Migdal–Eliashberg Theory

• Migdal–Eliashberg theory is a quantum many-body framework that extends BCS theory by including frequency-dependent electron–phonon interactions and strong-coupling effects.
• It employs coupled self-consistent equations for the superconducting gap and mass renormalization, using spectral functions and an effective Coulomb pseudopotential.
• The framework supports extensions such as anisotropic, multi-band, vertex-corrected, and nonadiabatic treatments, underpinning modern ab initio predictions for superconductors.

Migdal–Eliashberg Theory

Migdal–Eliashberg (ME) theory is a quantum many-body framework that extends BCS theory to incorporate frequency-dependent electron–phonon interactions and strong-coupling effects in conventional and unconventional superconductors. ME theory provides a set of coupled, self-consistent equations for the superconducting gap and mass renormalization, using as input the fully retarded electron–phonon spectral density and an effective Coulomb repulsion (the Coulomb pseudopotential). The theory is quantitatively accurate for a broad regime of strong-coupling phonon-mediated superconductors and can be generalized to include multi-band, anisotropic, and nonadiabatic effects. It forms the basis for modern first-principles calculations of superconducting properties and serves as a controlled starting point for treating fluctuations and competing instabilities.

1. Fundamental Equations and Ingredients

In the Matsubara (imaginary-frequency) formalism, the isotropic ME equations determine two central quantities: the anomalous (pairing) function $\varphi_n \equiv \varphi(i\omega_n)$ and the mass renormalization $Z_n \equiv Z(i\omega_n)$, where $\omega_n = \pi k_B T (2n + 1)$ are fermionic Matsubara frequencies. The equations (Durajski et al., 2016) read: $$\Delta(i\omega_n)\,Z(i\omega_n) = \pi T \sum_{m=-M}^{M} \Big[ K(i\omega_n, i\omega_m) - \mu^*\,\Theta(\omega_c-|\omega_m|) \Big] \frac{\Delta(i\omega_m)}{\sqrt{\omega_m^2 + \Delta(i\omega_m)^2}}$$

$$Z(i\omega_n) = 1 + \frac{\pi T}{\omega_n} \sum_{m=-M}^{M} K(i\omega_n, i\omega_m) \frac{\omega_m}{\sqrt{\omega_m^2 + \Delta(i\omega_m)^2}}$$

Here, $\Delta(i\omega_n) \equiv \varphi_n / Z_n$; $\mu^*$, the Coulomb pseudopotential, models the residual effective repulsion and is cutoff at $\omega_c \sim 10\,\omega_D$ (phonon scale). The electron–phonon kernel is: $$K(i\omega_n, i\omega_m) = 2\int_0^\infty d\Omega \frac{\Omega\,\alpha^2F(\Omega)}{\Omega^2 + (\omega_n-\omega_m)^2}$$ The central material-dependent inputs are:

• Eliashberg spectral function $\alpha^2F(\Omega)$: Encodes phonon density of states weighted by matrix elements; determines the retarded pairing interaction.
• Dimensionless electron-phonon coupling $\lambda$:

$$\lambda = 2\int_0^\infty d\Omega \frac{\alpha^2F(\Omega)}{\Omega}$$

• Logarithmic averaged phonon frequency $\omega_{\ln}$:

$$\omega_{\ln} = \exp\left[\frac{2}{\lambda} \int_0^\infty d\Omega\,\frac{\alpha^2F(\Omega)}{\Omega}\ln\Omega\right]$$

In the strong coupling regime, the characteristic ratio $2\Delta(0)/k_BT_c$ can greatly exceed the weak-coupling BCS value ($3.53$), reaching $4$–$5$ in materials such as H$_3$S, signaling strong-coupling and retardation effects (Durajski et al., 2016).

2. Migdal’s Theorem and the Validity Regime

Migdal’s theorem underpins ME theory by showing that vertex corrections to the electron–phonon interaction are suppressed by the small adiabatic parameter, $\Omega_D/E_F$, where $\Omega_D$ is a typical phonon frequency and $E_F$ is the electronic Fermi energy (Chowdhury et al., 2019): $$\text{Vertex correction} \sim \lambda \frac{\Omega_D}{E_F} \ll 1$$ This result allows neglect of vertex corrections for conventional metals, even at large $\lambda$, if $\Omega_D/E_F \ll 1$ (Bauer et al., 2011). The essential validity criterion is $P = \lambda(\omega_{\mathrm{ph}}/W) \ll 1$ ($W$ the conduction bandwidth). In physical electron–phonon superconductors, effective parameters are extracted experimentally or ab initio, not from bare interaction values, resolving discrepancies with low-dimensional or strong-coupling model studies. However, for materials near a polaronic or charge order transition, $\lambda_{\text{eff}}\gtrsim 3$–$4$ can invalidate the ME approach due to divergent self-energies and unphysical negative heat capacity (Yuzbashyan et al., 2022).

3. Extensions: Vertex Corrections, Anharmonicity, Nonadiabaticity

While standard ME theory neglects vertex corrections, extensions systematically incorporate them when the adiabatic approximation is less valid or when ab initio calculations indicate the need. Lowest-order local vertex corrections introduce nontrivial frequency and dynamical dependencies, shifting the effective $\mu^*$ and the relation between $\alpha^2F$ and the observable gap (Durajski et al., 2016). Anharmonic effects—arising from phonon-phonon interactions, Debye-Waller renormalizations, and mode softening—can also significantly affect $\alpha^2F$ via spectral broadening and suppressed $\lambda$ (Durajski et al., 2016). Nonadiabatic corrections, important for high-frequency modes or quasi-2D systems, further modify the effective interaction and Coulomb pseudopotential.

Recent theoretical work has mapped ME theory onto a classical spin chain, providing powerful new Monte Carlo solvers and analytical renormalization techniques. In this mapping, the nonlinear gap equations are recast as a ferromagnetic Heisenberg chain with site-dependent Zeeman fields, offering physical intuition into the sign-changing structure of the gap and the renormalization flow of interactions (including reproduction of the Morel–Anderson $\mu^*$ formula) (Yuzbashyan et al., 2022, Chou et al., 2023).

4. Anisotropy and Multi-Band Generalizations

The fully anisotropic ME formalism preserves the full $(k,i\omega_n)$-dependence of the gap and self-energy, crucial for materials with multiple Fermi sheets or strong momentum dependence, as in MgB$_2$ or n-doped graphene (Margine et al., 2014, Margine et al., 2012). The anisotropic equations require integrating over fine Brillouin zone meshes and depend on Wannier–Fourier interpolation of electron–phonon matrix elements (Margine et al., 2012). In the multi-band setting, the gap, renormalization, and Coulomb pseudopotential acquire matrix structure; careful treatment of the band-resolved $\alpha^2F_{jj'}(\omega)$ and $\lambda_{jj'}$ reveals enhancement of $T_c$ (by $4$–$8$\% in (111) diamond) and characteristic multiple gap structure (Romanin et al., 2020).

These first-principles workflows, implemented in modern codes such as EPW, enable ab initio prediction of $T_c$, gap anisotropy, and tunneling spectra for complex materials, including low-dimensional and layered systems, and superhydrides (Lucrezi et al., 2023).

5. Numerical Methods, Solution Workflows, and Analytic Continuation

A typical ME calculation involves:

1. Initial setup: Compute $\alpha^2F(\Omega)$ (e.g., via DFPT + Wannier interpolation), determine $\mu^*$ (fitted or from screened Coulomb matrix elements).
2. Matsubara solution: Solve coupled equations for $\Delta_n$ and $Z_n$ over $\sim\!\!10^3$–$10^4$ Matsubara points; for complex or multi-band systems, this may require advanced preconditioning, large memory, and sparse frequency grids (Lucrezi et al., 2023).
3. $T_c$ determination: For each approximation, $\mu^*$ is adjusted so that the gap vanishes at experimental or predicted $T_c$; convergence is set by relative changes $\ll 10^{-6}$.
4. Analytic continuation: Zero-temperature and spectroscopic properties are obtained by analytic continuation of the Matsubara-axis solution to the real axis, using Padé approximants or, more robustly, Nevanlinna analytic continuation utilizing auxiliary Green’s functions to guarantee causality and accuracy for real-frequency observables (Khodachenko et al., 2024).
5. Physical observables: Extract gap ratios, effective mass enhancements, density of states, and compare $T_c$ and $\Delta(0)$ with experiment or alternative models. Numerical studies confirm the need for self-consistency and inclusion of all relevant feedback effects, including phonon self-energy corrections, for accurate spectral and thermodynamic predictions (Nosarzewski et al., 2020).

6. Domain of Applicability, Competing Orders, and Breakdown

ME theory quantitatively describes strong-coupling, phonon-mediated superconductivity up to effective $\lambda \lesssim 3$ for typical $\omega_{\mathrm{ph}}/W \lesssim 0.05$ (Bauer et al., 2011, Esterlis et al., 2017). For higher $\lambda$, determinant QMC and DMFT benchmarks, as well as analytic considerations, show sharp crossover to competing charge-ordered, Peierls, or polaronic phases, with loss of Fermi-liquid coherence, softening of phonon modes, and collapse of the quasiparticle weight (Esterlis et al., 2017, Yuzbashyan et al., 2022). At strong coupling, ME theory predicts negative normal-state specific heat and vanishing quasiparticle lifetime, signaling instability to lattice-symmetry-broken phases with gaps in the single-particle density of states.

Near this breakdown, feedback effects (phonon softening, mode broadening) enhance effective coupling, promoting CDW or polaronic order, which competes with superconductivity, and ME theory no longer gives quantitatively correct $T_c$ or gaps. Precise boundaries depend on the bare versus effective parameters and the shape of $\alpha^2F(\Omega)$.

7. Extensions to Unconventional and Non-Fermi-Liquid Contexts

ME theory, though originally developed for conventional superconductors, is also formally exact in certain non-Fermi-liquid regimes with controlled expansions (e.g., large-$N$ or small $\epsilon$ in quantum-critical boson coupled to Fermi surface, or SYK-type interactions), even in the absence of coherent Landau quasiparticles (Chowdhury et al., 2019). In these non-BCS systems, the same structure of self-consistent equations governs pairing, with kernel $K(i\omega_n,i\omega_m)$ derived from critical bosonic propagators or local quantum-melonic limits. The essential requirement becomes the existence of a controlled expansion that suppresses vertex and crossing diagrams; metallic quantum criticality with dominant forward scattering often satisfies this criterion, although in 2+1D, shape fluctuations of the Fermi surface can destabilize the ME saddle, necessitating new theoretical approaches (Guo, 2023).

---

References:

" Migdal-Eliashberg equations - the effective model for superconducting state in H₃S" (Durajski et al., 2016) "Two-gap superconductivity in heavily n-doped graphene: ab initio Migdal-Eliashberg theory" (Margine et al., 2014) "Migdal-Eliashberg theory of multi-band high-temperature superconductivity in field-effect-doped hydrogenated (111) diamond" (Romanin et al., 2020) "Migdal-Eliashberg theory as a classical spin chain" (Yuzbashyan et al., 2022) "Monte Carlo solver and renormalization of Migdal-Eliashberg spin chain" (Chou et al., 2023) "Quantitative reliability of Migdal-Eliashberg theory for strong electron-phonon coupling" (Bauer et al., 2011) "Breakdown of the Migdal-Eliashberg theory and a theory of lattice-fermionic superfluidity" (Yuzbashyan et al., 2022) "Breakdown of Migdal-Eliashberg theory; a determinant quantum Monte Carlo study" (Esterlis et al., 2017) "Spectral properties and enhanced superconductivity in renormalized Migdal-Eliashberg theory" (Nosarzewski et al., 2020) "Nevanlinna Analytic Continuation for Migdal-Eliashberg Theory" (Khodachenko et al., 2024) "Full-bandwidth anisotropic Migdal-Eliashberg theory and its application to superhydrides" (Lucrezi et al., 2023) "The unreasonable effectiveness of Eliashberg theory for pairing of non-Fermi liquids" (Chowdhury et al., 2019) "Is the Migdal-Eliashberg Theory for 2+1D Critical Fermi Surface Stable?" (Guo, 2023) "Anisotropic Migdal-Eliashberg theory using Wannier functions" (Margine et al., 2012) "Nonlinear electron-phonon interactions in Migdal-Eliashberg theory" (Zappacosta et al., 6 Mar 2025).