 |
Center for Nonlinear Studies |
**NEW TIME** Laboratory data collection should be comprehensive but adaptable. An ideal system should provide a mechanism for specifying unambiguous machine-readable experiment that enable remote operation and replicability, presenting these as instructions to human operators and machines, capturing comprehensive data and metadata during experiment, and performing extraction/transformation/loading to facilitate machine learning. Yet, existing tools require significant development time that is incompatible with rapidly evolving scientific needs.In this talk, I will describe ESCALATE (Experiment Specification, Capture and Laboratory Automation Technology) an adaptable open source package for experiment description and data collection. As a specific example, I will describe its application to Robotic-Accelerated Perovskite Investigation and Discovery (RAPID). The first generation of RAPID uses inverse temperature crystallization (ITC) to grow halide perovskite single crystals for x-ray structure determination and bulk characterization using commercial liquid handling robots. All experiment plans for the syntheses are contributed remotely, by both human scientists and algorithms trained on the reaction data. More >40 compounds have been produced by the RAPID:ITC (compared to only 4 known ITC-grown halide perovskites prior to our work), including >10 new compounds or polymorphs. Incoming data collected by ESCALATE is used to automatically train machine learning models, evaluate model performance and feature influence, and quantify reproducibility. A live web dashboard communicates these insights to the scientist and management in visual form. I will conclude by describing case studies about new scientific insights extracted from the comprehensive RAPID dataset that have been enabled by this comprehensive dataset, and discuss ongoing deployments of ESCALATE to perovskite thin film and vapor diffusion synthesis experiments
Host: Ping Yang / Nick Lubbers