Microsoft Research has unveiled two AI tools, MatterGen and MatterSim, that take a fresh approach to developing new materials. Instead of searching through existing options, these tools can create and validate entirely new materials from scratch.
MatterGen creates new materials using a specialized diffusion algorithm that builds three-dimensional molecular structures. Its companion tool, MatterSim, then tests these materials by simulating how they'd perform in real-world conditions. Together, they help scientists design materials with specific chemical, mechanical, and electronic properties - a significant step forward from traditional screening methods.
The system is already showing promise in real-world testing. Working with the Shenzhen Institute of Advanced Technology, researchers successfully created a new material called TaCr2O6 that MatterGen proposed. The actual properties of the synthesized material matched the AI's predictions about 80 percent of the time.
MatterSim takes the process further by testing these materials under extreme conditions, simulating temperatures from absolute zero to 5,000 Kelvin and pressures up to 10 million atmospheres. The tool combines quantum mechanics principles with machine learning to handle these complex calculations.
"MatterGen generates thousands of candidates with user-defined constraints to propose new materials that meet specific needs," says Tian Xie, Principal Research Manager at Microsoft Research. "This represents a paradigm shift in how materials are designed."
Open access and integration
Microsoft has released MatterGen's source code under the MIT license, along with training and fine-tuning data. Both tools are now part of Microsoft's Azure Quantum Elements platform, where they're helping companies develop new materials for batteries, magnets, and fuel cells.
The company published its research findings in Nature today. These tools are part of Microsoft's "AI for Science" initiative, which began two and a half years ago.
