Generative Design of Crystalline Materials
Diffusion models that generate new stable crystal structures steered towards target chemistry and properties.
Tier Grades
Tier A: Frontier AI labs
feasible
Microsoft Research built MatterGen.
Tier B: Indonesian AI labs
stretch
Adapter fine-tuning on labelled property data is possible, but compute is not stated.
Tier C: Institution units such as ITB
stretch
Adapter fine-tuning is possible, but synthesis to confirm a design needs a materials lab.
Tier D: Organisations such as SEMAI (PoT, PoC, PoV)
not feasible
Generated crystals need synthesis and testing outside delivery scope.
Where this comes from
Tiers: a) frontier AI labs, b) Indonesian AI labs, c) institution units such as ITB, d) organisations such as SEMAI delivering proof of technology, concept and value. Each rationale cites the anchors below.
Anchors
- ComputeMatterGen: compute not stated in the abstract.
- DataMatterGen: fine-tuned with adapter modules on labelled property datasets; produces structures more than twice as likely to be novel and stable as prior models.
- Team size (proxy)Proxy only (author count of an exemplar paper, not a headcount): 21 authors on the MatterGen paper (arXiv 2312.03687).
- Time horizonMulti-year.
- Shelf-life signalsMatterGen reports a design with low supply-chain-risk chemistry, pointing to applied use.
- Pipeline roleTier A trains generative models; Tier C tests candidates; Tier B and Tier D have no observed role.
Where this comes from
Team size is a proxy (author counts of exemplar papers), not a headcount.
Who and What
Links
R&D TopicGenerative Design of Crystalline Materials
Who Runs Itnone recorded
Papers
- Crystal Diffusion Variational Autoencoder for Periodic Material GenerationTian Xie et al., ICLR 2022 (per arXiv comment), 2021 founding
- Fine-Tuned Language Models Generate Stable Inorganic Materials as TextNate Gruver et al., ICLR 2024 (per arXiv comment), 2024 state of the art
- FlowMM: Generating Materials with Riemannian Flow MatchingBenjamin Kurt Miller et al., ICML 2024, 2024 state of the art
- Crystal Structure Prediction by Joint Equivariant DiffusionRui Jiao et al., NeurIPS 2023 (per arXiv comment), 2023 state of the art
- MatterGen: a generative model for inorganic materials designClaudio Zeni et al., arXiv preprint, 2023 state of the art
Where this comes from
The papers to read first: where the direction began, its current best, and a survey when one exists.
Sources
- arxiv.org/abs/2312.03687 https://arxiv.org/abs/2312.03687
To cite this page: Project AURION, AI Use-Cases Research & Innovation, "Generative Design of Crystalline Materials", https://aurion.semai.id/topics/generative-crystal-materials-design/, records of October 2026.