Precision Metrology Meets Cosmology: Improved Constraints on Ultralight Dark Matter from Atom-Cavity Frequency Comparisons

We conduct frequency comparisons between a state-of-the-art strontium optical lattice clock, a cryogenic crystalline silicon cavity, and a hydrogen maser to set new bounds on the coupling of ultralight dark matter to standard model particles and fields in the mass range of ${10}^{-16}-{10}^{-21}\mathrm{eV}$. The key advantage of this two-part ratio comparison is the differential sensitivity to time variation of both the fine-structure constant and the electron mass, achieving a substantially improved limit on the moduli of ultralight dark matter, particularly at higher masses than typical atomic spectroscopic results. Furthermore, we demonstrate an extension of the search range to even higher masses by use of dynamical decoupling techniques. These results highlight the importance of using the best-performing atomic clocks for fundamental physics applications, as all-optical timescales are increasingly integrated with, and will eventually supplant, existing microwave timescales.

Figure 1

Atom-cavity frequency comparisons. (a) Schematic depiction of the three frequencies that compose the ratios for this dark matter search. Fractional fluctuations of these ratios are sensitive to fractional fluctuations of fundamental quantities via the inscribed equations. (b) Time-domain signal of the 12-day comparison for the $f_{\mathrm{Sr}}/f_c$ ratio (red) and the 33-day comparison of $f_{\mathrm{H}}/f_c$ (blue). For visual clarity, $f_{\mathrm{H}}/f_c$ has been decimated by a factor of 720. (c) Frequency-domain signal derived from (b) using the Lomb-Scargle technique to estimate the PSD. Data for $f_{\mathrm{Sr}}/f_c$ have a bin width of $3.4\times {10}^{-6}\mathrm{Hz}$ and for $f_{\mathrm{H}}/f_c$ have a bin width of $3.7\times {10}^{-7}\mathrm{Hz}$. Black lines indicate the noise model for $f_{\mathrm{Sr}}/f_c$ (dashed) and $f_{\mathrm{H}}/f_c$ (dot-dashed) ratios.

Figure 2

Dilatonic dark matter exclusion plot. (a) 95% confidence limits placed on the electromagnetic gauge modulus $d_e$, showing the improved limits set by $f_{\mathrm{Sr}}/f_c$ (red) and $f_{\mathrm{H}}/f_c$ (blue) ratios in the mass range above $1\times {10}^{-19}\mathrm{eV}$. The maximum projected sensitivity for a search of the same 11-day duration without data gaps is included (light green), highlighting the potential of this technique. Limits derived from previous spectroscopic searches (black lines) [38, 56, 57] and equivalence principle tests (purple lines) [58, 59] are included. (b) Demonstration of a significantly improved limit on the electron mass modulus ($d_{m_e}$) derived from the $f_{\mathrm{H}}/f_c$ (blue) ratio. Limits from equivalence principle tests (purple lines) [58, 59] are included. Shaded regions in both (a) and (b) are excluded at the 95% confidence level given the observed signal and noise models.

Figure 3

High-frequency search with dynamical decoupling. (a) Experimental sequence to interleave clock and spin-echo pulses. The dark time for both sequences is 1 s, with the number of echo pulses $N=1$, 3, and 5. (b) Time-domain signal for the three-pulse sequence over a single day of data collection, or 33 043 s. (c) Limits derived from each of the three pulse sequences demonstrating the flexibility in tailoring the sensitivity function for targeted searches. (d) Comparison of the limit derived from interpolating the low-frequency data of the clock sequence to higher frequencies (gray) compared to the dynamical decoupling sequence for $N=3$ (red) showing clear advantage for frequencies above the Nyquist frequency and the customized sensitivity for specific mass regions.