Thermal Conductivity Modulation in Carbon Nanotubes via Silicon Nanowire Encapsulation Investigated Using Neuroevolution Potential

ABSTRACT One‐dimensional van der Waals heterostructures based on single‐walled carbon nanotubes (SWCNT) have recently attracted increasing attention because of their structural flexibility and tunable thermal transport properties. Therefore, this study proposes a silicon nanowire (SiNW) encapsulation approach to construct a SiNW@CNT composite and systematically investigate its thermal transport behavior. Using an efficient neuroevolution potential model, we developed a high‐precision machine learning potential tailored for the SiNW@CNT structure. Using homogeneous non‐equilibrium molecular dynamics, equilibrium molecular dynamics, and heterogeneous nonequilibrium molecular dynamics simulations combined with spectral heat flux analysis, we found that SiNW encapsulation markedly reduces the thermal conductivity of SWCNT. The reduction in thermal conductivity becomes more pronounced as the SiNW filling ratio increases. At the maximum filling ratio, SiNW@CNT exhibits a thermal conductivity approximately 50% that of hollow SWCNTs. This reduction is attributed to SiNW encapsulation, which enhances phonon scattering within the SWCNT, shortens the phonon mean free path and lifetimes, and decreases overall thermal transport efficiency. In addition, as the system size increases, the thermal conductivity difference between SWCNT and SiNW@CNT widens, highlighting a clear size dependence and a transition from ballistic to diffusive transport. These findings provide a crucial theoretical basis for designing novel nanocomposites with tunable thermal conductivity.