Accelerating Materials Discovery: Learning a Universal Representation of Chemical Processes for Cross-Domain Property Prediction

TL;DR

This paper presents a universal process-graph representation and a multi-modal graph neural network that together enable cross-domain property prediction in materials discovery, significantly reducing experimental validation costs.

Contribution

The authors introduce a novel unified process-graph representation and a multi-modal GNN that generalizes across domains, enabling effective transfer learning for materials property prediction.

Findings

Model trained on 700,000 process graphs from 9,000 documents.

Pretrained model achieves strong performance on domain-specific datasets.

Universal representations transfer effectively with minimal additional data.

Abstract

Experimental validation of chemical processes is slow and costly, limiting exploration in materials discovery. Machine learning can prioritize promising candidates, but existing data in patents and literature is heterogeneous and difficult to use. We introduce a universal directed-tree process-graph representation that unifies unstructured text, molecular structures, and numeric measurements into a single machine-readable format. To learn from this structured data, we developed a multi-modal graph neural network with a property-conditioned attention mechanism. Trained on approximately 700,000 process graphs from nearly 9,000 diverse documents, our model learns semantically rich embeddings that generalize across domains. When fine-tuned on compact, domain-specific datasets, the pretrained model achieves strong performance, demonstrating that universal process representations learned at scale transfer effectively to specialized prediction tasks with minimal additional data.