Chemist Curtis Berlinguette leads a group that is creating a robot that can do an experiment just like a human being would in the lab. Not only can it make a material, it can then characterize it and make decisions on next steps to try based on its measurements. He’s a professor of chemistry at the University of British Columbia’s Stewart Blusson Quantum Matter Institute, and his goal is not only to accelerate development of new materials for clean technology, but also to free up scientists to think and be creative.
“I work on using robotics and automation, fused with machine learning, to try and find new materials for clean energy faster than we ever have before,” adds UBC graduate student Fraser Parlane.
“When I walk into the lab, I have all the tools available to me to discover the next material we need, say for solar cells, for some clean material application. But it’s a matter of finding the right combination and the right application of those materials that’s going to result in that champion next-generation material.”
Automating the search process lets Berlinguette and Parlane offload some of the routine tasks that normally take up a lot of time. There are millions upon millions of possible materials for chemists to synthesize and test. With urgent global demands for sustainable energy, the faster we can come up with new solutions, the better.
“Really, the problem ahead of us is: how fast can we search?” says Parlane. “How fast can we discover? How fast can we resolve the space to find that new material?”
Other researchers have automated chemical synthesis. The problem is that it still takes a lot of time to test millions of samples and process all the data that would be generated by testing every possible combination. That’s why the iterative process of testing as we go and making smart decisions on the next combinations to try helps explore possibilities in an organized way, instead of by brute force.
“The really fun thing about this robot is that it is actually learning on the fly,” explains Berlinguette. “It’s not building off any previous data. It’s not going in with any prior bias. It’s going out and making decisions based on the data that it’s collecting.”
And while machine learning and automation are taking care of routine tasks — the repetitive tasks that need to be done but that don’t add to our understanding — that leaves the chemists open to find patterns and meaning in the data. It gives them the time to be thoughtful and creative so they can design better experiments. These are all important advantages as society moves towards sustainable energy.