We present an autonomous physical vapor deposition system that integrates hardware automation,in-situ optical spectroscopy,and Bayesian machine learning into a complete self-driving laboratory framework making decisio...We present an autonomous physical vapor deposition system that integrates hardware automation,in-situ optical spectroscopy,and Bayesian machine learning into a complete self-driving laboratory framework making decisions on the fly.Using silver thin films as a model material,our platform efficiently navigates acomplex parameter space through active learning.By introducing a thin physical layer denoted as calibration layer,the machine learning models adapt to sample-specific conditions on the fly and reliably predict the deposition conditions to achieve user-specified optical properties.Moreover,from the high-throughput experimental data,the algorithm systematically captures the complex parameter-property relationships that are challenging to deduce by conventional trial-anderror methods.This study demonstrates the potential of self-driving laboratories for both reducing human labor and gaining new understanding of materials,providing a streamlined approach to enable self-driving physical vapor deposition systems.展开更多
基金supported by University of Chicago Big Idea Generator Seed Grantwith additional hardware support by the National Science Foundation(NSF CNS-2019131)+2 种基金Collaboration between the Yang and Chen groups was also supported by the National Science Foundation(NSF ECCS-2427944)This work made use of the shared facilities at the University of Chicago Materials Research Science and Engineering Center,supported by National Science Foundation under award number DMR-2011854This work made use of the Pritzker Nanofabrication Facility at the Pritzker School of Molecular Engineering at the University of Chicago,which receives support from Soft and Hybrid Nanotechnology Experimental(SHyNE)Resource(NSF ECCS-2025633),a node of the National Science Foundation’s National Nanotechnology Coordinated Infrastructure.
文摘We present an autonomous physical vapor deposition system that integrates hardware automation,in-situ optical spectroscopy,and Bayesian machine learning into a complete self-driving laboratory framework making decisions on the fly.Using silver thin films as a model material,our platform efficiently navigates acomplex parameter space through active learning.By introducing a thin physical layer denoted as calibration layer,the machine learning models adapt to sample-specific conditions on the fly and reliably predict the deposition conditions to achieve user-specified optical properties.Moreover,from the high-throughput experimental data,the algorithm systematically captures the complex parameter-property relationships that are challenging to deduce by conventional trial-anderror methods.This study demonstrates the potential of self-driving laboratories for both reducing human labor and gaining new understanding of materials,providing a streamlined approach to enable self-driving physical vapor deposition systems.
