Scalable Dark Matter Searches Using Integrated Photonics

TL;DR

This work presents a scalable search for wave-like dark matter in the 0.1–3 eV mass window using integrated photonics. By treating the DM field as a classical EM source, it shows how axion and dark-photon couplings can resonantly excite on-chip photonic modes, with phase matching achieved via periodic refractive-index modulation and Bound States in the Continuum to maximize coupling. To overcome DM coherence constraints, the authors propose a frequency-multiplexed architecture that monitors hundreds of distinct resonant frequencies in parallel, yielding broad mass coverage without sacrificing per-resonator sensitivity. Projected sensitivities reach $g_{a\gamma}\sim 10^{-11}\,\mathrm{GeV}^{-1}$ and $\chi\lesssim 10^{-14}$ across 0.1–3 eV, using wafer-scale resonator arrays (≈$10^6$ resonators per 15 cm wafer) and phase-matching schemes with low-noise single-photon detectors; DP DM searches can proceed without magnets as a near-term demonstration, while future expansion to axions would leverage available magnet bore volumes. Overall, the approach leverages mature photonics fabrication to open a new discovery space for dark matter in a mass range largely unexplored by traditional haloscope techniques.

Abstract

Dark matter (DM) with masses of order an electronvolt or below can have a non-zero coupling to electromagnetism while being compatible with cosmological observations. In these models, the ambient DM behaves as a new classical source in Maxwell's equations, which can excite potentially detectable electromagnetic (EM) fields in the laboratory. We propose a new integrated photonics-based approach to search for DM candidates in the 0.1 - few eV mass range. This approach offers a wide range of wavelength-scale devices like resonators and waveguides that are readily fabricated in large quantities, enabling a scalable and novel search. In particular, we demonstrate that refractive index-modulated resonators, such as etched/grooved microrings, or patterned slabs, support EM modes with efficient coupling to DM. When excited by DM, these modes are read out by coupling the resonators to a waveguide that terminates on a micron-scale-sized single photon detector, such as a single pixel of a low-noise charge-coupled device or a superconducting nanowire. We then estimate the sensitivity of this experimental concept in the context of axion-like particle and dark photon models of DM, demonstrating that nanophotonic confinement and scalability can extend dark matter sensitivity into previously unexplored parameter space.

Figures (10)

• Figure 1: $(a)$ Schematic of $N$ resonators with distinct frequencies $\omega_R$, $\omega_R(1+\delta)$, ..., $\omega_R(1+\delta)^{N-1}$ coupled to a single readout bus. DM sources $s_{\mathrm{DM}}$ driving each resonator have the same frequency, but potentially different phases, depending on the resonator separation $d$. The bus-resonator interaction is parametrized by coupling constants $\kappa$. Typical parameter choices are given in the box. $(b)$ Signal power comparison (computed using coupled mode theory) demonstrates the scaling advantage of frequency multiplexing. Left: Same-frequency resonators ($\omega_R$, critically coupled) exhibit power saturation, preventing $N$-scaling, when the linear size of the array exceeds the DM coherence length. Right: Frequency-diverse resonators $\omega_R(1+\delta)^{i-1}$ eliminate inter-resonator interference, enabling independent operation and broadband coverage. This analysis provides the foundation for the experimental architecture presented in \ref{['fig:setup']}.
• Figure 2: Conceptual architecture for wafer-scale DM detection. Periodically-modulated resonators (inset, with discretely varied refractive indicies $n_1$ and $n_2$) provide phase matching for efficient DM-photon coupling. Frequency-multiplexed readout enables parallel monitoring of hundreds of potential DM masses, while low dark count single-photon detectors provide the noise performance needed for discovery.
• Figure 3: Projected axion-photon coupling sensitivity assuming SNR=1 in \ref{['eq:snr']}. The bold colored lines show future sensitivity projections for different experimental configurations: SNSPD-based detectors (blue/green) covering 0.1--1.12 eV and CCD-based detectors (red) covering 1.12--2 eV mass ranges. Our approach can probe couplings $g_{a\gamma} < 10^{-10}$ GeV$^{-1}$, reaching into the theoretically motivated QCD axion band (yellow). Solid lines represent near-term projections with $Q=500$ resonators, while dashed lines show potential improvements with higher-$Q$ resonators over longer integration times. The green line demonstrates the exceptional sensitivity achievable in a narrow mass range with $Q=50000$ resonators and 1-year integration time. For context, current limits from CAST CAST:2017uph, stellar evolution constraints Ayala:2014peaDolan:2022kul (gray dashed line), and telescope-based searches Todarello:2023hdkGrin:2006awJanish:2023kviYin:2024lla (MUSE, JWST, WINERED) are shown as well, taken from the repository AxionLimits AxionLimits.
• Figure 4: Projected kinetic mixing sensitivity for DP DM detection using SNSPD-based detectors (blue/green) and CCD-based detectors (red) with two different experimental scales. Also shown are the existing limits from solar dark photon searches (Xenon1T An:2020bxd) and direct detection experiments (LAMPOST Chiles:2021gxk).
• Figure 5: The normalized amplitude $\left(\sqrt{V^-1\int_V \varepsilon|{\bf E}|^2}\right)^{-1}\int_0^{2\pi}\frac{d\phi}{2\pi} E_y(r,z) \times \frac{r}{R_0}$ for a fine-tuned $m=1,K=0$ confined mode in a cylindrical fiber Bragg grating is shown. This mode has a non-zero overlap factor $\eta$ and therefore couples to the DM source. We took a a periodic relative permittivity $\varepsilon= \frac{1}{2} (\varepsilon_1-\varepsilon_2) \left[1+\sin (2 \pi z/\Lambda)\right]+\varepsilon_2$ with $\varepsilon_2=1,\varepsilon_1=5$ in the core and $n_o=1$ in the cladding.