Machine Learning with Bluesky for Automated Setup, Acquisition and Analysis of Synchrotron Experiments

The increased flux resulting from the Advanced Photon Source Upgrade (APS-U) allows high-quality data to be collected faster than ever. Since battery experiments often seek to probe the dynamic behavior of materials, they are well suited to take advantage of this new generation of synchrotron sources. However, beam-time productivity is increasingly limited by the human time required in between measurements, limiting the extent to which the increased X-ray flux is useful. In an effort to overcome these limitations, the Spectroscopy group at APS has developed several tools to automate as much of this process as possible.<br/><br/>Operating a spectroscopy beamline across multiple X-ray edges can be a tedious process that requires re-configuration of multiple components when moving between elements of interest. Machine learning can now be used to automate much of this work, paving the way for more sophisticated operando experiments, or enabling higher-throughput measurements at multiple X-ray edges.<br/><br/>During data collection, real-time decisions must be made that influence data quality, such as acquisition time and step size. The Bluesky orchestration framework provides tools to making these decisions on-the-fly. These tools have been integrated into our beamline control system, and can be used to set data acquisition parameters in order to reach a certain level of data quality.<br/><br/>With increasing data rates come additional burdens on the researcher to analyze the results. This is especially true for X-ray emission experiments, where multiple emission lines must be identified and extracted from area detector images. The Spectroscopy group has developed several tools to automate this analysis and compare the results to physical models in order to extract richer insight into electronic structure.<br/><br/>These tools aim to remove much of the burden of operating a spectroscopy beamline at APS, leaving researchers free to focus on novel scientific discoveries.