advanced29 min studyLesson 500 in Physics

Phonon scattering and thermal transport

Connect heat-carrying phonon populations to group velocity and lifetime, distinguish normal and Umklapp scattering, and quantify the limits of a simple kinetic conductivity estimate.

On this page

What you’ll understand

  • Compute an isotropic kinetic estimate with consistent volumetric heat capacity and mean free path.
  • Distinguish crystal-momentum-conserving and resistive scattering processes.
  • State the quasiparticle, linear-response, and relaxation-time assumptions behind Fourier transport.

A solid can store vibrational energy without transporting it efficiently. Heat transport depends on how populations of moving phonon modes depart from local equilibrium, on their group velocities, and on how scattering relaxes that departure. The perfect harmonic model has no intrinsic phonon scattering, so a finite bulk conductivity requires additional physics or boundaries. We will derive a useful kinetic estimate, distinguish two classes of phonon collisions, and identify when a simple mean-free-path picture becomes inadequate.

Heat current comes from a population imbalance

At thermal equilibrium, modes traveling in opposite directions carry equal average energy currents, giving zero net heat flow. A small temperature gradient shifts their populations unevenly. In the relaxation-time approximation, each mode's departure from local Bose equilibrium is estimated from its group velocity, the gradient, and a transport lifetime. Summing the resulting energy currents gives a conductivity tensor. The derivative of frequency with respect to angular wavevector is the group velocity; using phase velocity instead can badly misrepresent modes near a flat zone-edge dispersion.

For an isotropic gray approximation with one representative speed and lifetime, the angular average of a squared directional velocity supplies a factor one-third. Conductivity becomes one-third of volumetric heat capacity times speed times mean free path. The heat capacity must be per volume, not the total heat capacity of an arbitrary sample. The resulting units are watts per metre kelvin. This estimate is useful for scale analysis, but a real material usually has broad branch-dependent distributions of speed, occupation, and scattering lifetime.

κij≃1V∑qνCqνvqν,ivqν,jτqν,κ≃13cVv2τ=13cVvℓ,ℓ=vτ\kappa_{ij}\simeq\frac1V\sum_{\mathbf q\nu}C_{\mathbf q\nu}v_{\mathbf q\nu,i}v_{\mathbf q\nu,j}\tau_{\mathbf q\nu},\qquad \kappa\simeq\frac13c_Vv^2\tau=\frac13c_Vv\ell,\qquad \ell=v\tau
Mode C has units J/K and c_V is volumetric heat capacity in J/(m³ K). Group velocity is in m/s, lifetime in s, mean free path in m, and conductivity in W/(m K). These are near-equilibrium quasiparticle relaxation-time approximations.↔ Scroll wide equations horizontally.

Normal collisions and Umklapp collisions play different roles

Anharmonic interactions allow phonons to combine or decay subject to energy conservation and crystal wavevector conservation modulo a reciprocal vector. A three-phonon combination obeys omega1 plus omega2 equal to omega3 and q1 plus q2 equal to q3 plus G. A process with G zero is normal; one with nonzero G is Umklapp when all wavevectors are represented in the first zone. The lattice absorbs the reciprocal-vector bookkeeping. This does not violate total physical momentum conservation when the complete crystal and environment are included.

Normal collisions conserve total phonon crystal momentum and can drive a distribution toward a drifting local equilibrium rather than a motionless one. They redistribute energy and can influence heat current, but should not generally be inserted as independent resistive events in a scalar relaxation-time sum. Umklapp, defects, and boundaries can relax the relevant directed transport. When normal collisions dominate over resistive scattering, collective hydrodynamic behavior can require a coupled collision treatment instead of the naive picture that every collision independently destroys the same fraction of current.

ω1+ω2=ω3,q1+q2=q3+G,τeff−1≃τB−1+τU−1for independent resistive channels\omega_1+\omega_2=\omega_3,\qquad \mathbf q_1+\mathbf q_2=\mathbf q_3+\mathbf G,\qquad \tau_{\mathrm{eff}}^{-1}\simeq\tau_B^{-1}+\tau_U^{-1}\quad\text{for independent resistive channels}
G is a reciprocal vector in the same two-pi convention used throughout the module. The final rate sum is an approximate Matthiessen rule for the stated boundary and Umklapp channels, not a universal collision-operator identity.↔ Scroll wide equations horizontally.

Local Fourier diffusion has a limited regime

Fourier's law relates heat flux to minus conductivity times temperature gradient. Combined with local energy conservation and constant coefficients, it gives a thermal diffusivity equal to conductivity divided by volumetric heat capacity. These are distinct quantities with different units. Local diffusion is appropriate when the spatial and temporal scales exceed relevant mean free paths and relaxation times. In a small sample or over short delays, ballistic propagation and boundary transmission can matter, making a bulk scalar conductivity insufficient to characterize the measured heat flow.

At low temperature boundary-limited mean free paths can be nearly temperature independent, so conductivity can inherit a bulk acoustic heat-capacity power law under additional assumptions. At higher temperature Umklapp scattering often lowers lifetimes, but an inverse-temperature conductivity is not a universal law. Strong disorder, mode localization, optical contributions, electrons, and explicit anharmonicity can change the behavior. The quasiparticle picture itself becomes doubtful when linewidths rival frequencies or paths are too short to define a clear wavevector. Distinguish a controlled model estimate from a measured material property or an exact transport solution.

A gray phonon conductivity estimate

A hypothetical isotropic solid has volumetric heat capacity cV=1.5×106 J m−3 K−1c_V=1.5\times10^6\,\mathrm{J\,m^{-3}\,K^{-1}}, representative group speed 3000 m/s, and resistive mean free path 20 nm. Estimate conductivity, lifetime, and heat flux for a temperature gradient of 1000 K/m along positive x.

  1. Convert the path to 20×10−9 m20\times10^{-9}\,\mathrm m. The gray kinetic formula gives κ=(1/3)(1.5×106)(3000)(20×10−9)=30 W m−1 K−1\kappa=(1/3)(1.5\times10^6)(3000)(20\times10^{-9})=30\,\mathrm{W\,m^{-1}\,K^{-1}}.

  2. The corresponding lifetime is τ=ℓ/v=(20×10−9)/3000=6.67×10−12 s\tau=\ell/v=(20\times10^{-9})/3000=6.67\times10^{-12}\,\mathrm s, or about 6.67 ps. This is a representative transport lifetime, not a measured lifetime for every branch.

  3. Fourier flux is jQ,x=−κ∂xT=−30(1000)=−3.0×104 W m−2j_{Q,x}=-\kappa\partial_xT=-30(1000)=-3.0\times10^4\,\mathrm{W\,m^{-2}}. Its negative direction carries energy toward lower temperature, consistent with the sign of the imposed gradient.

The result

The model estimates conductivity 30 W/(m K), lifetime 6.67 ps, and heat flux 30 kW/m² along negative x under local diffusive conditions.

Explore the model

Interactive lab

Interactive lab · heat

Thermal diffusivity controls spatial-mode decay

Reset and set BOTH Spatial mode n = 1 and Diffusivity κ = 0.5; Step twice to t = 1 and predict amplitude exp(−0.5). For a fresh comparison, Reset and set n = 1 and κ = 1; Step twice and predict exp(−1). Compare the exact amplitude readouts.

Thermal diffusivity controls spatial-mode decayThe exact periodic heat-equation mode is u(x,t) = exp(−κn²t) sin(nx), on [0,2π]. Playback advances physical time. The initial sine is shown separately, and the unnormalized L² norm is √π exp(−κn²t). This analytic instrument does not run a finite-difference solver or test discretization stability. Current readings: Current amplitude: 1; Spatial L² norm: 1.772; Decay rate · κn²: 3.6 /s. All parameters are adjustable with the labeled controls below.Periodic heat equation mode 3 at time 0: the amplitude is 1.0π/2π3π/22π−1−0.500.51x · periodic intervalTemperature perturbation uExact mode · n = 3
Current temperature modeInitial sineCurrent amplitude
Current amplitude
1
Spatial L² norm
1.772
Decay rate · κn²
3.6 /s
t = 0 s

The exact periodic heat-equation mode is u(x,t) = exp(−κn²t) sin(nx), on [0,2π]. Playback advances physical time. The initial sine is shown separately, and the unnormalized L² norm is √π exp(−κn²t). This analytic instrument does not run a finite-difference solver or test discretization stability. Parameter changes pause playback and recompute the model at the current elapsed time. Reset restores both parameters and returns time to zero. Playback starts paused. Adjust either parameter, then choose Play or Step.

Look forThis is one analytic periodic heat mode in instrument coordinates. Its κ is diffusivity, not thermal conductivity; it does not model phonon lifetimes, branches, ballistic flow, or a collision operator.

Check your understanding

Why should normal phonon collisions not automatically be counted as independent resistive rates?

Try it yourself

In an isotropic gray model, independent resistive boundary and Umklapp lifetimes are 20 ps and 30 ps. Take group speed 2000 m/s and cV=2×106 J m−3 K−1c_V=2\times10^6\,\mathrm{J\,m^{-3}\,K^{-1}}. Find the combined lifetime, mean free path, and conductivity.

Take a moment to make a prediction or work it out on paper. Open a hint when you need one.

Give me a hint

Add rates, not lifetimes. Apply the one-third angular factor only after finding the combined resistive lifetime.

Compare with the solution

The combined rate is 1/20+1/30=1/12 ps−11/20+1/30=1/12\,\mathrm{ps^{-1}}, giving τ=12 ps\tau=12\,\mathrm{ps}. The path is vτ=2000(12×10−12)=24 nmv\tau=2000(12\times10^{-12})=24\,\mathrm{nm}. Conductivity is (1/3)(2×106)(2000)(24×10−9)=32 W m−1 K−1(1/3)(2\times10^6)(2000)(24\times10^{-9})=32\,\mathrm{W\,m^{-1}\,K^{-1}}. The result assumes these resistive channels are independently approximated by rates and ignores mode dependence; adding a normal-collision rate in the same way would require additional justification.

Take it with you

  • Phonon transport combines heat-capacity weighting, group velocity, and transport relaxation.
  • Normal and Umklapp processes obey different crystal-momentum constraints and transport roles.
  • A gray conductivity estimate and a thermal diffusion equation require explicit scale and scattering assumptions.

Read further

Original sources and deeper explanations to follow at your own pace.

Optional progress, saved in this browser when available. Revisit any lesson at any time.

Lesson not marked complete.