Validation of large language model artificial intelligence-integrated optical microscopy analysis: application to in situ/operando electrochemistry

초록

Analyzing chemical states and monitoring their transformations in in situ and operando environments are indispensable in chemistry and materials science. However, achieving high-resolution analysis under operando conditions entails substantial instrumentation costs and limited availability. Herein, we propose and validate a large language model (LLM)-based artificial intelligence (AI)-integrated optical microscopy (OM) analysis, enabling researchers to conduct analyses with ease and at minimal cost. Using LLM AI, we address the limitations of optical microscopy (OM) in performing quantitative compositional analysis while retaining its low cost. The performance of the LLM AI-integrated OM analysis is validated by comparing results with those obtained from surface-sensitive and bulk-sensitive spectroscopy under pristine/ex situ conditions and in situ/operando environments. The effective analysis depth of the LLM AI-integrated OM is estimated to lie between those of the surface- and bulk-sensitive techniques, comparable to typical visible-light penetration depths. More advanced LLM AI versions are expected to further enhance analytical consistency and reproducibility. Our work can aid researchers worldwide by offering a cost-effective and accessible analytical methodology that broadens participation in the materials and chemical sciences.