Towards Autonomous Synchrotron Fourier Transform Infrared Microscopy

Résumé fourni par la source

Infrared (IR) absorption spectromicroscopy stands out as a non-destructive technique that delves into the chemical composition of samples down to the micron level [1]. This method relies on the principle that chemical bonds, characterized by dynamic dipole moments, resonate at specific frequencies within the IR spectrum (4,000-400 cm^-1). The stretching and bending motions of bonds in molecules, particularly those with varying electronegativities like O-H or C=O, result in distinct absorption patterns. These patterns act as unique identifiers for chemical groups, enabling the detection and analysis of chemical processes within a sample over time or under different conditions. Synchrotron Fourier Transform Infrared (SFTIR) spectromicroscopy enhances this process by using the intense photon output of a synchrotron source, offering significantly more detail than standard benchtop equipment. However, a full examination of a sample, especially over a large area with fine spatial resolution, is time-consuming, and requires over 24 hours to complete due to the probe-based nature of the measurement. To address this challenge, we have developed a method that significantly improves the efficiency of SFTIR spectromicroscopy. By integrating data collection with a Gaussian Process (GP)-based surrogate model, this approach provides a predictive framework that spans the entire sample area, including unmeasured regions [2,3]. This model, built on a multivariate normal distribution, allows for strategic selection of measurement locations that are most likely to reduce overall uncertainty. It also offers the ability to estimate the quality of the surrogate model itself, enabling informed decisions on when to conclude an experiment. Initial results have demonstrated the potential to accelerate the throughput of SFTIR spectromicroscopy by approximately 20 times, representing a substantial advancement in the field by maximizing the utility of available instrument time and facilitating the exploration of a wider range of samples.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.