Colour & QCD

QCD is the working language of every high-energy collider. At the LHC essentially all processes begin with coloured partons that scatter, radiate, and — because of confinement — dress into the collimated sprays of hadrons called jets, whose rates follow from the colour factors $C_F$ and $C_A$; the relative broadening of quark and gluon jets tests $C_A/C_F=\tfrac94$ directly, and $\alpha_s$ has been mapped from a few GeV to the TeV scale, world average $\alpha_s(M_Z)=0.1179\pm0.0009$. Where perturbation theory fails, lattice QCD discretises equation (1) on a spacetime grid and reproduces the proton mass to about one percent — showing most of it is gluonic field energy, not quark mass. In heavy-ion collisions at RHIC and the LHC, matter heated past the confinement transition becomes a quark–gluon plasma that flows as the most perfect fluid known. The open frontier is the problem Gell-Mann's theory posed: proving confinement — that the spectrum contains only colour singlets — is an unsolved Clay Millennium Prize problem.

Start at a single vertex. A red quark comes in, a blue quark goes out, and a gluon carries off the difference. Colour is conserved at the vertex just as electric charge is in QED, so the gluon must carry one unit of red and one of anti-blue — a coloured object living in the combination "colour $\otimes$ anticolour." Counting those combinations, $3\otimes\bar3=8\oplus1$, gives eight physical gluons, not nine: the colourless singlet would couple to colour-neutral matter and generate an unobserved long-range strong force, so nature keeps only the octet. The first figure makes the vertex and that counting one picture.

The laboratory does not show a gluon vertex directly; it shows the statistics of what quarks produce. The cleanest such number is the ratio $R$ of the rate for $e^+e^-$ to annihilate into hadrons versus into a muon pair. The virtual photon couples to each quark as it does to a muon, weighted by charge squared — but a quark comes in three colours, so every flavour is counted three times. $R$ is therefore three times the sum of squared quark charges, and it climbs in steps as the beam energy crosses each new flavour threshold. The measured steps sit right on the $N_c=3$ prediction; one colour would put them three times lower. The second figure is that measurement.

The whole theory follows from demanding that the quark fields be invariant under a colour rotation performed independently at every spacetime point. That single requirement — local $SU(3)$ invariance — fixes the Lagrangian, and everything below is a piece of it. For a quark field $\psi$ (a colour triplet):

where: $\psi_i$ — quark field with colour index $i=1,2,3$; $D_\mu$ — colour covariant derivative; $m$ — quark mass (colour-diagonal); $G^a_{\mu\nu}$ — gluon field-strength tensor, $a=1,\dots,8$. Reading: the first term is Dirac with the derivative promoted to $D_\mu$ (the quark–gluon coupling hides here); the second is the gluon kinetic term (the gluon self-coupling hides here). The covariant derivative makes local colour rotations harmless. It carries one gauge field $G^a_\mu$ per generator, its form dictated entirely by requiring $D_\mu\psi$ to transform the same way as $\psi$.

where: $g_s$ — strong coupling, with $\alpha_s\equiv g_s^2/4\pi\approx0.12$ at the $Z$ mass; $T^a$ — the eight $SU(3)$ generators, half the Gell-Mann matrices $\lambda^a$; $f^{abc}$ — totally antisymmetric structure constants; the eight $G^a_\mu$ are the gluon fields. Reading: the non-vanishing commutator is the whole story — $SU(3)$ is non-Abelian, the order of two colour rotations matters, and that is what forces the gluons to be coloured. The field-strength tensor is the non-Abelian generalisation of $F_{\mu\nu}=\partial_\mu A_\nu-\partial_\nu A_\mu$; because the generators do not commute, it acquires an extra quadratic-in-the-field term.

where: the first two terms mirror QED; the third, $\propto f^{abc}$, exists only because $SU(3)$ is non-Abelian. Reading: substituting (3) into the $-\tfrac14 G^a_{\mu\nu}G^{a\,\mu\nu}$ term of (1), the cross terms produce a triple-gluon vertex ($\propto g_s$) and a four-gluon vertex ($\propto g_s^2$) — gluons couple to gluons, the defining structural difference from photons. Expanding the covariant-derivative term of (1) isolates the interaction. Comparing with QED's $-iq\gamma^\mu A_\mu\psi$, the replacement rule and quark–gluon vertex factor read off at once:

where: $q$ — the single QED charge; $\tfrac12\lambda^a$ — the Gell-Mann matrix acting on the quark's colour wavefunction $c_i$, so the vertex current is $\bar u(p_3)\,c_j^\dagger\big(-\tfrac12 i g_s\lambda^a\gamma^\mu\big)c_i\,u(p_1)$. Reading: the Dirac structure $\gamma^\mu$ is untouched; the only change from QED is that the coupling is now a matrix in colour space — exactly what "colour is a vector, not a number" means at the level of a Feynman rule. The observable that turns colour into a countable integer is the ratio $R$. The photon couples democratically to any charged pair, so the ratio of hadronic to muonic annihilation counts available quark states weighted by charge:

where: $N_c=3$ — the number of colours; $Q_f$ — the electric charge of quark flavour $f$ in units of $e$; the sum runs over flavours light enough to be pair-produced at the given energy. Reading: $R$ is a direct headcount of colours — the factor $N_c$ multiplies the whole sum, so measuring $R$ measures $N_c$. Finally, the strength of one-gluon exchange between coloured particles is governed by the colour algebra through the quadratic Casimir — the QCD analogue of "$q_1q_2$" in the Coulomb potential:

where: $C_F=\tfrac43$ — Casimir for a quark (fundamental representation); $C_A=3$ — Casimir for a gluon (adjoint representation); these fix the colour factors of every QCD amplitude. Reading: $C_F$ sets how strongly a quark radiates gluons and $C_A$ how strongly a gluon does; the ratio $C_A/C_F=9/4$ is why gluon jets are broader than quark jets.

1 · The R ratio across the flavour thresholds (fully worked)

Predict $R=\sigma(e^+e^-\!\to\text{hadrons})/\sigma(e^+e^-\!\to\mu^+\mu^-)$ in the three energy regions: below the charm threshold, between charm and bottom, and above bottom. Use $N_c=3$ and compare with the ladder of measured plateaus.

Step 1. Below charm ($\sqrt s\lesssim3$ GeV) the open flavours are $u,d,s$ with charges $+\tfrac23,-\tfrac13,-\tfrac13$. The squared-charge sum is $\sum Q_f^2=\left(\tfrac23\right)^2+\left(\tfrac13\right)^2+\left(\tfrac13\right)^2=\tfrac49+\tfrac19+\tfrac19=\tfrac69=\tfrac23$. Multiplying by $N_c=3$: $R=3\cdot\tfrac23=\mathbf{2}$.

Step 2. Above the charm threshold ($\sqrt s\approx4$–$9$ GeV) add the charm quark, $Q_c=+\tfrac23$, contributing $\tfrac49$. Now $\sum Q_f^2=\tfrac23+\tfrac49=\tfrac{6}{9}+\tfrac49=\tfrac{10}{9}$, so $R=3\cdot\tfrac{10}{9}=\dfrac{10}{3}\approx\mathbf{3.33}$.

Step 3. Above the bottom threshold ($\sqrt s\gtrsim11$ GeV) add the bottom quark, $Q_b=-\tfrac13$, contributing $\tfrac19$. Then $\sum Q_f^2=\tfrac{10}{9}+\tfrac19=\tfrac{11}{9}$, so $R=3\cdot\tfrac{11}{9}=\dfrac{11}{3}\approx\mathbf{3.67}$.

Step 4. The three predicted plateaus $2,\ 3.33,\ 3.67$ track the data closely (small upward corrections of order $\alpha_s/\pi\approx0.04$ from gluon radiation raise each by a few percent, and narrow $c\bar c$, $b\bar b$ resonances spike above threshold). A single-colour world would predict $\tfrac23,\ \tfrac{10}{9},\ \tfrac{11}{9}$ — a factor of three low, decisively excluded. The staircase of $R$ is a direct count of $N_c=3$.

2 · The quark colour factor $C_F=\tfrac43$ from the Gell-Mann matrices

Show explicitly that $\sum_{a=1}^{8}(T^aT^a)=C_F\,\mathbb{1}$ with $C_F=\tfrac43$, using $T^a=\tfrac12\lambda^a$ and the normalisation $\mathrm{Tr}(\lambda^a\lambda^b)=2\delta^{ab}$.

Step 1. The operator $\sum_a T^aT^a$ commutes with every generator (it is the quadratic Casimir), so by Schur's lemma it is a multiple of the identity on the fundamental: $\sum_a T^aT^a=C_F\,\mathbb{1}$. It remains to fix $C_F$.

Step 2. Take the trace. On the right, $\mathrm{Tr}(C_F\mathbb{1})=C_F N_c=3C_F$. On the left, $\sum_a\mathrm{Tr}(T^aT^a)=\tfrac14\sum_a\mathrm{Tr}(\lambda^a\lambda^a)=\tfrac14(2\times8)=4$.

Step 3. Equate: $3C_F=4$, hence $C_F=\dfrac43$, matching $C_F=\dfrac{N_c^2-1}{2N_c}=\dfrac86=\dfrac43$. A quark's colour Casimir is $C_F=\tfrac43$; the gluon's is $C_A=N_c=3$, and $C_A/C_F=\tfrac94$ is why a gluon radiates more than twice as strongly as a quark.

3 · Why the $\Delta^{++}$ demands colour

The $\Delta^{++}$ baryon is $uuu$ with spin $\tfrac32$ (all three quark spins aligned) in a spatial ground state. Explain why, without colour, this would violate the Pauli principle, and how colour rescues it.

Step 1. The three $u$ quarks are identical fermions, so the total wavefunction — factorised as spatial $\times$ spin $\times$ flavour $\times$ colour — must be antisymmetric under exchange of any two.

Step 2. In the ground state the spatial part is symmetric ($L=0$), the spin part is symmetric ($S=\tfrac32$, all aligned), and the flavour part is symmetric ($uuu$). The product of these three is symmetric — the opposite of what Fermi statistics demands.

Step 3. The only remaining factor is colour, which must therefore be totally antisymmetric: $\tfrac{1}{\sqrt6}\,\epsilon_{ijk}\,|q_i q_j q_k\rangle=\tfrac{1}{\sqrt6}(rgb-rbg+gbr-grb+brg-bgr)$. Such a state exists precisely because there are three colours — $\epsilon_{ijk}$ needs exactly three indices. The $\Delta^{++}$ is consistent with Fermi statistics only if quarks carry a three-valued colour charge; this was an original 1964 motivation for colour, and the antisymmetric $\epsilon_{ijk}$ singlet is the colour structure of every baryon.

Colour and QCD are the summit of a mountain that one person did more than any other to build: the quark itself, the $SU(3)$ that classifies hadrons, the current algebra that made quarks respectable, and finally — with Fritzsch and Leutwyler — the colour gauge theory and its name.

Texts. Thomson, Modern Particle Physics, §10.1–10.2 (colour, the quark–gluon vertex, QCD) · Halzen & Martin, Quarks and Leptons, Ch. 2, 14 (colour, the $R$ ratio, QCD Feynman rules) · Griffiths, Introduction to Elementary Particles, 2nd ed., §8.1–8.3 (chromodynamics Feynman rules) · Peskin & Schroeder, §15.1–15.2 (non-Abelian gauge invariance and the field strength) · Schwartz, QFT and the Standard Model, Ch. 25–26 (Yang–Mills and QCD). Four results tie the article together and are worth keeping in view: (1) $R=3\sum_f Q_f^2$ gives $2,\ 10/3,\ 11/3$ across the $u\,d\,s$, $+c$, $+b$ thresholds; (2) $\sum_a T^aT^a=\tfrac43\mathbb{1}$ follows from $\mathrm{Tr}(\lambda^a\lambda^b)=2\delta^{ab}$; (3) the singlet colour factor $-\tfrac43$ and octet $+\tfrac16$ both follow from $\mathbf{T}_1\!\cdot\mathbf{T}_2=\tfrac12[C_{\text{tot}}-C_F-C_F]$; (4) the field strength (3) is obtained from $[D_\mu,D_\nu]=ig_sT^aG^a_{\mu\nu}$.

Prerequisites. The Local Gauge Principle · Feynman Rules of QED · SU(3) Flavour Symmetry · Evidence for Colour. Leads to. Gluons · QCD Vertices · Colour Factors · Asymptotic Freedom · Colour Confinement.

Ask the tutor: Want to see how the gluon self-coupling in that $f^{abc}$ term turns into asymptotic freedom, or how the colour factor $-\tfrac43$ becomes the confining potential that binds a proton?