The Autonomous Formulation Lab: Industrial Formulation Optimization Combining SAS & ML

by Dr Peter Beaucage, Dr Tyler Martin

Wednesday 18 Sept 2024, 11:30 → 13:00 Europe/Zurich

OHSA/E13

ZOOM ID: https://psich.zoom.us/j/61506069331

Abstract:
Societal need and regulations are driving reformulation of materials and products so that they reduce the pace of climate change and cause less harm to humanity and the environment. While scattering methods (SANS, SAXS, WAXS) are workhorse techniques for characterizing formulations, they are challenged to keep up with the resultant rapid pace of redesign. Consumer and industrial formulations often consist of dozens to hundreds of components with wildly varying and sometimes conflicting purposes and design requirements. With this large number of carefully balanced components, small perturbations to a formulation, for example replacing a petroleum-derived fragrance with a bio-derived one, can cause large changes in macroscopic properties and functionality. Approaches which can optimize material properties across a large number of composition parameters, generate large datasets, and reduce the time and cost of formulation discovery are needed to support this societally motivated product reformulation. Multimodal characterization and machine learning (ML) tools promise to greatly reduce the expense of exploring phase, stability, and property maps in highly multicomponent products. In this talk, we will describe the Autonomous Formulation Laboratory (AFL), a research program which seeks to enable the application of ML-driven autonomous techniques to the measurement of complex formulations through the development of instrumentation, datasets, standard challenges, and algorithms.

Bios:

Dr Tyler Martin is a staff member in the Materials Science and Engineering Division at NIST and a neutron beamline scientist for the nSoft consortium at the NIST Center for Neutron Research. Working closely with nSoft stakeholders, he leverages machine learning, molecular simulation, and liquid state theories to enhance neutron and x-ray scattering measurements of soft materials. Tyler co-leads the Autonomous Formulation Lab program, which combines machine learning with automated measurement with the goal of accelerating formulation discovery and optimization. Tyler’s Ph.D. at the University of Colorado focused on using simulation and theory to develop design rules for tailoring polymer nanocomposite morphology.

Dr Peter Beaucage is a staff scientist at the NIST Center for Neutron Research and co-leader of the Autonomous Formulation Laboratory project, which seeks to develop tools for the broad application of AI/ML in x-ray and neutron science in close collaboration with industry, government, and academic partners. His other interests include the development and application of resonant soft x-ray scattering (RSoXS), particularly for macromolecular solutions and biomolecules and the application of autonomous and high-throughput scattering to materials challenges in energy, water, and climate. Peter’s PhD at Cornell University focused on the development of quantum metamaterials, using block copolymers to produce 3D mesostructures in superconductors.