A team at UC Berkeley has won federal funding to use artificial intelligence to speed up the discovery of better catalysts, the molecular engines behind plastics, fuels, and medicines, potentially transforming how the chemical industry reduces energy consumption. The project, led by chemist John F. Hartwig, is one of just 278 teams nationwide selected for the U.S. Department of Energy's Genesis Mission, a historic initiative designed to double scientific productivity through AI and advanced computing.

Why Is Catalyst Discovery Such a Bottleneck for Green Chemistry?

Chemical manufacturing is one of the most energy-intensive industries on the planet. Every day, factories use extreme heat, pressure, and energy to create and break down complex molecules that become the plastics, fuels, and industrial materials we rely on. The problem is that designing better catalysts, the molecular machines that make these reactions possible, has traditionally been a slow, trial-and-error process. Researchers test countless combinations in the lab, hoping to stumble upon a more efficient reaction. This inefficiency means wasted energy and higher costs across the entire supply chain.

Hartwig's team has already pioneered some of the most widely used catalytic processes, discovering cleaner chemical reactions that drastically reduce the extreme conditions needed for molecular synthesis. Now, through the Genesis Mission, they're preparing to automate the science behind catalyst discovery itself.

How Does the AI Framework Solve Chemistry's "Small Data" Problem?

The breakthrough lies in solving a challenge that has plagued AI in chemistry: most machine learning models fail when training data is scarce. In chemistry, experimental data is expensive and time-consuming to generate, so researchers often have limited datasets to work with. Hartwig's team is combining two pretrained AI models to overcome this obstacle.

The framework pairs a large language model, which understands chemical knowledge from text, with a 3D quantum-physics model trained on over 500 million calculations. By merging text-based knowledge with 3D molecular structures, the AI can learn from small experimental datasets and explain why a catalyst works, allowing it to accurately predict cleaner chemical reactions. This transforms slow, trial-and-error chemistry into a fast, precise science.

Steps to Accelerate AI-Driven Catalyst Discovery

• Combine Multiple AI Models: Integrate large language models with quantum-physics models to leverage both textual chemical knowledge and 3D molecular structure understanding, enabling predictions even with limited experimental data.
• Reduce Lab Measurements: Use AI predictions to design custom catalysts for industrial reactions with fewer physical experiments, cutting the time and cost of the discovery process.
• Enable Local Deployment: Conduct the research in partnership with multiple institutions, like Lawrence Berkeley National Laboratory and Cornell University, to translate AI predictions into real-world energy savings across different industrial settings.

The project is being conducted in partnership with Lawrence Berkeley National Laboratory and Cornell University, bringing together multi-institutional expertise to translate AI predictions into tangible energy savings in real industrial settings.

"Addressing the world's most urgent energy challenges requires rethinking how we synthesize and recycle materials at the molecular level. By combining Professor Hartwig's work with advanced AI, we are turning traditional trial-and-error chemistry into predictive science that accelerates discoveries for efficient, industrial use of chemical feedstocks," stated Anne Baranger, Interim Dean of UC Berkeley's College of Chemistry.

Anne Baranger, Interim Dean, UC Berkeley College of Chemistry

What Does This Mean for Energy Efficiency in Manufacturing?

If successful, this approach could reshape how the chemical industry operates. By automating catalyst design, manufacturers could reduce the extreme heat and pressure required for reactions, directly lowering energy consumption and carbon emissions. The ripple effects would extend across industries that depend on chemical manufacturing, from pharmaceuticals to materials science.

The Genesis Mission itself represents a broader shift in how the U.S. government is approaching AI and scientific discovery. By investing in AI-driven research across multiple institutions, the Department of Energy is betting that artificial intelligence can accelerate solutions to some of humanity's most pressing challenges, including energy efficiency and sustainability.

Hartwig's team joins a cohort of researchers tackling similar problems across chemistry, materials science, and other fields. The selection of UC Berkeley's College of Chemistry as one of just 278 teams nationwide underscores the significance of this work and the potential for AI to fundamentally change how scientists approach discovery in energy-intensive industries.