Joseph Bennett, associate professor of chemistry and biochemistry, sees the integration of research and teaching as the heart of his work as a professor. Now he’s received a five-year National Science Foundation CAREER Award that will help him deepen the ways he weaves undergraduate students into the fabric of his cutting-edge materials research.
Bennett uses advanced computational methods to study two-dimensional materials—unusual, ultra-thin sheets of atoms that can exhibit strange (and useful) properties compared to typical 3D materials. In particular, he explores how scientists might design and manufacture new 2D materials with novel electric and magnetic properties that could enable better sensors, memory storage devices, and energy-efficient electronics.
Bennett’s current work with 2D materials builds on a 2025 study where his team identified dozens of promising candidate materials that their collaborators are now synthesizing in the lab. The CAREER project expands the search: Bennett’s group will computationally simulate materials that include different molecular building blocks, combined into different shapes, and containing different ions, in hopes of finding new materials that behave similarly to known materials but may also have unique, beneficial properties.
In the project’s final aim, “We go wild. We’re shooting for the home run,” Bennett says. They’ll be looking for electronic materials that remain stable in humid environments, an area where very little is known, but which could be a huge advance for biosensors.
Training students in the tools of modern chemistry
To design the new materials, Bennett’s group uses computers and databases containing the crystal structures of known materials alongside principles from inorganic chemistry and a powerful quantum mechanical modeling method used to predict the behavior of electrons in a material.
Scientific experiments like Bennett’s that happen inside a computer are common in modern chemistry, and their utility is likely to continue to rise as AI-fueled computing tools advance. However, if college curricula introduce computational chemistry tools to undergraduates at all, they tend to do so through black-box exercises where students plug parameters into a computer that spits out results without them understanding why.
Bennett is instead teaching his students using open-source tutorials he and his students developed, so learners can practice with real-world data from his lab. They experience the full workflow that Bennett and his student researchers undertake: setting up simulations, interpreting results, and spotting errors.
The approach gives students marketable skills in high-throughput computing, data analysis, and evaluation of results—including those produced by artificial intelligence tools. As a bonus, “If they like it in the class, they can join my research group,” Bennett says.
The CAREER award is also enabling the launch of a four-week summer training program open to undergraduates from any U.S. institution. Participants will receive data on the materials under study in Bennett’s lab, perform calculations themselves, and work with mentors to compile the findings into a manuscript draft. The experience is designed so students walk through the entire scientific process, including learning about peer review and publication.
“Keep them questioning”
Scientific funding agencies are increasingly looking for elements of computing, machine learning, and quantum-scale phenomena in the projects they support, Bennett explains. As a 2026 fellow of the Molecular Sciences Software Institute (MolSSI), which supports educators who make modern computational chemistry accessible to undergraduates, Bennett is already embedding those skills in courses. Combining that work with the new, dedicated summer program, will offer students early, hands-on experience at UMBC with the tools shaping the new era of computation-driven materials discovery.
Bennett is candid that the learning curve is steep—for students and for himself. He is refreshing his programming skills in the Python language and in Jupyter notebooks, an interface ideal for integrating coding, data visualization, and tutorial instruction. He’s also deliberately incorporating imperfect data sets so students encounter the same friction researchers face when code won’t compile, a file type is wrong, or any number of other challenges arise. The goal is not to give them cookie-cutter exercises that turn out perfectly every time, but to teach them to grapple with setbacks and develop resilient problem-solving skills.
“I really want to help cultivate a growth mindset in all these students,” Bennett says. “Just keep them questioning, keep them learning, and keep them doing.”
