AI-Driven Expansion and Application of the Alexandria Database

2025-12-09

AI Summary

Efficient identification and expansion of thermodynamically stable inorganic compounds remains a key challenge in materials discovery. Method: We propose a multi-stage, AI-driven materials discovery paradigm integrating the Matra-Genoa generative model, Orb-v2 interatomic potential, and ALIGNN graph neural network to establish an end-to-end generation–screening–validation pipeline; we publicly release the sAlex25 dataset and a GRACE force-field fine-tuning protocol. Contributions/Results: Our approach achieves 99% accuracy in identifying near-stable compounds, reproduces experimentally observed structural disorder rates (37–43%)—a first for AI-generated data—and uncovers novel physical insights, including sublinear scaling of convex hull connectivity. It yields 1.3 million DFT-validated structures, including 74,000 newly predicted stable materials, expanding the ALEXANDRIA database to 5.8 million structures and 175,000 convex-hull compounds, thereby substantially improving benchmark performance.

Technology Category

• Search and Optimization: Metareasoning and Metaheuristics
• Machine Learning: Deep Neural Architectures and Foundation Models
• Natural Language Processing: Generation

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Web Mining and Content Analysis: Web data generation and simulation
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

We present a novel multi-stage workflow for computational materials discovery that achieves a 99% success rate in identifying compounds within 100 meV/atom of thermodynamic stability, with a threefold improvement over previous approaches. By combining the Matra-Genoa generative model, Orb-v2 universal machine learning interatomic potential, and ALIGNN graph neural network for energy prediction, we generated 119 million candidate structures and added 1.3 million DFT-validated compounds to the ALEXANDRIA database, including 74 thousand new stable materials. The expanded ALEXANDRIA database now contains 5.8 million structures with 175 thousand compounds on the convex hull. Predicted structural disorder rates (37-43%) match experimental databases, unlike other recent AI-generated datasets. Analysis reveals fundamental patterns in space group distributions, coordination environments, and phase stability networks, including sub-linear scaling of convex hull connectivity. We release the complete dataset, including sAlex25 with 14 million out-of-equilibrium structures containing forces and stresses for training universal force fields. We demonstrate that fine-tuning a GRACE model on this data improves benchmark accuracy. All data, models, and workflows are freely available under Creative Commons licenses.

Problem

• Develops a multi-stage AI workflow for high-accuracy materials discovery
• Expands a materials database with millions of DFT-validated stable compounds
• Analyzes structural and stability patterns to improve predictive models

Innovation

• Multi-stage AI workflow for materials discovery
• Combines generative models and neural networks
• Expands database with DFT-validated stable compounds