PNNL

Abstract

This project aims to develop a ‘self-driving’ NMR spectrometer using an adaptive control software framework that was created for this purpose, NMRSmartControl, which updates experimental parameters as it increments experimental conditions (e.g. temperature, elapsed time) to ensure efficient and accurate collection of large-scale dynamical NMR datasets. A closed-loop automation cycle for collection of variable-temperature NMR dynamics measurements was created, with feed-forward of adjusted parameters from simple optimization experiments to set up a ‘production’ experiment at a given condition, which are then used to predict experimental acquisition parameters for the subsequent condition in the series. The platform is designed to facilitate future integration with the SciLink multi-agent AI workflow for theory-in-the-loop, with the aim to transition from adaptive control for a set experimental plan to fully autonomous experimentation based on a user-defined objective.