can play a key role in data-driven materials discovery. Here we show that a crystal diffusion variational autoencoder (CDVAE) is capable of generating two-dimensional (2D) materials of high chemical and structural diversity and formation energies mirroring the training structures. Specifically, we train the CDVAE on 2615 2D materials with energy above the convex hull $\Delta H_{\mathrm{hull}}<0.3$ eV/atom, and generate 5003 materials that we relax using density functional theory (DFT). We also generate 14192 new crystals by systematic element substitution of the training structures. We find that the generative model and lattice decoration approach are complementary and yield materials with similar stability properties but very different crystal structures and chemical compositions. In total we find 11630 predicted new 2D materials, where 8599 of these have $\Delta H_{\mathrm{hull}}<0.3$ eV/atom as the seed structures, while 2004 are within 50 meV of the convex hull and could potentially be synthesized. The relaxed atomic structures of all the materials are available in the open Computational 2D Materials Database (C2DB). Our work establishes the CDVAE as an efficient and reliable crystal generation machine, and significantly expands the space of 2D materials.

References in corpus (13)

• A real-space grid implementation of the Projector Augmented Wave method
• Two-dimensional materials from high-throughput computational exfoliation of experimentally known compounds
• AFLOW: An automatic framework for high-throughput materials discovery
• The Computational 2D Materials Database: High-Throughput Modeling and Discovery of Atomically Thin Crystals
• Data-driven materials science: status, challenges and perspectives
• Low Hole Effective Mass p-type Transparent Conducting Oxides: Identification and Design Principles
• AiiDA: Automated Interactive Infrastructure and Database for Computational Science
• Recent Progress of the Computational 2D Materials Database (C2DB)
• An invertible crystallographic representation for general inverse design of inorganic crystals with targeted properties
• Constrained crystals deep convolutional generative adversarial network for the inverse design of crystal structures
• High-throughput discovery of novel cubic crystal materials using deep generative neural networks
• Crystal Diffusion Variational Autoencoder for Periodic Material Generation
• High-throughput exploration of alloying as design strategy for thermoelectrics

Cited by in corpus (18)

• Inverse Design of Next-generation Superconductors Using Data-driven Deep Generative Models
• Developments and applications of the OPTIMADE API for materials discovery, design, and data exchange
• Artificial Intelligence in Materials Science and Engineering: Current Landscape, Key Challenges, and Future Trajectorie
• A Self-Improvable Polymer Discovery Framework Based on Conditional Generative Model
• Generative diffusion model for surface structure discovery
• Layer group classification of two-dimensional materials
• Magnetic order in the computational 2D materials database (C2DB) from high throughput spin spiral calculations
• Cascading symmetry constraint during machine learning-enabled structural search for sulfur induced Cu(111)-$(\sqrt{43}\times \sqrt{43})$ surface reconstruction
• Ab initio property characterisation of thousands of previously unknown 2D materials
• Orbital magnetization in two-dimensional materials from high-throughput computational screening
• Materials discovery acceleration by using condition generative methodology
• Self-Supervised Generative Models for Crystal Structures
• A Framework for Identifying Non-van der Waals 2D Materials
• Massive Discovery of Low-Dimensional Materials from Universal Computational Strategy
• Strategies to search for two-dimensional materials with long spin qubit coherence time
• Discovery of 2D Materials via Symmetry-Constrained Diffusion Model
• High-Throughput Prediction of Exfoliable Non-van der Waals Materials from a Universal Potential
• Gradient-based grand canonical optimization enabled by graph neural networks with fractional atomic existence