Experiment Specification, Capture and Laboratory Automation Technology (ESCALATE): a software pipeline for automated che
- PDF / 1,181,214 Bytes
- 14 Pages / 612 x 792 pts (letter) Page_size
- 60 Downloads / 184 Views
Artificial Intelligence Research Letter
Experiment Specification, Capture and Laboratory Automation Technology (ESCALATE): a software pipeline for automated chemical experimentation and data management Ian M. Pendleton and Gary Cattabriga, Department of Chemistry, Haverford College, 370 Lancaster Avenue, Haverford, Pennsylvania 19041, USA Zhi Li, Molecular Foundry, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, California 94720, USA Mansoor Ani Najeeb, Department of Chemistry, Haverford College, 370 Lancaster Avenue, Haverford, Pennsylvania 19041, USA Sorelle A. Friedler, Department of Computer Science, Haverford College, 370 Lancaster Avenue, Haverford, Pennsylvania 19041, USA Alexander J. Norquist , Department of Chemistry, Haverford College, 370 Lancaster Avenue, Haverford, Pennsylvania 19041, USA Emory M. Chan , Molecular Foundry, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, California 94720, USA Joshua Schrier , Department of Chemistry, Fordham University, 441 E. Fordham Road, The Bronx, New York, 10458, USA Address all correspondence to Joshua Schrier at [email protected] (Received 15 January 2019; accepted 22 May 2019)
Abstract Applying artificial intelligence to materials research requires abundant curated experimental data and the ability for algorithms to request new experiments. ESCALATE (Experiment Specification, Capture and Laboratory Automation Technology)—an ontological framework and opensource software package—solves this problem by providing an abstraction layer for human- and machine-readable experiment specification, comprehensive and extensible (meta-) data capture, and structured data reporting. ESCALATE simplifies the initial data collection process, and its reporting and experiment generation mechanisms simplify machine learning integration. An initial ESCALATE implementation for metal halide perovskite crystallization was used to perform 55 rounds of algorithmically-controlled experiment plans, capturing 4336 individual experiments.
Introduction Chemistry and materials science are entering a new data-driven age,[1–3] in which planning algorithms select experiments to be conducted by humans or performed autonomously using laboratory robotics.[4–6] Laboratory automation has been an ongoing endeavor for nearly a quarter century with seminal demonstrations of high-throughput materials research performed by Xiang et al. in 1995.[7] Subsequent research, predominantly in combinatorial chemistry, focused on the development of high-throughput techniques targeting new material syntheses[8] and methods of characterization[9] that have been the topic of several comprehensive reviews.[10–17] Recent advances in machine learning and artificial intelligence allow for extracting further physical insights latent within these results.[18,19] Important themes include comprehensive capture and analysis of “successful” and “failed” experiments,[20–23] adaptively modifying the experiment plans as data are collected,[4,24–29] machine-learned characterization of experimental o
Data Loading...
