Follow-up Search for a Tentative Dark Photon Signal Near 19.5 $μ$eV using ORGAN-Q infrastructure

TL;DR

The study tests a tentative dark photon signal reported near $f_X \simeq 4.71$ GHz ($m_X \simeq 19.5\,\mu\mathrm{eV}$) with kinetic mixing $\varepsilon \sim 6.5\times10^{-15}$. Using the ORGAN-Q infrastructure in a magnet-free, narrowband cavity search, the team targets the TM$_{010}$ mode around $f_0 \approx 4.71018$ GHz and conducts a $\sim 13$-day integration with a standard haloscope analysis pipeline. No excess consistent with the DP template is observed, and a 95% CL upper limit of $\varepsilon < 6.5\times10^{-15}$ is set within roughly ±15 linewidths of the reported mass, excluding the TASEH signal at 99.92% CL under the random-polarization assumption. The results constrain DP parameter space in this mass range and validate ORGAN-Q’s capability for high-sensitivity, low-field dark matter searches.

Abstract

A recent independent dark photon (DP) focused reanalysis of existing data from the TASEH axion haloscope experiment reported a tentative DP dark matter signal with local significance $\sim 4.7σ$ at $f_X \simeq$ 4.71 GHz, corresponding to $m_X \simeq 19.5~μ$eV and kinetic mixing $ε\sim 6.5\times 10^{-15}$. Motivated by this report, we performed a dedicated, narrowband follow-up experiment to confirm or refute the signal with a cryogenic microwave cavity operated \emph{without} a magnetic field, leveraging the ORGAN-Q dilution refrigeration and receiver chain. Scanning a window centered on the reported frequency over a live time of $T_{int}\sim 13$ days, we find no excess consistent with a dark photon signal as reported, and set $95\%$ C.L. exclusion on $ε$ in a narrow mass range around $\sim 19.5~μ$eV, excluding a signal of the strength and frequency reported to 99.92$\%$ confidence. We discuss the experiment and present the exclusion limits.

Figures (4)

• Figure 1: Schematic of the experimental setup includes a cryogenic resonant cavity at 25 mK, coupled to a low-noise HEMT amplifier chain for signal readout (blue line). The amplified signal is mixed down using an image-rejection mixer (IRM) before being digitized by a Moku Pro. A variable-temperature hot load and switch enable calibrated noise-injection for system noise characterisation. Microwave transmission and reflection paths are measured using a vector network analyzer (VNA). Cryogenic piezoelectric actuators (grey) control movement of the antenna (coupled directly to a cryogenic circulator) and the sapphire tuning stub (blue).
• Figure 2: Left: simplified COMSOL simulation of the cavity used in the experiment. The tuning stub, and strongly coupled antenna can be seen. The simulation was used to inform the sensitivity, and compared against experimental results. Top right: the simulated frequency against tuning stub position along the z-axis. The '0' position was set such that the mode was at the frequency of interest, and the cavity was tuned around this position. Bottom right: the simulated geometry factor divided by the polarisation angle normalisation against frequency. The polarisation normalised geometry factor presented here is equivalent to the form factor typically presented in axion haloscopes.
• Figure 3: (a) Distribution of normalised grand spectrum bins $\delta^g_k/\sigma^g_k$ following Savitzky–Golay (SG) filtering, vertical combination of overlapping spectra, and horizontal re-binning according to the Maxwell–Boltzmann signal template. The observed narrowing, quantified by $\xi^g = 0.90$, is attributed to the negative correlations introduced by the SG filter. (b) Signal-to-noise ratio (SNR) histograms for a synthetic DP signal embedded in Gaussian white noise, with one dataset passed through the SG filter (black) to simulate the analysis pipeline and the other left unfiltered (blue). The corresponding attenuation of a DP signal is quantified by the ratio of the mean SNRs, $\eta^{\text{SG}} = 7.60/9.37 = 0.81$.
• Figure 4: Exclusion limits on the dark photon kinetic mixing parameter $\epsilon$ for the ORGAN follow-up experiment assuming the random polarization scenario. The 95% confidence level limits are shown in blue, assuming a local dark matter density of $\rho_{\text{DM}}=0.45\,\text{GeV cm}^{-3}$. The shaded band represents the associated uncertainty in the limit. For comparison, the claimed signal reported by the TASEH collaboration is shown in red, with the horizontal line denoting the reported mixing strength of $\epsilon\simeq6.5\times10^{-15}$ and the vertical band marking the signal frequency $\pm 1$ expected dark photon linewidth.