Maxwell's electrodynamics is the classical limit of QED (quantum electrodynamics). Using Maxwell's equations, the electrostatic (Coulomb) potential of a point charge is obtained as $\Phi \propto \frac{1}{r}$. But, in QED, the Coulomb potential is modified due to the radiative corrections. The QED result for the electrostatic potential of a point charge-like electron (i.e., radiative corrections to Coulomb's law) is found to be as $^1$

$$\Phi (r) = \frac{e}{r}\left( {1 + \frac{\alpha }{{4\sqrt \pi }}\frac{{{e^{ - 2m{ _e}r}}}}{{{{(m{_e}r)}^{3/2}}}}} \right),\qquad({m_e}r \gg 1) \tag 1$$


$$\Phi (r) = \frac{e}{r}\left[ {1 + \frac{{2\alpha }}{{3\pi }}\left( {\ln \left( {\frac{1}{{m{ _e}r}}} \right) - \gamma - \frac{5}{6}} \right)} \right],\qquad({m_e}r \ll 1) \tag 2$$

where $e$ and $m_e$ are the electric charge and the mass of the electron respectively, $\alpha$ is the fine-structure constant and $\gamma$ is the Euler–Mascheroni constant. The second term is responsible for the electric polarization of the vacuum with the order-$\alpha$ vacuum polarization diagram as

enter image description here

On the other hand, Euler and Heisenberg constructed an effective Lagrangian in their famous paper$^2$ for taking into account vacuum polarization to one loop. In the weak field limit, the Euler-Heisenberg (EH) Lagrangian reduces to

$$\mathcal{L}_{\mathrm{eff}} = -\frac14 F_{\mu\nu} F^{\mu\nu}+\frac{\alpha^2}{90 m_e^4}\left[\left(F_{\mu\nu} F^{\mu\nu}\right)^2 +\frac{7}{4}\left(F_{\mu\nu} \tilde{F}^{\mu\nu}\right)^2\right] +\mathcal{O}(m_e^{-6}), \tag 3$$ or equivalently $$\mathcal{L}_{\mathrm{eff}} = \frac12 \left(\mathbf{E}^2 - \mathbf{B}^2\right)+\frac{2\alpha^2}{45 m_e^4}\left[\left(\mathbf{E}^2 - \mathbf{B}^2\right)^2 +7\left(\mathbf{E} \cdot \mathbf{B}\right)^2\right]+\mathcal{O}(m_e^{-6}). \tag 4$$

In this effective field theory, the electrostatic potential is found to be as $$\Phi (r) = \frac{e}{{{r}}} - \frac{{8{\alpha ^2}}}{{225m_e^4}}\frac{{{e^3}}}{{{r^5}}} + O({\alpha ^3}). \tag 5$$

However, I have difficulty interpreting these results.

  1. Correction in the electric potential formula (5) in EH theory is higher order in the fine structure constant ($\alpha$) than the correction in the electric potential formulas (1) and (2) in QED. What does this mean? Do they not describe the same physics? (I know that Maxwell equations in EH theory are no longer linear.)

  2. As far as I know, EH theory as nonlinear electrodynamics gives us the QED $β$-function, the prediction of light-by-light scattering and its cross-section, Schwinger pair creation, etc. Can we claim that the EH theory could describe some results of QED (e.g., quantum mechanical nonlinearity of the fields) without incorporating any quantum mechanical (QED) treatment? To what extent is this claim true?

$^1$ Berestetskii, V. B.; Lifshitz, E. M.; Pitaevskii, L., Quantum Electrodynamics, Section 114 (1980)

$^2$ Heisenberg, W., and H. Euler. "Consequences of Dirac theory of the positron." arXiv preprint physics/0605038 (2006)

  • 1
    $\begingroup$ To be fair, these two cases are expansions in different parameters, aren't they? I mean the QED calculation is not an electron mass expansion. $\endgroup$ Jun 11, 2021 at 21:26
  • 2
    $\begingroup$ Is the Coulomb potential defined the same way in both cases? I can think of at least three reasonable ways to define "Coulomb potential" in QED: (1) consider $\langle\vec E(x)\rangle$ with a background pointlike charge at the origin; (2) consider the ground-state energy vs the distance between two background pointlike charges; (3) consider electron scattering from an external pointlike charge. I don't see why these should all give the same result, even without any further approximations. Defns 1 and 2 can be used in EH theory, but 3 can't. Did B,L,P use defn 3 for the QED case? $\endgroup$ Jun 12, 2021 at 20:16
  • 1
    $\begingroup$ @ChiralAnomaly - Thanks for your attention. Based on your category, It seems B.,L.,P. has used Defn. 1, i.e., starting with renormalizing the photon-self energy (1-loop vacuum polarization diagrams) for the photon propagator and then adding the radiative correction to the electric potential by using the definition of Coulomb law for a point like charge (Do we use the same technical terms?) I think Peskin $ Schroeder has implemented Defn. 2 in your category (but not sure since I'm not familiar with this book). $\endgroup$
    – SG8
    Jun 13, 2021 at 11:24
  • 1
    $\begingroup$ The EH Lagrangian doesn't contain any matter field by definition. It is a theory where you are at energies that are too low to produce any electrons. (1) is how the interaction between two electron is changed by QED effects. (5) has nothing to do with how to electrons interact and just describes how photons interact among themselves. Where did you find (5)? $\endgroup$
    – FrodCube
    Jun 15, 2021 at 13:25
  • 1
    $\begingroup$ I thought I knew how to answer but thinking about it I realized it’s not that obvious. I found a comment about this at the end of page 8 here but I am not sure I really understand it. I guess I’ll wait for somebody else to answer. $\endgroup$
    – FrodCube
    Jun 17, 2021 at 20:40

1 Answer 1


EH Lagrangian is derived in the approximation of constant (or slowly-varying) electromagnetic fields. It is an effective Lagrangian only containing E and B fields, but not their derivatives.

Even if you consider external point charges (and not the electrons themselves, which you have integrated out,) it would be wrong to derive the corrected Coulomb potential from this Lagrangian.

One would have to include all terms in the effective action containing derivatives of E and B, to obtain the fully corrected Coulomb potential. This is unrealistically hard, but of course to the lowest order, only a few additional higher-derivative terms would be enough.

What you observed is that the EH Lagrangian by itself is not enough. In other words, the answer (1), (2) is more correct. It is derived using different techniques, for example, by computing the loop corrections to the photon two-point function, as you noted. Such a computation usually bypasses explicitly writing the effective action. However, if you were to do that, you would discover that the leading $O(\alpha)$ effect is not captured by the EH Lagrangian, which as you noticed starts contributing only at the sub-leading order.

  • $\begingroup$ Thanks for this answer. I should ponder more on what you wrote but I think it makes sense. $\endgroup$
    – SG8
    Dec 18, 2023 at 15:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.