WHAT THEY LEARNED: Malia Wenny ’17

The chemistry major and environmental science minor used machine learning to search for compounds that could most efficiently function in solar cells.

Artificial intelligence and solar energy are two “buzz phrases” in modern science and technology. In her senior thesis, “Probing the synthesis-structure relationship in organohalide perovskites,” Malia Wenny ’17 used one to further the other.

The chemistry major and environmental science minor worked with two advisors, Associate Professors of Chemistry Alex Norquist and Joshua Schrier, to use machine learning in the analysis of organohalide perovskites, a type of compound that can convert sunlight into electricity. Machine learning is a form of artificial intelligence in which computers are fed data, and can “learn” on their own to identify patterns or perform other tasks. Wenny’s project used this process to learn about the most efficient organohalide perovskites.

“Going in, I didn’t know much about these compounds or even how to describe and talk about them,” said Wenny, who also concentrated in scientific computing and who is now pursuing a Ph.D. in chemistry at Harvard. “I also was just beginning my work with machine learning and computer science. Now, I feel much more comfortable describing to people what I study, why I study it, and what progress I have made.”

What did you learn working on your thesis?
I’ve performed computational methods that allow us to study individual compounds in great detail, such as electronic structure calculations and non-covalent interactions index calculations. I’ve also “zoomed out” and used machine-learning models to extract information.

from many compounds at once. I’ve also learned about how computational work feeds into synthetic (wet lab) chemistry. We can use the information we get from machine learning to guide us as we try to make new compounds. This means I also got to do a fair amount of work in the lab to make new perovskites.

What are the implications for your thesis research?
The main implication of my thesis research is that we need more data to make meaningful progress in studying organohalide perovskites. Right now, Josh and Alex are working on using robots to speed up the synthesis of organohalide perovskites. In the long term, if we develop a good robotic system and combine it with a machine-learning model, we can search for the “best” organohalide perovskite and how to make it. Other researchers would then be able to take that compound, test its ability as a solar cell, and possibly change the field of photovoltaics!

“What They Learned” is a blog series exploring the thesis work of recent graduates.