The Abelian Higgs Mechanism

Derivation

The Abelian Higgs Mechanism

D-383 Home PU-402 Threads symmetry · force · fields · matter Depends on Goldstone's Theorem, Local U(1) Gauge Invariance and Minimal Coupling, Proca Field and Its Degrees of Freedom

Statement

For a complex scalar field \(\phi\) with a spontaneously broken global \(U(1)\) symmetry, promoting that symmetry to a local (gauged) \(U(1)\) with gauge field \(A_\mu\) removes the massless Goldstone boson from the physical spectrum: it is absorbed as the longitudinal polarisation of \(A_\mu\), which thereby acquires a mass \(m_A = e v\), where \(v\) is the vacuum expectation value and \(e\) the gauge coupling. No physical degree of freedom is created or destroyed — the two real scalar components plus two transverse photon states rearrange into one massive scalar (the Higgs) plus three massive vector states.

Why it matters

The Abelian Higgs model is the minimal, fully soluble prototype of mass generation without breaking gauge invariance. It resolves the apparent paradox that gauge invariance forbids a bare mass term \(m_A^2 A_\mu A^\mu\), yet the electroweak \(W\) and \(Z\) bosons are massive: the mass arises dynamically from the vacuum structure of a scalar sector, not from an explicit term in the Lagrangian.

It is also the field-theory face of a laboratory phenomenon: inside a superconductor the photon becomes massive (the Meissner effect and finite penetration depth), which is precisely this mechanism with the Cooper-pair condensate playing the role of \(\phi\). The same two lines of algebra thus underpin both the Standard Model and macroscopic superconductivity.

Assumptions

The scalar potential has a degenerate circle of minima at \(|\phi| = v/\sqrt{2} \neq 0\).If the minimum were at \(\phi = 0\) the symmetry would be unbroken, \(A_\mu\) would stay massless, and there would be no Goldstone mode to absorb.

The gauge coupling \(e\) is nonzero.At \(e = 0\) the symmetry is only global; the Goldstone boson remains a physical massless particle (Goldstone's theorem) and the photon stays massless — no mixing occurs.

The theory is quantised in a gauge where the Goldstone field is removed (unitary gauge), or equivalently one tracks the \(R_\xi\) gauge bookkeeping.If one works in a fixed covariant gauge without care, the would-be Goldstone reappears as an unphysical pole and a spurious ghost; degree-of-freedom counting only closes once gauge redundancy is fixed.

Spacetime is flat and translationally invariant, so a constant \(\phi = v/\sqrt{2}\) is an exact stationary configuration.On a curved or finite background the condensate becomes position dependent, the vector mass becomes a local quantity, and the clean plane-wave mode counting is modified.

Derivation

\[ \mathcal{L} = (D_\mu \phi)^{*}(D^\mu \phi) - V(\phi) - \tfrac{1}{4} F_{\mu\nu}F^{\mu\nu}, \qquad D_\mu = \partial_\mu + i e A_\mu \]

Start from the gauge-invariant Lagrangian: the covariant derivative \(D_\mu\) makes \(\mathcal{L}\) invariant under \(\phi \to e^{i\alpha(x)}\phi\), \(A_\mu \to A_\mu - \tfrac{1}{e}\partial_\mu\alpha\). A

\[ V(\phi) = -\mu^2\,\phi^{*}\phi + \lambda\,(\phi^{*}\phi)^2, \qquad \mu^2 > 0,\ \lambda > 0 \]

Choose the Mexican-hat potential. With \(\mu^2>0\) the origin is a maximum, forcing a nonzero minimum. A

\[ \frac{\partial V}{\partial (\phi^{*}\phi)} = 0 \ \Rightarrow\ |\phi|^2 = \frac{\mu^2}{2\lambda} \equiv \frac{v^2}{2}, \qquad v = \sqrt{\frac{\mu^2}{\lambda}} \]

Minimise \(V\): stationarity fixes the modulus of \(\phi\) on a circle of radius \(v/\sqrt{2}\); the phase is undetermined, which is the flat Goldstone direction. A

\[ \phi(x) = \frac{1}{\sqrt{2}}\big(v + h(x)\big)\,e^{\,i\,\theta(x)/v} \]

Parametrise fluctuations in polar form: \(h\) is the radial (Higgs) mode, \(\theta\) the angular (would-be Goldstone) mode. This separates the massive and massless directions cleanly. B

\[ \phi \to e^{-i\theta(x)/v}\phi = \frac{1}{\sqrt{2}}\big(v + h(x)\big), \qquad A_\mu \to A_\mu' = A_\mu + \frac{1}{e v}\partial_\mu\theta \]

Perform a gauge transformation with \(\alpha(x) = -\theta(x)/v\) — unitary gauge. Because \(\theta\) is pure gauge, it can be rotated away entirely; it does not disappear, it is shifted into \(A_\mu'\). C

\[ D_\mu \phi = \frac{1}{\sqrt{2}}\Big(\partial_\mu h + i e (v + h)\,A_\mu'\Big) \]

Substitute the unitary-gauge field into \(D_\mu\phi\). The angular field is gone; \(A_\mu'\) now couples directly to the constant \(v\). B

\[ (D_\mu\phi)^{*}(D^\mu\phi) = \tfrac{1}{2}\partial_\mu h\,\partial^\mu h + \tfrac{1}{2}e^2 (v+h)^2 A_\mu' A'^\mu \]

Expand the modulus squared; cross terms \(\propto \partial_\mu h\, A'^\mu (v+h)\) vanish because \(\partial_\mu h\) is real and the \(iA_\mu'\) piece is imaginary, so the product is a total-derivative-free real combination with no linear mixing. B

\[ \tfrac{1}{2}e^2 (v+h)^2 A_\mu' A'^\mu = \underbrace{\tfrac{1}{2}e^2 v^2\, A_\mu' A'^\mu}_{\text{mass term}} + e^2 v\, h\, A_\mu' A'^\mu + \tfrac{1}{2}e^2 h^2 A_\mu' A'^\mu \]

Expand \((v+h)^2 = v^2 + 2vh + h^2\). The \(v^2\) piece is quadratic in \(A'\) with no derivatives — a genuine gauge-boson mass term; the rest are Higgs–photon interactions. A

\[ \tfrac{1}{2}m_A^2\, A_\mu' A'^\mu = \tfrac{1}{2}e^2 v^2\, A_\mu' A'^\mu \ \Rightarrow\ \boxed{\,m_A = e v\,} \]

Read off the mass by matching to the Proca form \(\tfrac{1}{2}m_A^2 A_\mu A^\mu\). The gauge field is now massive. A

\[ V = \text{const} + \mu^2 h^2 + \dots \ \Rightarrow\ m_h^2 = 2\mu^2 = 2\lambda v^2 \]

Expand \(V\) about the minimum to second order in \(h\): the radial curvature gives the physical Higgs mass. The angular curvature is zero — that flat direction became the longitudinal photon. B

\[ \underbrace{2}_{\text{transverse }A} + \underbrace{2}_{\text{complex }\phi} \;=\; \underbrace{3}_{\text{massive vector}} + \underbrace{1}_{\text{Higgs }h} \]

Count degrees of freedom before and after. A massless vector carries 2 polarisations, a massive (Proca) vector carries 3; the missing longitudinal state is exactly the one real scalar removed from \(\phi\). Total is conserved: \(4=4\). C

Result

\[ m_A = e v, \qquad m_h = \sqrt{2\lambda}\,v, \qquad 2+2 = 3+1 \]

Reading. Gauging a spontaneously broken \(U(1)\) gives the gauge boson a mass equal to the coupling times the vacuum expectation value. The would-be Goldstone boson is not a physical particle: it has been "eaten" to supply the longitudinal polarisation that a massive vector must carry. The surviving physical scalar is the radial Higgs excitation, with mass set by the potential's curvature. Degrees of freedom are exactly conserved.

Units check. In natural units \([e]\) is dimensionless and \([v] = \text{mass}\), so \([m_A] = [e][v] = \text{mass}\) — correct. Likewise \([\lambda]\) is dimensionless in four dimensions, giving \([m_h] = [v] = \text{mass}\). Restoring \(\hbar, c\): \(v\) has units of energy, and \(m_A c^2 = e\,v\) is an energy, so \(m_A = e v / c^2\) is a mass.

Limiting cases

\(e \to 0\): \(m_A \to 0\), the photon stays massless and \(\theta\) re-emerges as a genuine massless Goldstone boson — global spontaneous breaking, Goldstone's theorem restored.
\(v \to 0\) (unbroken phase, \(\mu^2 \to 0^+\)): \(m_A \to 0\) and \(m_h \to 0\); symmetry is restored and both fields are massless — the mechanism switches off.
\(\lambda \to \infty\) at fixed \(v\): \(m_h \to \infty\), the Higgs decouples and one recovers a nonlinear sigma model / Stückelberg description of a massive photon.
Superconductor analogue: \(m_A^{-1}\) equals the London penetration depth \(\lambda_L\); a large condensate \(v\) means a short penetration depth and strong Meissner screening.

Breaks when

Non-Abelian gauge group without a suitable representation. The counting \(2+2=3+1\) is specific to \(U(1)\). For \(SU(2)\times U(1)\) some generators may remain unbroken (leaving a massless photon), and the number of eaten Goldstones equals the number of broken generators — the naive Abelian bookkeeping fails.
Lower dimensions with strong fluctuations. In \(d \le 2\) the Coleman–Mermin–Wagner theorem forbids a genuine continuous-symmetry-breaking vacuum expectation value; long-wavelength phase fluctuations destroy the ordered state, so a sharp \(v\) and hence a well-defined \(m_A = ev\) do not exist.
Explicit symmetry breaking present. If the potential contains a term that tilts the Mexican hat (e.g. a linear \(\phi + \phi^{*}\) piece), the angular mode acquires a mass and is a pseudo-Goldstone; it is no longer a pure gauge direction and cannot be cleanly gauged away.
Finite temperature above the critical point. For \(T > T_c\) thermal corrections restore the symmetry, \(v(T) \to 0\), and the gauge boson mass melts — the electroweak phase transition and superconducting \(T_c\) are exactly this.

Failure modes

"The Goldstone boson is destroyed." It is not annihilated; it is re-labelled. Its single real degree of freedom becomes the longitudinal polarisation of \(A_\mu\). Counting is conserved, never violated.
Adding a bare mass term \(m^2 A_\mu A^\mu\) "by hand." This breaks gauge invariance explicitly and spoils renormalisability and unitarity at high energy. The whole point is that the mass emerges from \(v\), not from an inserted term.
Confusing \(v\) with \(\langle h \rangle\). After shifting, \(\langle h \rangle = 0\); the vacuum expectation value lives in the constant \(v\), not in the fluctuation field. Writing a Higgs propagator with a nonzero one-point function is a sign of this error.
Using \(m_A = \tfrac{1}{2}ev\) or \(m_A^2 = e^2v^2/2\). The factor comes from matching \(\tfrac{1}{2}e^2v^2 A^2\) to \(\tfrac{1}{2}m_A^2 A^2\), giving \(m_A^2 = e^2 v^2\), i.e. \(m_A = ev\) with no extra \(\tfrac{1}{2}\).
Forgetting the transverse-longitudinal split. Claiming a massless vector has 3 polarisations, or a massive one has 2. The jump from 2 to 3 is the entire content of the mechanism.

Discussion

The deepest point is that gauge symmetry is never actually "broken" here. A local symmetry is a redundancy of description, not a physical symmetry, and Elitzur's theorem guarantees it cannot be spontaneously broken. What happens is that the gauge is fixed by the vacuum choice, and the redundant phase field \(\theta\) — which was pure gauge all along — is repackaged into the physical longitudinal photon. The phrase "spontaneous gauge-symmetry breaking" is a convenient abuse; the rigorous statement is that a gauge-invariant condensate \(\langle\phi^{*}\phi\rangle \neq 0\) endows the vector with a mass.

The mechanism is smoothly connected to the massive vector field one could write down via the Stückelberg trick: a Proca field plus a compensating scalar. In the limit \(m_h \to \infty\) the Abelian Higgs model reduces exactly to Stückelberg's theory, showing that the Higgs is the "ultraviolet completion" that keeps the massive photon unitary and renormalisable up to arbitrary energies. Without it, longitudinal-vector scattering amplitudes grow with energy and violate unitarity around \(4\pi v\) — this is the theoretical bound that guaranteed something Higgs-like had to appear at the electroweak scale.

In the \(R_\xi\) gauges the would-be Goldstone \(\theta\) is not removed but retained with a gauge-dependent mass \(\sqrt{\xi}\,m_A\), accompanied by Faddeev–Popov ghosts of the same mass. Physical S-matrix elements are \(\xi\)-independent: the Goldstone pole, the ghost pole, and the unphysical longitudinal-vector pole cancel in every gauge-invariant observable. This machinery is what makes the massive-vector theory perturbatively renormalisable (the 't Hooft–Veltman result), a fact that is completely obscured in unitary gauge where the propagator \(\sim (g_{\mu\nu} - k_\mu k_\nu/m_A^2)\) grows at large \(k\) and hides the good high-energy behaviour.

Common misconceptions. The Higgs field does not "give mass to everything" through this mechanism alone — fermion masses arise from separate Yukawa couplings, and most of the proton's mass is QCD binding energy, not Higgs. The mechanism described here specifically explains gauge-boson mass and the fate of the Goldstone boson, nothing more.

Worked examples

Example A — Gauge-boson mass from a given vacuum expectation value and coupling. A toy Abelian Higgs model has \(v = 246\ \text{GeV}\) (the electroweak scale) and gauge coupling \(e = 0.35\). Find \(m_A\).

\[ m_A = e v \]

Symbolic result from step 9. A

\[ m_A = 0.35 \times 246\ \text{GeV} = 86\ \text{GeV} \]

Insert numbers; the coupling is dimensionless so the answer carries the units of \(v\). A

\[ m_A \approx 86\ \text{GeV} \]

Reading. This is deliberately close to the physical \(W\) mass (\(80.4\ \text{GeV}\)): with the true \(W\)–\(Z\) coupling structure the same arithmetic reproduces the electroweak boson masses. The mass tracks \(v\) linearly, so the boson mass and the vacuum scale are locked together.

Example B — Photon mass and penetration depth in a superconductor. In a superconductor the effective condensate gives a photon mass corresponding to a London penetration depth \(\lambda_L = 50\ \text{nm}\). Find the photon rest energy \(m_A c^2\) and comment on the required condensate scale.

\[ m_A = \frac{\hbar}{c\,\lambda_L} \quad\Longleftrightarrow\quad \lambda_L = \frac{1}{m_A} \ (\text{natural units}) \]

The massive photon's Compton wavelength is the penetration depth; this is the mechanism's real-world identity. B

\[ m_A c^2 = \frac{\hbar c}{\lambda_L} = \frac{197.3\ \text{eV·nm}}{50\ \text{nm}} \]

Use \(\hbar c = 197.3\ \text{eV·nm}\); the length in the denominator, energy comes out. B

\[ m_A c^2 = 3.95\ \text{eV} \]

Divide. A

\[ m_A c^2 \approx 3.9\ \text{eV} \]

Reading. Inside the superconductor the photon behaves as a particle of rest energy \(\sim 4\ \text{eV}\), which is why static magnetic fields are screened over \(\sim 50\ \text{nm}\) — the Meissner effect is the Abelian Higgs mechanism in a solid. The corresponding condensate scale \(v = m_A/e\) is set by the Cooper-pair density.

Problems

Starting from \(V(\phi) = -\mu^2|\phi|^2 + \lambda|\phi|^4\), derive \(v^2 = \mu^2/\lambda\) and show \(m_h^2 = 2\mu^2\).
Solution
Set \(u=|\phi|^2\); \(dV/du = -\mu^2 + 2\lambda u = 0 \Rightarrow u = \mu^2/2\lambda\), so \(|\phi|^2 = v^2/2\) with \(v^2 = \mu^2/\lambda\). Writing \(\phi = (v+h)/\sqrt2\) (real gauge), \(V = -\tfrac{\mu^2}{2}(v+h)^2 + \tfrac{\lambda}{4}(v+h)^4\). The coefficient of \(\tfrac12 h^2\) is \(V''\) at \(h=0\): \(V'' = -\mu^2 + 3\lambda v^2 = -\mu^2 + 3\mu^2 = 2\mu^2\). Hence \(m_h^2 = 2\mu^2 = 2\lambda v^2\).
Verify by explicit polarisation counting that a massless vector in 4D has 2 physical states and a massive (Proca) vector has 3, using the little-group argument.
Solution
A massive particle can be boosted to rest; its little group is the rotation group \(SO(3)\), and a spin-1 state fills the \(j=1\) representation with \(2j+1 = 3\) states (\(m_j = -1,0,+1\)). A massless particle has no rest frame; its little group is \(ISO(2)\), whose finite-dimensional physical representations are labelled by helicity, giving only the two \(h = \pm 1\) states (the longitudinal \(h=0\) is projected out by gauge invariance). The difference \(3-2=1\) is the eaten Goldstone.
Show that in unitary gauge no term linear in the fluctuation fields \(h\) or \(A_\mu'\) survives, confirming that \(v\) is a true stationary point.
Solution
The potential expansion has no linear \(h\) term because \(V'(v)=0\) by the minimisation condition (Problem 1). The kinetic term gives \(\tfrac12 e^2(v+h)^2 A'^2\); the piece linear in fields would need one field only, but every term here has \(\ge 2\) fields (\(v^2 A'^2\) is quadratic in \(A'\), \(2vh A'^2\) is cubic, etc.). The cross term \(\partial_\mu h\,A'^\mu\) is absent because \((D_\mu\phi)^*(D^\mu\phi)\) yields \(\tfrac12(\partial h)^2 + \tfrac12 e^2(v+h)^2 A'^2\) with the mixed real×imaginary product cancelling. Hence no tadpoles; \(v\) is stationary.
A gauge boson is measured at \(m_A = 91.2\ \text{GeV}\) (the \(Z\)) with an effective coupling \(g_{\text{eff}} = 0.74\). Estimate the vacuum expectation value \(v\) and compare with \(246\ \text{GeV}\).
Solution
From \(m_A = g_{\text{eff}} v\), \(v = m_A/g_{\text{eff}} = 91.2/0.74 = 123\ \text{GeV}\). This is half the electroweak \(v = 246\ \text{GeV}\); the factor of 2 reflects the physical \(Z\) mass being \(m_Z = \tfrac12\sqrt{g^2+g'^2}\,v\), i.e. the correct combination of couplings differs from a single effective \(g\). The estimate is the right order of magnitude and illustrates \(v \sim m_A/\text{coupling}\).
In the \(R_\xi\) gauge the would-be Goldstone has mass \(\sqrt{\xi}\,m_A\). For \(m_A = 80.4\ \text{GeV}\), find the Goldstone mass in Feynman gauge (\(\xi=1\)) and Landau gauge (\(\xi=0\)), and state why it is unobservable.
Solution
Goldstone mass \(= \sqrt{\xi}\,m_A\). Feynman gauge \(\xi=1\): mass \(= 80.4\ \text{GeV}\), equal to \(m_A\). Landau gauge \(\xi=0\): mass \(=0\), a massless would-be Goldstone. The value is gauge-dependent, which immediately signals it is not a physical particle: in every gauge-invariant S-matrix element the Goldstone pole is exactly cancelled by the unphysical longitudinal-vector and Faddeev–Popov ghost contributions, so no measurement can isolate it. Only \(m_A\) and \(m_h\) are gauge-independent observables.