Machine learning-assisted study on the thermal transport properties of two-dimensional M₃(C₆O₆)₂ (M = Fe, Co, Ni) materials

Two-dimensional metal-organic frameworks (MOF) are widely used in electronic devices and energy storage due to their large surface area, abundant active sites, and tunable sizes. A deeper understanding of the thermal transport properties of two-dimensional MOF materials is essential for these applications. In this work, we systematically studied the thermal transport properties of M₃(C₆O₆)₂(M = Fe, Co, Ni) by using a machine learning interatomic potential method combined with the phonon Boltzmann transport equation. The results show that the lattice thermal conductivities of Fe₃(C₆O₆)₂, Co₃(C₆O₆)₂, and Ni₃(C₆O₆)₂at room temperature are 4.0 W mK^-1, 5.5 W mK^-1, and 5.8 W mK^-1, respectively. The differences in thermal conductivity primarily arise from variations in phonon relaxation times, which can be elucidated by examining the three-phonon scattering phase space. Further analysis of bond strengths reveals that the strong bonding between Fe and O impedes phonon propagation through the oxygen atoms, resulting in lower lattice thermal conductivity. Our work provides a fundamental reference for understanding thermal transport in two-dimensional MOF.