Millicharged Particles: Models & Detection

• Millicharged particles (MCPs) are hypothetical states with an electric charge εe (with ε ≪ 1) that emerge in hidden-sector theories and dark matter models.
• They are produced by diverse mechanisms—from meson decays and Drell–Yan processes to cosmic-ray collisions—enabling a broad range of experimental probes.
• Detection strategies include electron recoils, ionization in trackers, optical phase shifts, and geomagnetic signals, offering complementary insights across laboratory and astrophysical environments.

Millicharged particles (MCPs) are hypothetical states carrying an electric charge $Q_\chi=\epsilon e$ with $\epsilon\ll 1$. In the recent literature they appear in several distinct but overlapping roles: as generic hidden-sector matter with an effective electromagnetic coupling induced by kinetic mixing, as subdominant or dominant dark-matter candidates, as relativistic decay products of heavier dark states, and as ultralight coherent fields. The phenomenology is correspondingly broad. MCPs can be produced in meson decays, Drell–Yan processes, cosmic-ray collisions, stellar plasmas, core-collapse supernovae, or heavy dark-matter decays; they can be probed through electron recoils, deep-inelastic scattering, ionization in trackers and scintillators, coherent optical phase shifts, or even monochromatic geomagnetic signals (Xu, 2022, Plestid et al., 2020, Gan et al., 2023, Arza et al., 24 Jan 2025).

1. Definition and model realizations

The minimal phenomenological definition is a particle $\chi$ with electric charge $Q_\chi=\epsilon e$, where $\epsilon$ is a small dimensionless millicharge parameter. Several papers treat $\epsilon$ and $m_\chi$ as the only low-energy parameters, without committing to a unique ultraviolet completion (Plestid et al., 2020, Gorbunov et al., 2022).

A standard realization introduces an additional unbroken $U(1)'$ gauge symmetry with a massless hidden photon $A'_\mu$ kinetically mixed with the Standard Model photon. In that case the interaction structure can be written as

$\mathcal{L}=\sum_q e_q\,\bar q\gamma^\mu q\,A_\mu-\frac14F'_{\mu\nu}F'^{\mu\nu} +\bar\chi(i\slashed D-m_\chi)\chi-\frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu}, \qquad D_\mu=\partial_\mu-ig_\chi A'_\mu,$

and, after diagonalizing the kinetic terms, $\epsilon\ll 1$0 acquires an effective electromagnetic charge $\epsilon\ll 1$1 (Xu, 2022). A closely related cosmological analysis distinguishes two broad cases: “pure” MCPs without an accompanying dark photon, and MCPs with a massless dark photon generated by kinetic mixing; the two cases have different reheating and $\epsilon\ll 1$2 phenomenology (Gan et al., 2023).

Other works adopt the direct interaction

$\epsilon\ll 1$3

for a Dirac fermion, which is sufficient for supernova production and detector scattering calculations (Cui et al., 9 Jun 2026). For ultralight bosonic MCP dark matter, the relevant effective theory is scalar QED,

$\epsilon\ll 1$4

with $\epsilon\ll 1$5 (Arza et al., 24 Jan 2025).

A recurrent misconception is that MCPs require a dark photon. The literature does not support that as a universal statement: some analyses treat the millicharge as phenomenological, while others derive it from kinetic mixing (Plestid et al., 2020, Gan et al., 2023).

2. Production mechanisms across environments

MCP production mechanisms are highly environment-dependent. In the early universe, the “cosmic millicharge background” is the irreducible population produced from the Standard Model plasma during reheating and subsequent radiation domination. In that setting, searches for MCPs can potentially allow an upper bound on the reheating temperature down to $\epsilon\ll 1$6 MeV, and cosmology can distinguish scenarios with and without an accompanying dark photon (Gan et al., 2023).

In stellar plasmas, the dominant channel depends on temperature and plasma frequency. For late stages of stellar evolution, three regimes were identified: plasmon decay for $\epsilon\ll 1$7, Compton-like scattering for $\epsilon\ll 1$8 and $\epsilon\ll 1$9, and electron-positron annihilation at higher temperatures; semi-analytical fits were derived for direct implementation in stellar-evolution codes (Fiorillo et al., 2 Apr 2026). In core-collapse supernovae, plasmon decay $\chi$0 and $\chi$1 provide intense sources, and the finite MCP mass induces a time-of-flight delay relative to the neutrino burst (Cui et al., 9 Jun 2026).

Atmospheric cosmic rays furnish a permanent terrestrial source. High-energy protons striking the atmosphere produce mesons, and electromagnetic or Dalitz decays such as $\chi$2 and $\chi$3 yield an ambient isotropic MCP flux at Earth (Plestid et al., 2020). Neutrino-beam facilities generate analogous sources in proton targets: $\chi$4, $\chi$5, $\chi$6, and $\chi$7 decays dominate at lower masses, while Drell–Yan production extends the reach to the GeV scale (Magill et al., 2018).

Collider production at $\chi$8 machines is especially simple. Near threshold,

$\chi$9

has total cross section

$Q_\chi=\epsilon e$0

so the non-relativistic regime $Q_\chi=\epsilon e$1 is both kinematically distinctive and experimentally useful for direct ionization searches (Gorbunov et al., 2022).

A distinct class of source models assumes a heavy dark matter component $Q_\chi=\epsilon e$2 gravitationally captured in the Sun or Earth and decaying via $Q_\chi=\epsilon e$3. The resulting MCPs are relativistic, with $Q_\chi=\epsilon e$4, and can be searched for in neutrino telescopes or air-shower observatories (Xu, 2022, Xu, 2024, Xu, 2024, Xu, 24 Nov 2025).

3. Scattering, transport, and detector response

For many laboratory searches the key observable is elastic scattering on electrons. In neutrino experiments, the MCP–electron differential cross section is

$Q_\chi=\epsilon e$5

which reduces at small momentum transfer to the familiar $Q_\chi=\epsilon e$6 enhancement. In terms of an electron recoil threshold, the integrated cross section scales approximately as

$Q_\chi=\epsilon e$7

making low-threshold detectors especially favorable (Magill et al., 2018). Super-K recasts use the same Coulomb-dominated feature and define a windowed cross section over the allowed recoil range; because the MCP flux also scales as $Q_\chi=\epsilon e$8, the total event rate scales as $Q_\chi=\epsilon e$9 (Plestid et al., 2020).

At higher energies, hidden-photon models predict effective electromagnetic neutral-current deep-inelastic scattering on nuclei, with

$\epsilon$0

One fit used for IceCube analyses is

$\epsilon$1

over $\epsilon$2 (Xu, 2022). When the running electromagnetic coupling is included, the fitted form becomes

$\epsilon$3

over $\epsilon$4 (Xu, 24 Nov 2025).

Propagation through matter depends sensitively on $\epsilon$5. Ionization losses scale as $\epsilon$6; for $\epsilon$7, the stopping power in rock is about $\epsilon$8 MeV per km, which is small enough that downward-going cosmic-ray-produced MCPs with energies above a few hundred MeV can reach km-deep detectors (Plestid et al., 2020). In contrast, for solar-, terrestrial-core-, or Earth-crossing scenarios, attenuation is usually modeled through the interaction length

$\epsilon$9

with exponential survival factors in the Sun, Earth, air, or ice (Xu, 2022, Xu, 2024, Xu, 2024).

Direct ionization signatures can also be exploited in trackers. For non-relativistic MCPs at $\epsilon$0 colliders, the mean number of ionizing collisions per unit length scales as $\epsilon$1, and the track curvature obeys

$\epsilon$2

which allows slow, weakly ionizing, curved trajectories to be reconstructed near threshold (Gorbunov et al., 2022).

A different detection paradigm is coherent optics. For Earth-bound MCP dark matter in interferometers, the effective interaction Hamiltonian is proportional to the photon number operator, producing a phase shift

$\epsilon$3

which can be read out with Mach–Zehnder or single-arm squeezed-light interferometers (Chen et al., 2022, Nugroho, 2024). For ultralight bosonic MCP dark matter in the geomagnetic field, the observable is instead a monochromatic quasi-static magnetic field at angular frequency $\epsilon$4 (Arza et al., 24 Jan 2025).

4. Collider, beam-dump, and neutrino-beam searches

Beam and collider experiments have established a heterogeneous but increasingly precise laboratory program. A notable result is that neutrino experiments provide leading constraints in mass windows that are not optimally covered by traditional collider searches. Using electron-scattering data, LSND sets new constraints for $\epsilon$5 MeV and MiniBooNE for $\epsilon$6 MeV; projected sensitivities show that SBND and MicroBooNE can provide the leading bounds in the $\epsilon$7 MeV regime, while DUNE and SHiP can probe $\epsilon$8 MeV–$\epsilon$9 GeV well beyond existing bounds (Magill et al., 2018).

Cosmic-ray production strengthens this picture for underground detectors. A fully quantitative treatment of atmospheric production found new limits from Super-K that are the best in sensitivity reach for $m_\chi$0 GeV and competitive up to $m_\chi$1 GeV, while also constraining strongly interacting dark-matter scenarios independently of the MCP abundance (Plestid et al., 2020).

At $m_\chi$2 colliders, the standard search channel is mono-photon plus missing energy, but near-threshold direct observation in the tracker can outperform missing-energy methods. For a $m_\chi$3–$m_\chi$4 factory collecting $m_\chi$5 fb$m_\chi$6 in one year at a given energy point, direct observation of tracker deposits can probe the MCP charge down to $m_\chi$7 for MCP masses in an $m_\chi$8 MeV vicinity of each energy beam value; this mass region is explicitly described as unreachable with the searches for missing energy and single photon (Gorbunov et al., 2022). This directly contradicts the common assumption that collider sensitivity to MCPs is exhausted by invisible or mono-photon signatures.

Far-forward LHC searches add another ingredient. A Geant4-based study of secondary production in the TAXN absorber showed that, for the proposed FORMOSA detector, hadronic and electromagnetic showers in downstream infrastructure can enhance the expected signal yield by approximately $m_\chi$9 for $U(1)'$0 (Adhikary et al., 13 May 2026). This implies that realistic sensitivity projections for forward MCP searches must include secondary production, not only primary production at the interaction point.

5. Searches using the Sun, the Earth, and transient astrophysical sources

Neutrino telescopes and air-shower observatories probe MCPs in source geometries that are specific but powerful. In a solar-capture scenario where TeV-scale dark matter $U(1)'$1 is trapped in the Sun and decays to relativistic MCPs, IceCube can directly detect secondary cascades in the range $U(1)'$2 when $U(1)'$3, and six years of null data exclude

$U(1)'$4

for the stated benchmark assumptions (Xu, 2022).

If the heavy dark matter is captured in the Earth instead, IceCube probes a different parameter region. One Earth-core analysis found direct detectability at $U(1)'$5 energies when

$U(1)'$6

and a ten-year null result excludes

$U(1)'$7

(Xu, 2024). A later analysis with a running electromagnetic coupling shifts the detectability window to

$U(1)'$8

and quotes a new excluded region

$U(1)'$9

(Xu, 24 Nov 2025). These exclusions are explicitly scenario-dependent rather than universal MCP bounds.

At much higher energies, the Pierre Auger Observatory can search for upward-going MCPs from superheavy dark-matter decay. Under the benchmark assumptions of the cited study, Auger is sensitive in the secondaries’ energy range $A'_\mu$0 for

$A'_\mu$1

and fourteen years of null data rule out

$A'_\mu$2

(Xu, 2024).

Transient astrophysical sources add a complementary strategy. For core-collapse supernovae, MCP masses induce a time-of-flight delay relative to the neutrino burst, opening a clean search window after the neutrino signal has passed. For $A'_\mu$3 and sub-MeV to MeV-scale masses, more than 10 events per year can be detected at XENONnT, JUNO, DUNE, and Hyper-Kamiokande, and the search can improve upon the existing SN cooling bound on $A'_\mu$4 by up to an order of magnitude (Cui et al., 9 Jun 2026).

An even more unusual probe operates in the ultralight regime. Bosonic millicharged dark matter in the Earth’s magnetic field generates a monochromatic quasi-static magnetic signal with angular frequency twice the MCP mass. Recasts of SuperMAG and SNIPE Hunt data constrain the effective charge for bosonic MCP dark matter in the mass ranges $A'_\mu$5 and $A'_\mu$6, and these bounds surpass current stellar cooling constraints by more than ten orders of magnitude (Arza et al., 24 Jan 2025).

6. Cosmology, stellar evolution, and model dependence

Cosmology does not impose a single MCP narrative. In reheating studies, the irreducible cosmic millicharge background depends on whether MCPs come with a dark photon. Without a dark photon, overproduction and MCP–baryon interactions dominate the constraints; with a dark photon, $A'_\mu$7 becomes central. The identified parameter regions imply that accelerator and other experiments can probe reheating scenarios and, in favorable cases, set an upper bound on the reheating temperature down to $A'_\mu$8 MeV (Gan et al., 2023). This suggests that MCP phenomenology can act as a probe of the post-inflationary thermal history, not merely of hidden sectors.

Stellar evolution provides a separate and comparatively model-minimal lever arm. For pre-supernova objects with $A'_\mu$9 and $\mathcal{L}=\sum_q e_q\,\bar q\gamma^\mu q\,A_\mu-\frac14F'_{\mu\nu}F'^{\mu\nu} +\bar\chi(i\slashed D-m_\chi)\chi-\frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu}, \qquad D_\mu=\partial_\mu-ig_\chi A'_\mu,$0, MCP emissivities fall into three regimes—plasmon decay, Compton-like scattering, and electron-positron annihilation—and semi-analytical fits now exist for direct use in stellar-evolution codes (Fiorillo et al., 2 Apr 2026). This is important because several earlier constraints were derived for different plasma conditions; late-stage stellar evolution fills an intermediate regime between red-giant/white-dwarf bounds and core-collapse supernova bounds.

Two common misconceptions require qualification. First, MCP bounds are not universally transferable across models. The Sun-, Earth-core-, and Auger-based exclusions rely on specific assumptions about a heavy dark-matter species $\mathcal{L}=\sum_q e_q\,\bar q\gamma^\mu q\,A_\mu-\frac14F'_{\mu\nu}F'^{\mu\nu} +\bar\chi(i\slashed D-m_\chi)\chi-\frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu}, \qquad D_\mu=\partial_\mu-ig_\chi A'_\mu,$1, its lifetime, its capture cross section, the exclusive decay $\mathcal{L}=\sum_q e_q\,\bar q\gamma^\mu q\,A_\mu-\frac14F'_{\mu\nu}F'^{\mu\nu} +\bar\chi(i\slashed D-m_\chi)\chi-\frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu}, \qquad D_\mu=\partial_\mu-ig_\chi A'_\mu,$2, and the use of a massless hidden photon interaction model (Xu, 2022, Xu, 2024, Xu, 2024). Second, MCPs are not only “missing-energy particles.” The current literature supports direct detection channels ranging from ionization tracks in $\mathcal{L}=\sum_q e_q\,\bar q\gamma^\mu q\,A_\mu-\frac14F'_{\mu\nu}F'^{\mu\nu} +\bar\chi(i\slashed D-m_\chi)\chi-\frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu}, \qquad D_\mu=\partial_\mu-ig_\chi A'_\mu,$3 colliders, to electron recoils in neutrino detectors, to coherent optical phase shifts, to geomagnetic spectral lines (Gorbunov et al., 2022, Plestid et al., 2020, Chen et al., 2022, Arza et al., 24 Jan 2025).

Taken together, the recent arXiv literature portrays MCPs as a unifying target across collider physics, neutrino detectors, underground rare-event searches, stellar evolution, cosmology, and precision magnetometry. The most robust general statement is not that a single experiment dominates, but that MCP sensitivity is strongly conditioned by production environment, kinematic regime, and ultraviolet realization. This suggests that future progress will come less from one “best” search than from cross-comparison among complementary probes that are sensitive to different combinations of $\mathcal{L}=\sum_q e_q\,\bar q\gamma^\mu q\,A_\mu-\frac14F'_{\mu\nu}F'^{\mu\nu} +\bar\chi(i\slashed D-m_\chi)\chi-\frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu}, \qquad D_\mu=\partial_\mu-ig_\chi A'_\mu,$4, $\mathcal{L}=\sum_q e_q\,\bar q\gamma^\mu q\,A_\mu-\frac14F'_{\mu\nu}F'^{\mu\nu} +\bar\chi(i\slashed D-m_\chi)\chi-\frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu}, \qquad D_\mu=\partial_\mu-ig_\chi A'_\mu,$5, abundance, mediator structure, and source geometry.