Professor helps lead global effort to accelerate materials discovery

University of Colorado Boulder Professor Hendrik Heinz is a co-principal investigator in FAIRmat II, a newly funded €33 million international consortium working to accelerate the discovery, design and development of new materials using artificial intelligence (AI) and data science. The consortium is coordinated by Humboldt-Universität zu Berlin (HU Berlin).

CU Boulder sat down with Heinz to talk about the significance of the consortium.

Why is FAIRmat II important?

FAIRmat II enables researchers to collaborate more effectively to develop materials and technologies that benefit industry and society.

How is this accomplished?

Although AI has the potential to transform materials research, its success depends on access to reliable, well-organized scientific data that can be shared across laboratories and research methods. FAIRmat developed the NOMAD ecosystem, a shared digital platform that allows scientists to collect, organize, share and analyze data from experiments and computer simulations. The platform already hosts more than 20 million publicly accessible data entries.

What role does AI play in this?

Our goal is to use AI to perform much of the data analysis, dramatically accelerating materials discovery. For example, screening 1,000 compounds could take 100 employees 10 years. With high-quality data and AI tools, researchers could potentially narrow down promising materials in a week or even a day.

Beyond speeding up the process, we also want to develop AI tools that can predict material properties even when scientific data is incomplete.

What specific expertise will your research group bring to the consortium?

We have developed computer models and tools that help us understand and predict how materials behave at the molecular level. Building on this work, we will use AI to improve how simulation data is processed and analyzed, helping researchers more efficiently predict material properties and design new materials.

We are also jointly developing data standards to a format that can be called, read and utilized by AI agents.

Furthermore, by identifying key datasets and research problems, CU Boulder researchers could collaborate with FAIRmat’s team of programmers to develop and test code that improves the infrastructure for materials design. This collaboration would provide access to computational infrastructure that CU Boulder would not have the resources to develop independently.

What will FAIRmat II’s connection to CU Boulder be?

We plan to send graduate students to Berlin to contribute to these efforts. Postdocs from my team may also be involved. There’s also interest for research experiences for undergraduate students to be part of this consortium. 

Could FAIRmat II benefit industry?

These tools could significantly reduce the time and laboratory work required to discover new materials. For example, a company developing longer-lasting batteries could use AI to identify 100 promising compounds within a month, then narrow them down to a few candidates for laboratory testing.

How might FAIRmat II impact our daily lives?

It can lead to materials discovery with many applications, such as perovskites for solar cells and lightweight materials used in space exploration. It could lead to materials for LEDs and semiconductors. Really, there is no limit. 

What do you see for the future?

There is and will be a continuous stream of new data deposited by university research teams. 

We will be able to compare computer simulations with laboratory experiments to better understand how accurately our models predict the behavior of materials. With more reliable data and better predictions, we will be able to accelerate the development of new materials and products. It could be new membranes for purifying water. It could be an EV battery that has a longer lifetime. Or flexible solar panels on jackets that could charge your cell phone while you're walking. The possibilities are endless.