Abstract: Adsorption represents a fundamental physicochemical process, and adsorption experiments are being increasingly integrated into undergraduate laboratory curricula. However, conventional teaching approaches often confine students to passively executing predetermined experimental procedures, limiting their comprehension of critical aspects such as adsorbent selection and microscopic adsorption mechanisms. To address these limitations, we propose an AI-enhanced digital experimental framework that combines generative AI models with molecular dynamics (MD) simulations. This innovative approach enables students to predict adsorption performance pre-experiment and engage in interactive pre- and post-experimental discussions through human-computer dialogue. The experimental workflow begins with students utilizing ChatGPT as an intelligent assistant for literature review, data organization, adsorbent screening, and personalized preparation via interactive dialogue. Subsequently, MD simulations are employed to predict adsorbent performance and investigate adsorption mechanisms. Experimental validation then follows to examine adsorption kinetics and thermodynamics in comparison with simulation predictions. Finally, students can engage in reflective discussions with ChatGPT regarding experimental observations and conceptual questions. This comprehensive digital teaching methodology encompasses interactive preparation, numerical simulation of adsorption processes, experimental verification, and AI-assisted reflection, effectively transforming students from passive executors into active learners and researchers. The approach presents a novel paradigm for modernizing and digitizing traditional experimental pedagogy.

Key words: Adsorption experiment, Digital teaching, Artificial intelligence (AI), Molecular dynamics (MD) simulations, ChatGPT