QCD Theta Term in Strong Interactions

• The QCD theta term is a CP-violating, dimension-four operator arising from the topological structure of gauge fields that labels distinct vacuum sectors.
• It influences key observables such as topological susceptibility and hadronic electric dipole moments through nonperturbative effects and chiral effective theories.
• Recent investigations employing lattice QCD, chiral perturbation theory, and geometric/Berry-phase frameworks elucidate its role in CP violation and offer diverse methodological insights.

The QCD theta term is the CP-odd topological term that can be added to the strong-interaction Lagrangian,

$$\mathcal L_\theta = -\,\theta\,\frac{g^2}{32\pi^2}\,F^a_{\mu\nu}\tilde F_a^{\mu\nu},$$

or, in trace conventions,

$$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$

Its local density $q(x)\propto F\tilde F$ is a total derivative, yet its spacetime integral

$Q=\int d^4x\,q(x)$

is sensitive to the nontrivial topology of gauge fields and labels topological sectors. For that reason the term is invisible in perturbation theory but physically consequential nonperturbatively, where it controls CP violation, vacuum structure, topological susceptibility, and a wide range of low-energy observables (Kim et al., 2013, Bhattacharya et al., 2023, Ai et al., 2024).

1. Operator structure, topology, and discrete symmetries

In explicit component form the non-Abelian field strength is

$$F_{\mu\nu}^a=\partial_\mu A_\nu^a-\partial_\nu A_\mu^a+g f^{abc}A_\mu^bA_\nu^c, \qquad \tilde F_a^{\mu\nu}=\frac12 \epsilon^{\mu\nu\rho\sigma}F_{\rho\sigma}^a,$$

so the topological density is

$$q(x)=\frac{g^2}{32\pi^2}F_{\mu\nu}^a\tilde F_a^{\mu\nu}.$$

Equivalent normalizations using $\mathrm{Tr}\,F\tilde F$ also appear in the literature, reflecting convention rather than substance (Kim et al., 2013, Thacker, 2013).

The theta term is Lorentz and gauge invariant, but it is odd under $P$ and $T$, and therefore CP-violating for $\theta\neq 0 \mod \pi$ (Zhang et al., 3 Jul 2025). In perturbation theory $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$0, with $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$1 a Chern–Simons current, so the term does not modify local classical equations of motion. Nonperturbatively, however, finite-action gauge configurations carry integer Pontryagin index, and the path integral decomposes into sectors weighted by $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$2 (Kim et al., 2013, Gamboa, 30 May 2025).

This topological character underlies the standard statement that the theta term is the unique dimension-four CP-violating operator in low-energy QCD (Bhattacharya et al., 2023). It also underlies the strong-CP problem: experimental bounds on hadronic EDMs require the physical angle to be extremely small; values quoted in the cited literature include $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$3, $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$4, and $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$5 (Kim et al., 2013, Xiong, 2013, Bhattacharya et al., 2023).

2. Vacuum angle, axial anomaly, and topological sectors

The conventional vacuum picture is that QCD possesses multiple sectors labeled by integer $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$6, and the physical vacuum is a $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$7-vacuum,

$$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$8

or, equivalently in the Euclidean path integral,

$$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$9

(Kim et al., 2013, Gamboa, 30 May 2025). Because $q(x)\propto F\tilde F$0, physics is $q(x)\propto F\tilde F$1-periodic in $q(x)\propto F\tilde F$2 (Cai et al., 2016, Thacker, 2013).

Via the axial anomaly, the theta term can be rotated into complex quark masses. For one light flavor this yields

$q(x)\propto F\tilde F$3

which makes the CP-odd pseudoscalar component explicit (Kim et al., 2013). More generally the observable parameter is

$q(x)\propto F\tilde F$4

not the bare $q(x)\propto F\tilde F$5 alone (Bhattacharya et al., 2023).

At small quark mass, low-energy effective theory and holographic constructions recover the standard $q(x)\propto F\tilde F$6-dependence of the vacuum energy and topological susceptibility. In V-QCD, the vacuum energy is multi-branched and becomes $q(x)\propto F\tilde F$7-periodic only after taking the minimum over branches, while the topological susceptibility interpolates between the chiral regime,

$q(x)\propto F\tilde F$8

and the pure-Yang–Mills limit at large quark mass (Jarvinen, 2016). A related topological-insulator-inspired picture interprets $q(x)\propto F\tilde F$9 as an order parameter for topological polarization, with quasi-vacua separated by domain walls across which the local value of $Q=\int d^4x\,q(x)$0 jumps by $Q=\int d^4x\,q(x)$1 (Thacker, 2013).

A distinct, nonstandard canonical-quantization analysis on $Q=\int d^4x\,q(x)$2 reaches a different conclusion for pure Yang–Mills: after gauge fixing and restricting to a properly normalizable Hilbert space, the Hamiltonian can be made $Q=\int d^4x\,q(x)$3-independent, and observables derived from the constrained Hilbert space do not violate CP (Ai et al., 2024). That result is explicitly limited to pure Yang–Mills and does not analyze the full $Q=\int d^4x\,q(x)$4 problem with dynamical quarks.

3. Low-energy realization in chiral effective theory

At hadronic scales, the theta term induces CP-odd pion–nucleon and nucleon–photon operators. In two-flavor chiral EFT, a careful vacuum-alignment analysis removes pion tadpoles and yields a hierarchy of CP-odd couplings tied to isospin breaking (Mereghetti et al., 2010). The central structural result is that the $Q=\int d^4x\,q(x)$5-induced $Q=\int d^4x\,q(x)$6 spurion and the strong isospin-breaking $Q=\int d^4x\,q(x)$7 spurion are components of the same chiral $Q=\int d^4x\,q(x)$8 vector, so low-order time-reversal-violating couplings are related to charge-symmetry-breaking ones by a universal factor

$Q=\int d^4x\,q(x)$9

(Mereghetti et al., 2010).

For the CP-odd pion–nucleon sector the leading operators are commonly written as

$$F_{\mu\nu}^a=\partial_\mu A_\nu^a-\partial_\nu A_\mu^a+g f^{abc}A_\mu^bA_\nu^c, \qquad \tilde F_a^{\mu\nu}=\frac12 \epsilon^{\mu\nu\rho\sigma}F_{\rho\sigma}^a,$$0

For the $$F_{\mu\nu}^a=\partial_\mu A_\nu^a-\partial_\nu A_\mu^a+g f^{abc}A_\mu^bA_\nu^c, \qquad \tilde F_a^{\mu\nu}=\frac12 \epsilon^{\mu\nu\rho\sigma}F_{\rho\sigma}^a,$$1 source, one chiral analysis gives

$$F_{\mu\nu}^a=\partial_\mu A_\nu^a-\partial_\nu A_\mu^a+g f^{abc}A_\mu^bA_\nu^c, \qquad \tilde F_a^{\mu\nu}=\frac12 \epsilon^{\mu\nu\rho\sigma}F_{\rho\sigma}^a,$$2

with the noteworthy ratio

$$F_{\mu\nu}^a=\partial_\mu A_\nu^a-\partial_\nu A_\mu^a+g f^{abc}A_\mu^bA_\nu^c, \qquad \tilde F_a^{\mu\nu}=\frac12 \epsilon^{\mu\nu\rho\sigma}F_{\rho\sigma}^a,$$3

which is much larger than a naive isospin estimate because $$F_{\mu\nu}^a=\partial_\mu A_\nu^a-\partial_\nu A_\mu^a+g f^{abc}A_\mu^bA_\nu^c, \qquad \tilde F_a^{\mu\nu}=\frac12 \epsilon^{\mu\nu\rho\sigma}F_{\rho\sigma}^a,$$4 is anomalously small (Bsaisou et al., 2012). A more recent HB$$F_{\mu\nu}^a=\partial_\mu A_\nu^a-\partial_\nu A_\mu^a+g f^{abc}A_\mu^bA_\nu^c, \qquad \tilde F_a^{\mu\nu}=\frac12 \epsilon^{\mu\nu\rho\sigma}F_{\rho\sigma}^a,$$5PT analysis using updated hadronic input quotes

$$F_{\mu\nu}^a=\partial_\mu A_\nu^a-\partial_\nu A_\mu^a+g f^{abc}A_\mu^bA_\nu^c, \qquad \tilde F_a^{\mu\nu}=\frac12 \epsilon^{\mu\nu\rho\sigma}F_{\rho\sigma}^a,$$6

(Mulder et al., 10 Feb 2025).

These couplings feed directly into EDM observables. For a nucleon,

$$F_{\mu\nu}^a=\partial_\mu A_\nu^a-\partial_\nu A_\mu^a+g f^{abc}A_\mu^bA_\nu^c, \qquad \tilde F_a^{\mu\nu}=\frac12 \epsilon^{\mu\nu\rho\sigma}F_{\rho\sigma}^a,$$7

where $$F_{\mu\nu}^a=\partial_\mu A_\nu^a-\partial_\nu A_\mu^a+g f^{abc}A_\mu^bA_\nu^c, \qquad \tilde F_a^{\mu\nu}=\frac12 \epsilon^{\mu\nu\rho\sigma}F_{\rho\sigma}^a,$$8 is the CP-odd electromagnetic form factor (Bhattacharya et al., 2023, Bhattacharya et al., 2021). For the deuteron, the two-nucleon contribution induced by the theta term has been computed up to and including next-to-next-to-leading order, with the result

$$F_{\mu\nu}^a=\partial_\mu A_\nu^a-\partial_\nu A_\mu^a+g f^{abc}A_\mu^bA_\nu^c, \qquad \tilde F_a^{\mu\nu}=\frac12 \epsilon^{\mu\nu\rho\sigma}F_{\rho\sigma}^a,$$9

and the dominant uncertainty comes from the CP- and isospin-violating $$q(x)=\frac{g^2}{32\pi^2}F_{\mu\nu}^a\tilde F_a^{\mu\nu}.$$0 coupling (Bsaisou et al., 2012).

A complementary controlled toy model, deformed QCD on $$q(x)=\frac{g^2}{32\pi^2}F_{\mu\nu}^a\tilde F_a^{\mu\nu}.$$1, makes the topological origin of these low-energy effects explicit. There the topological susceptibility in pure glue is a positive contact term,

$$q(x)=\frac{g^2}{32\pi^2}F_{\mu\nu}^a\tilde F_a^{\mu\nu}.$$2

and the $$q(x)=\frac{g^2}{32\pi^2}F_{\mu\nu}^a\tilde F_a^{\mu\nu}.$$3-like mode cancels it in the chiral limit, realizing the Witten–Veneziano mechanism microscopically (Thomas et al., 2011).

4. Lattice QCD, topology, and the sign problem

In Euclidean spacetime the theta term appears as an imaginary contribution to the action, so the Boltzmann weight becomes complex and importance sampling encounters the theta-induced sign problem (Sasaki et al., 2012, Cai et al., 2016). One standard decomposition is

$$q(x)=\frac{g^2}{32\pi^2}F_{\mu\nu}^a\tilde F_a^{\mu\nu}.$$4

where $$q(x)=\frac{g^2}{32\pi^2}F_{\mu\nu}^a\tilde F_a^{\mu\nu}.$$5 is the partition function at fixed topological charge (Cai et al., 2016). A detailed large-volume analysis shows that even exact knowledge of the infinite-volume fixed-topological-density function $$q(x)=\frac{g^2}{32\pi^2}F_{\mu\nu}^a\tilde F_a^{\mu\nu}.$$6 need not suffice to reconstruct $$q(x)=\frac{g^2}{32\pi^2}F_{\mu\nu}^a\tilde F_a^{\mu\nu}.$$7 in regions where the curvature of $$q(x)=\frac{g^2}{32\pi^2}F_{\mu\nu}^a\tilde F_a^{\mu\nu}.$$8 is negative; in such regions the interplay of the sign problem and the thermodynamic limit becomes structural rather than merely numerical (Cai et al., 2016).

A model strategy for alleviating the sign problem is to perform an axial transformation that moves $$q(x)=\frac{g^2}{32\pi^2}F_{\mu\nu}^a\tilde F_a^{\mu\nu}.$$9 into a light-quark $\mathrm{Tr}\,F\tilde F$0 mass term, then define a reference theory that drops the P-odd mass and reweights observables. In three-flavor PNJL and EPNJL studies this reference theory leaves the P-even condensates nearly unchanged at $\mathrm{Tr}\,F\tilde F$1, while nonzero $\mathrm{Tr}\,F\tilde F$2 makes the chiral transition sharper and can render it first order even at $\mathrm{Tr}\,F\tilde F$3 in the EPNJL model (Sasaki et al., 2012).

Direct lattice determinations of theta-induced EDMs remain difficult. A 2+1+1-flavor HISQ calculation found

$\mathrm{Tr}\,F\tilde F$4

for the continuum topological susceptibility at $\mathrm{Tr}\,F\tilde F$5 MeV, but concluded that present lattice QCD calculations do not provide a reliable estimate of the contribution of the $\mathrm{Tr}\,F\tilde F$6-term to the nucleon electric dipole moments; a factor of ten higher statistics was judged necessary for significantly better control of the systematics (Bhattacharya et al., 2021). A later lattice study emphasized that unresolved excited-state contaminations are a major source of systematic uncertainty in the extraction of the nEDM from the topological term, and quoted a clover-based topological susceptibility

$\mathrm{Tr}\,F\tilde F$7

consistent with chiral expectations (Bhattacharya et al., 2023).

5. Geometric, Berry-phase, and domain-based reinterpretations

Several recent and earlier works recast the theta term in geometric language. One approach derives the $\mathrm{Tr}\,F\tilde F$8-term from Fujikawa’s Jacobian together with an adiabatic Berry connection $\mathrm{Tr}\,F\tilde F$9 on the space of gauge configurations. In that construction the effective action reduces to standard QCD with a theta term after imposing

$P$0

and setting $P$1, while the functional Berry phase

$P$2

is interpreted as a holonomy over gauge-configuration space (Gamboa, 30 May 2025).

A related program interprets $P$3 itself as a Berry phase analogous to electric polarization in topological insulators. In the two-dimensional $P$4 or $P$5 setting, the Berry connection of Dirac Bloch states over a Brillouin zone defines

$P$6

and the four-dimensional generalization identifies $P$7-domain walls with $P$8-dimensional Chern–Simons membranes across which $P$9 jumps by $T$0 (Thacker, 2013). This line of work frames $T$1 as an order parameter for topological polarization rather than only as a constant coupling.

A different domain-based construction motivated by heavy-ion phenomenology treats $T$2 as a local quantity $T$3 and attempts to parametrize generalized $T$4-vacua by the gauge-dependent dimension-two condensate $T$5. In that model, surfaces of constant $T$6 define “gauge slices” associated with different local $T$7 values, and a parity-odd domain is the union of such slices (Kim et al., 2013). This suggests a geometric organization of metastable CP-odd regions, but it is explicitly a gauge-dependent model rather than a first-principles gauge-invariant formulation.

Another proposal replaces a constant theta parameter by the phase $T$8 of the quark condensate and couples its derivative to the Chern–Simons current,

$T$9

In that framework topological defects such as vortices make $\theta\neq 0 \mod \pi$0 nontrivial, and the derivative coupling is argued to reproduce $\theta\neq 0 \mod \pi$1-problem physics while avoiding a fundamental constant $\theta\neq 0 \mod \pi$2 (Xiong, 2013). This suggests an alternative effective description of theta-like effects, but the equivalence to standard QCD $\theta\neq 0 \mod \pi$3 physics is model dependent.

6. Effective theta in external environments and phenomenological probes

Beyond the vacuum, several works study effective or local theta phenomena. In relativistic heavy-ion collisions, metastable parity-odd domains have been argued to require a localized $\theta\neq 0 \mod \pi$4, with effective values of order $\theta\neq 0 \mod \pi$5 in a domain of a few fermis, far larger than the vacuum EDM bound (Kim et al., 2013). Within the generalized-vacuum model just mentioned, such domains are associated with instanton-liquid regions and evolving unions of gauge slices (Kim et al., 2013).

A lattice study of QCD in CP-odd electromagnetic backgrounds found that external fields with $\theta\neq 0 \mod \pi$6 induce an effective QCD theta term,

$\theta\neq 0 \mod \pi$7

with

$\theta\neq 0 \mod \pi$8

for $\theta\neq 0 \mod \pi$9 MeV and preliminary indications of $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$00 at $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$01 MeV (Bonati et al., 2013). The same study verified the anomaly relation between SU(3) and U(1) contributions to Dirac zero modes in such backgrounds (Bonati et al., 2013).

In dense isospin-asymmetric quark matter, a two-flavor NJL model with a $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$02-dependent KMT determinant finds that $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$03 suppresses the conventional chiral and pion condensates $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$04, enhances the pseudoscalar and scalar-isovector condensates $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$05, lowers the critical isospin chemical potential, and at $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$06 produces a first-order transition at

$$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$07

with CP restoration (Zhang et al., 3 Jul 2025). The same work reports that, when $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$08 is interpreted as a static axion background, the equation of state of nonstrange quark stars is stiffened and their maximum masses and radii increase (Zhang et al., 3 Jul 2025).

Finally, precision EDM searches probe the theta term indirectly. A recent HB$$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$09PT analysis of CP-odd semileptonic electron–nucleus interactions induced by $$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$10 in paramagnetic molecules derived

$$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$11

from HfF$$S_\theta=\theta\,\frac{g^2}{32\pi^2}\int d^4x\,\mathrm{Tr}\,G_{\mu\nu}\tilde G^{\mu\nu}.$$12 measurements (Mulder et al., 10 Feb 2025). This is weaker than neutron-based limits, but it establishes paramagnetic molecular experiments as a distinct probe of strong CP violation and suggests that a further improvement of one to two orders of magnitude would make them competitive with the neutron and diamagnetic-atom program (Mulder et al., 10 Feb 2025).