Researchers at the University of Chicago Pritzker School of Molecular Engineering have developed a fully automated laboratory system that uses robotics and artificial intelligence to optimize the process of creating thin metal films. These films are essential components in electronics, optics, and quantum technologies. Traditionally, scientists spend months adjusting variables such as temperature and composition through trial-and-error methods.
The new “self-driving” lab system is designed to handle this work independently. According to Yuanlong Bill Zheng, first author of the study and current Ph.D. student at UChicago PME, “We wanted to free researchers from the tedious, repetitive labor of setting up and tweaking these experiments. Our system automates the entire loop—running experiments, measuring the results and then feeding those results back into a machine-learning model that guides the next attempt.”
Assistant Professor Shuolong Yang, senior author of the research published in npj Computational Materials, said: “I think in the future this kind of approach will be used more widely across the whole field of hard material synthesis and, eventually, complex quantum material synthesis. It points to a very intriguing futuristic mode of manufacturing.”
The system centers on physical vapor deposition (PVD), a method where materials like silver are vaporized and condensed onto surfaces as ultra-thin layers. PVD is sensitive to many factors—including environmental conditions—which makes consistent results difficult when done manually.
Zheng worked with UChicago College students Connor Blake and Layla Mravac to build a robotic setup capable of managing each stage of PVD autonomously. They collaborated with Assistant Professor Dr. Yuxin Chen and Ph.D. student Fengxue Zhang from UChicago’s Computer Science Department to create a machine learning algorithm that predicts optimal parameters for producing specific thin films.
“A researcher can tell the model what they want to come out at the end, and the machine learning model will guide the system through a sequence of experiments to achieve it,” said Zheng.
To address inconsistencies caused by subtle differences between experimental runs or trace gases in equipment chambers, each experiment begins with a calibration layer that helps account for unique conditions during every run.
“Researchers have long struggled with irreproducibility in physical vapor deposition, where tiny variations in hidden variables make it hard to get the same result twice,” explained Zheng. “Those inconsistencies end up in the training data as noise and can be detrimental for the machine learning model. Our high-throughput automated setup captured these variations in a systematic, quantitative way.”
For testing purposes, researchers tasked their system with growing silver films possessing specific optical properties—a challenge even though silver is well understood scientifically. The self-driving lab achieved target outcomes within an average 2.3 attempts per test case. Overall experimentation was completed within several dozen runs—a process that would typically require weeks if performed manually by human teams.
The cost for building this automated setup was under $100,000—significantly less than commercial efforts attempting similar automation for film synthesis.
With these results as proof-of-concept, researchers aim to adapt their method for more complex materials relevant to advanced electronics and quantum devices.
“This is just a prototype, but it shows how AI and robotics can transform not only how we make thin films, but how we approach materials discovery across the board,” said Yang.
The study received support from both university grants and National Science Foundation funding sources.