Thermal transport in low-dimensional materials is of great fundamental and applied interest due to their unusual and widely tunable properties, as demonstrated by numerous studies of two- and one-dimensional (1D) crystals. In grim contrast, low-dimensional amorphous materials remain largely unexplored, despite their potentially unique characteristics and applications. Here, we theoretically explore thermal transport in quasi-1D single-walled amorphous carbon nanotubes (a-CNTs) by combining homogeneous nonequilibrium molecular dynamics and lattice dynamics, based on a custom-trained high-accuracy machine-learned potential. For both zigzag and armchair a-CNTs of different diameters, the quantum-corrected thermal conductivity (κ) consistently drops by about an order of magnitude as the degree of disorder increases from 0.01 to 0.08, which is quantified by the concentration of Stone-Wales defects. Spectral analysis reveals the predominant role of the low-frequency (<3 THz) vibrational modes, the mean free paths (MFPs) of which are long but finite, thus leading to converged κ as the tube length reaches several micrometers. Further, lattice dynamics calculations yield much larger MFPs for the longitudinal acoustic modes compared to the transverse acoustic modes, which is primarily attributed to the distinctly higher group velocity of the former. Moreover, the κ suppression appears to be dictated by the structural disorder while anharmonicity barely contributes. Our work expands the understanding of thermal transport in low-dimensional amorphous materials and may help promote their eventual application.

Close

PrevNext

View figure in article

• Thermal conductivity
• Thermal properties
• Amorphous materials
• Nanotubes
• Machine learning
• Molecular dynamics