Scientists have created a technique enabling artificial intelligence models to generate materials with exotic quantum properties that could advance quantum computing applications.
The Massachusetts Institute of Technology team developed SCIGEN (Structural Constraint Integration in GENerative model), which guides popular AI materials models to create substances with specific geometric structures linked to quantum phenomena.
Existing generative materials models from companies including Google, Microsoft and Meta typically focus on producing stable materials rather than those with unique quantum characteristics. This limitation has hindered progress in fields like quantum computing, where researchers have identified only twelve candidates for quantum spin liquid materials despite a decade of intensive study.
“The models from these large companies generate materials optimized for stability,” says Mingda Li, MIT’s Class of 1947 Career Development Professor. “Our perspective is that’s not usually how materials science advances. We don’t need 10 million new materials to change the world, we just need one really good material.”
The research, published in Nature Materials, demonstrates how SCIGEN constrains diffusion models to follow specific geometric rules during material generation. The technique blocks outputs that fail to align with desired structural patterns, steering AI towards materials with properties suitable for quantum applications.
Testing revealed significant potential for accelerating materials discovery. The SCIGEN-equipped model generated over 10 million candidate materials featuring Archimedean lattices, geometric patterns associated with quantum phenomena. Following stability screening, one million materials remained viable for further analysis.
Advanced simulations conducted using Oak Ridge National Laboratory supercomputers examined 26,000 selected materials, revealing magnetic properties in 41% of the structures. The research team successfully synthesised two previously unknown compounds, TiPdBi and TiPbSb, with experimental results confirming the AI model’s predictions.
“We wanted to discover new materials that could have a huge potential impact by incorporating these structures that have been known to give rise to quantum properties,” says Okabe, the paper’s first author.
The breakthrough addresses a critical bottleneck in quantum computing research. Quantum spin liquids could enable stable, error-resistant qubits essential for quantum operations, but no confirmed quantum spin liquid materials currently exist.
“There’s a big search for quantum computer materials and topological superconductors, and these are all related to the geometric patterns of materials,” Xie says.
Future developments may incorporate additional design constraints including chemical and functional requirements alongside geometric specifications.
