Oscillating Resonances: Imprints of ultralight dark matter at colliders

TL;DR

This work shows that ultralight dark matter can imprint oscillating resonances at colliders through mediator-based UV completions, producing time-dependent mediator masses or kinetic mixing. It develops theoretical models and loop-induced couplings that link ULDM to variations in α and m_e, and demonstrates how collider searches can be reinterpreted to probe an oscillating background. The authors introduce peak-reconstruction methods using mass-binned data and time-domain FFT-based analyses to recover the central resonance and its absolute normalization, and they quantify the impact of such oscillations on current and future experiments like Belle II, LHCb, and SHiP. The findings indicate that oscillating-resonance signatures constitute a powerful, complementary probe of ULDM that can reach parameter space not excluded by atomic clocks or equivalence-principle tests, making time-stamped collider data particularly valuable for discovery.

Abstract

In models where ultralight fields constitute dark matter, the misalignment mechanism leads to coherent, low-amplitude oscillations in fundamental constants. This effect arises from effective operators that couple dark matter to Standard Model fields. We present different models that can induce these effective operators by integrating out a mediator field. For mediator masses within the reach of collider searches, an alternative way to discover ultralight dark matter is to search for a resonance. Due to being a mediator to dark matter, the mass of the mediator oscillates. The resonance therefore, should not appear as a single isolated peak, but is smeared out once data is averaged over an oscillation cycle or more. Remarkably, the oscillation period and amplitude are in the range of current and future collider searches, even though constraints from atomic clocks probing variations of the fine-structure constant and the electron mass are very strong. We recast existing searches and projections from Belle II, LHCb and SHiP for an `oscillating resonance', and discuss how the periodicity of the signal can be used to reconstruct the peak from mass-binned data. We further show that time-stamped data would allow to unfold the signal via a fast Fourier transform and determine the significance of the signal for different background levels. The discovery of an oscillating resonance at a collider, with characteristics as predicted in ultralight dark matter scenarios, would constitute a powerful probe of dark matter's underlying nature.

Figures (13)

• Figure 1: Leading-order Feynman diagrams that induce dark matter couplings to SM fermions and photons for a vector mediator $X$. The analogous diagrams induce couplings to photons and fermions in the scalar mediator model by replacing $X$ by $S$.
• Figure 2: The loop-induced photon and electron couplings for a dark photon $X$ in the ULDM background. The collider bounds (LHCb and Belle-II) correspond to a 10% mass modulation and $g_X = 4\pi$. The shaded lines denote the latest bounds from precision sensors such as atomic clocks (Rb/Cs, BACON), atomic spectroscopy (Dy/Dy), clock-cavity comparisons (H/Si, Sr/Si) and EP violation experiment (MICROSCOPE).
• Figure 3: The loop-induced photon and electron couplings for a scalar $S$ in the ULDM background. The collider bounds correspond to Belle and Belle II limits from $e^+ e^- \to S e^+ e^- \to\ell^+\ell^- e^+ e^-$ for 10% mass modulation. The shaded lines denote the latest bounds from precision sensors such as atomic clocks (Rb/Cs, BACON), atomic spectroscopy (Dy/Dy), clock-cavity comparisons (H/Si, Sr/Si) and EP violation experiment (MICROSCOPE).
• Figure 4: Normalised invariant mass distributions for a static dark photon (black) and in the presence of an oscillating DM background (blue) for a central mass $m_X^0=500\ {\rm MeV}$ and modulation amplitude $\delta m_X=20\ {\rm MeV}$. The gray band indicates the detector resolution around the central mass. In the oscillating case, the resonance is spread across the modulation window, diluting the signal within any single mass bin. The red marker highlights the tallest bin (local maximum) in the modulated distribution, which is located near the edge of the modulation window rather than at the central mass.
• Figure 5: Ratio of the kinetic mixing parameter in the oscillating case to the static case $\epsilon_{\rm osc}/\epsilon_{\rm static}$ for a 10 GeV dark photon at Belle II, shown as a function of the ULDM mass $m_\phi$. Different coloured curves correspond to different values of the ULDM energy scale $\Lambda$, quantifying the DM coupling strength with the SM. The grey shaded region denotes oscillation periods longer than the experimental run-time ($T_{\rm osc} \gtrsim\ 3\ {\rm years}$, where the modulation is effectively static). The coloured markers indicate the ULDM masses at which the modulation is 50% for each $\Lambda$. At larger $m_\phi$, the oscillation period becomes short compared to the data-taking time, and the effect of modulation averages out, causing the ratio to approach unity.