Abstract:Billions of organic molecules are known, but only a tiny fraction of the functional inorganic materials have been discovered, a particularly relevant problem to the community searching for new quantum materials. Recent advancements in machine-learning-based generative models, particularly diffusion models, show great promise for generating new, stable materials. However, integrating geometric patterns into materials generation remains a challenge. Here, we introduce Structural Constraint Integration in the GENerative model (SCIGEN). Our approach can modify any trained generative diffusion model by strategic masking of the denoised structure with a diffused constrained structure prior to each diffusion step to steer the generation toward constrained outputs. Furthermore, we mathematically prove that SCIGEN effectively performs conditional sampling from the original distribution, which is crucial for generating stable constrained materials. We generate eight million compounds using Archimedean lattices as prototype constraints, with over 10% surviving a multi-staged stability pre-screening. High-throughput density functional theory (DFT) on 26,000 survived compounds shows that over 50% passed structural optimization at the DFT level. Since the properties of quantum materials are closely related to geometric patterns, our results indicate that SCIGEN provides a general framework for generating quantum materials candidates.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to effectively generate new materials with specific geometric patterns and symmetries in quantum material design. Although currently, machine - learning - based generative models have made significant progress in the field of material design, most of the new materials generated by these models are still limited to the distribution in the database, and it is difficult to generate materials with specific constraints. Specifically, existing generative models face challenges in integrating geometric patterns (such as triangular lattices, honeycomb lattices, kagome lattices, etc.) into the material generation process. This limits scientists' ability to explore new quantum materials with unique properties. To overcome this challenge, the authors propose SCIGEN (Structural Constraint Integration in the GENerative model), which is a method that can integrate structural constraints in any pre - trained generative diffusion model. SCIGEN guides the generation process towards an output that meets the constraint conditions by using the diffused constraint structure to mask the denoising structure in each diffusion step. This method can not only effectively perform conditional sampling from the original distribution, but also generate new materials that are stable and conform to specific geometric patterns without retraining or fine - tuning the base model. The paper demonstrates the effectiveness of SCIGEN by generating millions of materials and performing multi - stage stability pre - screening. Among them, more than 10% of the materials passed the preliminary screening, and high - throughput density functional theory (DFT) calculations further indicate that more than 50% of the materials passed structural optimization at the DFT level. These results show that SCIGEN provides a general framework for generating quantum material candidates with specific geometric patterns.

Structural Constraint Integration in Generative Model for Discovery of Quantum Material Candidates

Generative Design of Crystal Structures by Point Cloud Representations and Diffusion Model

High‐Throughput Discovery of Novel Cubic Crystal Materials Using Deep Generative Neural Networks

Scalable Diffusion for Materials Generation

Generative Organic Electronic Molecular Design Informed by Quantum Chemistry

Generative Design of inorganic compounds using deep diffusion language models

Physics guided deep learning for generative design of crystal materials with symmetry constraints

Exploration of crystal chemical space using text-guided generative artificial intelligence

Generative Hierarchical Materials Search

Unleashing the power of novel conditional generative approaches for new materials discovery

Spectroscopy-Guided Discovery of Three-Dimensional Structures of Disordered Materials with Diffusion Models

Establishing baselines for generative discovery of inorganic crystals

Generative Inverse Design of Crystal Structures via Diffusion Models with Transformers

Discovery of 2D Materials via Symmetry-Constrained Diffusion Model

MatterGen: a generative model for inorganic materials design

Generative Chemical Transformer: Neural Machine Learning of Molecular Geometric Structures from Chemical Language via Attention

A deep generative modeling architecture for designing lattice-constrained perovskite materials

Quantum Generative Models for Small Molecule Drug Discovery

Diffusion Generative Models for Designing Efficient Singlet Fission Dimers

Atomic structure generation from reconstructing structural fingerprints

Crystal Structure Generation Based On Material Properties