[A photo of experimentation. Credit to Pixabay]

Researchers from the Pacific Northwest National Laboratory (PNNL) and Argonne National Laboratory are using artificial intelligence (AI) to discover new solvents for redox flow batteries, advancing energy storage technology.

The groundbreaking research represents one of the growing cases where experts are successfully applying AI to accelerate scientific discoveries.

The team’s focus is on identifying chemical solutions that can effectively store and release energy in these specialized batteries, which are crucial for improving energy efficiency.

Developing the most efficient battery requires testing numerous solvents for their ability to produce and store energy.

Without the help of AI, analyzing the solubility of these solvents would take hours of laboratory work.

To address this challenge, the team of experts employed AI technology to test multiple solvent combinations.

However, while AI is a powerful tool, it often needs additional feedback to be as accurate as possible.

To prevent errors, the team gathered supplementary data to fill in gaps in the AI algorithm.

Vijay Murugesan, a materials scientist from PNNL, said, “Our approach is incredibly efficient. We’re leveraging the speed of high-throughput and human intuition to better train AI. When we went to George Crabtree with the idea to use PNNL’s high-throughput capability for electrolyte discovery, he challenged us to think bigger and collaborate with the AI team. Through his inspiration, we learned that together we can produce impactful results faster by integrating AI models and robotic platforms.”

In a separate but related development, Zhichu Ren, a PhD student in MIT’s Department of Materials Science and Engineering, has taken AI applications in chemistry a step further.

Ren has developed an AI-powered lab assistant called CRESt that can perform tasks such as mixing chemicals, recording data, and suggesting ideas to researchers.

Workign with Ren’s mentor, Ju Li, the team applied CRESt to a project converting carbon dioxide to formate, a material that powers fuel cells to generate electricity.

Li’s team needed to find the most efficient process for powering the fuel cell by testing combinations of materials, a typically time-consuming process.

CRESt features an active learning program, continuously learning from the researchers’ experiments.

The program assists the team by suggesting promising combinations of chemicals, while also guiding the researchers when it detects uncertainty about specific data points or experiments.

Li explained, “We had to figure out what to mix and then choose the processing parameters, but by using CRESt, we can do the least number of tests possible to achieve optimal results.”

Despite AI’s current reliance on human assistance for research projects, experts predict that as the power of AI continues to grow, it will become increasingly capable of rapid analysis and idea generation.