Compton Scattering

A photon billiards an electron

By the early 1920s most of the physics community had accepted Einstein's photoelectric photon — the quantum of electromagnetic energy E = hf. What was still controversial was whether that energy quantum carried mechanical momentum. Maxwell's wave theory predicted radiation pressure (and so a momentum density u/c), but the photon-as-particle picture demanded that an individual photon hit an individual electron in a billiards-like collision, and nobody had cleanly seen that.

Arthur Compton, working at Washington University in St. Louis, fired the K-α line of molybdenum (λ = 71 pm, energy 17.4 keV) at a graphite block and measured the wavelength of X-rays scattered at various angles with a Bragg crystal spectrometer. He saw two peaks: an unshifted line at the original wavelength, and a shifted line whose Δλ grew with the scattering angle. The unshifted peak came from photons bouncing off tightly bound inner-shell electrons (effectively scattering from the whole atom). The shifted peak — the Compton peak — came from photons bouncing off loosely bound outer electrons, which behave essentially as free particles.

The shift fit a single equation:

Δλ = λ' − λ = (h / m_e c) · (1 − cos θ)

where θ is the scattering angle. The proportionality constant h/(m_e c) = 2.426 × 10⁻¹² m is now called the Compton wavelength. Δλ depends only on θ — not on the incident wavelength, not on the target material, not on photon intensity. That universality is a fingerprint of two-body relativistic kinematics. There is no wave-only explanation that gives this equation.

Conservation of energy and momentum

Compton's formula falls out of one application each of energy and momentum conservation, treating the photon as a relativistic particle with E = pc = hc/λ. Set up the collision with the photon coming in along x with energy E = hc/λ and the electron initially at rest. After the collision the photon scatters at angle θ with energy E' = hc/λ', and the electron recoils at angle φ with energy K and momentum p_e.

Energy conservation:

hc/λ + m_e c² = hc/λ' + √((p_e c)² + (m_e c²)²)

Momentum conservation along x and y:

h/λ = (h/λ') cos θ + p_e cos φ
0   = (h/λ') sin θ − p_e sin φ

Square and add the two momentum equations to eliminate φ, square the energy equation to eliminate p_e, and after a few lines of algebra you get the Compton formula:

1/λ' − 1/λ = (1 − cos θ) / λ_C
⇔
λ' − λ = λ_C (1 − cos θ),  λ_C ≡ h/(m_e c)

Notice the formula contains no Planck constant in disguise; h shows up only inside the Compton wavelength, which is itself a length set by combining h, the electron mass, and c. The result is purely kinematic — dynamics enters only through which scattering angles are populated.

Worked example: 1 MeV gammas off a free electron

Suppose a 1 MeV gamma photon (λ = hc/E = 1.240 pm) Compton-scatters at 60° from a free electron. The wavelength shift is:

Δλ = λ_C · (1 − cos 60°)
   = 2.426 pm · (1 − 0.5)
   = 1.213 pm

λ' = λ + Δλ = 1.240 + 1.213 = 2.453 pm
E' = hc/λ' = 1240 / 2.453 = 505 keV

So the photon retains barely half its initial energy. The electron recoils with kinetic energy 1 MeV − 0.505 MeV = 0.495 MeV — comparable to its rest mass, so it is mildly relativistic. At θ = 180° (head-on backscatter):

Δλ_max = 2 λ_C = 4.85 pm
λ' = 6.09 pm,  E' = 204 keV
electron recoil = 796 keV (~γ ≈ 2.5)

Above an incident energy of a few hundred keV, large-angle Compton scattering deposits most of the photon's energy into the electron — the basis of every Compton-suppression scheme in gamma-ray spectroscopy. The famous "Compton edge" in a NaI(Tl) spectrum — the high-energy cutoff of the Compton continuum — is exactly E_γ − E'(180°), the maximum kinetic energy a single Compton-scattered electron can carry away inside the detector.

Compton vs Thomson scattering

The classical theory of light scattering by a free electron is Thomson scattering: the electron oscillates in the incident E-field, re-radiating at the same frequency. The cross-section is the famous σ_T = (8π/3) r_e² = 6.65 × 10⁻²⁹ m², where r_e = e²/(4πε₀ m_e c²) = 2.82 fm is the classical electron radius. There is no wavelength shift, scattering is symmetric about θ = 90°, and the cross-section is independent of frequency.

This works fine when hf ≪ m_e c². At visible light frequencies (hf ≈ 2 eV vs 511 keV electron rest energy), the photon's momentum is negligible compared to a free electron's possible kinetic states; Thomson scattering reproduces the data exactly. As hf approaches and exceeds m_e c², the recoil becomes nonnegligible and the scattering picks up the Compton shift, the cross-section drops, and the angular distribution becomes forward-peaked. The exact differential cross-section in the relativistic regime is the Klein–Nishina formula (1929):

dσ/dΩ = (½) r_e² (λ/λ')² [(λ/λ') + (λ'/λ) − sin² θ]

Klein–Nishina reduces to Thomson when λ' → λ (low energy). At 1 MeV the total cross-section has dropped from σ_T = 0.665 b to σ_KN ≈ 0.21 b — a factor of three suppression. At 1 GeV it is below 10 mb.

Wavelength shifts at characteristic angles and energies

Scattering angle θ	Δλ / λ_C	Δλ (pm)	For 100 keV in: λ' (pm)	λ' for 1 MeV in (pm)
0°	0	0	12.40	1.240
30°	0.134	0.325	12.73	1.565
60°	0.500	1.213	13.62	2.453
90°	1.000	2.426	14.83	3.666
120°	1.500	3.639	16.04	4.879
150°	1.866	4.527	16.93	5.767
180°	2.000	4.852	17.25	6.092

The 100 keV column shows that even at maximum back-scatter the relative shift is only 39 % — Compton scattering is mild. The 1 MeV column shows it is severe: a 180° backscatter cuts the photon energy by a factor of five.

Where Compton scattering shows up

• PET and SPECT imaging. 511 keV annihilation photons interact in NaI or LSO crystals partly through photoelectric absorption (full-energy peak) and partly through Compton scattering (Compton continuum). Coincidence-pair selection cuts the Compton background; modern time-of-flight PET resolves the residual ambiguity within ~200 ps.
• Compton-imaging gamma-ray telescopes. The COMPTEL instrument on the Compton Gamma Ray Observatory (1991–2000) reconstructed the sky in 1–30 MeV gammas by recording two-step Compton interactions: first scatter in a low-Z scintillator, second absorption in a high-Z one, and then back-projecting the cone consistent with the measured Δλ at the first vertex.
• Inverse Compton in astrophysical jets. Synchrotron-self-Compton models for blazars treat the same relativistic electron population that emits synchrotron photons as a target population that up-scatters those same photons into the GeV–TeV band. The two-bump SED of Mrk 421 is the canonical example.
• Cosmic-microwave-background distortions. The Sunyaev–Zel'dovich effect: hot electrons in galaxy-cluster atmospheres inverse-Compton-scatter CMB photons, distorting the blackbody spectrum and producing an angle-dependent intensity dip below ~217 GHz and a bump above. A direct probe of cluster gas pressure independent of redshift.
• Material analysis: Compton-profile spectroscopy. Measuring the broadening of the shifted Compton line in a real solid (rather than a sharp line, you get a Doppler-broadened distribution) reveals the projected electron-momentum distribution n(p_z), used in solid-state research to map the Fermi surface of metals and the chemical environment of light-element compounds.

Compton's original 1923 setup

Compton's apparatus was small enough to fit on a tabletop. A molybdenum X-ray tube produced K-α photons at 17.4 keV (λ = 0.71 Å). These struck a small block of carbon (graphite). Scattered X-rays exited at a chosen angle — Compton measured 45°, 90° and 135° — and were Bragg-diffracted off a calcite crystal whose orientation could be tuned to select wavelength. Detection used an ionization chamber.

The Bragg condition 2d sin θ_B = nλ converts the scattered wavelength to an angular position on the rotating crystal. The unshifted peak appeared at the original Bragg angle; the shifted peak appeared at slightly larger angle, growing with the carbon-block scattering angle by exactly λ_C(1 − cos θ). At θ = 90° the predicted Δλ = 0.024 Å, and Compton measured 0.022 ± 0.001 Å. Within his stated uncertainty, the photon-particle prediction was perfect.

The experiment was technically straightforward but conceptually devastating: it demanded that anyone doing X-ray science treat the photon as a billiard ball with momentum p = h/λ, not a wavefront. Within a year Heisenberg's matrix mechanics and Schrödinger's wave mechanics were on the way; Compton's clean-room data was one of the foundations they were trying to explain.

Variants and extensions

• Inverse Compton scattering. A relativistic electron up-scatters a low-energy photon to higher energy. Dominant cooling mechanism for energetic electrons in cosmic environments and the production mechanism for synchrotron-self-Compton TeV emission from blazars.
• Klein–Nishina formula. The full QED differential cross-section for Compton scattering, valid for any photon energy. Reduces to Thomson at low energy and falls as 1/E at high energy.
• Double Compton scattering. Higher-order QED process in which a single photon scatters off an electron and emits a second photon simultaneously. Negligible in most contexts but sets the lower bound on the Compton-suppression resolution.
• Doppler-broadened Compton profile. In real solids, the target electrons have a momentum distribution. The Compton peak is broadened, and the broadening profile encodes the electronic structure of the material. A standard probe of Fermi surfaces.
• Compton-imaging gamma telescopes (COMPTEL, MEGAlib, all-sky Compton imagers). Use the kinematics of two-step Compton interactions to reconstruct gamma-ray source positions in the 0.5–30 MeV band, where photoelectric absorption is too weak to use lensing or coded apertures.

Common pitfalls

• Confusing Δλ and ΔE. The Compton formula is exact in wavelengths but not in energies. Photon energy is hc/λ, so a small Δλ at long λ is a small ΔE, but at short λ a small Δλ is a large fractional ΔE. For 1 MeV gammas, Δλ at 90° is 2.4 pm out of 1.24 pm — the shifted wavelength has tripled.
• Treating bound electrons as free. Compton's clean formula assumes the target electron is at rest and unbound. Inner-shell electrons (K, L) of high-Z atoms behave more like the whole atom and produce the unshifted Rayleigh peak. The Compton shift is suppressed when the photon energy is below the electron's binding energy.
• Forgetting which mass to use. Δλ uses the rest mass of the recoiling particle. For atomic Compton you would use the proton or atomic mass and the shift is 1836× or more times smaller. For Compton off a positron the shift is the same as for an electron because m_e = m_e+.
• Assuming Klein–Nishina = Thomson at all energies. Below 50 keV they are within a few percent. Above 100 keV you must use Klein–Nishina; using σ_T at 1 MeV overestimates the cross-section by 3×, giving wrong absorption coefficients in shielding calculations.
• Reading the Compton edge as a peak. A NaI gamma-ray spectrum shows a continuum (the Compton continuum) ending at the Compton edge, not a sharp Compton peak. The peak in the spectrum is the photoelectric full-energy peak; the edge is below it by the maximum Compton-recoil energy E_γ²(2 + E_γ/m_e c²)⁻¹.