First Limits on Axion Dark Matter from a DALI Prototype

Abstract

We report a pilot dark-matter search with a cryogenic, magnetized, scaled-down DALI prototype. An analysis of 36 hours of data reveals no statistically significant excess attributable to axionlike particles. We therefore set new exclusion limits in the 6.883--6.920 GHz band, reaching an axion-photon coupling sensitivity of $g_{aγγ}\lesssim 1.27\times10^{-11}\,\mathrm{GeV}^{-1}$ at 28.54 $μ$eV. These results consolidate the DALI approach and motivate a next-stage haloscope to explore a broader mass range with upgraded instrumentation.

Figures (5)

• Figure 1: Schematic view of the DALI(PoP) prototype. A tunnel yoke holds an array of Nd blocks that generates a dipole-like magnetic field. Ambient axions are converted into microwave photons, which are collected by the antenna. The resonator, composed of 20 ZrO$_2$ layers and enclosed by a copper mirror, enhances the weak axion-induced signal to an observable level. The detected signal is subsequently processed by the readout chain shown below (from left to right: amplifier, band-pass filter, attenuator, amplifier, downconverter, low-pass filter, analog-to-digital converter, and data storage). The haloscope is housed in a Faraday cage to suppress external spurious signals (not shown). The magnet is segmented along its longitudinal axis.
• Figure 2: Field simulation vs. measurements. The magnet bore is approximately $30$ (transverse)$\times 11$ (height)$\times 40$ (longitudinal) cm. Two one-dimensional profiles are extracted from the 3D magnetic-field simulations: a center--longitudinal cut at half height (blue dashed) and a center--transverse cut (red dashed). The field generated by the magnets installed inside the steel yoke was simulated using Elmer FEMelmer, employing Gmshgmsh to generate the three-dimensional meshes. The dotted lines indicate the resonator size (10$\times$10$\times$14.5 cm) in its centered configuration, shown for reference. Control Hall-probe measurements are shown with 3% vertical error bars and horizontal probe-positioning bars; the right inset displays three measurements along the bore height at the longitudinal center. The simulated field distribution is in good agreement with the measurements.
• Figure 3: Quality factor and coupling factor. Measured quality factor data is shown in blue. The full width at half maximum for a fit, represented by a blue dashed line, is $\sim$20 MHz. In the present prototype configuration the receiver is undercoupled. The observed optical power coupling factor $\beta$ is depicted in red, while broadband data ($\sim$5--8 GHz) is smoothed with a SG baseline using a window size $W=201$ and polynomial order $d=1$, shown by a red dashed line as a reference. Throughout this work, $Q_L$ and $\beta$ are taken from the measured data, not from the fitted or smoothed curves.
• Figure 4: Normalized, correlation-corrected grand spectrum (left), with detection threshold $\alpha=3$ (blue). Narrow gaps indicate masked frequency bins identified from dedicated interference-control data. We label the four main masked regions as S1--S4, all attributed to instrumental origin—cf. Table \ref{['TableII']}. Inset: histogram of normalized power excesses with the best-fitting Gaussian (red dashed), showing its mean ($\mu$) and standard deviation ($\sigma$).
• Figure 5: New exclusion region (blue) at 90% CL. Narrow gaps indicate masked bins and vetoed neighborhoods around above-threshold excursions (Table \ref{['TableII']}). Other laboratory-based constraints in the vicinity of the newly excluded region are also shown PhysRevLett.133.221005PhysRevD.97.092001PhysRevLett.118.061302PhysRevLett.121.261302.