Hiding a Light Vector Boson from Terrestrial Experiments: A Chargephobic Dark Photon

TL;DR

This work generalizes light vector mediator phenomenology by allowing a massive gauge boson $X$ to couple to an arbitrary linear combination of the SM electromagnetic and $B-L$ currents, parametrized by a dark mixing angle $\varphi$ and mass $m_X$ in the MeV–GeV range. The authors identify a distinctive chargephobic limit $\varphi=3\pi/4$ where tree-level couplings to electrons and protons vanish, rendering most terrestrial searches ineffective; constraints therefore arise predominantly from neutrino scattering, supernova cooling, and early-Universe $\Delta N_{\rm eff}$ calculations, with RG running reintroducing tiny couplings at higher energies. They perform a comprehensive, model-independent analysis across cosmology, astrophysics, and collider/beam-dump experiments, rederiving and recasting major bounds for general $\varphi$, and show that supernova and $\Delta N_{\rm eff}$ bounds provide robust coverage across the parameter space. The study also discusses meson-dominated hadronic decays, RG evolution of mixing, and future probes like SHiP and REDTOP, which can explore hadronic channels and hadronic-portal production inaccessible to charged-lepton–driven searches. Overall, a chargephobic vector boson can evade most current terrestrial constraints while remaining testable via neutrino/CMB/BBN and next-generation experiments, illustrating the importance of complementary probes for flavor-universal anomaly-free vectors.

Abstract

We calculate the terrestrial, astrophysical and cosmological constraints on a light vector boson that couples to an arbitrary combination of the electromagnetic and $B-L$ currents of the Standard Model. The dark photon and a vector boson coupling to $B-L$ are special cases of our generalized flavor-universal anomaly-free vector boson, requiring just one additional parameter (the "dark mixing angle" corresponding to the linear combination of the electromagnetic and $B-L$ currents) beyond that of the overall coupling strength and the vector boson mass, where we focus on the range $1\, {\rm MeV}$ to $60\, {\rm GeV}$. We perform a detailed investigation of a unique combination where the vector boson couplings to electrically charged leptons and protons are highly suppressed: the "chargephobic dark photon". A chargephobic vector boson is very weakly constrained by current terrestrial experiments including beam dumps and collider experiments, since they rely on couplings to electrons and protons. Instead, neutrino scattering experiments (such as COHERENT), astrophysical sources (supernova emission), and cosmology ($ΔN_{\rm eff}$) provide the strongest constraints due to the nonzero couplings of the chargephobic vector boson to neutrinos and neutrons. Indeed, we find that supernova emission and $ΔN_{\rm eff}$ provide constraints throughout the space of dark mixing angles, demonstrating their importance to provide model-independent constraints. For nearly all of the parameter space, a chargephobic vector boson is the most weakly constrained anomaly-free vector boson that couples to flavor-independent or flavor-dependent combinations of Standard Model currents. Finally, we highlight the importance of future experiments, including SHiP, that are able to probe new regions of the chargephobic parameter space due to the significantly improved detector capabilities.

Figures (17)

• Figure 1: Charges $x_i$ defined in Eq. (\ref{['electricChargeUX']}). The proton, neutron, and $\pi^{+}$ charges are equal and opposite to the electron, neutrino, and $\pi^{-}$ respectively. We also show the scaled Cesium charge $x_{\rm Cs}/A_{\rm Cs}$ which is relevant for $\text{CE}\nu\text{NS}$ experiments. The vertical dotted lines correspond to specific combinations of the electromagnetic and $B-L$ currents that are of interest, as shown in Fig. \ref{['angleParamSpaceFig']}.
• Figure 2: The dark mixing angle "compass". Each arrow corresponds to a dark mixing angle $\varphi$ that has a vanishing tree-level coupling to at least one SM state. The compass is used in subsequent plots to easily illustrate which dark mixing angles are being considered.
• Figure 3: Terresterial experiment constraints, astrophysical constraints, and cosmological constraints on the dark photon ($\varphi=0$) and $B-L$ case ($\varphi=\pi/2$), see text for full details. For $B-L$, there are additional constraints that arise due to non-zero couplings to neutrinos, in particular those labeled CHARM, COHERENT, CONUS, NA64, and TEXONO. For the dark photon case, the constraints labeled BBN, LESNe, SN1987A $\gamma$, 511 keV line, Fireball PVO/GW170817 were taken directly from Ref. Caputo:2025avc. Since the branching fraction to charged leptons in the $B-L$ case is comparable to the dark photon case, we anticipate the same processes that led to the bounds Caputo:2025avc to also apply to $B-L$. We have taken the liberty to overlay these dark photon bounds from Caputo:2025avc as unlabeled light-shaded regions with dashed-lines in the $B-L$ case. This is illustrative of where one might expect these constraints to appear, however, a detailed calculation for $B-L$ is beyond the scope of the paper.
• Figure 4: The central results of our paper: the current constraints (left) and anticipated future sensitivity (right) on a chargephobic dark photon ($\varphi=3\pi/4$). The absence of the vast majority of terrestrial experiments (compare with Fig. \ref{['DPandBmLCurrentConstraintPlots']}) arises due to the specific couplings that vanish for chargephobic dark photon: charged leptons and protons. See text for full details.
• Figure 5: Tree level cross section ratio of $X$ to $A'$ production rate. All values in the table are multiplied by a factor of $g_X^2/(\varepsilon e)^2$; we set this extra term to one in the plot. The ratio also sets energies in both models equal when necessary.