Deep Inelastic Scattering

Why deep inelastic scattering matters

Before 1968 the proton was widely regarded as an elementary particle — the positive twin of the electron, dressed up with an anomalous magnetic moment. Elastic electron-proton scattering (measured by Hofstadter in the 1950s) had shown the proton has a finite size of about 0.8 fm through its form factors, but that only said the charge was smeared out; it said nothing about lumps inside. Deep inelastic scattering changed the question. By transferring enough momentum to resolve distances far smaller than the proton, the SLAC-MIT experiments caught the electron ricocheting off hard, pointlike grains buried in the proton. That was the direct experimental birth of the quark as a real particle rather than a bookkeeping device from Gell-Mann and Zweig's 1964 classification scheme.

The impact was foundational. DIS gave us: (1) the parton model and the picture of the proton as a swarm of quarks, antiquarks and gluons; (2) the momentum-fraction distributions f(x) that every LHC prediction still relies on today; (3) the discovery that quarks carry only ~half the proton's momentum, forcing the existence of neutral gluons; and (4) via scaling violations, the first quantitative test of QCD and its defining property, asymptotic freedom.

The kinematics: Q², ν and x

An electron of four-momentum k scatters to k′, exchanging a single virtual photon that carries four-momentum q = k − k′ to the proton (four-momentum P, mass M). Three Lorentz-invariant quantities describe the collision:

Q² = −q²  = 4 E E′ sin²(θ/2)     (spacelike momentum transfer, > 0)
ν  = P·q / M = E − E′             (energy lost by the electron, lab frame)
x  = Q² / (2 M ν)                 (Bjorken scaling variable, 0 < x ≤ 1)

Symbols and units: Q² is minus the squared four-momentum transfer, in GeV² (natural units, ℏ = c = 1); large Q² resolves small distances since the probed length ~ ℏc/√Q² ≈ 0.197 GeV·fm / √Q². E, E′ are the incoming and scattered electron energies (GeV) and θ the scattering angle. ν is the electron's energy loss (GeV). M = 0.938 GeV/c² is the proton mass. x is dimensionless: the fraction of the proton's momentum carried by the struck quark in the infinite-momentum frame. Elastic scattering (proton stays intact) sits exactly at x = 1; the "deep inelastic" region has x well below 1 and Q² ≫ M².

The cross-section and structure functions

The double-differential cross-section for unpolarized electron-proton DIS, mediated by one photon, is

d²σ/(dx dQ²) = (4π α² / x Q⁴) · [ (1 − y − x²y²M²/Q²) F₂(x,Q²) + x y² F₁(x,Q²) ]

where α ≈ 1/137 is the fine-structure constant, y = ν/E is the fractional energy loss (inelasticity, 0 to 1), and F₁, F₂ are the two structure functions. Everything we cannot compute from first principles about the proton's guts is bundled into F₁ and F₂. The 1/Q⁴ prefactor is the Mott/Rutherford falloff of scattering off a point charge; the physics of the target lives entirely in the structure functions.

In the naive parton model, the proton is a beam of free, pointlike quarks and F₂ becomes a simple sum:

F₂(x) = Σ_i  e_i² · x · f_i(x)

Here the sum runs over quark flavors i, e_i is the quark's electric charge in units of e (+2/3 for up/charm, −1/3 for down/strange), and f_i(x) is the parton distribution function (PDF) — the number density of quarks of flavor i carrying momentum fraction x. Because the photon couples to charge squared, up quarks (charge² = 4/9) count four times as heavily as down quarks (charge² = 1/9). This is why the neutron and proton structure functions differ, and comparing them isolates the up- and down-quark distributions.

The Callan-Gross relation: quarks are spin-½

The two structure functions are not independent. For spin-1/2 partons they obey the Callan-Gross relation:

2 x F₁(x) = F₂(x)

Physically, F₁ is generated purely by the magnetic (spin) coupling, which vanishes for spinless targets. If the partons had spin 0, we would measure F₁ = 0 and a longitudinal cross-section; if they were spin-1/2 Dirac particles, F₁ and F₂ lock together as above. The data came down firmly on the spin-1/2 side: measurements gave the ratio R = σ_L/σ_T ≈ 0.2 and small, consistent with 2xF₁ ≈ F₂. That is how DIS certified not only that partons exist but that they are Dirac fermions — exactly the quarks.

Bjorken scaling — the pointlike signature

James Bjorken predicted in 1969 that in the deep inelastic limit (Q² → ∞, ν → ∞, x fixed), the structure functions would stop depending on Q² and become functions of x alone:

F₂(x, Q²)  →  F₂(x)        (Bjorken scaling)

This is the crux of the whole story. A target with internal size or soft structure imposes a form factor F(Q²) that suppresses hard collisions as Q² grows — that is exactly what elastic scattering off the whole proton shows. A pointlike charge has no form factor and no preferred scale, so the response can only depend on the dimensionless x. When SLAC saw F₂ essentially flat in Q² across a decade of Q² values, it meant the electron was hitting objects with no measurable size. Feynman interpreted this immediately in terms of his "partons," and the partons were identified with quarks.

Scaling violations — the evidence for QCD

Scaling is not exact. Look closely and F₂ drifts logarithmically with Q²: at small x it slowly rises with Q², at large x it falls. These scaling violations are not a failure of the parton picture — they are precisely what QCD predicts. As Q² increases, the virtual photon resolves finer detail and catches quarks in the act of radiating gluons (and gluons splitting into quark-antiquark pairs). The evolution is governed by the DGLAP equations (Dokshitzer-Gribov-Lipatov-Altarelli-Parisi):

dq(x, Q²) / d ln Q²  =  (α_s(Q²) / 2π) ∫ₓ¹ (dz/z) P(x/z) q(z, Q²)

where α_s(Q²) is the running strong coupling (which shrinks at large Q² — asymptotic freedom) and P(x/z) are the splitting functions giving the probability that a parton of momentum fraction z radiates down to x. The measured pattern of scaling violations matches DGLAP evolution quantitatively over many orders of magnitude in Q², most spectacularly at the HERA electron-proton collider (1992–2007). This agreement is one of the cleanest confirmations that QCD is the correct theory of the strong force.

The missing momentum — where gluons hide

Add up how much momentum the charged quarks carry by integrating the momentum-weighted distributions:

∫₀¹ x · [ q(x) + q̄(x) ] dx  ≈  0.5

Only about half. The quarks the photon can see account for roughly 50% of the proton's momentum; the other half is carried by something electrically neutral that the photon is blind to. That something is the gluon, the carrier of the strong force. DIS thus gave an indirect but quantitative discovery of gluons years before they were seen directly (as three-jet events at PETRA in 1979). Neutrino DIS, which couples through the weak charged current, later confirmed the quark charges and the gluon momentum fraction independently.

Key numbers and comparisons

Quantity	Value / relation	Meaning
Proton mass M	0.938 GeV/c²	Sets the ν→x conversion
Proton charge radius	≈ 0.84 fm	From elastic form factor
Fine-structure constant α	≈ 1/137	Photon coupling in cross-section
Up-quark charge	+2/3 e	Weight 4/9 in F₂
Down-quark charge	−1/3 e	Weight 1/9 in F₂
Bjorken x range	0 < x ≤ 1	x = 1 is elastic limit
Quark momentum fraction	≈ 0.5	Rest is neutral gluons
Callan-Gross	2xF₁ = F₂	Confirms spin-½ partons
R = σ_L/σ_T	≈ 0.2 (small)	Would be large for spin-0

Elastic vs. deep inelastic — a comparison

Feature	Elastic e-p scattering	Deep inelastic scattering
Final state	Proton stays intact	Proton shatters into hadrons
Kinematics	x = 1 (one constraint)	x < 1, independent x and Q²
Q² dependence	Steep form-factor falloff	Nearly flat (Bjorken scaling)
Probes	Whole proton charge cloud	Individual pointlike quarks
Describes target by	Form factors G_E, G_M	Structure functions F₁, F₂
Historic result	Proton has size ~0.8 fm	Proton is not fundamental

Worked example: computing x at SLAC

Take a representative SLAC-era beam: an incoming electron of E = 20 GeV scatters to E′ = 12 GeV at an angle θ = 10°. Then the energy loss is ν = E − E′ = 8 GeV, and the momentum transfer is

Q² = 4 E E′ sin²(θ/2) = 4 · 20 · 12 · sin²(5°) ≈ 960 · (0.0872)² ≈ 7.3 GeV²

so the Bjorken variable is

x = Q² / (2 M ν) = 7.3 / (2 · 0.938 · 8) ≈ 0.49

A struck quark carrying about half the proton's momentum. Now change only Q² by taking a different angle at the same x — the key discovery was that F₂ came out essentially the same value. That flatness in Q², repeated across many bins, is Bjorken scaling in action, and it is why a smeared-out proton was ruled out in favor of hard pointlike quarks.

History: SLAC 1968 and the Nobel Prize

The experiments ran on the 2-mile linear accelerator at the Stanford Linear Accelerator Center (SLAC), using the 8 GeV and 20 GeV magnetic spectrometers in End Station A. Jerome Friedman and Henry Kendall (MIT) and Richard Taylor (SLAC) led the collaboration; the first inelastic results appeared in 1968–1969. The unexpectedly weak Q² dependence baffled everyone until James Bjorken supplied the scaling prediction and Richard Feynman recast it in the intuitive language of partons — free, pointlike constituents seen in a fast-moving proton. Bjorken and Emmanuel Paschos worked out the parton-model structure functions; Callan and Gross derived the spin-1/2 relation. Friedman, Kendall and Taylor shared the 1990 Nobel Prize in Physics "for their pioneering investigations concerning deep inelastic scattering of electrons on protons and bound neutrons, which have been of essential importance for the development of the quark model in particle physics."

Common misconceptions

• "The electron hits the whole proton." In DIS it does not — at high Q² the collision is with a single quark; the rest of the proton is a spectator that then hadronizes. That is why the process is inelastic (the proton breaks up).
• "Bjorken scaling means F₂ is exactly constant in Q²." Only approximately. The residual logarithmic Q² dependence (scaling violations) is real, calculable, and is the evidence for QCD — do not dismiss it as noise.
• "x is the same as the scattering angle." No. x = Q²/(2Mν) is a Lorentz invariant; the same x can be reached with many different beam energies and angles. Its physical meaning (momentum fraction) only appears in the infinite-momentum frame.
• "Quarks carry all the proton's momentum." They carry only about half. Missing this is missing the discovery of gluons — the neutral half the photon cannot see.
• "DIS measured the quark masses." It measured charges (via the 4:1 up:down weighting) and spin (via Callan-Gross), and mapped the momentum distributions f(x). Quark masses are not what F₂ delivers.
• "Elastic and deep inelastic are the same experiment at different energies." They are kinematically distinct: elastic is pinned to x = 1 with the proton intact; DIS occupies the x < 1, high-Q² region where the proton fragments. The two regimes tell opposite stories — size versus pointlike substructure.