DiffSyn: a generative diffusion approach to materials synthesis planning

doi:10.1038/s43588-025-00949-9

DiffSyn: a generative diffusion approach to materials synthesis planning

2026-02-02

0

PRE

AI

Rafael Gómez‐Bombarelli

E

Elsa Olivetti *

DOI：10.1038/s43588-025-00949-9

Original

Original request for help

Save

Abstract

En 中文

The synthesis of crystalline materials, such as zeolites, remains a notable challenge owing to a high-dimensional synthesis space, intricate structure–synthesis relationships and time-consuming experiments. Here, considering the ‘one-to-many’ relationship between structure and synthesis, we propose DiffSyn, a generative diffusion model trained on over 23,000 synthesis recipes that span 50 years of literature. DiffSyn generates probable synthesis routes conditioned on a desired zeolite structure and an organic template. DiffSyn a chieves state-of-the-art performance by capturing the multi-modal nature of structure–synthesis relationships. We apply Diffsny to differentiate among competing phases and generate optimal synthesis routes. As a proof of concept, we synthesize a UFI material using DiffSyn-generated synthesis routes. These routes, rationalized by density functional theory binding energies, resulted in the successful synthesis of a UFI material with a high Si/AlICP of 19.0, which is expected to improve thermal stability. A generative AI approach is developed for predicting materials synthesis recipes—a complex challenge in materials science. Using this approach, the authors experimentally synthesized a material using AI-generated synthesis recipes.

Keywords:

Computational methods

Porous materials

Computer Science

general

DiffSyn: a generative diffusion approach to materials synthesis planning

Abstract

Journal

Organization