Guided diffusion for the discovery of new superconductors
9 affiliations
- 1Department of Physics, University of Florida, Gainesville, FL, USA
- 2Quantum Theory Project, University of Florida, Gainesville, FL, USA
- 3Department of Materials Science and Engineering, University of Florida, Gainesville, FL, USA
- 4Department of Aerospace Engineering and Mechanics, University of Minnesota, Minneapolis, MN, USA
- 5Department of Chemistry, University of Florida, Gainesville, FL, USA
- 6Center for Neural Science, New York University, New York, NY, USA
- 7Center for Soft Matter Research, Department of Physics, New York University, New York, NY, USA
- 8Courant Institute of Mathematical Sciences, New York University, New York, NY, USA
- 9Simons Center for Computational Physical Chemistry, New York University, New York, NY, USA
Abstract
The inverse design of materials with specific desired properties, such as high-temperature superconductivity, represents a formidable challenge in materials science due to the vastness of chemical and structural space. We present a guided diffusion framework to accelerate the discovery of novel superconductors. A DiffCSP foundation model is pretrained on the Alexandria Database and fine-tuned on 7183 superconductors with first-principles-derived labels. Employing classifier-free guidance, we sample 200,000 structures, which lead to 34,027 unique candidates. A multistage screening process that combines machine learning and density functional theory (DFT) calculations to assess stability and electronic properties, identifies 773 candidates with DFT-calculated T c > 5 K. Notably, our generative model demonstrates effective property-driven design. Our computational findings were validated against experimental synthesis and characterization performed as part of this work, which highlighted challenges in sparsely charted chemistries and disorder. This end-to-end workflow accelerates superconductor discovery while underscoring the challenge of predicting and synthesizing experimentally realizable materials.
Related
- GCMDiff: a condition-guided diffusion framework for pharmacophore-fragment drug discovery and target-specific molecular design
- A Scoping Review of Diffusion-Based De Novo Molecular Generation: Architectures, Evaluation, and Open Challenges for Computational Drug Discovery
- Machine-learning-guided discovery of selective CK2 inhibitors with reduced CDK off-target activity
- Green Extraction and Molecular Docking-Guided Discovery of Asiaticoside as a Multi-Target Modulator of the SIRT1/HMGB1/p53-p21 Axis in H2O2-Induced Senescent Endothelial Cells
- Bioinformatics-guided discovery and functional verification of fungal polyester hydrolases for polycaprolactone degradation
BibTeX
@article{prakash2026guided,
title = {{Guided diffusion for the discovery of new superconductors}},
author = {Pawan Prakash and Jason B. Gibson and Z. Li and Gabriele Di Gianluca and Juan Esquivel and Eric G. Fuemmeler and Benjamin Geisler and Jung Soo Kim and Adrian Roitberg and Ellad B. Tadmor and Mingjie Liu and Stefano Martiniani and Gregory R. Stewart and J. J. Hamlin and P. J. Hirschfeld and Richard G. Hennig},
journal = {npj Computational Materials},
year = {2026},
month = may,
doi = {10.1038/s41524-026-02117-7},
url = {https://doi.org/10.1038/s41524-026-02117-7},
note = {CC BY},
}
RIS
TY - JOUR
TI - Guided diffusion for the discovery of new superconductors
AU - Pawan Prakash
AU - Jason B. Gibson
AU - Z. Li
AU - Gabriele Di Gianluca
AU - Juan Esquivel
AU - Eric G. Fuemmeler
AU - Benjamin Geisler
AU - Jung Soo Kim
AU - Adrian Roitberg
AU - Ellad B. Tadmor
AU - Mingjie Liu
AU - Stefano Martiniani
AU - Gregory R. Stewart
AU - J. J. Hamlin
AU - P. J. Hirschfeld
AU - Richard G. Hennig
PY - 2026
DA - 2026/05/28
JO - npj Computational Materials
DO - 10.1038/s41524-026-02117-7
UR - https://doi.org/10.1038/s41524-026-02117-7
AB - The inverse design of materials with specific desired properties, such as high-temperature superconductivity, represents a formidable challenge in materials science due to the vastness of chemical and structural space. We present a guided diffusion framework to accelerate the discovery of novel superconductors. A DiffCSP foundation model is pretrained on the Alexandria Database and fine-tuned on 7183 superconductors with first-principles-derived labels. Employing classifier-free guidance, we sample 200,000 structures, which lead to 34,027 unique candidates. A multistage screening process that combines machine learning and density functional theory (DFT) calculations to assess stability and electronic properties, identifies 773 candidates with DFT-calculated T c > 5 K. Notably, our generative model demonstrates effective property-driven design. Our computational findings were validated against experimental synthesis and characterization performed as part of this work, which highlighted challenges in sparsely charted chemistries and disorder. This end-to-end workflow accelerates superconductor discovery while underscoring the challenge of predicting and synthesizing experimentally realizable materials.
SN - 2057-3960
ER -