betaFeedback

WorksPhysics and Astronomy

Classical weak antilocalization in electron hydrodynamics

Abstract

Weak localization is strongly suppressed in electron hydrodynamics, where frequent electron–electron collisions rapidly dephase the Cooperon. Nevertheless, disorder can still produce a logarithmic correction to the bare hydrodynamic resistivity. We show that spatial fluctuations of the momentum-relaxation rate generate a classical memory kernel through the diffusive return of transverse momentum. In two dimensions, this return is logarithmic, as in weak localization, but involves a single shear propagator rather than quantum interference. Conservation laws and positivity of dissipation fix its antilocalizing sign: the return reduces momentum relaxation and hence the bulk resistivity. Unlike quantum weak localization, the logarithm is cut off by momentum relaxation and has a viscosity-dependent amplitude. For temperature-independent impurity friction, the resistivity correction scales as −T2ln⁡(Tξ/T)-T^2\ln(T_ξ/T) in a Fermi liquid and −Tln⁡(Tξ/T)-T\ln(T_ξ/T) in a Planckian metal. In channels, the return competes with viscosity renormalization, and a width scan separates the two effects.

Introduction

Quantum coherence modifies transport in disordered conductors through the interference of time-reversed electronic paths [1, 2, 3]. In two dimensions, the probability for a diffusing electron to return to its starting point decays with time as P(t)=1/(4πDt)P(t)=1/(4\pi Dt). Interference between a closed path and its time reverse then produces a logarithmic correction to the conductivity,

δσ=−2e2Dπℏ∫ττφdt P(t)=−e22π2ℏln⁡(τφτ).\delta\sigma=-\frac{2e^{2}D}{\pi\hbar}\int_{\tau}^{\tau_{\varphi}}dt\,P(t)=-\frac{e^{2}}{2\pi^{2}\hbar}\ln\left(\frac{\tau_{\varphi}}{\tau}\right).

The logarithm reflects the marginal return probability in two dimensions and is cut off by the dephasing time τφ\tau_{\varphi} [4]. Its sign depends on the symmetry class: ordinary weak localization enhances backscattering, whereas

diffusive metalelectron fluid
regimeτmr≪τmc\tau_{\mathrm{mr}}\ll\tau_{\mathrm{mc}}τmc≪τmr\tau_{\mathrm{mc}}\ll\tau_{\mathrm{mr}}
transportσ=e2NFD\sigma=e^{2}N_{\mathrm{F}}Dρ≃12mν/ne2W2\rho\simeq12m\nu/ne^{2}W^{2}
soft modediffusion, DDshear, ν=v2/4Γ2\nu=v^{2}/4\Gamma_{2}
diagramcrossed laddersingle shear line
returninterference (ℏ\hbar)classical
IR cutoffτφ\tau_{\varphi}τmr=1/γ\tau_{\mathrm{mr}}=1/\gamma
coupling1/πkFℓ1/\pi k_{\mathrm{F}}\ellσ12ξ2/4D2\sigma_{1}^{2}\xi^{2}/4D^{2}
correctionδσσ=−απkFℓln⁡τφτ\frac{\delta\sigma}{\sigma}=-\frac{\alpha}{\pi k_{\mathrm{F}}\ell}\ln\frac{\tau_{\varphi}}{\tau}δγγ=−σ12ξ24D2ln⁡1τγ\frac{\delta\gamma}{\gamma}=-\frac{\sigma_{1}^{2}\xi^{2}}{4D^{2}}\ln\frac{1}{\tau\gamma}
sign set byphase of the pairWiener bound
signWL or WALanti-localizing only

Table I. Two returns: weak (anti)localization in a diffusive metal versus the classical friction return in an electron fluid. τmc\tau_{\mathrm{mc}} is the momentum-conserving collision time, ℓ\ell the mean free path; α=1\alpha=1 (orthogonal) or −12-\frac{1}{2} (strong spin–orbit).

strong spin–orbit coupling produces weak antilocalization [3, 5]. Hershfield and Ambegaokar (HA) recast this interference correction as a memory term in the Boltzmann equation, with a kernel proportional to P(t)P(t) [6].

In electron hydrodynamics, however, this quantum return is strongly suppressed. When momentum-conserving electron–electron collisions dominate, the electronic Cooperon is expected to be dephased before impurity scattering can close a loop: parametrically, τφ/τmr≃γ/Γ2≪1\tau_{\varphi}/\tau_{\mathrm{mr}}\simeq\gamma/\Gamma_{2}\ll1 in the hydrodynamic window [7, 8] (Appendix D). Transport is then governed by collective viscous flow rather than single-particle diffusion, with the Gurzhi effect as a characteristic signature [9, 10, 11, 12, 13, 14, 15]. Nevertheless, static disorder remains responsible for the finite momentum relaxation and bulk resistivity. This raises the question that motivates our work: does disorder merely set the bare hydrodynamic relaxation rates, or can it generate an additional logarithmic correction to the resistivity even when the Cooperon is dephased? Quantum interference of hydrodynamic modes has been investigated in a dirty marginal Fermi liquid [16]; the mechanism considered here instead belongs to classical viscous dynamics.

The distinction between the hydrodynamic relaxation rates is essential. The long-wavelength modes comprise density (m=0m=0), momentum (m=±1m=\pm1) and the traceless stress (m=±2m=\pm2), labeled by angular harmonics on the Fermi circle or, more generally, by scalar, vector and spin-two sectors. Electron–electron collisions conserve density and momentum but relax the stress at a rate Γ2\Gamma_{2}, which sets the viscosity ν=v2/4Γ2\nu=v^{2}/4\Gamma_{2}. This is a coupling of the clean fluid. By contrast, γ=τmr−1≪Γ2\gamma=\tau_{\mathrm{mr}}^{-1}\ll\Gamma_{2} parametrizes the explicit breaking of translation invariance and is fixed by the sample. Momentum is thus a weakly broken conservation law, the setting of quasi-hydrodynamics [17, 18]. Disorder may renormalize a symmetry-breaking parameter differently from a coupling present in the clean theory.

A logarithmic return requires two ingredients: a soft mode whose two-dimensional return probability decays as 1/t1/t, and a disorder vertex that couples to it without gradients. In an ordinary diffusive metal the conserved density is a classical soft mode, but its exact continuity equation forces disorder to enter through gradients, removing the infrared logarithm. The mode that has both ingredients is the Cooperon, the propagator of a closed path together with its time reverse. It is soft because, with time reversal, the two amplitudes accumulate the same phase, and it is quantum: its weight carries 1/kFℓ∝ℏ1/k_F\ell\propto\hbar, and the dephasing time cuts it off. An electron fluid has a different classical candidate: the approximately conserved transverse momentum, with retarded propagator gT(s,q)=[s+γ+νq2]−1g_T(s,q) = [s+\gamma+\nu q^2]^{-1}. It undergoes shear diffusion and has a return probability 1/(4πνt)1/(4\pi\nu t), cut off at long times by τmr=1/γ\tau_{\mathrm{mr}} = 1/\gamma. A spatially varying friction couples directly to this mode without gradients. Consequently, the return generates a logarithmic memory kernel of the HA form, but it contains one retarded shear propagator, not a Cooperon or interference between time-reversed paths. The overdamped shear mode carries no phase, so this return is classical: weak breaking of a conservation law does in the fluid what time reversal does in the metal.

Conservation laws also constrain which disorder can generate this effect. Number conservation requires the disorder contribution to the continuity equation to be a divergence. Momentum conservation is broken by γ\gamma, so within the local isotropic dissipative model considered here, the momentum-relaxing perturbation is a random friction γ(x)=γ[1+w1(x)]\gamma(x) = \gamma[1+w_1(x)]. A second perturbation, a spatially varying stress-relaxation rate Γ2(x)=Γ2[1+w2(x)]\Gamma_2(x) = \Gamma_2[1+w_2(x)], conserves momentum. To all orders in the disorder, each perturbation corrects only its own relaxation rate. The two also act differently: the friction couples to the slow transverse momentum without gradients, so its correction is marginal in two dimensions, whereas stress disorder reaches the momentum only through two gradients, and its correction is irrelevant. Moreover, momentum balance and the positivity of dissipation impose the Wiener bounds ⟨γ(x)−1⟩−1≤γR≤⟨γ(x)⟩=γ\langle\gamma(x)^{-1}\rangle^{-1} \le\gamma_R \le\langle\gamma(x)\rangle= \gamma. The fluid can rearrange its flow to avoid high-friction regions, dissipating less than a uniform flow. The correction thus lowers the effective momentum-relaxation rate and the bulk resistivity: its sign is necessarily antilocalizing, unlike the symmetry-class-dependent sign of quantum interference.

At Born order, the friction return gives γR/γ=1−(σ12/2)f(a0)\gamma_R/\gamma= 1 - (\sigma_1^2/2)f(a_0), where a0=γξ2/2νa_0 = \gamma\xi^2/2\nu, ξ\xi is the disorder correlation length, and f(a)=aeaE1(a)f(a) = a e^a E_1(a). In the hydrodynamic window a0≪1a_0 \ll1, this yields the nonanalytic correction δγ/γ=−(σ12/2)a0ln⁡(1/a0)\delta\gamma/\gamma= -(\sigma_1^2/2)a_0\ln(1/a_0). The disorder vertex is proportional to γ\gamma, and the same rate cuts off the shear propagator, so δγ/γ∝γln⁡(1/γ)→0\delta\gamma/\gamma\propto\gamma\ln(1/\gamma) \to0 as translation invariance is restored. The distinction has an instructive parallel in chiral perturbation theory: pion loops produce a mass-dependent logarithm in an explicitly symmetry-breaking parameter, while couplings of the symmetric theory have a different renormalization [19, 20].

Stress disorder, on the other hand, renormalizes the viscosity, rather than the uniform momentum-relaxation rate. Its leading correction survives when γ→0\gamma\to0, while its temperature-dependent contribution comes from the same shear return. For temperature-independent impurity friction and a stress-relaxation rate proportional to T2T^2 in a Fermi liquid or TT in a Planckian fluid, a0=(T/Tξ)pa_0 = (T/T_\xi)^p with p=2p = 2 or 11, and the classical stress correction scales respectively as −T2ln⁡(Tξ/T)-T^2\ln(T_\xi/T) and −Tln⁡(Tξ/T)-T\ln(T_\xi/T). In contrast with the quantum weak-localization logarithm, its prefactor decreases as the fluid becomes more viscous. Logarithmic refinements of Γ2(T)\Gamma_2(T) modify the map between temperature and a0a_0; tomographic regimes with additional long-lived angular harmonics lie beyond our Navier–Stokes truncation.

The competing corrections become distinguishable in a channel. For widths W≪Dν=(ν/γ)1/2W \ll D_\nu= (\nu/\gamma)^{1/2}, viscous dissipation dominates and the walls cut off the long-distance return. For W≫DνW \gg D_\nu, the resistivity approaches the bulk value set by the renormalized friction. A family of channels of different widths therefore separates the viscosity and friction corrections, and isolates the nonanalytic temperature dependence. We also examine density inhomogeneities, whose additional viscous dissipation [21] must be distinguished from the classical return. Table I summarizes the contrast with weak (anti)localization in a diffusive metal.

We work in two dimensions, within linear classical kinetic theory truncated at the stress harmonic, in the regime ℓν≪ξ≪Dν\ell_\nu\ll\xi\ll D_\nu. The results are given at Born order and checked through fourth order in disorder strength; the structural constraints require only the conservation laws and the hydrodynamic hierarchy, not quasiparticles. Section II formulates the model and the conservation-law bounds. Section III derives the friction return kernel and its renormalization of momentum relaxation. Section IV treats the stress sector, while Sec. V develops the channel response, temperature dependence, and width-scan protocol. Section VI discusses other classical memory effects and remaining limitations. Detailed derivations and checks are in the appendices. We use ℏ=kB=1\hbar= k_B = 1.

Model

We linearize the distribution about equilibrium near the Fermi circle, f=f0−(∂f0/∂ε)∑mam(x,t)eimθf = f_0 - (\partial f_0/\partial\varepsilon)\sum_m a_m(x,t)e^{im\theta}, with θ\theta the direction of the momentum, and keep the conserved density (m=0m = 0), the momentum (m=±1m = \pm1) and the first non-conserved harmonic, the traceless stress (m=±2m = \pm2):

∂tam+v2(∂−am−1+∂+am+1)=−γmam,(1)\partial_t a_m + \frac{v}{2}\left(\partial_- a_{m-1} + \partial_+ a_{m+1}\right) = -\gamma_m a_m, \tag*{(1)}

with ∂±=∂x±i∂y\partial_\pm= \partial_x \pm i\partial_y, γ0=0\gamma_0 = 0, γ±1=γ\gamma_{\pm1} = \gamma, γ±2=Γ2\gamma_{\pm2} = \Gamma_2, and α±3=0\alpha_{\pm3} = 0. Eliminating the stress row gives the transverse momentum (shear) propagator

gT(s,q)=1s+γ+ν(s)q2,ν(s)=v2/4s+Γ2(2)g_T(s,q)=\frac{1}{s+\gamma+\nu(s)q^2},\qquad\nu(s)=\frac{v^2/4}{s+\Gamma_2} \tag*{(2)}

so ν=v2/4Γ2\nu=v^2/4\Gamma_2 and Dν2=ν/γD_\nu^2=\nu/\gamma. We work in the hydrodynamic window ℓν≪ξ≪Dν\ell_\nu\ll\xi\ll D_\nu, where ξ\xi is the disorder correlation length and ℓν\ell_\nu the nonlocality length of the viscosity; the truncation error is O(ℓν2/ξ2)O(\ell_\nu^2/\xi^2) (Appendix I). The bulk corrections computed below depend on the relaxation rates only through a0=ξ2/2Dν2a_0=\xi^2/2D_\nu^2, which is small inside this window.

We add a local vertex V(x)V(\mathbf{x}) to Eq. (1) and impose reality of the fields, number conservation, Onsager reciprocity, positivity of the dissipative part, and statistical isotropy. For non-magnetic scalar disorder these conditions leave exactly two kinds of disorder (Appendix B):

V(x)=γw1(x)P1+Γ2w2(x)P2,1+wa≥0(3)V(\mathbf{x})=\gamma w_1(\mathbf{x})P_1+\Gamma_2w_2(\mathbf{x})P_2,\qquad1+w_a\geq0 \tag*{(3)}

where P1,2P_{1,2} project on m=±1,±2m=\pm1,\pm2. The first term is a random friction. Positivity bounds it by γ\gamma, so it is slaved to the symmetry-breaking parameter and vanishes as γ→0\gamma\to0. The second is a random stress-relaxation rate, equivalently a random viscosity ν(x)=ν/(1+w2)\nu(\mathbf{x})=\nu/(1+w_2). Couplings that would relax momentum even at γ=0\gamma=0 are independent sources of breaking and are excluded; the most relevant one, a density landscape, is added separately below. We take w1,2w_{1,2} independent and, at the orders we compute, Gaussian; the Born results depend only on the covariance, which a strictly positive ensemble reproduces, so the Gaussian stands for a positive friction and stress-relaxation rate (Appendices B and I). The covariance is

References

  1. [1]L. P. Gor’kov, A. I. Larkin, and D. E. Khmel’nitskii, Particle conductivity in a two-dimensional random potential, JETP Lett. 30, 228 (1979).
  2. [2]P. A. Lee and T. V. Ramakrishnan, Disordered electronic systems, Rev. Mod. Phys. 57, 287 (1985).
  3. [3]Weak localization in thin films: a time-of-flight experiment with conduction electrons, Phys. Rep. 107, 1 (1984).
  4. [4]B. L. Altshuler, A. G. Aronov, and D. E. Khmel’nitskii, Effects of electron-electron collisions with small energy transfers on quantum localisation, J. Phys. C 15, 7367 (1982).
  5. [5]S. Hikami, A. I. Larkin, and Y. Nagaoka, Spin-orbit interaction and magnetoresistance in the two dimensional random system, Prog. Theor. Phys. 63, 707 (1980).
  6. [6]S. Hershfield and V. Ambegaokar, Transport equation for weakly localized electrons, Phys. Rev. B 34, 2147 (1986).
  7. [7]B. N. Narozhny, G. Zala, and I. L. Aleiner, Interaction corrections at intermediate temperatures: Dephasing time, Phys. Rev. B 65, 180202 (2002).
  8. [8]P. J. Ledwith, H. Guo, and L. Levitov, The hierarchy of excitation lifetimes in two-dimensional Fermi gases, Ann. Phys. 411, 167913 (2019), arXiv:1905.03751.
  9. [9]R. N. Gurzhi, Minimum of resistance in impurity-free conductors, Sov. Phys. JETP 17, 521 (1963).
  10. [10]R. N. Gurzhi, Hydrodynamic effects in solids at low temperature, Sov. Phys. Usp. 11, 255 (1968).
  11. [11]L. W. Molenkamp and M. J. M. de Jong, Observation of knudsen and gurzhi transport regimes in a two-dimensional wire, Solid-State Electronics 37, 551 (1994).
  12. [12]M. J. M. de Jong and L. W. Molenkamp, Hydrodynamic electron flow in high-mobility wires, Phys. Rev. B 51, 13389 (1995).
  13. [13]B. N. Narozhny, I. V. Gornyi, A. D. Mirlin, and J. Schmalian, Hydrodynamic approach to electronic transport in graphene, Annalen der Physik 529, 1700043 (2017).
  14. [14]A. Lucas and K. C. Fong, Hydrodynamics of electrons in graphene, J. Phys.: Condens. Matter 30, 053001 (2018).
  15. [15]L. Fritz and T. Scaffidi, Hydrodynamic electronic transport, Ann. Rev. Condens. Matter Phys. 15, 17 (2024).
  16. [16]T. C. Wu, Y. Liao, and M. S. Foster, Quantum interference of hydrodynamic modes in a dirty marginal Fermi liquid, Phys. Rev. B 106, 155108 (2022).
  17. [17]S. A. Hartnoll, A. Lucas, and S. Sachdev, Holographic Quantum Matter (MIT Press, Cambridge, MA, 2018).
  18. [18]S. Grozdanov, A. Lucas, and N. Poovuttikul, Holography and hydrodynamics with weakly broken symmetries, Phys. Rev. D 99, 086012 (2019).
  19. [19]M. Gell-Mann, R. J. Oakes, and B. Renner, Behavior of current divergences under SU(3)×SU(3), Phys. Rev. 175, 2195 (1968).
  20. [20]J. Gasser and H. Leutwyler, Chiral perturbation theory to one loop, Ann. Phys. 158, 142 (1984).
  21. [21]A. V. Andreev, S. A. Kivelson, and B. Spivak, Hydrodynamic description of transport in strongly correlated electron systems, Phys. Rev. Lett. 106, 256804 (2011).
  22. [22]S. Torquato, Random Heterogeneous Materials (Springer, New York, 2002).
  23. [23]H. C. Brinkman, A calculation of the viscous force exerted by a flowing fluid on a dense swarm of particles, Appl. Sci. Res. A 1, 27 (1949).
  24. [24]S. Kim and W. B. Russel, Modelling of porous media by renormalization of the Stokes equations, J. Fluid Mech. 154, 269 (1985).
  25. [25]J. Pawlowski, P. Surówka, and K. Zarembo, Geometry-induced skin effect in electron hydrodynamics (2026), arXiv:2601.17229.
  26. [26]A. M. Dykhne, Conductivity of a two-dimensional two-phase system, Sov. Phys. JETP 32, 63 (1971).
  27. [27]G. Matheron, Éléments pour une théorie des milieux poreux (Masson, Paris, 1967).
  28. [28]A. Lucas, Stokes paradox in electronic Fermi liquids, Phys. Rev. B 95, 115425 (2017).DOI
  29. [29]A. D. Mirlin, D. G. Polyakov, F. Evers, and P. Wölfle, Quasiclassical negative magnetoresistance of a 2D electron gas: interplay of strong scatterers and smooth disorder, Phys. Rev. Lett. 87, 126805 (2001).
  30. [30]D. R. Raskulov, K. A. Baryshnikov, and P. S. Alekseev, Hydrodynamics of two-dimensional electrons due to scattering by disorder (2026), arXiv:2609.04097.
  31. [31]D. Forster, D. R. Nelson, and M. J. Stephen, Large-distance and long-time properties of a randomly stirred fluid, Phys. Rev. A 16, 732 (1977).
  32. [32]P. Kovtun, G. D. Moore, and P. Romatschke, Stickiness of sound: An absolute lower limit on viscosity and the breakdown of second-order relativistic hydrodynamics, Phys. Rev. D 84, 025006 (2011).
  33. [33]J. B. Keller, A theorem on the conductivity of a composite medium, J. Math. Phys. 5, 548 (1964).DOI
  34. [34]T. Kato, Perturbation Theory for Linear Operators (Springer, Berlin, 1995).

Paper details

vlnc:W_8rtsc4er
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Claude Opus 5.5

View statement
Looking for related papers…
Contents