Long-Lived Dark Photons in New Physics

• Long-lived dark photons are hypothetical massive vector bosons from an extra U(1) symmetry that exhibit macroscopic decay lengths due to feeble kinetic mixing.
• They are produced via meson decays, proton bremsstrahlung, and exotic Higgs decays, leading to signatures like displaced vertices, lepton jets, and non-pointing photons.
• Search strategies employ precision vertexing, timing detectors, and far-forward experiments, with current limits constraining the kinetic mixing parameter and mass range.

A long-lived dark photon is a hypothetical massive vector boson associated with an extra U(1) gauge symmetry, interacting with the Standard Model (SM) via suppressed kinetic mixing and possessing a macroscopic proper decay length—typically millimeters to kilometers in the rest frame. These particles are ubiquitous in extensions of the SM with hidden sectors and play a central role in contemporary searches for new physics at accelerator and beam-dump experiments. The distinctive phenomenology of long-lived dark photons arises from both their feeble couplings (kinetic mixing parameter $\epsilon \ll 1$), which suppress visible decays, and their possible production in rare SM or exotic transitions, enabling experimental searches through displaced or non-pointing decay signatures.

1. Theoretical Motivation and Model Framework

Long-lived dark photons arise in models that augment the SM gauge structure by an extra Abelian gauge group U(1)$_D$, with gauge boson $A'_\mu$ (often denoted $\gamma_D$). The Lagrangian contains the renormalizable kinetic-mixing operator: $$\mathcal{L} \supset -\frac{1}{4}F_{\mu\nu}F^{\mu\nu} - \frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} + \frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu} + \frac12 m_{A'}^2 A'_\mu A'^\mu$$ where $F_{\mu\nu}$ and $F'_{\mu\nu}$ are the SM photon and dark-photon field strengths, and $\epsilon \ll 1$ is the kinetic mixing parameter. After diagonalization and field redefinition, $A'$ acquires an effective coupling $\epsilon e$ to the SM electromagnetic current $_D$0, but with all SM fields otherwise neutral under U(1)$_D$1.

The proper decay width to a charged lepton pair $_D$2 is

$_D$3

with similar expressions for decays into hadrons above threshold, typically computed by multiplying the dimuon width by the experimentally measured $_D$4-ratio, $_D$5.

The proper lifetime is given by $_D$6. For representative parameters, e.g., $_D$7 MeV and $_D$8, one finds $_D$9 (Collaboration et al., 2018, Ferber et al., 2022). The lifetime scales as $A'_\mu$0, such that small $A'_\mu$1 and sub-GeV $A'_\mu$2 yield decay lengths spanning the millimeter to kilometer regime.

2. Production Mechanisms and Kinematics

2.1 Meson Decays and Proton Beam Facilities

At hadron colliders and fixed-target experiments, long-lived dark photons are often produced in association with neutral meson decays: $A'_\mu$3 Further production arises from proton bremsstrahlung and Drell-Yan (DY) quark–antiquark annihilation. Proton bremsstrahlung is modeled using the Weizsäcker-Williams approximation, with substantial theoretical uncertainty stemming from the modeling of proton elastic and inelastic form factors, and the choice of virtuality regulator $A'_\mu$4 (Kyselov et al., 2024).

2.2 Collider Production: Higgs Decays and Exotic Portals

Long-lived dark photons are extensively searched for in exotic Higgs decays, such as $A'_\mu$5 (Higgs portal), often via intermediate dark-Higgs states, or more complex hidden-sector models. The partial width of $A'_\mu$6 proceeds via the mixing parameter(s) (e.g., a Higgs mixing angle or portal coupling) and is affected by kinematics, e.g., via phase-space factors depending on $A'_\mu$7 and $A'_\mu$8: $A'_\mu$9 (Jeanty et al., 2022, 2206.12181).

In the context of hadronic colliders (e.g., LHC), production through gluon–gluon fusion (ggF), vector-boson fusion (VBF), and associated production with electroweak bosons (WH, ZH) is explored, each offering distinct topologies and backgrounds (2206.12181, Collaboration, 2023).

3. Experimental Signatures and Search Strategies

The signature of a long-lived dark photon is a displaced decay to SM charged particle pairs—$\gamma_D$0, $\gamma_D$1, or hadrons—at measurable distances from the primary interaction point, yielding 'lepton jets', narrow clusters of tracks, or non-pointing photon signals.

3.1 Displaced Vertices

Vertex detectors at $\gamma_D$2 and hadron colliders can resolve decay vertices with radial displacement as small as $\gamma_D$3m (D'Onofrio et al., 2019, collaboration et al., 2017, Ferber et al., 2022). Event selection usually requires:

• Identification of a high-momentum photon (prompt or radiative) or triggered event topology.
• Reconstruction of a displaced, two-pronged vertex in the inner tracker or vertex detector with invariant mass consistent with $\gamma_D$4.
• Suppression of backgrounds from material-induced conversions, Dalitz decays, and long-lived SM hadrons via kinematic, topological, and vertex-quality cuts.

3.2 Lepton Jets and Calorimeter/Muon System Signatures

At the LHC, decays in the calorimeter or muon systems are interpreted as 'dark-photon jets' (LJs), requiring specialized reconstruction algorithms (e.g., clustering stand-alone muon tracks, energy-deposit patterns with low EM fraction, and isolation cuts) (2206.12181, Collaboration, 2023). Dedicated neural-network taggers distinguish cosmic-ray and beam-induced backgrounds from signal.

For ultra-long-lived scenarios, signals may appear as displaced photons in the HCAL (hadronic calorimeter), e.g., in future Higgs factories (Jiang et al., 30 Oct 2025).

3.3 Timing and Surface/Lifetime-Frontier Detectors

Precision timing detectors (e.g., CMS-MTD) can resolve delayed decays via nanosecond-scale time-of-flight differences (Du et al., 2021, Du et al., 2019). Surface or far-forward detectors (e.g., MATHUSLA, FASER, FACET) exploit long baselines to catch decays occurring hundreds of meters from the IP, operating in ultra-low-background environments (Collaboration et al., 2018, Du et al., 2021, Araki et al., 2020). Event selection here is optimized for energetic, collimated decay products and includes geometric acceptance and energy thresholds.

4. Experimental Limits and Projected Sensitivities

The coverage in the dark photon parameter space—mass $\gamma_D$5 vs. kinetic mixing $\gamma_D$6—depends strongly on the detector geometry, integrated luminosity, and decay length regime of sensitivity.

4.1 Current and Future Collider Results

• ATLAS: Excludes $\gamma_D$7 at 95% CL for $\gamma_D$8 and $\gamma_D$9 GeV, corresponding to $$\mathcal{L} \supset -\frac{1}{4}F_{\mu\nu}F^{\mu\nu} - \frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} + \frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu} + \frac12 m_{A'}^2 A'_\mu A'^\mu$$0, through a combination of ggF, WH, and VBF searches (2206.12181, Collaboration, 2023). Exclusion is set in both branching fraction and $$\mathcal{L} \supset -\frac{1}{4}F_{\mu\nu}F^{\mu\nu} - \frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} + \frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu} + \frac12 m_{A'}^2 A'_\mu A'^\mu$$1 space.
• LHCb: First displaced-vertex dedicated search in the low-mass window $$\mathcal{L} \supset -\frac{1}{4}F_{\mu\nu}F^{\mu\nu} - \frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} + \frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu} + \frac12 m_{A'}^2 A'_\mu A'^\mu$$2 MeV, with no excess observed and 90% confidence limits on $$\mathcal{L} \supset -\frac{1}{4}F_{\mu\nu}F^{\mu\nu} - \frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} + \frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu} + \frac12 m_{A'}^2 A'_\mu A'^\mu$$3 set as a function of $$\mathcal{L} \supset -\frac{1}{4}F_{\mu\nu}F^{\mu\nu} - \frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} + \frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu} + \frac12 m_{A'}^2 A'_\mu A'^\mu$$4 and $$\mathcal{L} \supset -\frac{1}{4}F_{\mu\nu}F^{\mu\nu} - \frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} + \frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu} + \frac12 m_{A'}^2 A'_\mu A'^\mu$$5 (collaboration et al., 2017). Sensitivity is limited by the available sample size, but further improvements are expected with increased Run 3 luminosity and upgraded vertex reconstruction.
• Belle II: Displaced-vertex searches sensitive to $$\mathcal{L} \supset -\frac{1}{4}F_{\mu\nu}F^{\mu\nu} - \frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} + \frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu} + \frac12 m_{A'}^2 A'_\mu A'^\mu$$6 GeV for $$\mathcal{L} \supset -\frac{1}{4}F_{\mu\nu}F^{\mu\nu} - \frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} + \frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu} + \frac12 m_{A'}^2 A'_\mu A'^\mu$$7 in $$\mathcal{L} \supset -\frac{1}{4}F_{\mu\nu}F^{\mu\nu} - \frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} + \frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu} + \frac12 m_{A'}^2 A'_\mu A'^\mu$$8 mode, and for $$\mathcal{L} \supset -\frac{1}{4}F_{\mu\nu}F^{\mu\nu} - \frac{1}{4}F'_{\mu\nu}F'^{\mu\nu} + \frac{\epsilon}{2}F_{\mu\nu}F'^{\mu\nu} + \frac12 m_{A'}^2 A'_\mu A'^\mu$$9 in $F_{\mu\nu}$0 and $F_{\mu\nu}$1, utilizing the clean environment and optimized vertexing (Ferber et al., 2022).
• FASER and FASER2: In Run 3 (150 fb$F_{\mu\nu}$2), FASER probes $F_{\mu\nu}$3 few $F_{\mu\nu}$4–$F_{\mu\nu}$5 for $F_{\mu\nu}$6–$F_{\mu\nu}$7; FASER2 (HL-LHC, 3 ab$F_{\mu\nu}$8) extends coverage up to $F_{\mu\nu}$9 GeV and down to $F'_{\mu\nu}$0 (Collaboration et al., 2018). With scalar portal production, FASER2 can reach $F'_{\mu\nu}$1 as low as $F'_{\mu\nu}$2 provided the scalar decays promptly (Araki et al., 2020).
• Heavy Photon Search (HPS): The first displaced-vertex search in $F'_{\mu\nu}$3–$F'_{\mu\nu}$4 MeV region, with no signal observed and sensitivity projected to improve to $F'_{\mu\nu}$5 in future high-statistics runs (Adrian et al., 2022).
• SHiP: Electromagnetic shower–induced production (cascade and positron annihilation $F'_{\mu\nu}$6) gives a $F'_{\mu\nu}$7–$F'_{\mu\nu}$8 increase in event rate over meson-only production, allowing sensitivity to $F'_{\mu\nu}$9 down to $\epsilon \ll 1$0 for $\epsilon \ll 1$1–$\epsilon \ll 1$2 MeV (Zhou et al., 2024).
• LHeC and FCC-he: Sensitivity to $\epsilon \ll 1$3 at $\epsilon \ll 1$4 MeV for $\epsilon \ll 1$5 ab$\epsilon \ll 1$6, with reach improving by a factor $\epsilon \ll 1$7 at FCC-he (3 ab$\epsilon \ll 1$8) (D'Onofrio et al., 2019).
• ILC: For $\epsilon \ll 1$9 ab$A'$0 at $A'$1 GeV, the projected Higgs branching ratio reach is $A'$2–$A'$3, corresponding to $A'$4–$A'$5 for $A'$6–$A'$7 GeV (Jeanty et al., 2022).
• Higgs Factories (Z-pole): Displaced photon searches in the HCAL at future $A'$8 colliders spanning $A'$9–$\epsilon e$0 m can target $\epsilon e$1–$\epsilon e$2 for $\epsilon e$3–$\epsilon e$4 GeV, outperforming monophoton searches by up to an order of magnitude in the long-lifetime regime (Jiang et al., 30 Oct 2025).

4.2 Theoretical Uncertainties

Modeling of proton-induced production carries uncertainties up to a factor $\epsilon e$5 in the bremsstrahlung (ISR) flux and $\epsilon e$630–50% in Drell-Yan production, directly impacting the sensitivity contours in $\epsilon e$7–$\epsilon e$8 space (Kyselov et al., 2024).

Tables below summarize the sensitivities of selected experiments for minimal kinetic-mixing dark photon models:

Experiment	Mass Range (GeV)	$\epsilon e$9 Sensitivity
FASER (Run 3)	$_D$00–$_D$01	$_D$02 – $_D$03
FASER2 (HL-LHC)	$_D$04–$_D$05	$_D$06 – $_D$07
Belle II	$_D$08–$_D$09	$_D$10 – $_D$11
ATLAS (Higgs→$_D$12)	$_D$13–$_D$14	$_D$15 – $_D$16
SHiP	$_D$17–$_D$18	$_D$19
HPS (future)	$_D$20–$_D$21	$_D$22

5. Complementary Probes and Model Variations

5.1 Hidden Sector and Enhanced Production

Extensions with additional hidden-sector states (e.g., dark fermions $_D$23) and extra gauge bosons can lead to alternate production mechanisms, notably hidden-sector radiation—where copious dark photons are generated off a dark fermion line—substantially increasing reach in large mass and small kinetic-mixing domains (Du et al., 2021, Du et al., 2019). This channel is especially potent for far-forward detectors (FACET, FASER2), as enhancement factors of $_D$24–$_D$25 relative to the minimal model are observed.

5.2 Cosmology and Dark Matter

Scenarios wherein the dark photon is cosmologically long-lived and stable (or decays outside the detector) provide viable dark matter candidates, especially in freeze-in scenarios at low reheating temperatures. The relevant parameter space is delimited by relic abundance requirements, invisible Higgs decay constraints (e.g., $_D$26), and searches for displaced photon signatures from mediator decays (Arias et al., 21 Jul 2025). Collider and cosmology constraints together exclude most of the "thermalized mediator" regime, isolating a viable window for feeble portal couplings, $_D$27–$_D$28 GeV, $_D$29–$_D$30 GeV, and $_D$31 m.

5.3 Theoretical and Experimental Systematics

A primary challenge stems from theoretical uncertainties in modeling production cross sections (especially proton bremsstrahlung and Drell-Yan at proton accelerators), as well as modeling of hadronic decays below the perturbative QCD threshold. These uncertainties shift exclusion and sensitivity contours by factors up to 10, and dedicated tools such as SensCalc facilitate propagating these uncertainties in experimental analyses (Kyselov et al., 2024).

6. Summary and Outlook

Long-lived dark photons are a primary focus of current and future lifetime-frontier searches at colliders, beam-dump facilities, and fixed-target experiments. Their phenomenology is controlled by the mass $_D$32 and kinetic-mixing $_D$33, with lifetime $_D$34 scaling as $_D$35. Experimental strategies span prompt decays, displaced-vertex searches, and non-pointing or delayed photon signatures, each probing distinct regions of parameter space. Synergy among complementary detectors (e.g., deep vertexing, large decay volumes, precision timing, beam-dump setups, and Higgs factories) is essential to maximally cover the allowed model space.

The continual extension of detector capabilities (timing layers, far-forward detectors, ultra-low-background environments), reduction of theoretical uncertainties (especially in production rates), and accumulation of larger integrated luminosity will further increase the reach for long-lived dark photons, potentially illuminating the structure of hidden sectors and their connections to the SM.