Rapid prediction of single-site adsorbate probability distributions in metal–organic frameworks using graph neural networks

Key Points

• This study aims to demonstrate the use of machine learning to generate adsorbate probability distributions for metal-organic frameworks.
• Utilized graph neural networks for rapid prediction of adsorbate probability distributions.
• Bypassed traditional atomistic simulations to enhance efficiency.
• Extracted adsorption binding sites reliably from the predicted distributions.
• Successfully generated adsorbate probability distributions quickly using machine learning techniques.
• Extracted binding sites from the distributions with high reliability.