Tree-level calculations give leading-order predictions in powers of the coupling constant$\alpha = e^2/(4\pi) \approx 1/137$. But quantum corrections arise from loop diagrams in which virtual particles circulate in closed loops. These corrections are suppressed by additional powers of $\alpha$, yet they produce measurable effects and are essential for precision physics.

At one-loop order in QED, there are exactly three primitive divergent diagrams:

Vertex correction $\delta\Gamma^\mu$: a virtual photon connecting the two fermion legs at the $e\bar\psi\gamma^\mu\psi A_\mu$ vertex
Vacuum polarization $\Pi^{\mu\nu}(q)$: a virtual $e^+e^-$ loop inserted into the photon propagator
Electron self-energy $\Sigma(p)$: a virtual photon emitted and reabsorbed by the electron

By the power-counting theorem, these three diagrams contain all the ultraviolet (UV) divergences of QED at one loop. Every other one-loop diagram is UV-finite. This is the hallmark of a renormalizable theory.

💡Why Only Three Divergent Diagrams?

The superficial degree of divergence of a QED diagram depends only on the number of external fermion lines $E_f$ and external photon lines $E_\gamma$:

D = 4 - (3/2)E_f - E_\gamma

Only diagrams with D $\geq$ 0 can diverge. With the constraint that QED conserves charge and has Furry's theorem (odd-photon vertices vanish), only three structures remain: the vertex (D = 0), vacuum polarization (D = 2), and self-energy (D = 1).

6.2 UV Divergences and the Need for Regularization

Loop integrals involve integrating over all possible momenta of virtual particles. At high momenta (the ultraviolet regime), these integrals generically diverge. Consider the generic one-loop structure:

$$\int \frac{d^4k}{(2\pi)^4} \frac{N(k)}{(k^2 - m^2 + i\epsilon)((k-p)^2 - m^2 + i\epsilon)} \sim \int_0^\Lambda \frac{k^3 \, dk}{k^{4-n}} \to \infty$$

where the power of divergence depends on the numerator structure. To make sense of these integrals, we must regularize them — introduce a mathematical device that renders the integral finite while preserving gauge invariance and Lorentz symmetry.

The most powerful regularization scheme is dimensional regularization, where we analytically continue from 4 to $d = 4 - \varepsilon$ spacetime dimensions. Divergences then appear as poles in $1/\varepsilon$:

$$\int \frac{d^dk}{(2\pi)^d} \frac{1}{(k^2 - \Delta)^n} = \frac{i(-1)^n}{(4\pi)^{d/2}} \frac{\Gamma(n - d/2)}{\Gamma(n)} \left(\frac{1}{\Delta}\right)^{n-d/2}$$

The Gamma function $\Gamma(n - d/2)$ has poles when $n - d/2$ is a non-positive integer, which is precisely where UV divergences lurk. For $d = 4 - \varepsilon$:

$$\Gamma\!\left(\frac{\varepsilon}{2}\right) = \frac{2}{\varepsilon} - \gamma_E + O(\varepsilon), \qquad \gamma_E \approx 0.5772$$

💡Why Dimensional Regularization?

A hard momentum cutoff $\Lambda$ breaks gauge invariance — it treats some directions in momentum space differently from others. Dimensional regularization preserves all symmetries of QED (gauge invariance, Lorentz invariance) because the Lagrangian has the same form in any dimension. The price is that we must work with $d$-dimensional integrals and Dirac algebra, which introduces the arbitrary mass scale $\mu$.

6.3 The Renormalization Program

The key insight of renormalization is that we never measure the "bare" parameters$m_0$, $e_0$ of the Lagrangian. We measure physical (renormalized) quantities. The bare Lagrangian is rewritten by splitting fields and parameters:

$$\psi_0 = \sqrt{Z_2}\,\psi_R, \quad A_0^\mu = \sqrt{Z_3}\,A_R^\mu, \quad m_0 = m + \delta m, \quad e_0 = Z_e \, e$$

The renormalization constants $Z_2$, $Z_3$, $Z_e$, and $\delta m$ are chosen to absorb all UV divergences. This introduces counterterms into the Lagrangian:

$$\mathcal{L}_{\text{QED}} = \mathcal{L}_{\text{renormalized}} + \mathcal{L}_{\text{counterterms}}$$

where the counterterm Lagrangian is:

$$\mathcal{L}_{\text{ct}} = (Z_2 - 1)\bar\psi_R(i\!\not\!\partial - m)\psi_R - \delta m \, Z_2 \bar\psi_R\psi_R - \tfrac{1}{4}(Z_3 - 1)F_{\mu\nu}F^{\mu\nu} - (Z_1 - 1)e\bar\psi_R\gamma^\mu\psi_R A_\mu$$

The counterterms generate new Feynman rules (denoted by crosses on diagrams) that exactly cancel the UV-divergent parts of loop integrals, leaving finite physical predictions.

Renormalization Conditions

To fix the finite parts of the counterterms, we impose on-shell renormalization conditions:

The electron propagator has a pole at $\not\!p = m$ with residue 1
The photon propagator has a pole at $q^2 = 0$ with residue 1
The $e\bar\psi\gamma^\mu\psi A_\mu$ vertex equals $-ie\gamma^\mu$ when all particles are on-shell at zero momentum transfer

6.4 The One-Loop Vertex Correction

The vertex correction modifies the $e\bar\psi\gamma^\mu\psi A_\mu$ interaction. At one loop, a virtual photon connects the incoming and outgoing electron lines at the vertex. The corrected vertex function is:

$$-ie\Gamma^\mu(p', p) = -ie\gamma^\mu + (-ie)^3 \int \frac{d^4k}{(2\pi)^4} \frac{\gamma^\nu (\not\!k + \not\!p' + m)\gamma^\mu (\not\!k + \not\!p + m)\gamma_\nu}{[(k+p')^2 - m^2][(k+p)^2 - m^2][k^2 - \mu_\gamma^2]}$$

where $\mu_\gamma$ is a small fictitious photon mass introduced to regulate infrared (IR) divergences. We use Feynman parametrization to combine the three denominators:

$$\frac{1}{ABC} = 2\int_0^1 dx\,dy\,dz\;\delta(x+y+z-1)\;\frac{1}{[xA + yB + zC]^3}$$

After shifting the loop momentum $k \to \ell = k + xp' + yp$ and performing the$d$-dimensional integral, the result involves the Dirac structure:

$$\Gamma^\mu(p', p) = \gamma^\mu F_1(q^2) + \frac{i\sigma^{\mu\nu}q_\nu}{2m} F_2(q^2)$$

where $q = p' - p$ is the momentum transfer and $\sigma^{\mu\nu} = \frac{i}{2}[\gamma^\mu, \gamma^\nu]$. The two form factors are:

$F_1(q^2)$: the Dirac form factor, with $F_1(0) = 1$ (enforced by the Ward identity / renormalization condition)
$F_2(q^2)$: the Pauli form factor, which is UV-finite and gives the anomalous magnetic moment

The UV-divergent piece resides entirely in $F_1$ and is canceled by the vertex counterterm $\delta Z_1$. The finite remainder of $F_1$ depends on $q^2$and contributes to charge form factor measurements.

💡Vertex Correction and the Magnetic Moment

The tree-level vertex $\gamma^\mu$ predicts $g = 2$ for the electron (Dirac's result). The one-loop correction generates $F_2(0) \neq 0$, shifting $g$ away from 2. Schwinger's famous result for the leading correction is:

$F_2(0) = \frac{\alpha}{2\pi} \approx 0.00116$

This was the first triumph of radiative corrections — a purely quantum effect with no classical analogue, confirmed by experiment to extraordinary precision.

6.5 Schwinger's Calculation of $F_2(0)$

The Pauli form factor $F_2(q^2)$ is UV-finite, so we can compute it directly in$d = 4$. After Feynman parametrization and evaluation of the Dirac algebra in the numerator, the $\sigma^{\mu\nu}q_\nu$ piece yields:

$$F_2(q^2) = \frac{\alpha}{2\pi}\int_0^1 dx\,dy\,dz\;\delta(x+y+z-1)\;\frac{2m^2 z(1-z)}{m^2(1-z)^2 - q^2 xy - \mu_\gamma^2 z}$$

At zero momentum transfer $q^2 = 0$ and taking $\mu_\gamma \to 0$(the IR regulator drops out of $F_2$):

$$F_2(0) = \frac{\alpha}{2\pi}\int_0^1 dx\,dy\,dz\;\delta(x+y+z-1)\;\frac{2z}{1-z}$$

Performing the $x$ integral (using the delta function to set $x = 1 - y - z$), then integrating over $y$ from 0 to $1-z$, and finally over $z$:

$$F_2(0) = \frac{\alpha}{2\pi} \int_0^1 dz\;\frac{2z(1-z)}{1-z} = \frac{\alpha}{2\pi}\int_0^1 dz\;2z = \frac{\alpha}{2\pi}$$

This gives the anomalous magnetic moment:

$$\boxed{a_e \equiv \frac{g-2}{2} = F_2(0) = \frac{\alpha}{2\pi} \approx 0.0011614}$$

This is Schwinger's 1948 result. It agreed with experiment immediately, providing the first concrete evidence that the renormalization program of QED is physically correct.

Key Concepts (Page 1)

• Loop diagrams give quantum corrections suppressed by powers of $\alpha$
• QED has three primitive one-loop divergences: vertex, vacuum polarization, self-energy
• Dimensional regularization ($d = 4 - \varepsilon$) preserves gauge invariance
• Renormalization absorbs UV divergences into redefinitions of $m$, $e$, and field normalization
• The vertex function decomposes into Dirac ($F_1$) and Pauli ($F_2$) form factors
• $F_2(0) = \alpha/(2\pi)$ is Schwinger's anomalous magnetic moment result

← QED

Page 1 of 4

Chapter 6: Radiative Corrections

Vacuum Polarization →

Quantum Field Theory Course

Radiative Corrections & Renormalization

🔗Course Connections

📚Prerequisites

Video Lecture

6.1 Beyond Tree Level: Loop Diagrams