Migdal's Theorem and Vertex-Correction Validity

Migdal’s result is a parametric statement, not permission to omit vertex corrections in every electron–phonon problem. In a broad-band metal with a regular Fermi surface, typical phonon energy $\omega_{\mathrm{ph}}\ll E_F$, nonsingular momentum transfer, and moderate effective coupling, the leading vertex correction is smaller than the electron self-energy contribution by an adiabatic parameter of order $\lambda\omega_{\mathrm{ph}}/E_F$. Low density, flat or narrow bands, forward-focused coupling, van Hove kinematics, and strong coupling can remove this hierarchy.

Required background. Retarded electron–phonon interactions supplies the microscopic fields. Fermi-surface kinematics supplies the electronic scale and phase space.

Helpful background. Current vertices and Ward consistency distinguishes a small pairing vertex correction from conservation-required response vertices.

Statement with hypotheses

Consider electrons with dispersion $\xi_{\mathbf k}$ coupled to phonons,

$$H_{e\text{-}ph}=\sum_{\mathbf k\mathbf q\sigma} g_{\mathbf q} c^\dagger_{\mathbf k+\mathbf q,\sigma}c_{\mathbf k\sigma} (b_{\mathbf q}+b^\dagger_{-\mathbf q}).$$

Let $E_F$ mean the distance from the chemical potential to the relevant band edge or curvature scale, not automatically the full bandwidth. For a three-dimensional metal with typical momentum transfer of order $k_F$, regular density of states, phonon scale $\omega_{\mathrm{ph}}$, and dimensionless coupling $\lambda$ not parametrically large, the first crossed vertex obeys schematically

$$\frac{\delta\Gamma}{\Gamma_0} =O\!\left(\lambda\frac{\omega_{\mathrm{ph}}}{E_F}\right).$$

The small ratio comes from the mismatch between slow ionic and fast electronic scales plus Fermi-surface phase space. It is not simply the ionic mass ratio, though for ordinary acoustic phonons $\omega_D/E_F$ often scales as $(m/M)^{1/2}$. Migdal 1958, pp. 996–999 derives the suppression in the adiabatic regime.

This is an asymptotic theorem about a stated family of diagrams and kinematics. It does not prove that the self-energy is small: the retained rainbow diagrams can produce a mass enhancement $m^\ast/m\simeq1+\lambda$. Nor does it eliminate vertex insertions required by a Ward identity for a conserved response.

Power-counting sketch

The leading vertex triangle contains two electron denominators and one phonon denominator. Linearizing near the Fermi surface, the internal electronic poles occupy a narrow angular and energy region. After the frequency contour is performed, the nonadiabatic part is controlled by the recoil of an electron across the Fermi surface, $v_Fq\sim E_F$ for generic $q\sim k_F$, whereas the phonon transfers only $\omega_{\mathrm{ph}}$. Relative to the one-loop self-energy, this leaves the ratio $\omega_{\mathrm{ph}}/E_F$.

The argument changes for $q\ll k_F$: then $v_Fq$ rather than $E_F$ enters the denominator, and the limits $q\to0$ and $\omega\to0$ may not commute. A strongly forward-focused coupling can therefore have nonadiabatic corrections even when $\omega_{\mathrm{ph}}/E_F$ is numerically small. The same warning applies near a van Hove point, where $v_F$ vanishes and the density of states is singular. Pietronero, Strässler, and Grimaldi 1995, §§II–III provides an explicit nonadiabatic vertex analysis.

A material-level validity test

Before invoking Migdal control, report:

1. the band-resolved $E_F$ or curvature scale for every Fermi pocket that matters;
2. the phonon spectrum and the part weighted by $\alpha^2F(\omega)$;
3. the coupling $\lambda$ and whether polarons or lattice instabilities are nearby;
4. the momentum structure $g_{\mathbf q}$, including forward or nesting peaks;
5. density-of-states singularities, narrow bands, and nonadiabatic pockets; and
6. an observable sensitivity test with at least the leading vertex correction or an independent method.

A broad band elsewhere in the material cannot control a shallow pocket that dominates pairing. Likewise, fitting tunnelling data with an Eliashberg kernel does not retrospectively prove the small parameter.

The validity map places this check before the retarded self-consistent equations.

The generic broad-band branch suppresses crossed electron–phonon vertices by an adiabatic ratio. Low density, narrow or flat bands, forward scattering, and strong coupling leave that branch and require explicit vertex tests. Original schematic, not to scale.

The paired-matter claim test matrix records this control separately from evidence for a microscopic mechanism.

Exercise

Compare two pockets. A material has $\omega_{\mathrm{ph}}=25\,\mathrm{meV}$ and $\lambda=1$. One pocket has $E_F=1\,\mathrm{eV}$; another has $E_F=40\,\mathrm{meV}$. Estimate the nominal Migdal parameter.

Solution

The estimates are $0.025$ and $0.625$. The first supports an adiabatic expansion if momentum structure and band regularity are benign. The second does not: an explicit nonadiabatic or vertex-corrected treatment is required. Averaging the two Fermi energies would conceal the uncontrolled shallow pocket.

References

• Migdal, A. B. (1958). “Interaction between electrons and lattice vibrations in a normal metal.” Soviet Physics JETP 7, 996–1001; English translation of ZhETF 34, 1438. Official JETP archive.
• Pietronero, L., Strässler, S., and Grimaldi, C. (1995). “Nonadiabatic superconductivity. I. Vertex corrections for the electron–phonon interactions.” Physical Review B 52, 10516–10529. doi:10.1103/PhysRevB.52.10516.