Unraveling the Humidity-Induced Phase Transition in CALF-20 via Machine Learning Potentials.

Area of Science:

• Materials Science
• Computational Chemistry
• Chemical Engineering

Background:

• CALF-20 metal-organic framework (MOF) exhibits high CO2 selectivity, but its humidity-induced phase transitions are mechanistically unclear.
• Understanding water and CO2 competition at the molecular level is crucial for MOF applications.

Purpose of the Study:

• To elucidate the molecular mechanism of humidity-induced phase transitions in CALF-20 using machine learning.
• To investigate the competitive adsorption of water and CO2 within the MOF structure.

Main Methods:

• Development of a machine learning potential (MLP) with first-principles accuracy.
• Molecular dynamics simulations to study water dynamics and phase transitions.
• Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) for experimental validation.

Main Results:

• Water coordination at Zn nodes forms hydrogen-bonded networks, driving the open-pore to closed-pore phase transition.
• CO2 preadsorption disrupts these water networks, inhibiting water cluster formation.
• MLP simulations accurately reproduced experimental adsorption isotherms and X-ray diffraction patterns.

Conclusions:

• Provides a molecular-level understanding of humidity-induced phase transitions in flexible MOFs.
• Demonstrates the disruptive effect of CO2 on water adsorption dynamics in CALF-20.
• Establishes a simulation framework for modeling guest-responsive behavior in porous materials.