Colour confinement

Every other charge in physics can be isolated. Rub a balloon and you carry off a net electric charge; a single electron sits in a Penning trap for months. Colour is the one exception. Half a century of searches — Millikan-style oil drops, deep-sea sediments, Moon rock, high-energy collisions — has never turned up a free quark, or any object carrying a net colour charge, or the fractional electric charge $\pm\tfrac13 e$ , $\pm\tfrac23 e$ that a bare quark would wear. Colour confinement is the working name for this brute fact: coloured objects are permanently locked inside colour-singlet combinations, and nothing with non-zero colour ever propagates as a free particle. The mechanism is written into the gluon itself. Unlike the photon, which is electrically neutral and lets the electric field of a charge fan out as a $1/r^2$ Coulomb field, the gluon carries colour. Gluons therefore attract one another through their own triple- and quartic-gluon vertices, and the colour field between two separating quarks does not spread — it collapses into a narrow flux tube of roughly constant energy per unit length. Pull the quarks apart and the stored energy grows linearly, $V(r)\simeq\kappa r$ with a string tension $\kappa\approx0.9\ \mathrm{GeV/fm}$ , a constant restoring force of about $1.4\times10^{5}\ \mathrm{N}$ — fifteen tonnes of weight — that never dies away. Long before the quarks reach macroscopic separation the tube snaps, converting its energy into fresh quark–antiquark pairs. That is why $e^{+}e^{-}\to q\bar q$ produces back-to-back jets of colourless hadrons and never two lone quarks. There is still no analytic proof that QCD confines — it is one of the Clay Millennium Prize problems — but lattice simulations demonstrate it numerically and reproduce $\kappa$ from first principles.

Key takeaways

No free quark or fractionally charged particle has ever been seen; coloured objects exist only inside colour singlets — mesons ( $q\bar q$ ), baryons ( $qqq$ ) and antibaryons.
Confinement traces to gluon self-interaction: gluons carry colour, attract one another, and squeeze the colour field into a flux tube of constant energy density.
The static potential is Coulombic at short range and linear at long range, $V(r)=-\tfrac{4}{3}\dfrac{\alpha_s}{r}+\kappa r$ , with string tension $\kappa\approx0.9\ \mathrm{GeV/fm}\approx0.18\ \mathrm{GeV^2}$ .
One-gluon exchange binds the singlet ( $\mathbf T_1\!\cdot\!\mathbf T_2=-\tfrac43$ , attractive) and repels the octet ( $+\tfrac16$ ) — group theory alone selects the colourless channel.
The tube stores so much energy that it breaks into new $q\bar q$ pairs at $r\approx1.2\ \mathrm{fm}$ — the origin of hadronisation and jets.
Confinement has no analytic proof; the rigorous criterion is the Wilson-loop area law $\langle W\rangle\sim e^{-\sigma A}$ , verified on the lattice.
Key idea: because gluons carry colour and attract one another, the field between separating quarks collapses into a tube of constant energy density, so the energy grows without bound with distance and only colour-singlet combinations can ever be free.

The core insight

Because gluons carry colour and attract each other, the colour field between separating quarks collapses into a tube of constant energy density, making the stored energy grow linearly and without bound with distance — so no coloured object can be isolated and only colour-neutral (singlet) combinations exist as free particles.

Where this lives today

Confinement is simultaneously routine engineering and a frontier mystery. On the practical side, lattice QCD collaborations compute the static potential, the string tension, and the light-hadron spectrum to percent accuracy on the world's largest supercomputers, feeding the precision predictions that underpin flavour physics and the search for new phenomena at the LHC. In hadron spectroscopy the linear Regge trajectories $J=\alpha'M^2+\alpha_0$ — a direct fingerprint of the flux-tube string — organise hundreds of measured resonances, and newly discovered tetraquark and pentaquark states are being tested for whether they too are ordinary colour singlets or genuinely exotic bindings. At the extreme of temperature, heavy-ion collisions at RHIC and the LHC melt hadrons into a quark–gluon plasma above $T_c\approx155\ \mathrm{MeV}$ — the deconfined phase that filled the microsecond-old universe — probed through jet quenching and quarkonium suppression. And overarching it all sits the Yang–Mills mass-gap problem: a rigorous proof that pure $SU(3)$ gauge theory confines and has a finite mass gap carries a one-million-dollar Clay Prize and remains, half a century after Wilson, unclaimed.

The physical picture

The contrast with electromagnetism is the whole story. Two opposite electric charges share a dipole field whose lines bow outward through all of space; the energy needed to separate them saturates at a finite value, so you can always tear the pair apart and keep the halves. Between a quark and an antiquark the gluon field refuses to spread. Because gluons are themselves coloured, the field lines attract sideways and bundle into a thin tube — a "string" — whose energy per unit length is fixed. Doubling the separation doubles the tube, doubles the energy, and the force between the ends never falls off. The first figure sets the two pictures side by side and attaches the number that makes confinement inescapable.

Figure

QED versus QCD: the electric dipole field spreads through space and its energy saturates, but the gluon field between a quark and antiquark collapses into a flux tube of constant energy per length, giving a linear potential V ≈ κr with κ ≈ 0.9 GeV/fm ≈ 1.4×10⁵ N

Because the force never dies, separating a quark to infinity would cost infinite energy — so it never happens. What happens instead is that the tube stores enough energy to pay for a new quark–antiquark pair out of the vacuum, snaps in the middle, and caps each broken end with a fresh partner. The short-distance behaviour is entirely different: there the coupling is weak (asymptotic freedom) and the potential is an ordinary Coulomb-like $-\tfrac43\alpha_s/r$ . Stitching the two regimes together gives the Cornell potential, the single most useful curve in quarkonium physics, and the second figure plots it with real units and marks where the string breaks.

Figure

The Cornell potential V(r) = −(4/3)α_s/r + κr with α_s ≈ 0.3 and κ ≈ 0.9 GeV/fm: Coulombic at short range, linear at long range, and breaking near r ≈ 1.2 fm where the stored energy ≈ 1.1 GeV materialises a new quark–antiquark pair

The core equations

The static potential between a heavy quark and antiquark is the quantitative face of confinement. Perturbative one-gluon exchange supplies the short-distance Coulomb term; the non-perturbative flux tube supplies the linear term. Together they are the Cornell potential, which fits the charmonium and bottomonium spectra with only two parameters.

V(r)=-\frac{4}{3}\,\frac{\alpha_s}{r}+\kappa\,r \qquad (1)

where: $\alpha_s$ — the strong coupling (run at the relevant scale, $\alpha_s\approx0.3$ for charmonium); $\tfrac43$ — the colour factor $C_F$ of the attractive singlet channel; $\kappa$ — the string tension, $\kappa\approx0.9\ \mathrm{GeV/fm}\approx0.18\ \mathrm{GeV^2}$ ; $r$ — the quark separation. Reading: at small $r$ the $1/r$ term dominates and quarks are nearly free; at large $r$ the $\kappa r$ term dominates and the force $\mathrm{d}V/\mathrm{d}r\to\kappa$ becomes constant — confinement. Confinement permits only colour singlets to exist as free particles, and there is exactly one singlet in each allowed hadron channel. The meson singlet is the symmetric colour–anticolour sum; the baryon singlet is the totally antisymmetric combination of three colours.

|M\rangle=\tfrac{1}{\sqrt3}\bigl(r\bar r+g\bar g+b\bar b\bigr),\qquad |B\rangle=\tfrac{1}{\sqrt6}\,\varepsilon_{ijk}\,q^{i}q^{j}q^{k}=\tfrac{1}{\sqrt6}\bigl(rgb-rbg+gbr-grb+brg-bgr\bigr) \qquad (2)

where: $r,g,b$ — the three colour states of a quark, $\bar r,\bar g,\bar b$ — the anticolours; $\varepsilon_{ijk}$ — the totally antisymmetric Levi-Civita symbol on the three colours. Reading: both states are annihilated by every $SU(3)$ colour raising and lowering operator — the defining property of a singlet — so they carry no net colour and can propagate freely. Which multi-quark combinations contain a singlet is fixed by the $SU(3)$ tensor product decompositions. Only the meson and baryon channels yield one; diquarks and $qq\bar q$ do not, which is why they are never seen as free states.

\mathbf 3\otimes\bar{\mathbf 3}=\mathbf 8\oplus\mathbf 1,\qquad \mathbf 3\otimes\mathbf 3\otimes\mathbf 3=\mathbf{10}\oplus\mathbf 8\oplus\mathbf 8\oplus\mathbf 1,\qquad \mathbf 3\otimes\mathbf 3=\mathbf 6\oplus\bar{\mathbf 3} \qquad (3)

where: boldface numbers are $SU(3)$ colour multiplet dimensions; $\mathbf 1$ is the singlet. Reading: $q\bar q$ and $qqq$ each contain exactly one $\mathbf 1$ (allowed hadrons), while $qq$ ( $\mathbf 6\oplus\bar{\mathbf 3}$ ) and $qq\bar q$ contain no $\mathbf 1$ and so cannot form isolated particles. The rigorous, non-perturbative criterion for confinement is the behaviour of the Wilson loop — the trace of the gauge field carried around a closed rectangular contour of spatial width $r$ and temporal extent $T$ . An area law signals a linearly confining potential.

\langle W(C)\rangle\sim e^{-\sigma\,A}=e^{-\sigma\,rT}\ \Longrightarrow\ V(r)=\sigma\,r \qquad (4)

where: $A=rT$ — the area enclosed by the loop; $\sigma$ — the string tension (the same $\kappa$ as above); a perimeter law $e^{-cP}$ instead would signal a screened, deconfined phase. Reading: because the expectation value falls with the enclosed area rather than its perimeter, the energy of a static pair grows linearly with separation — this is the definition of confinement that lattice QCD actually tests.

Derivation AWhy the colour singlet binds and the octet does not

The goal is to show, from group theory alone, that one-gluon exchange makes the colour-singlet channel of a quark–antiquark pair attractive while the octet channel is repulsive — so the strong force does not merely permit colourless hadrons, it actively selects them. We compute the colour factor $\mathbf T_1\!\cdot\!\mathbf T_2$ that multiplies the Coulomb potential using the Casimir trick, evaluate it in the singlet and octet channels, and read off the sign and magnitude that appear as the $\tfrac43$ in the Cornell potential.

Step 1. One-gluon exchange between two colour charges gives a potential $V(r)=(\mathbf T_1\!\cdot\!\mathbf T_2)\,\alpha_s/r$ , where $\mathbf T_i$ is the vector of $SU(3)$ generators acting on particle $i$ (for an antiquark, in the conjugate representation). The entire channel dependence sits in the number $\mathbf T_1\!\cdot\!\mathbf T_2$ . Why: the spatial and coupling parts are common to all colour channels, so a single group-theory factor decides whether a given channel attracts or repels.

Step 2. Use the Casimir identity. Writing the total colour generator $\mathbf T=\mathbf T_1+\mathbf T_2$ , square it: $\mathbf T^2=\mathbf T_1^2+\mathbf T_2^2+2\,\mathbf T_1\!\cdot\!\mathbf T_2$ , so $\mathbf T_1\!\cdot\!\mathbf T_2=\tfrac12\bigl[C_2(\text{total})-C_2(1)-C_2(2)\bigr]$ . The quadratic Casimir is $C_2(\mathbf 3)=C_2(\bar{\mathbf 3})=\tfrac43$ for a single quark or antiquark, $C_2(\mathbf 1)=0$ , and $C_2(\mathbf 8)=3$ . Why: the Casimir converts an awkward operator dot-product into a difference of known numbers labelling the representations of the pair and its constituents.

Step 3. Evaluate the singlet channel, where the $q\bar q$ pair combines to total colour $\mathbf 1$ : $\mathbf T_1\!\cdot\!\mathbf T_2=\tfrac12\bigl[0-\tfrac43-\tfrac43\bigr]=-\tfrac43$ . The potential is $V=-\tfrac43\,\alpha_s/r$ , negative — attractive. This is precisely the coefficient appearing in equation (1). Why: a negative colour factor means the exchanged gluon lowers the energy as the pair draws together, so the colourless channel is the one that binds.

Step 4. Evaluate the octet channel, where the pair combines to $\mathbf 8$ : $\mathbf T_1\!\cdot\!\mathbf T_2=\tfrac12\bigl[3-\tfrac43-\tfrac43\bigr]=+\tfrac16$ . The potential $V=+\tfrac16\,\alpha_s/r$ is positive — repulsive. So the coloured octet configuration is energetically disfavoured relative to the singlet by a factor of eight in the coefficient and by its opposite sign. Why: the same exchange that binds the singlet pushes the octet apart, which is the microscopic reason free coloured pairs are not merely forbidden by fiat but dynamically suppressed. *Verified: $C_2(\mathbf 3)=\tfrac43$ , $C_2(\mathbf 8)=3$ ; singlet $\tfrac12(0-\tfrac43-\tfrac43)=-\tfrac43$ ; octet $\tfrac12(3-\tfrac83)=+\tfrac16$ ; ratio and signs match Halzen–Martin §2.15 and the Cornell coefficient.*

Go deeper · Derivation B · 2 minThe flux tube, the linear potential, and where the string breaks

Derivation BThe flux tube, the linear potential, and where the string breaks

This derivation builds the confining term $\kappa r$ from a simple physical model of the colour field and then shows that the same linear growth guarantees the field cannot stretch forever — it must break and hadronise. We treat the field between separated quarks as a tube of fixed cross-sectional area carrying constant energy density, integrate to get the potential, differentiate to get the constant force, and finally balance the stored energy against the rest energy of a light quark–antiquark pair to find the breaking distance that underlies jet formation.

Step 1. Model the colour field between a widely separated $q$ and $\bar q$ as a tube of constant cross-section $\mathcal A$ carrying a constant energy density $u$ . The stored energy is $V(r)=u\,\mathcal A\,r\equiv\kappa r$ , defining the string tension $\kappa=u\mathcal A$ as energy per unit length. Why: the non-Abelian self-attraction of gluons is what keeps $\mathcal A$ from growing with $r$ ; with a fixed tube, energy can only accumulate linearly.

Step 2. The force between the ends is $F=-\mathrm{d}V/\mathrm{d}r=-\kappa$ , a constant independent of separation. Numerically $\kappa\approx0.9\ \mathrm{GeV/fm}=0.9\times(1.602\times10^{-10}\ \mathrm J)/(10^{-15}\ \mathrm m)\approx1.4\times10^{5}\ \mathrm N$ . Why: a distance-independent restoring force is the sharp signature that separates confinement from every $1/r^2$ force in physics — you cannot outrun it by pulling harder.

Step 3. Because $V(r)\to\infty$ as $r\to\infty$ , isolating a quark costs infinite energy and is impossible. Instead, once the tube has stored enough energy to create a light $q'\bar q'$ pair — roughly $2m_{q'}$ in constituent terms, of order $1\ \mathrm{GeV}$ — it becomes energetically favourable for the vacuum to pop such a pair into the middle of the tube. Why: the vacuum will always take the cheaper option; making two shorter strings each capped by a new quark costs less than continuing to stretch one long one.

Step 4. Estimate the breaking distance from $\kappa r_\ast\approx1\ \mathrm{GeV}$ : $r_\ast\approx(1\ \mathrm{GeV})/(0.9\ \mathrm{GeV/fm})\approx1.1\text{–}1.2\ \mathrm{fm}$ , about one hadron diameter. The original $q\bar q$ and the new $q'\bar q'$ regroup into two colour-singlet hadrons. Why: this is exactly hadronisation — the reason a struck quark in $e^{+}e^{-}\to q\bar q$ emerges as a collimated jet of hadrons rather than as a bare quark, and the reason no experiment has ever caught a free colour charge in flight. *Verified: $F=\kappa\approx0.9\ \mathrm{GeV/fm}\approx1.4\times10^{5}\ \mathrm N$ ; breaking distance $r_\ast\sim1.2\ \mathrm{fm}$ matches lattice string-breaking studies; $\kappa\approx0.18\ \mathrm{GeV^2}$ consistent with the Regge-slope value below.*

Go deeper · Derivation C · 2 minWhy the gauge group is SU(3) and not U(3): the fate of the ninth gluon

Derivation CWhy the gauge group is SU(3) and not U(3): the fate of the ninth gluon

The aim here is to explain why there are exactly eight gluons and not nine, and why that number is not an accident but a requirement for confinement to make sense. A naive count of colour–anticolour combinations gives nine; we show that the ninth combination is a colour singlet, that a singlet gluon would mediate an unconfined, long-range strong force between ordinary hadrons, and that its absence in nature is precisely the statement that the gauge group is the traceless $SU(3)$ rather than the full $U(3)$ .

Step 1. Gluons carry a colour and an anticolour, so there are naively $3\times3=9$ combinations. Group-theoretically these decompose as $\mathbf 3\otimes\bar{\mathbf 3}=\mathbf 8\oplus\mathbf 1$ : an octet of eight coloured gluons plus one colour-singlet combination $\tfrac1{\sqrt3}(r\bar r+g\bar g+b\bar b)$ . Why: the ninth state is not an independent coloured gluon but the symmetric, colour-neutral trace — exactly the extra generator that distinguishes $U(3)$ from $SU(3)$ .

Step 2. Ask what this singlet gluon would do. Being colourless, it would not couple to itself and would not be confined; it would couple democratically to the total colour charge of any object — including colour-singlet hadrons — and so mediate a long-range, Coulomb-like strong force between neutrons, protons and atoms. Why: a confined force is possible only because its carriers are themselves charged and self-interacting; a neutral carrier behaves like the photon and produces an infinite-range force.

Step 3. No such long-range strong force is observed: nuclear forces are short-ranged and there is no "strong photon" acting between distant hadrons. Nature therefore excludes the singlet gluon, leaving exactly the eight generators of $SU(3)$ . Why: promoting colour to the traceless group $SU(3)$ rather than $U(3)$ removes the singlet by hand, and this is the choice the data force upon us.

Step 4. The eight surviving gluons all carry colour and self-interact through the triple- and quartic-gluon vertices, which is what makes the field lines attract and form the flux tube of the previous derivation. Why: the very feature that eliminates the ninth gluon — non-Abelian, self-coupled gauge fields — is the same feature that produces confinement, tying the group choice and the mechanism together. *Verified: $\mathbf 3\otimes\bar{\mathbf 3}=\mathbf 8\oplus\mathbf 1$ (dimension $9=8+1$ ); the singlet $\tfrac1{\sqrt3}(r\bar r+g\bar g+b\bar b)$ is annihilated by all $SU(3)$ ladder operators; $SU(3)$ has $3^2-1=8$ generators.*

Limiting cases

Case	Result	Meaning
Short distance $r\to0$	$V\approx-\tfrac43\,\alpha_s/r$	Coulombic, asymptotic freedom — quarks nearly free, perturbation theory valid
Long distance $r\to\infty$	$V\approx\kappa r$ , $F\to\kappa$	linear confinement — constant force $\approx1.4\times10^{5}\ \mathrm N$ , no isolation possible
String-breaking $\kappa r\sim2m_q$	new $q\bar q$ pair created	hadronisation; struck quarks emerge as jets, not free quarks
High temperature $T>T_c\approx155\ \mathrm{MeV}$	$\sigma\to0$ , deconfinement	quark–gluon plasma; colour charges screened, not liberated as singlets
Wilson loop, confined phase	$\langle W\rangle\sim e^{-\sigma A}$ (area law)	rigorous criterion; perimeter law instead means screening/deconfinement

Common trapA state with I₃ colour = Y colour = 0 must be colourless

Not so — vanishing colour-isospin and colour-hypercharge are necessary but nowhere near sufficient. Two of the eight gluons carry $I_3^{c}=Y^{c}=0$ (they are the diagonal combinations $\tfrac1{\sqrt2}(r\bar r-g\bar g)$ and $\tfrac1{\sqrt6}(r\bar r+g\bar g-2b\bar b)$ ), yet they are fully coloured members of the octet and are confined like every other gluon. Being a true colour singlet is a much stronger condition: the state must be annihilated by all the colour raising and lowering operators, not merely have zero eigenvalue under the two diagonal (Cartan) generators. Only the symmetric combination $\tfrac1{\sqrt3}(r\bar r+g\bar g+b\bar b)$ passes that test — which is exactly why it is the meson singlet and not a gluon. The lesson: "colour-neutral" means invariant under the whole $SU(3)$ , not just uncharged under its two commuting generators.

Worked examples

1 · The string tension from three directions (fully worked)

Take the string tension $\kappa\approx0.9\ \mathrm{GeV/fm}$ . Find (a) the confining force in newtons, (b) the separation at which the tube breaks, and (c) cross-check $\kappa$ against the measured Regge slope $\alpha'\approx0.93\ \mathrm{GeV^{-2}}$ .

Step 1. Convert the force. The force between the tube ends is $F=\kappa$ . Using $1\ \mathrm{GeV}=1.602\times10^{-10}\ \mathrm J$ and $1\ \mathrm{fm}=10^{-15}\ \mathrm m$ , $F=0.9\times\dfrac{1.602\times10^{-10}\ \mathrm J}{10^{-15}\ \mathrm m}=\mathbf{1.4\times10^{5}\ N}$ . That is the weight of roughly 15 tonnes, and it does not diminish with distance.

Step 2. Find the breaking distance. The tube snaps once its stored energy $\kappa r$ can pay for a light $q\bar q$ pair, about $1\ \mathrm{GeV}$ : $r_\ast\approx\dfrac{1\ \mathrm{GeV}}{0.9\ \mathrm{GeV/fm}}\approx\mathbf{1.2\ fm}$ . Beyond this the vacuum inserts a new pair and two colourless hadrons emerge — the microscopic act of hadronisation.

Step 3. Convert $\kappa$ to natural units. Using $\hbar c=0.1973\ \mathrm{GeV\,fm}$ , $\kappa=0.9\ \mathrm{GeV/fm}\times0.1973\ \mathrm{GeV\,fm}=0.178\ \mathrm{GeV^2}$ , so $\sqrt{\kappa}\approx\mathbf{0.42\ GeV}=420\ \mathrm{MeV}$ — the natural energy scale of confinement, close to $\Lambda_{\mathrm{QCD}}$ .

Step 4. Cross-check with Regge phenomenology. A rotating relativistic string predicts hadron masses on linear Regge trajectories $J=\alpha' M^2+\alpha_0$ with slope $\alpha'=1/(2\pi\sigma)$ . Inverting, $\sigma=\dfrac{1}{2\pi\alpha'}=\dfrac{1}{2\pi\times0.93\ \mathrm{GeV^{-2}}}=\mathbf{0.171\ GeV^2}$ , i.e. $\approx0.87\ \mathrm{GeV/fm}$ . Independent hadron-spectroscopy data give the same $\kappa\approx0.9\ \mathrm{GeV/fm}$ as the potential model — a striking consistency for a quantity with no free parameters.

2 · Counting the colour singlets

Verify from $SU(3)$ that $q\bar q$ and $qqq$ each admit exactly one colour singlet, while $qq$ admits none, and confirm the octet channel is repulsive.

Step 1. Mesons: $\mathbf 3\otimes\bar{\mathbf 3}=\mathbf 8\oplus\mathbf 1$ (dimensions $9=8+1$ ). There is exactly one singlet, $\tfrac1{\sqrt3}(r\bar r+g\bar g+b\bar b)$ — the colour wavefunction of every meson.

Step 2. Baryons: $\mathbf 3\otimes\mathbf 3\otimes\mathbf 3=\mathbf{10}\oplus\mathbf 8\oplus\mathbf 8\oplus\mathbf 1$ (dimensions $27=10+8+8+1$ ). Again exactly one singlet, the totally antisymmetric $\tfrac1{\sqrt6}\varepsilon_{ijk}q^iq^jq^k$ — which is why baryons need three quarks of three different colours ("red + green + blue = white").

Step 3. Diquarks: $\mathbf 3\otimes\mathbf 3=\mathbf 6\oplus\bar{\mathbf 3}$ (dimensions $9=6+3$ ). No singlet appears, so a $qq$ pair — and likewise $qq\bar q$ — cannot be a free colour-neutral particle. This is the group-theoretic reason the observed hadrons are only $q\bar q$ and $qqq$ (plus their antistates and, more recently, singlet-combining tetra- and pentaquarks).

Step 4. Bind versus repel. From Derivation A, the singlet colour factor is $\mathbf T_1\!\cdot\!\mathbf T_2=-\tfrac43$ (attractive) and the octet is $+\tfrac16$ (repulsive). So the singlet is not just allowed but energetically preferred, and coloured configurations are actively pushed apart — the double reason confinement leaves only singlets behind.

Physics of the portrait

Confinement is the one place in the Standard Model where the decisive progress came not from a pencil-and-paper proof but from a computer. The person who made that possible reformulated gauge theory on a grid so that the question "does QCD confine?" could finally be asked precisely.

Origins · 1 minKenneth Wilson and how this was discovered

Kenneth Wilson

1936–2013 · lattice gauge theory and the confinement criterion

Kenneth Wilson came to confinement from an unlikely direction: the theory of phase transitions. His renormalization-group analysis of critical phenomena — for which he won the 1982 Nobel Prize — taught him to think about physics at every length scale at once, and in 1974 he turned that instinct on quark confinement. Perturbation theory was useless there, because confinement is inherently strong-coupling, so Wilson did something radical: he replaced continuous spacetime with a discrete lattice of points and formulated $SU(3)$ gauge theory directly on it, with the gauge field living on the links between sites. On this lattice he defined the object now called the Wilson loop and proved that, in the strong-coupling limit, its expectation value falls off as the exponential of the area it encloses — an area law — which translates immediately into a linearly rising potential $V(r)=\sigma r$ between static quarks. That was the first mathematically sharp criterion for confinement, and it converted a vague hypothesis into a computable quantity. Lattice QCD, the entire field that grew from Wilson's 1974 paper, now calculates the string tension, the hadron spectrum, and the deconfinement temperature from first principles on supercomputers, and reproduces the $\kappa\approx0.9\ \mathrm{GeV/fm}$ measured in quarkonium. A full analytic proof that continuum Yang–Mills theory has a mass gap and confines remains open — it is a Clay Millennium Prize problem — but every numerical test passes, and the framework in which those tests are run is Wilson's.

History

1964

— Gell-Mann and Zweig propose quarks; their fractional charges are never seen in isolation

1965

— Han and Nambu introduce a three-valued colour charge, the seed of the SU(3) gauge symmetry

1973

— Gross, Wilczek and Politzer find asymptotic freedom; Fritzsch, Gell-Mann and Leutwyler formulate QCD

1974

— Wilson defines lattice gauge theory and the area-law criterion for confinement

2005

— RHIC reports a deconfined quark–gluon plasma, the strongly coupled high-temperature phase of QCD

Going further

Texts. Thomson, Modern Particle Physics, §10.4 (colour confinement), §10.4.1 (confinement and hadronic states), §10.4.2 (hadronisation and jets), §9.6.4 ( $SU(3)$ singlet wavefunctions) · Halzen & Martin, Quarks and Leptons, §2.15 (colour factors) and Ch. 14 (QCD) · Griffiths, Introduction to Elementary Particles, 2nd ed., §8.4–8.5 · Peskin & Schroeder, An Introduction to Quantum Field Theory, §17 (Wilson loops and confinement) · Schwartz, Quantum Field Theory and the Standard Model, Ch. 25–26.

Provenance. The colour $SU(3)$ gauge structure and the eight gluons are recalled by name from aq-pp-colour-qcd and aq-pp-gluons; the singlet-counting machinery from aq-pp-su3-flavour; the experimental case for colour from aq-pp-colour-evidence. The Coulomb-plus-linear potential is developed quantitatively in aq-pp-heavy-quark-potential, the short-distance running that justifies the $-\tfrac43\alpha_s/r$ term in aq-pp-asymptotic-freedom and aq-pp-qcd-vertices, and the string-breaking picture that produces jets in aq-pp-jets-fragmentation (forward references, not re-derived here). A verifier should confirm: (1) the colour factors $-\tfrac43$ (singlet) and $+\tfrac16$ (octet) from the Casimirs; (2) $\kappa\approx0.9\ \mathrm{GeV/fm}\approx1.4\times10^{5}\ \mathrm N\approx0.18\ \mathrm{GeV^2}$ and $\sqrt\kappa\approx420\ \mathrm{MeV}$ ; (3) the Regge cross-check $\sigma=1/(2\pi\alpha')\approx0.17\ \mathrm{GeV^2}$ ; (4) the decompositions $\mathbf 3\otimes\bar{\mathbf 3}=\mathbf 8\oplus\mathbf 1$ , $\mathbf 3\otimes\mathbf 3\otimes\mathbf 3=\mathbf{10}\oplus\mathbf 8\oplus\mathbf 8\oplus\mathbf 1$ , $\mathbf 3\otimes\mathbf 3=\mathbf 6\oplus\bar{\mathbf 3}$ .

Prerequisites. Colour and QCD · Gluons · $SU(3)$ Flavour · Evidence for Colour. Leads to. Jets and Fragmentation · Heavy-Quark Potential · QCD Vertices · Asymptotic Freedom.

Ask the tutor: Ask me to derive the $-\tfrac43$ colour factor step by step, estimate where a flux tube breaks, or explain why lattice QCD's area law is the sharp definition of confinement.

Has colour confinement actually been proven?

Not analytically, and this is a genuine open problem. There is no mathematical proof that the QCD Lagrangian implies that all physical states are colour singlets; establishing that pure $SU(3)$ Yang–Mills theory has a mass gap and confines is one of the seven Clay Millennium Prize problems, worth a million dollars. What we do have is overwhelming numerical evidence: lattice QCD, which puts the theory on a discrete spacetime grid and evaluates it on supercomputers, exhibits the area law, computes a non-zero string tension, and reproduces the observed hadron spectrum — all hallmarks of confinement. So confinement is established as a property of QCD in the same sense a well-tested numerical result is established, just not yet as a theorem.

How is confinement different from asymptotic freedom?

They are two ends of the same coupling. Asymptotic freedom says the strong coupling $\alpha_s$ decreases at short distances (high energies), so quarks probed at tiny separations behave almost as free particles — this is the $-\tfrac43\alpha_s/r$ Coulombic part of the potential and what deep-inelastic scattering sees. Confinement is the opposite regime: at large distances (low energies) $\alpha_s$ grows, the flux tube forms, and the potential rises linearly. The same running coupling that fades at short range becomes strong at long range. Remarkably, asymptotic freedom is proven perturbatively while confinement is not — they are the calculable and the still-uncalculable faces of one theory.

Do free gluons exist, since they mediate the force?

No — gluons carry colour, so they are confined exactly like quarks and never appear as free particles. But QCD does predict bound states made purely of gluons, called glueballs, which are colour singlets built from two or more gluons and are perfectly allowed. Lattice QCD predicts the lightest glueball (a scalar) around $1.5\text{–}1.7\ \mathrm{GeV}$ , but glueballs mix with ordinary $q\bar q$ mesons of the same quantum numbers, making them notoriously hard to identify unambiguously in experiments. So "a free gluon" is forbidden, while "a colourless ball of glue" is expected and actively hunted.

Where do jets come from, physically?

Directly from string breaking. When a quark is knocked out of a hadron — say in $e^{+}e^{-}\to q\bar q$ — it and its partner fly apart, and a flux tube stretches between them. As the separation grows the tube stores energy at the constant rate $\kappa$ , and once that energy reaches about a GeV the tube snaps by creating a new $q\bar q$ pair from the vacuum. This repeats, cascading the original quark's energy into many colour-singlet hadrons, all travelling in roughly the same direction as the parent quark. The collimated spray you see in a detector — a jet — is therefore a fossil record of a quark that was never allowed to escape as itself. Two-jet events map the underlying $q\bar q$ ; three-jet events revealed the gluon.

Where does most of the proton's mass come from, then?

Overwhelmingly from confinement, not from the quark masses. The up and down quark masses sum to only a few MeV, yet the proton weighs $938\ \mathrm{MeV}$ . The rest is the energy of the confined gluon field and the relativistic motion of the quarks trapped inside — the QCD binding energy made visible through $E=mc^2$ . In this sense over 98% of the mass of ordinary matter, and hence of you, is stored confinement energy rather than Higgs-generated quark mass. Confinement is not an obscure technicality; it is where most of the mass in the visible universe actually resides.

Is a quark–gluon plasma a gas of free quarks?

Not in the sense of liberated colour charges. Above a critical temperature $T_c\approx155\ \mathrm{MeV}$ the hadrons dissolve and colour charge becomes deconfined — quarks and gluons can move over distances larger than a hadron because their colour fields are screened by the surrounding hot medium, much as charges are screened in a plasma. But the plasma as a whole is still colour-neutral, and as it cools and expands it re-hadronises into ordinary colourless particles before anything reaches a detector. So deconfinement relaxes confinement over a droplet of hot matter; it never delivers an isolated, free-flying quark to the laboratory.

Why colour singlets specifically, and not just electric-charge neutrality?

Because the confining force couples to colour, not to electric charge. Confinement is a statement about the $SU(3)$ colour gauge symmetry: only states invariant under the full colour group — annihilated by every colour generator — have no flux tube attached and can propagate freely. Electric neutrality is a separate, weaker condition tied to a different (Abelian) symmetry; a colour-octet state can be electrically neutral yet still coloured and still confined, and conversely a proton is a colour singlet but carries electric charge $+e$ perfectly happily. The rule that permits free particles is colour-singletness, full stop, and the "necessary but not sufficient" trap in the warning box is exactly the failure to appreciate this.