Mechanistic Insights into Water Filling Effects on Thermal Transport of Carbon Nanotubes from Machine Learning Molecular Dynamics

Zhiqiang Li,Jian Wang,Haoyu Dong,Yanguang Zhou,Linhua Liu,Jia-Yue Yang
DOI: https://doi.org/10.1016/j.ijheatmasstransfer.2024.126152
IF: 5.2
2024-01-01
International Journal of Heat and Mass Transfer
Abstract:Understanding the behaviors of filling water in narrow carbon nanotubes (CNTs) is critical for a wide range of emerging technologies, including energy conversion and water desalination. Further probing the vibrational coupling between CNT and water contributes to investigating the corresponding thermal properties. However, classical molecular dynamics (MD) fails to exactly characterize the complex CNT-Water interactions due to the inaccuracy of empirical potentials. To cope with this, we develop an accurate and highly efficient neuro-evolution potential (NEP) model for CNT-Water systems and perform extensive MD simulations to investigate the effect of filling water on CNT thermal conductivity. The trained NEP obtains a force accuracy of 69.43 meV/Å and an atomic energy precision of 0.785 meV/atom. Anticipatedly, the structural and vibrational characteristics of the NEP-driven MD simulations show an excellent agreement with ab initio MD simulations. We computationally prove that the thermal conductivities of CNT filled with water are reduced by 23 % and 28 % for different filling ratios, which lies in the transport blockage of the phonon at low frequency. We further find that water filling has less effect on the ultra-long mean free path (MFP). The maximum MFP of empty CNT is 42.94 μm, and the CNT filled with water can be as large as 39.29 μm. Consequently, the thermal conductivity of CNT filled with water is firmly size dependent, accompanied by a rise in thermal conductivity from 88.37 to 436.96 W/m∙K as the size goes from 107.0 Å to 2148.0 Å. Besides, the thermal conductivity presents a different temperature dependence before and after the quantum correction due to the overestimation of the heat capacity. This work sheds light on precise assessment and detailed insights into the thermal properties of CNT filling water, which would considerably affect the following areas including electronic devices cooling, wastewater treatment and adsorption-driven steam power.
What problem does this paper attempt to address?