COMPTON SCATTERING

By 1920, physicists had two contradictory pictures of light and no clean test to separate them. Einstein's 1905 photoelectric paper insisted that light comes in discrete quanta — photons — each carrying energy $E = hf$. But the photoelectric effect alone could be explained by a semi-classical model in which only the metal's absorption is quantised; light itself might still be a continuous wave. The community was divided, and the Nobel Prize committee was stalling.

X-rays offered a sharper probe. When a high-energy X-ray beam strikes a material like graphite, it scatters in all directions. The question is: what wavelength comes out?

Classical wave theory gives an unambiguous answer. An incoming electromagnetic wave sets electrons into forced oscillation at the wave's own frequency $f_0$. Oscillating charges reradiate at that same frequency. Therefore the scattered wave must have the same wavelength as the incoming wave, regardless of angle. The prediction is $\lambda' = \lambda_0$, full stop.

In 1922, at Washington University in St. Louis measured the scattered X-rays with a Bragg spectrometer — a precision crystal spectrometer that reads wavelength from diffraction angle — and found something the wave picture flatly forbids: the scattered X-ray had a longer wavelength than the incoming one. The shift depended on the scattering angle $\theta$. At $\theta = 0$ the wavelength was unchanged. At $\theta = 90°$ it was shifted by about 2.4 pm. At $\theta = 135°$ the shift was larger still.

Wave theory said: no shift, ever. The data said: shifted, angle-dependent, reproducible. Something in the classical picture was catastrophically wrong.

The resolution came when Compton did something bold: he treated the photon as a mechanical particle obeying special relativity. A photon carries energy $E_\gamma = hf = hc/\lambda$. Einstein had proposed in 1905 that a photon also carries momentum — a proposal the photoelectric data supported but did not force. For a massless particle on the null cone, the relativistic energy-momentum relation $E^2 = (pc)^2 + (m_e c^2)^2$ collapses to $E = pc$, so the photon's momentum is:

Now treat the scattering as a two-body elastic collision in special relativity. An incoming photon (energy $E_\gamma = hc/\lambda_0$, momentum $p_\gamma = h/\lambda_0$ directed along the x-axis) strikes an electron at rest (energy $m_e c^2$, momentum zero). After the collision, the photon exits at angle $\theta$ with wavelength $\lambda'$, and the electron recoils at some angle $\phi$ with relativistic kinetic energy it did not have before.

Conservation laws supply three equations — energy, $x$-momentum, $y$-momentum — and the electron's mass-shell condition $E_e^2 - (p_e c)^2 = (m_e c^2)^2$ supplies the fourth constraint needed to close the system. Working through the algebra (eliminate the electron's recoil angle and momentum, square to use the mass-shell condition) yields a single clean formula for the wavelength shift. The full derivation is a standard SR exercise; the four-momentum approach in §04.1 packages it most elegantly — the Four-momentum conservation condition $p_\gamma^\mu + p_e^\mu = p_{\gamma'}^\mu + p_{e'}^\mu$ is four equations in one.

The prefactor $h / m_e c$ is the Compton shift — the electron Compton wavelength $\lambda_C \approx 2.426\,\text{pm}$. It is the natural length scale of the scattering: at $\theta = 90°$ the shift is exactly $\lambda_C$; at $\theta = 180°$ (back-scatter) it is $2\lambda_C$.

Three things about this formula are remarkable. First, $\Delta\lambda$ is independent of the incoming wavelength $\lambda_0$. Whether you use soft X-rays at 100 pm or hard X-rays at 10 pm, the shift at any given angle is always the same $\lambda_C(1 - \cos\theta)$. This wavelength-independence is the formula's clearest signature — it is structurally impossible in the wave picture, which would have to make $\Delta\lambda$ scale with $\lambda_0$ for any resonance-based mechanism. Second, $\Delta\lambda \geq 0$ always: the scattered photon never comes out bluer than it went in, because the electron recoil absorbs a positive amount of kinetic energy. Third, the formula contains $h$ and $m_e$ — quantum and mechanical constants together — in a ratio that is a pure length. The Compton wavelength is where quantum mechanics and special relativity first meet in a single number.

Compton published his measurements in Physical Review in May 1923. Using molybdenum K-alpha X-rays ($\lambda_0 = 0.071\,\text{nm} = 71\,\text{pm}$, produced by bombarding a molybdenum target and selecting the K-alpha line with a crystal monochromator) scattered off graphite, he measured the outgoing wavelength at six angles spanning 0° to 135°. At each angle the spectrometer resolved two peaks: one at $\lambda_0$ (classical component — photons scattering off tightly bound inner electrons that do not recoil) and one at $\lambda' > \lambda_0$ (the shifted Compton component — photons scattering off nearly-free outer electrons). The shifted peak tracked Compton's formula precisely.

The gap between the two curves — the cyan classical line sitting at $\lambda' = 71\,\text{pm}$ and the amber Compton curve climbing to $\sim 75\,\text{pm}$ at $\theta = 135°$ — is not a subtle systematic. At 90° the discrepancy is 2.4 pm; Compton's spectrometer resolution was better than 0.5 pm. The data was unambiguous.

Simultaneously and independently, Peter Debye derived the same formula from momentum-kinematics arguments. Compton did not know about Debye's derivation when he submitted; both papers appeared in 1923. The effect carries Compton's name because his data accompanied the formula. The Nobel committee awarded Compton the 1927 prize in Physics specifically for this discovery.

Before 1923, Einstein's photon was a theoretical proposal supported by one experiment (the photoelectric effect) in one regime (UV/visible light on metals). Critics could argue that the photoelectric effect merely required quantised energy absorption in the metal, not a discretised photon. The photon did not have to carry momentum; it did not have to scatter; it did not have to obey Newton's billiard-ball kinematics in a relativistic wrapper.

After Compton 1923, none of that was tenable. The Compton shift formula requires that the photon carry a definite momentum $p = h/\lambda$, that this momentum is conserved in a two-body collision with an electron, and that the collision is governed by special-relativistic four-momentum conservation. A photon that merely delivers quantised energy in bulk cannot produce an angle-dependent wavelength shift that is exactly $\lambda_C(1 - \cos\theta)$. The formula's very structure — the factor $(1 - \cos\theta)$ — is the kinematic signature of a two-body elastic collision. It does not arise from any wave-scattering model.

The four-momentum picture is the cleanest formulation. As developed in §04.1, the Four-momentum $p^\mu = (E/c, \vec{p})$ packages energy and momentum into one four-vector that transforms as a Lorentz four-vector under boosts. The mass-shell condition $p_\mu p^\mu = m^2 c^2$ is a Lorentz scalar — invariant in every frame. For a photon, $p_\mu p^\mu = 0$: the photon lives on the null cone of the Minkowski metric. Four-momentum conservation $p_\gamma^\mu + p_e^\mu = p_{\gamma'}^\mu + p_{e'}^\mu$ is four equations, all Lorentz-covariant, that encode both energy and momentum conservation automatically. The Compton shift formula is what you get when you apply this machinery to a massless-particle–massive-particle collision and then ask what the scattered photon's wavelength must be.

had proposed the momentum $p = h/\lambda$ in 1909 and 1917, but the proposal lacked a direct mechanical test. Compton provided it. The photon was now a particle in the full sense: it carries energy, carries momentum, scatters off other particles at angles governed by four-momentum conservation, and obeys a mass-shell condition that makes it a null vector in Minkowski spacetime. The 1923 data was the mechanical birth certificate of the photon.

From here, the relativistic dynamics of §04 are complete for point particles and photons. The next module, §05, pushes into applied SR: GPS as relativity, the Sagnac effect, relativistic jets, and the barn-pole paradox. Beyond §05, §06 takes the geometry of §03 and asks what happens when spacetime itself is curved by every massive object in it — and that question leads all the way to general relativity.