Chern–Simons Portal Models

• Chern–Simons portal models are defined by topological gauge couplings, such as BF and Chern–Simons terms, that link visible and hidden sectors to enable charge–flux transmutation.
• They use scalar mixing and anomaly-induced mechanisms across 2+1 and 3+1 dimensions to facilitate interactions in systems like vortex dynamics, interface electrodynamics, and dark-matter models.
• Numerical analyses reveal that the mixing terms produce distinctive magnetic field profiles, anyonic soliton structures, and non-decoupling effects observable in collider experiments.

The expression Chern–Simons portal model denotes a family of constructions in which otherwise distinct sectors communicate through Chern–Simons or BF-type topological couplings, often supplemented by scalar mixing. In the $2+1$-dimensional formulation most directly associated with the phrase, a visible Maxwell–Higgs sector and a hidden Chern–Simons–Higgs sector are coupled by a BF gauge mixing term and a Higgs portal, so that vortex excitations in one sector induce electric charge and magnetic flux in the other (Ireson et al., 2016). In $3+1$ dimensions, the same phrase is also used for Standard Model extensions with a new massive vector boson $X_\mu$ coupled to electroweak gauge bosons by anomaly-induced Chern–Simons-like operators, yielding a vector portal that is topological rather than kinetically mixed (Gorkavenko et al., 2024). Related realizations occur in parity-invariant planar gauge theories, interface electrodynamics, non-Abelian Higgs-portal systems, higher-gauge theories, and dark-matter models (Lima et al., 2022).

1. Terminological scope and defining structures

A Chern–Simons portal is not a single unique Lagrangian. The common feature is a portal interaction built from a topological tensor structure, typically $\epsilon^{\mu\nu\rho}$ in $2+1$ dimensions or $\epsilon^{\mu\nu\lambda\rho}$ in $3+1$ dimensions, that couples gauge sectors directly rather than through ordinary kinetic mixing. In the Abelian $2+1$-dimensional model, the portal is explicitly

$$L_{mix}=\xi \varepsilon^{\mu\alpha\beta}A_\mu \partial_\alpha B_\beta+\zeta (|\phi|^2-\phi_0^2)(|\eta|^2-\eta_0^2),$$

combining BF gauge mixing with a Higgs portal (Ireson et al., 2016). In the Standard Model extension, the portal is

$$\mathcal{L}_{CS}=c_W \,\epsilon^{\mu\nu\lambda\rho} X_\mu W_\nu \partial_\lambda W_\rho + c_{\gamma}\cos\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda A_\rho + c_Z\sin\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda Z_\rho,$$

with a new $3+1$0 vector boson $3+1$1 and dimensionless Wilson coefficients $3+1$2 (Nourbakhsh et al., 4 Sep 2025).

This portal logic differs from both standard kinetic mixing and a pure Higgs portal. The $3+1$3-dimensional BF coupling is described as the lower-dimensional analogue of kinetic mixing, but with distinct topological properties: it ties magnetic flux to electric charge and can generate anyonic statistics (Ireson et al., 2016). The $3+1$4-dimensional Standard Model construction is likewise presented as a vector portal, but explicitly “not via kinetic mixing”; instead it proceeds through anomaly-induced gauge–gauge interactions (Gorkavenko et al., 2024). In several realizations the portal is combined with scalar-sector mixing, but the defining interaction remains the Chern–Simons or BF term rather than the scalar coupling alone (Ipiña et al., 2018).

A recurrent misconception is that Chern–Simons portal models are necessarily parity-violating or necessarily fermion-coupled. The parity-invariant Maxwell–Chern–Simons $3+1$5 model shows that mixed Chern–Simons couplings can be arranged into a parity-even structure (Lima et al., 2022). Conversely, the Standard Model extension with a GeV-scale Chern–Simons boson emphasizes that there is no direct tree-level interaction between the new boson and Standard Model fermions in the minimal setup (Gorkavenko et al., 2024).

2. The Abelian $3+1$6 BF portal

The most literal realization of a Chern–Simons portal is the $3+1$7 theory with a visible Maxwell–Higgs sector $3+1$8 and a hidden Chern–Simons–Higgs sector $3+1$9 (Ireson et al., 2016). The visible gauge dynamics is governed by a Maxwell term and quartic Higgs potential, the hidden gauge dynamics by a Chern–Simons term and a sixth-order Higgs potential, and the two sectors are coupled by the BF term and the Higgs portal. The matter content is segregated: $X_\mu$0 is charged only under $X_\mu$1, and $X_\mu$2 only under $X_\mu$3. Without the mixing terms, the sectors are completely decoupled.

The portal acts simultaneously in the gauge and scalar sectors. The gauge equations show that the curl of $X_\mu$4 acts as a source for $X_\mu$5, and vice versa, with strength $X_\mu$6; the scalar equations contain cross-couplings proportional to $X_\mu$7, so that symmetry breaking in one sector shifts masses in the other (Ireson et al., 2016). This suggests a precise operational meaning for the phrase “portal”: topological objects, electric charges, magnetic fluxes, and scalar mass parameters are transmitted across the visible–hidden split.

The central nonperturbative objects are static, axially symmetric vortices with independent winding numbers $X_\mu$8 and $X_\mu$9 in the two sectors. Because of the BF and Chern–Simons couplings, both temporal gauge components $\epsilon^{\mu\nu\rho}$0 and $\epsilon^{\mu\nu\rho}$1 must be nonzero for finite-energy solutions. The resulting Gauss-law relations are

$\epsilon^{\mu\nu\rho}$2

with quantized fluxes

$\epsilon^{\mu\nu\rho}$3

For vortices,

$\epsilon^{\mu\nu\rho}$4

These relations encode the portal effect in its sharpest form: a hidden-sector vortex with flux $\epsilon^{\mu\nu\rho}$5 necessarily carries visible-sector electric charge $\epsilon^{\mu\nu\rho}$6 (Ireson et al., 2016).

The same model admits an $\epsilon^{\mu\nu\rho}$7 supersymmetric extension. The supersymmetric completion introduces an additional real scalar $\epsilon^{\mu\nu\rho}$8, which cannot consistently be set to zero, and produces an effective scalar potential in which the couplings are no longer independent. The BPS system includes

$\epsilon^{\mu\nu\rho}$9

together with first-order equations for the magnetic fields and covariant-holomorphicity conditions $2+1$0, $2+1$1 (Ireson et al., 2016). In this regime the vortices are described as BPS anyonic solitons whose electric and magnetic properties are controlled by both sectors.

Numerically, the gauge mixing $2+1$2 produces several characteristic effects. The scalar profiles $2+1$3 and $2+1$4 are almost insensitive to $2+1$5, but the Maxwell magnetic field can develop a ring-shaped maximum away from the origin, and the Maxwell electric field, which vanishes at $2+1$6, becomes nontrivial with two rings of opposite sign when $2+1$7 (Ireson et al., 2016). This is one of the clearest demonstrations that Chern–Simons dynamics in one sector can be transmitted into a Maxwell sector through BF mixing.

3. Related low-dimensional and non-Abelian generalizations

Several models extend or reinterpret the portal structure while retaining the same topological mechanism. In the parity-invariant Maxwell–Chern–Simons $2+1$8 theory, the mixed term

$2+1$9

is combined with two charged scalars $\epsilon^{\mu\nu\lambda\rho}$0 whose charge assignments make them bifundamental-like under the two Abelian groups (Lima et al., 2022). Finite-energy vortices are labeled by two integers $\epsilon^{\mu\nu\lambda\rho}$1, with fluxes

$\epsilon^{\mu\nu\lambda\rho}$2

and mutual charge–flux attachment

$\epsilon^{\mu\nu\lambda\rho}$3

The angular momentum is

$\epsilon^{\mu\nu\lambda\rho}$4

This model is significant because it shows that the portal effect—charge in one sector tied to flux in the other—can be realized while preserving parity (Lima et al., 2022).

A different realization occurs in interface electrodynamics. There the action consists of a piecewise Maxwell bulk term plus localized surface interactions

$\epsilon^{\mu\nu\lambda\rho}$5

supported on two planes $\epsilon^{\mu\nu\lambda\rho}$6 (Pis'mak et al., 2014). The model is explicitly local, gauge invariant, and renormalizable. Its observable consequence is not vortex transmutation but modified boundary conditions: transmission and reflection coefficients depend on the Chern–Simons interaction strength, while Snell’s law is preserved and parallel and perpendicular polarization components are mixed (Pis'mak et al., 2014). This is a surface-confined Chern–Simons portal between electromagnetic regions rather than between particle sectors.

The non-Abelian $\epsilon^{\mu\nu\lambda\rho}$7 Chern–Simons–Higgs model coupled to an uncharged triplet $\epsilon^{\mu\nu\lambda\rho}$8 through

$\epsilon^{\mu\nu\lambda\rho}$9

illustrates a Higgs-portal version of the same idea (Ipiña et al., 2018). Here the Chern–Simons–Higgs vortex suppresses $3+1$0 in its core, making $3+1$1 condense there and form a halo. The model exhibits three parameter regions: ordinary Chern–Simons–Higgs vortices, vortices with a $3+1$2-halo, and a region with no vortex solutions (Ipiña et al., 2018). The portal effect is therefore expressed as competing vacuum structure bound to a topological defect.

At a more formal level, $3+1$3-dimensional Chern–Simons theory based on a crossed module uses a higher connection $3+1$4, with $3+1$5 a $3+1$6-valued $3+1$7-form and $3+1$8 an $3+1$9-valued $2+1$0-form, and action

$2+1$1

(Zucchini, 2021). The theory is fully gauge invariant on closed $2+1$2-manifolds, while with boundary its gauge variation is a boundary term. Depending on boundary conditions, level quantization occurs and surface charges obey a nontrivial Poisson-bracket algebra, described as a higher counterpart of the familiar WZNW current algebra (Zucchini, 2021). A plausible implication is that higher-form versions of Chern–Simons portals can be organized systematically through boundary symmetry rather than only through bulk effective operators.

4. Anomaly-induced portals to the Standard Model

In $2+1$3 dimensions the Chern–Simons portal is typically formulated as an extension

$2+1$4

with a new massive vector boson $2+1$5 associated with $2+1$6 (Nourbakhsh et al., 4 Sep 2025). Heavy chiral fermions charged under both $2+1$7 and Standard Model gauge groups generate mixed anomalies; after they are integrated out, the low-energy theory contains Chern–Simons-like couplings between $2+1$8 and electroweak gauge bosons. A key point emphasized in both the general Standard Model extension and the HL-LHC study is that these coefficients are non-decoupling: they are not suppressed by $2+1$9 in the manner of generic dimension-6 operators (Gorkavenko et al., 2024).

The minimal low-energy interaction is the Chern–Simons gauge Lagrangian written above. In the short Standard Model extension, $$L_{mix}=\xi \varepsilon^{\mu\alpha\beta}A_\mu \partial_\alpha B_\beta+\zeta (|\phi|^2-\phi_0^2)(|\eta|^2-\eta_0^2),$$0 is described as a Stueckelberg field, there is no direct interaction between $$L_{mix}=\xi \varepsilon^{\mu\alpha\beta}A_\mu \partial_\alpha B_\beta+\zeta (|\phi|^2-\phi_0^2)(|\eta|^2-\eta_0^2),$$1 and Standard Model fermions at tree level, and the only tree-level portal to the visible sector is through electroweak gauge bosons (Gorkavenko et al., 2024). After integrating out heavy chiral fermions, loop-induced flavor-changing couplings to down-type quarks appear,

$$L_{mix}=\xi \varepsilon^{\mu\alpha\beta}A_\mu \partial_\alpha B_\beta+\zeta (|\phi|^2-\phi_0^2)(|\eta|^2-\eta_0^2),$$2

with

$$L_{mix}=\xi \varepsilon^{\mu\alpha\beta}A_\mu \partial_\alpha B_\beta+\zeta (|\phi|^2-\phi_0^2)(|\eta|^2-\eta_0^2),$$3

The same work also stresses an unresolved issue: for same-flavor fermions and leptons, loop diagrams with the $$L_{mix}=\xi \varepsilon^{\mu\alpha\beta}A_\mu \partial_\alpha B_\beta+\zeta (|\phi|^2-\phi_0^2)(|\eta|^2-\eta_0^2),$$4 vertex are divergent within the minimal effective Lagrangian, so decays such as $$L_{mix}=\xi \varepsilon^{\mu\alpha\beta}A_\mu \partial_\alpha B_\beta+\zeta (|\phi|^2-\phi_0^2)(|\eta|^2-\eta_0^2),$$5 are not yet reliably calculable in that setup (Gorkavenko et al., 2024).

The collider-oriented study adds a phenomenological decay portal,

$$L_{mix}=\xi \varepsilon^{\mu\alpha\beta}A_\mu \partial_\alpha B_\beta+\zeta (|\phi|^2-\phi_0^2)(|\eta|^2-\eta_0^2),$$6

and focuses on the effective parameter space $$L_{mix}=\xi \varepsilon^{\mu\alpha\beta}A_\mu \partial_\alpha B_\beta+\zeta (|\phi|^2-\phi_0^2)(|\eta|^2-\eta_0^2),$$7 under the assumption $$L_{mix}=\xi \varepsilon^{\mu\alpha\beta}A_\mu \partial_\alpha B_\beta+\zeta (|\phi|^2-\phi_0^2)(|\eta|^2-\eta_0^2),$$8 (Nourbakhsh et al., 4 Sep 2025). The benchmark masses are $$L_{mix}=\xi \varepsilon^{\mu\alpha\beta}A_\mu \partial_\alpha B_\beta+\zeta (|\phi|^2-\phi_0^2)(|\eta|^2-\eta_0^2),$$9, and the signal process is

$$\mathcal{L}_{CS}=c_W \,\epsilon^{\mu\nu\lambda\rho} X_\mu W_\nu \partial_\lambda W_\rho + c_{\gamma}\cos\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda A_\rho + c_Z\sin\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda Z_\rho,$$0

The signature is one prompt electron, large missing transverse energy, at least one jet, and two displaced muons forming a displaced secondary vertex (Nourbakhsh et al., 4 Sep 2025).

The HL-LHC analysis uses $$\mathcal{L}_{CS}=c_W \,\epsilon^{\mu\nu\lambda\rho} X_\mu W_\nu \partial_\lambda W_\rho + c_{\gamma}\cos\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda A_\rho + c_Z\sin\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda Z_\rho,$$1 collisions at $$\mathcal{L}_{CS}=c_W \,\epsilon^{\mu\nu\lambda\rho} X_\mu W_\nu \partial_\lambda W_\rho + c_{\gamma}\cos\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda A_\rho + c_Z\sin\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda Z_\rho,$$2 TeV, integrated luminosity $$\mathcal{L}_{CS}=c_W \,\epsilon^{\mu\nu\lambda\rho} X_\mu W_\nu \partial_\lambda W_\rho + c_{\gamma}\cos\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda A_\rho + c_Z\sin\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda Z_\rho,$$3, and pile-up with an average of $$\mathcal{L}_{CS}=c_W \,\epsilon^{\mu\nu\lambda\rho} X_\mu W_\nu \partial_\lambda W_\rho + c_{\gamma}\cos\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda A_\rho + c_Z\sin\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda Z_\rho,$$4 interactions per bunch crossing. Signal and backgrounds are generated at LO with MadGraph5_aMC@NLO, showered with Pythia 8, clustered with anti-$$\mathcal{L}_{CS}=c_W \,\epsilon^{\mu\nu\lambda\rho} X_\mu W_\nu \partial_\lambda W_\rho + c_{\gamma}\cos\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda A_\rho + c_Z\sin\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda Z_\rho,$$5 and $$\mathcal{L}_{CS}=c_W \,\epsilon^{\mu\nu\lambda\rho} X_\mu W_\nu \partial_\lambda W_\rho + c_{\gamma}\cos\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda A_\rho + c_Z\sin\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda Z_\rho,$$6, and passed through Delphes with the CMS_PhaseII_200PU card. A BDT is trained on kinematic, invariant-mass, angular, and displacement observables, with dominant discriminants identified as $$\mathcal{L}_{CS}=c_W \,\epsilon^{\mu\nu\lambda\rho} X_\mu W_\nu \partial_\lambda W_\rho + c_{\gamma}\cos\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda A_\rho + c_Z\sin\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda Z_\rho,$$7, $$\mathcal{L}_{CS}=c_W \,\epsilon^{\mu\nu\lambda\rho} X_\mu W_\nu \partial_\lambda W_\rho + c_{\gamma}\cos\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda A_\rho + c_Z\sin\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda Z_\rho,$$8, $$\mathcal{L}_{CS}=c_W \,\epsilon^{\mu\nu\lambda\rho} X_\mu W_\nu \partial_\lambda W_\rho + c_{\gamma}\cos\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda A_\rho + c_Z\sin\theta_W\,\epsilon^{\mu\nu\lambda\rho} X_\mu Z_\nu \partial_\lambda Z_\rho,$$9, $3+1$00, $3+1$01, and $3+1$02 (Nourbakhsh et al., 4 Sep 2025).

The resulting expected $3+1$03 confidence-level exclusions are quoted in the $3+1$04 plane. For $3+1$05, the analysis excludes $3+1$06 at $3+1$07 GeV and $3+1$08 at $3+1$09 GeV, with sensitivity improving with increasing $3+1$10 over the $3+1$11–$3+1$12 GeV range (Nourbakhsh et al., 4 Sep 2025). The abstract summarizes the reach as constraints on the $3+1$13–$3+1$14 coupling down to $3+1$15, while electroweak precision data and LEP single-photon limits still require $3+1$16 to be much smaller than $3+1$17 (Nourbakhsh et al., 4 Sep 2025).

5. Dark-matter realizations

A further specialization is the $3+1$18 portal to Chern–Simons dark matter, where the dark matter candidate is a massive vector boson $3+1$19 of a dark $3+1$20, coupled to a mediator $3+1$21 through

$3+1$22

(Arcadi et al., 2017). Two mechanisms connect the mediator to the Standard Model. In scenario I the mediator is the gauge boson of a second Abelian group and couples to hypercharge through kinetic mixing $3+1$23. In scenario II the connection is instead a second Chern–Simons interaction,

$3+1$24

The dark-matter phenomenology is unusually portal-specific. Direct detection is suppressed because the effective DM–quark operator is axial and derivative,

$3+1$25

so the scattering is spin-dependent and momentum suppressed (Arcadi et al., 2017). The paper gives the estimate

$3+1$26

far below present spin-dependent limits.

Relic abundance is controlled by the usual thermal condition $3+1$27. In scenario I, annihilation channels include $3+1$28, $3+1$29, $3+1$30, $3+1$31, $3+1$32, $3+1$33, and $3+1$34; most are $3+1$35-wave or $3+1$36-wave suppressed, whereas $3+1$37 is $3+1$38-wave and often dominant (Arcadi et al., 2017). In scenario II the main channels are $3+1$39, $3+1$40, and $3+1$41, with the last again providing the dominant $3+1$42-wave contribution when kinematically open. The allowed parameter space therefore clusters near resonance regions $3+1$43 or in the regime $3+1$44, where $3+1$45 is available (Arcadi et al., 2017).

Indirect detection is correspondingly weak except in the TeV regime. The study concludes that direct-detection searches are not promising, that indirect-detection experiments furnish complementary limits for TeV-scale masses, especially with the CTA, and that mono-jet and dilepton searches at the LHC are important mainly in the kinetic-mixing realization (Arcadi et al., 2017). The same work also provides a UV completion with heavy chiral fermions and Stueckelberg-type scalars, where the Chern–Simons coefficient $3+1$46 is generated radiatively and anomaly cancellation fixes the allowed charge assignments (Arcadi et al., 2017).

6. Conceptual unification, misconceptions, and open problems

Across these realizations, the most persistent structural theme is charge–flux transmutation across sectors. In the Abelian BF portal this is explicit in $3+1$47 and $3+1$48 (Ireson et al., 2016). In the parity-invariant Maxwell–Chern–Simons model it appears as $3+1$49 and $3+1$50 (Lima et al., 2022). In interface electrodynamics it reappears as boundary conditions in which jumps of $3+1$51 and $3+1$52 are proportional to magnetic and electric components induced by the surface Chern–Simons term (Pis'mak et al., 2014). This suggests that the defining physical content of a Chern–Simons portal is not merely gauge-field mixing, but topological conversion laws between excitations in different sectors.

A second unifying feature is non-decoupling. In Standard Model extensions, the anomaly-induced Chern–Simons coefficients survive integration out of very heavy fermions and can remain observable at collider scales (Gorkavenko et al., 2024). In low-dimensional soliton models, the portal remains visible in the global charges of vortices, in ring-shaped magnetic profiles, and in anyonic angular momentum (Ireson et al., 2016). In higher-gauge theory, nontriviality migrates to the boundary: gauge variation on a manifold with boundary becomes a boundary term, and surface charges obey a higher current algebra (Zucchini, 2021). A plausible implication is that Chern–Simons portals are best understood as topological infrastructures tying infrared observables to ultraviolet anomaly or boundary data.

Several open issues remain model-specific. The minimal Standard Model extension with only the electroweak Chern–Simons operators does not yet yield a finite diagonal effective coupling to same-flavor fermions and leptons, so the total lifetime and visible branching fractions of the light $3+1$53 boson are not fully under theoretical control in that setup (Gorkavenko et al., 2024). The HL-LHC displaced-vertex analysis evades that difficulty by introducing a phenomenological axial lepton coupling, which is technically consistent with the chosen search strategy but not identical to the minimal fermion-free construction (Nourbakhsh et al., 4 Sep 2025). In the dark-matter context, the most experimentally accessible kinetic-mixing regime tends to require couplings larger than their loop-motivated natural estimates, whereas the more topological second-CS scenario is harder to probe (Arcadi et al., 2017).

The term Chern–Simons portal model therefore denotes a class rather than a single theory. Its common content is a topological channel—BF, mixed Chern–Simons, surface Chern–Simons, or anomaly-induced electroweak Chern–Simons-like interaction—through which visible and hidden sectors, bulk regions, or competing order parameters exchange gauge, scalar, and topological information. In different dimensions this yields electrically charged vortices, polarization-mixing interfaces, halo-bearing non-Abelian vortices, higher boundary current algebras, long-lived GeV-scale vectors at colliders, or vector dark matter with suppressed direct detection (Ireson et al., 2016).