The drastic reduction in thermal conductivity observed in bundled single-walled carbon nanotubes (SWCNTs) has eluded accurate theoretical prediction. We show that Bose-Einstein phonon statistics are essential to reproduce this effect. Using a machine-learning-driven quantum-statistical framework, combining a neuroevolution potential with anharmonic lattice dynamics and the Boltzmann transport equation (ALD-BTE) calculations, we quantitatively predict an 81% reduction for a 5 μm seven-SWCNT bundle, matching experiments. The reduction originates from two distinct mechanisms enabled by quantum statistics: (1) the breaking of rotational symmetry, which directly quenches high-conductivity, symmetry-sensitive phonon modes like the twist mode; and (2) the introduction of abundant intertube phonon modes, which expands the three-phonon scattering phase space and elevates scattering rates universally. These insights, inaccessible to classical simulations, resolve a long-standing experimental-theoretical discrepancy and provide a predictive framework for designing SWCNT-based thermal materials.
Tao et al. (Wed,) studied this question.