Active Learning Molecular Dynamics-Assisted Insights into Ultralow Thermal Conductivity of Two-Dimensional Covalent Organic Frameworks
International Journal of Heat and Mass Transfer/International journal of heat and mass transfer(2024)
摘要
Two-dimensional covalent organic frameworks (2D COFs) are novel materials with high porosities and large surface areas that are highly sought after for separation technologies and energy storage. The exothermic issues of gas adsorption and storage processes need to be emphasized, which requires an in-depth understanding of heat transfer mechanisms. Limited by experimental conditions, imprecision of empirical potentials and large computational cost, the thermal transport mechanism inside 2D COFs still remains elusive. To this end, a neuro-evolution potential (NEP) was developed to investigate the thermal properties of representative 2D COFs, e.g., COF-5. The trained NEP achieves a total energy accuracy of ∼0.785 meV/atom and a force accuracy of ∼69.43 meV/Å, respectively. As expected, the NEP-driven MD simulations display remarkable consistency with ab initio MD (AIMD) simulations in terms of structural and vibrational properties. Based on homogeneous non-equilibrium molecular dynamics (HNEMD) and equilibrium molecular dynamics (EMD) simulations, the in-plane and cross-plane thermal conductivity of monolayer COF-5 is estimated to be 1.12 ± 0.08 W/m·K and 0.50 ± 0.04 W/m·K, respectively. The spectral decomposition method further quantifies the contribution of different phonon modes to thermal conductivity, highlighting the important contribution of low-frequency (below 5 THz) phonons to thermal conductivity. It finds that the ultralow thermal conductivity origins from energy localization on the benzene ring, vibration mismatch and the short mean free path. This work provides fundamental workflow for evaluating the thermal properties of 2D COFs, whose findings contribute to the thermal design of related devices, e.g., off-gas treatment, chemical catalysis, and solar seawater desalination.
更多查看译文
关键词
Active learning,Neuro-evolution potential,Thermal conductivity,Two-dimensional covalent organic frameworks,Workflow
AI 理解论文
溯源树
样例
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要