Lamb Shift

Definition

The Lamb shift is a small energy splitting between two hydrogen levels that the best pre-1947 theory said should be identical. Specifically, the 2s₁⁄₂ state lies about 1057 MHz higher than the 2p₁⁄₂ state, even though Dirac's relativistic equation puts them at exactly the same energy.

That "1057 MHz" is an energy expressed as a frequency through E = hf. In other units it is roughly 4.4 microelectronvolts — about a millionth of the energy that binds the electron in the n = 2 shell. It is a whisper. And explaining that whisper rebuilt physics.

The cause is the quantum electrodynamic (QED) vacuum. Empty space is not empty — it seethes with fluctuating electromagnetic fields and virtual particles. The electron interacts with this churn, and the interaction nudges its energy. Because an s-electron and a p-electron sample the vacuum differently near the nucleus, they are nudged by different amounts, and the degeneracy breaks.

How it works

Start with what Dirac theory says. Solve the Dirac equation for an electron in the hydrogen Coulomb potential, and the energy levels come out depending only on the principal quantum number n and the total angular momentum j — never on the orbital angular momentum l by itself. The 2s₁⁄₂ state has n = 2, l = 0, j = 1⁄2. The 2p₁⁄₂ state has n = 2, l = 1, j = 1⁄2. Same n, same j. Dirac says: same energy. Degenerate.

Now turn on the vacuum. In QED the electromagnetic field has a ground state that is not quiet — it has zero-point fluctuations in every mode. The bound electron is constantly buffeted by these fluctuating fields, so its position does a tiny random jitter, sometimes called Zitterbewegung-flavored smearing. Instead of sitting at a sharp point, the electron is effectively spread over a small fuzzy ball.

A smeared charge feels an averaged potential. Near the nucleus the Coulomb potential −e²/r is sharply peaked, so averaging it over the jitter weakens the binding right where it is strongest. The effect is largest exactly at the nucleus — and only s-states have a nonzero wavefunction amplitude there. The 2s electron, with finite density at r = 0, gets shifted upward; the 2p electron, whose density vanishes at the nucleus, barely notices. The levels split.

The full QED accounting has three named pieces:

• Electron self-energy. The electron emits and reabsorbs virtual photons (it interacts with its own field). This is the dominant term and pushes the 2s level up by the bulk of the shift.
• Vacuum polarization (Uehling term). A virtual electron–positron pair briefly screens the nucleus, modifying the potential at very short range. This contributes a small piece of the opposite sign for s-states.
• Anomalous magnetic moment. The same vacuum coupling that gives the electron its g − 2 also feeds a small spin-orbit-like contribution.

Add them up with the right signs and the answer is ≈ 1057 MHz — exactly what Lamb and Retherford measured.

A worked example — Bethe's back-of-the-envelope

In June 1947, on a train home from the Shelter Island conference, Hans Bethe did the first estimate. Here is the spirit of it with concrete numbers.

The mean-square jitter of the electron driven by vacuum fluctuations scales as:

⟨(Δr)²⟩  ≈  (2α/π) · (ℏ/mc)² · ln(K_max / K_min)

where α ≈ 1/137 is the fine-structure constant, ℏ/mc ≈ 3.9 × 10⁻¹³ m is the electron's reduced Compton wavelength, and the logarithm runs over the range of photon energies that matter. The smearing changes the energy of an s-state by an amount proportional to ⟨(Δr)²⟩ times the curvature of the potential at the nucleus:

ΔE  ≈  (1/6) ⟨(Δr)²⟩ · ⟨∇²V⟩  =  (1/6) ⟨(Δr)²⟩ · 4π e² |ψ(0)|²

For the 2s state of hydrogen, |ψ(0)|² = 1/(8πa₀³) with a₀ the Bohr radius. The naive integral over photon energies diverges at the high end (K_max → ∞), which is the famous infinity. Bethe's move: cut it off at the electron rest energy mc² ≈ 511 keV, because above that the non-relativistic treatment is meaningless and the divergence is really part of the electron's observed mass. With that cutoff the logarithm becomes a finite number of order 7–8, and the arithmetic yields:

ΔE(2s − 2p)  ≈  1040 MHz   (Bethe, 1947, non-relativistic)
ΔE(2s − 2p)  ≈  1057 MHz   (full QED + experiment)

Getting 1040 out of 1057 from a train-ride estimate is one of the great calculations in physics. It proved the vacuum was the answer, and it showed that the infinity was tameable.

The numbers in context

Effect in hydrogen	Origin	Typical size (frequency)	Described by
Gross structure (n levels)	Coulomb binding	~10¹⁵ Hz (eV scale)	Bohr / Schrödinger
Fine structure (spin-orbit)	Relativity + spin	~10 GHz (α² of gross)	Dirac equation
Lamb shift (2s–2p)	QED vacuum fluctuations	≈ 1057 MHz	QED radiative correction
Hyperfine (21 cm line)	Nuclear spin moment	1420 MHz	Magnetic dipole coupling
Electron g − 2	QED vacuum (same family)	shifts moment by ~0.1%	QED radiative correction
2s natural linewidth	Metastable lifetime ~0.12 s	~1 Hz (very narrow)	Two-photon decay

The Lamb shift sits in a beautiful gap: smaller than fine structure (which Dirac already explained), comparable to but distinct from hyperfine structure (which comes from the proton, not the vacuum), and produced by the same vacuum coupling that gives the electron its anomalous magnetic moment.

How Lamb and Retherford actually measured it

The genius of the 1947 experiment was avoiding optical spectroscopy entirely. Optically, the 2s–2p gap would be buried in the Doppler width of an emission line. Instead they used a microwave/radio-frequency resonance method born from wartime radar.

• Produce a beam of hydrogen atoms and excite some to the metastable 2s₁⁄₂ state, which lives ~0.12 s — astronomically long for an excited atom, because it cannot decay to 1s by a single dipole photon.
• Detect those metastable atoms downstream by the electrons they eject from a metal target.
• Bathe the beam in tunable microwaves. When the frequency matches the 2s → 2p transition (the Lamb shift, ~1 GHz), atoms are driven into the short-lived 2p state, which decays to 1s almost instantly. The metastable beam is depleted, and the detector signal drops.
• The microwave frequency at the dip is the Lamb shift. Reading it off gave ≈ 1000 MHz immediately, refined to ~1057 MHz.

Measuring an energy difference directly as a microwave frequency is exquisitely precise — far better than subtracting two optical lines. That precision is what made the small disagreement with Dirac theory undeniable.

Why this gave us renormalization

The self-energy integral diverges. Sum the electron's interaction with all vacuum photon modes and you get infinity, because there is no upper limit on photon energy. Before 1947 this was treated as an embarrassing flaw in quantum field theory. The Lamb shift turned it into a method.

Bethe's argument, sharpened by Kramers and then made rigorous by Schwinger, Feynman, Tomonaga, and Dyson, runs like this. A free electron also has a divergent self-energy — but we never see it separately, because that infinity is already baked into the mass we measure on the scale. The number m in our equations is the "bare" mass; the physical mass is bare mass plus the (infinite) self-energy. So when we compute the energy of a bound electron, we should subtract off the free-electron self-energy, keeping only the difference. The infinities cancel, and the leftover is finite — and it is the Lamb shift.

This subtraction — absorbing divergences into the definitions of measured quantities (mass, charge) — is renormalization. The Lamb shift was its first triumphant test: a finite prediction where there had been an infinity, agreeing with experiment. Everything in the Standard Model that we can calculate, from g − 2 to particle scattering at the LHC, rests on the renormalization program the Lamb shift validated.

Variants and regimes

• Heavier hydrogen-like ions. The Lamb shift scales steeply with nuclear charge Z (roughly as Z⁴ for the leading self-energy, even faster for higher-order terms). In hydrogen-like uranium (Z = 92) the 1s Lamb shift reaches ~460 eV — a swing of about a hundred million in magnitude versus hydrogen — making heavy ions a stringent test of QED in ultra-strong fields.
• Muonic hydrogen. Replace the electron with a muon (207× heavier) and the orbit shrinks ~200-fold, so the muon spends far more time inside the proton. The Lamb shift in muonic hydrogen is dominated by the proton's finite size, which is how the "proton radius puzzle" was uncovered around 2010.
• Helium and few-electron systems. Lamb shifts in helium fine structure provide independent high-precision determinations of the fine-structure constant α.
• Higher levels and circular Rydberg states. The shift falls off rapidly with n and with l, since both reduce the wavefunction density at the nucleus.
• Bound-state QED today. Modern calculations include two-loop self-energy and recoil corrections; theory and experiment now agree to better than a part in 10⁵, limited mainly by knowledge of the proton charge radius.

Common pitfalls and misconceptions

• "The Lamb shift is just fine structure." No. Fine structure splits different j and is contained in Dirac theory. The Lamb shift lifts a degeneracy that Dirac theory leaves intact (same n, same j, different l). It is a genuinely new, QED-only effect.
• "It comes from the proton." That is hyperfine structure. The Lamb shift is overwhelmingly about the electron and the vacuum; the proton enters only as a small finite-size correction (large only for muonic atoms).
• "Vacuum fluctuations are a metaphor." They produce a measured 1057 MHz number that matches calculation to five digits. The Casimir effect and g − 2 corroborate the same vacuum. It is as real as any energy in physics.
• "The infinities mean QED is broken." The infinities are an artifact of using unobservable bare parameters. Renormalization removes them systematically, and the Lamb shift was the proof it works.
• "The 2s level goes up because the electron gains energy from the photons." More precisely, the smearing weakens the binding near the nucleus, raising the energy (making it less bound). It is a change in how the electron samples the potential, not a simple energy deposit.
• "You can see it in an ordinary spectrometer." Doppler broadening swamps a 1 GHz splitting on an optical line. It took a radio-frequency resonance method (and later Doppler-free laser spectroscopy) to resolve it.

Why it matters — applications and legacy

• It launched modern QED. The Lamb shift, together with the electron g − 2, forced the development of renormalized quantum electrodynamics — still the most precisely tested theory in science.
• It validated renormalization. The technique that tames infinities everywhere in particle physics earned its first experimental confirmation here.
• Fundamental constants. Lamb-shift and related measurements feed the determination of α and the Rydberg constant to extreme precision.
• The proton radius puzzle. Muonic-hydrogen Lamb-shift spectroscopy revealed a discrepancy in the proton's measured size, driving a decade of experiment and theory.
• Strong-field QED. Lamb shifts in highly charged ions probe quantum electrodynamics in the most intense electric fields available in the lab.
• Tests for new physics. Any deviation between measured and calculated Lamb shifts constrains exotic forces and particles, since QED's prediction is so sharp.

Derivation analysis — where the precision comes from

The leading self-energy shift for an s-state in hydrogen can be written schematically as:

ΔE_self  =  (4 α⁵ / 3π n³) · mc² · [ ln(1/α²) + small terms ]   (s-states)

Two features drive the answer. First, the α⁵ scaling: the shift is suppressed by five powers of the fine-structure constant relative to the rest energy, which is precisely why 13.6 eV of binding produces only a 4 µeV splitting. Second, the large logarithm ln(1/α²) ≈ ln(137²) ≈ 9.8 enhances the bare α⁵ estimate; this log is the finite remnant of Bethe's cutoff, the survivor of the cancelled infinity.

To reach modern accuracy you keep going: the Bethe logarithm (a numerically evaluated average excitation energy, ≈ 2.98 for 2s), two-loop QED terms of order α²(Zα)⁴, recoil corrections of order m/M (the proton is not infinitely heavy), and the finite nuclear size term proportional to the proton charge radius squared. The first three are pure theory and known to many digits; the last is the dominant uncertainty. That is why precise hydrogen and muonic-hydrogen Lamb-shift measurements double as the best probes of the proton's radius — the QED part is so clean that whatever is left over is the nucleus.