Heat conduction in low-dimensional systems - carbon nanotubes, one-dimensional nonlinear lattices, and nanostructures with interfaces - is governed by phonons whose mean free paths (MFPs) range from nanometers to microns or more. Simulations on such systems consistently find mode-resolved MFPs that grow roughly inversely with frequency at low frequency, in conflict with the inverse-square frequency dependence of standard perturbative (Umklapp) scattering theory, and the extreme low-frequency asymptotic form of the MFP spectrum remains unmeasured and underived. Because this low-frequency tail controls the length scaling of thermal conductivity, the field cannot yet say whether the thermal conductivity of a defect-free carbon nanotube converges, grows logarithmically, or diverges as a power law: current first-principles modeling, molecular dynamics studies, and mode-coupling analyses give mutually incompatible answers as late as 2026. The open problem is to construct a microscopic, mode-resolved account of phonon MFPs and the anharmonic scattering channels that limit them, sufficient to fix the low-frequency MFP asymptotics, explain the observed inverse-frequency scaling and its deviations, and derive the ballistic-to-diffusive crossover in a way that reconciles equilibrium and nonequilibrium simulations with each other and with experiment.

Background

A phonon - a quantized lattice vibration - transports heat in insulating solids, and the efficiency of this transport is set by the phonon mean free path (MFP), the average distance a phonon travels between successive scattering events. When a sample is much shorter than the relevant MFPs, heat flow is ballistic: every phonon crosses the sample without scattering, and the heat current saturates at a value fixed by the number of propagating modes (the Landauer picture, in which the heat current equals a transmission function times the number of available modes). When the sample is much longer than the MFPs, transport is diffusive and obeys Fourier's law with a thermal conductivity κ\kappa; the crossover between the two regimes is controlled by the frequency-resolved MFP spectrum Λ(ω)\Lambda(\omega). For most bulk crystals the MFP spectrum is now routinely computed from anharmonic lattice dynamics based on the Boltzmann transport equation (a kinetic equation for the phonon distribution whose collision term is treated perturbatively in the anharmonicity). Low-dimensional conductors strain this framework in two ways.

First, in quasi-one-dimensional systems such as carbon nanotubes, theory has long recognized a 'problem of long waves': acoustic phonons of very low frequency (long wavelength) have vanishing scattering rates, so their MFPs can exceed microns and the thermal conductivity grows with sample length. Mingo and Broido's Boltzmann-transport calculations found that the conductivity of a defect-free nanotube diverges with length unless scattering processes beyond lowest-order three-phonon terms are included, and that it keeps growing up to lengths of order 10 μm10\,\mu\mathrm{m} at room temperature. Non-equilibrium molecular dynamics (NEMD, direct simulation of a sample with hot and cold thermostats) by Sääskilahti and coworkers determined the frequency-dependent MFPs in isotopically pure nanotubes from the length-dependence of the spectral transmission, finding Λ(ω)∝ω−0.97\Lambda(\omega)\propto\omega^{-0.97} over a wide range, MFPs above 10 μm10\,\mu\mathrm{m} below 0.5 0.5\,THz, and mode-resolved relaxation times τ∝ω−1.2\tau\propto\omega^{-1.2} (longitudinal/twist modes) and ω−1.1\omega^{-1.1} (flexural modes) - none of which match the ω−2\omega^{-2} Umklapp scaling (Umklapp scattering is a three-phonon process in which the crystal momentum change is transferred to the lattice as a whole) that perturbation theory predicts; they also could not extract the true ω→0\omega\to 0 asymptotic form and left the convergence of κ\kappa open. A parallel exactly-solvable line uses one-dimensional nonlinear lattices - the Fermi-Pasta-Ulam β\beta (FPU-β\beta) chain, whose particles interact through quadratic-plus-quartic nearest-neighbor springs, and the ϕ4\phi^4 chain, whose particles sit in quartic on-site potentials - where heat conduction is known to be anomalous: for momentum-conserving chains such as FPU-β\beta, κ\kappa diverges with system size rather than saturating (reviewed by Lepri, Livi, and Politi). In these strongly nonlinear models the relevant carriers are renormalized phonons, quasiparticles whose frequencies are shifted from the harmonic values by the anharmonicity itself (the basis of effective phonon theory). Direct numerical measurement of their MFPs - by the 'triggering wave' and 'tuning fork' methods, in which a weak external drive excites a chosen wave number and the decay of the response yields the MFP at that wave number - shows the MFP is inversely proportional to the renormalized frequency, again contradicting the perturbative ω−2\omega^{-2} law, with an additional unexplained long-wavelength factor ∼k−0.70\sim k^{-0.70} in momentum-conserving lattices.

Second, at nanoscale interfaces between different materials, the thermal boundary conductance (the interface analog of thermal conductivity, describing heat transfer across the interface) involves the same mode-resolved MFP physics, and simple Matthiessen summations - combining scattering rates of different processes as if they were independent resistances, the non-spectral form of which errs by tens of percent in the intermediate regime - must be replaced by spectral, geometry-dependent MFPs to reproduce full Monte Carlo simulations of the Boltzmann equation, as shown by Davier and coworkers. Across the nanotube and nonlinear-lattice settings the same question recurs: which phonon modes and which anharmonic scattering channels actually limit the heat current, what is the true low-frequency asymptotic form of their MFPs, and how does that spectrum translate into the observed length scaling of the conductance - the ballistic-to-diffusive crossover. The question is presently unsettled and actively contested: state-of-the-art atomistic modeling claims a finite asymptotic nanotube conductivity reached only at macroscopic lengths, molecular dynamics work attributes apparent non-diffusive behavior to thermostat artifacts, and mode-coupling theory (a non-perturbative treatment of correlated scattering) supported by new simulations claims a power-law divergence.

Problem Statement

Construct a microscopic, mode-resolved account of phonon mean free paths (the average distance a phonon - a quantized lattice vibration - travels between successive scattering events, resolved by phonon mode and frequency) and of the anharmonic scattering channels that limit them in low-dimensional heat conductors - defect-free carbon nanotubes, one-dimensional nonlinear lattices of the FPU-β\beta type (chains whose nearest-neighbor interactions are quadratic-plus-quartic in the stretch) and the ϕ4\phi^4 type (chains whose particles sit in quartic on-site potentials) - that (i) fixes the low-frequency (long-wavelength) asymptotic form of the mode-resolved MFP Λ(ω)\Lambda(\omega) as ω→0\omega\to 0, (ii) identifies which scattering channels (e.g., flexural and optical modes, higher-order multi-phonon processes, mode-coupling correlations) produce the observed inverse-frequency scaling Λ∝1/ω\Lambda\propto 1/\omega rather than the perturbative ω−2\omega^{-2}, and explains the additional long-wavelength factors seen in momentum-conserving lattices (lattices whose dynamics conserve total momentum, such as the FPU-β\beta chain), and (iii) derives from that spectrum the length scaling of the thermal conductivity κ\kappa across the ballistic-to-diffusive crossover - settling whether κ\kappa of a pristine quasi-one-dimensional conductor converges to a finite limit, grows logarithmically, or diverges as a power law, at what length scales each regime is entered, and why current equilibrium simulations, nonequilibrium simulations, first-principles modeling, mode-coupling theory (a non-perturbative treatment of correlated scattering), and experiment disagree. The single unifying objective is a predictive mode-resolved theory of the MFP spectrum and its scattering channels, with the length-scaling of κ\kappa and the reconciliation of the conflicting computational and experimental claims as direct consequences.

The object class of the account is the FPU-β\beta chain, the ϕ4\phi^4 chain and a physical quasi-one-dimensional carbon nanotube, the three settings named in the Problem Statement, addressed jointly; parts (i)-(iii) are posed together as aspects of a single account, and the length scales in (iii) follow from the MFP spectrum Λ(ω)\Lambda(\omega). The statement fixes no single material, temperature or parameter regime. The nanoscale-interface material named in the objective and in the Background motivates the same mode-resolved MFP physics.

References

  1. K. Sääskilahti, J. Oksanen, S. Volz, J. Tulkki, 'Frequency-dependent phonon mean free path in carbon nanotubes from nonequilibrium molecular dynamics', Physical Review B 91, 115426 (2015). DOI: 10.1103/PhysRevB.91.115426, https://doi.org/10.1103/PhysRevB.91.115426 (arXiv:1411.2838, https://arxiv.org/abs/1411.2838)
  2. Nianbei Li, Junjie Liu, Changqin Wu, Baowen Li, 'Temperature and frequency dependent mean free paths of renormalized phonons in nonlinear lattices', New Journal of Physics 20, 023006 (2018). DOI: 10.1088/1367-2630/aaa4a8, https://doi.org/10.1088/1367-2630/aaa4a8 (arXiv:1510.03190, https://arxiv.org/abs/1510.03190)
  3. B. Davier, P. Dollfus, N. D. Le, S. Volz, J. Shiomi, J. Saint-Martin, 'Revisiting thermal conductivity and interface conductance at the nanoscale', International Journal of Heat and Mass Transfer 183, 122056 (2022). DOI: 10.1016/j.ijheatmasstransfer.2021.122056, https://doi.org/10.1016/j.ijheatmasstransfer.2021.122056 (arXiv:2006.14027, https://arxiv.org/abs/2006.14027)
  4. Natalio Mingo, David Broido, 'Length Dependence of Carbon Nanotube Thermal Conductivity and the "Problem of Long Waves"', Nano Letters 5, 1221-1225 (2005). DOI: 10.1021/nl050714d, https://doi.org/10.1021/nl050714d
  5. Giuseppe Barbalinardo, Zekun Chen, Haikuan Dong, Zheyong Fan, Davide Donadio, 'Ultrahigh Convergent Thermal Conductivity of Carbon Nanotubes from Comprehensive Atomistic Modeling', Physical Review Letters 127, 025902 (2021). DOI: 10.1103/PhysRevLett.127.025902, https://doi.org/10.1103/PhysRevLett.127.025902 (arXiv:2103.10633, https://arxiv.org/abs/2103.10633)
  6. Daniel Bruns, Alireza Nojeh, A. Srikantha Phani, Jörg Rottler, 'Comment on "Ultrahigh Convergent Thermal Conductivity of Carbon Nanotubes from Comprehensive Atomistic Modeling"', Physical Review Letters 128, 259601 (2022). DOI: 10.1103/PhysRevLett.128.259601, https://doi.org/10.1103/PhysRevLett.128.259601
  7. Giuseppe Barbalinardo, Zekun Chen, Haikuan Dong, Zheyong Fan, Davide Donadio, 'Barbalinardo et al. Reply', Physical Review Letters 128, 259602 (2022). DOI: 10.1103/PhysRevLett.128.259602, https://doi.org/10.1103/PhysRevLett.128.259602
  8. I-Ling Chang, Chia-Shing Li, Guan-Shiung Wang, Chi-Lin Wu, Chih-Wei Chang, 'Does equilibrium or nonequilibrium molecular dynamics correctly simulate thermal transport properties of carbon nanotubes?', Physical Review Materials 4, 036001 (2020). DOI: 10.1103/PhysRevMaterials.4.036001, https://doi.org/10.1103/PhysRevMaterials.4.036001
  9. Yuanyang Ren, Yang Wang, Kai Wu, David Cubero, 'Heat superdiffusion in carbon nanotubes', Applied Physics Letters 129, 052203 (2026). DOI: 10.1063/5.0338597, https://doi.org/10.1063/5.0338597
  10. Hu-Wei Jia, Wen-Jun Liu, Yue-Hong Wu, Kou-Han Ma, Lei Wang, Ning-Hua Tong, 'Thermal broadening of the phonon spectral function in classical lattice models: Projective truncation approximation', Physical Review B 111, 045153 (2025). DOI: 10.1103/PhysRevB.111.045153, https://doi.org/10.1103/PhysRevB.111.045153 (arXiv:2411.06384, https://arxiv.org/abs/2411.06384)
  11. Sha Liu, Junjie Liu, Peter Hänggi, Changqin Wu, Baowen Li, 'Triggering waves in nonlinear lattices: Quest for anharmonic phonons and corresponding mean-free paths', Physical Review B 90, 174304 (2014). DOI: 10.1103/PhysRevB.90.174304, https://doi.org/10.1103/PhysRevB.90.174304 (arXiv:1403.3598, https://arxiv.org/abs/1403.3598)
  12. S. Lepri, R. Livi, A. Politi, 'Thermal conduction in classical low-dimensional lattices', Physics Reports 377, 1-80 (2003). DOI: 10.1016/S0370-1573(02)00558-6, https://doi.org/10.1016/S0370-1573(02)00558-6
  13. B.-W. Huang, T. Hsiao, K.-H. Lin, D.-W. Chiou, C.-W. Chang, 'Length-dependent thermal transport and ballistic thermal conduction', AIP Advances 5, 053202 (2015). DOI: 10.1063/1.4914584, https://doi.org/10.1063/1.4914584