Dark Photon Cavity Haloscope
A microwave bell so quiet it might already be ringing in time with the dark sector — and we are still arguing whether it is.
Searches for kinetic mixing between an ordinary photon and a hypothetical massive 'dark photon' (γ′) that may form some or all of the dark matter halo around our galaxy. A high-quality-factor cavity is tuned across candidate dark-photon masses (typically 1 µeV–100 µeV, equivalent to a few hundred MHz to tens of GHz). At the resonant mass, a tiny ordinary-photon field appears inside the cavity even though no ordinary radiation is being driven — the dark-photon halo bleeds energy in through kinetic mixing. A near-quantum-limited microwave amplifier reads it out.
If a dark photon γ′ exists, the Standard Model photon and γ′ mix with a small parameter ε (typically 10⁻⁹–10⁻¹³ in regions of interest). A galactic halo of cold dark photons of mass m_γ′ acts like an external oscillating field at frequency f = m_γ′c²/h. Inside a microwave cavity tuned to that frequency, this field drives a real ordinary-photon mode whose amplitude is proportional to ε, the cavity volume, and √Q. The cavity is cooled to mK, shielded with mu-metal and superconducting cans, and read out with a Josephson parametric amplifier (JPA) or SNAIL/TWPA at the standard quantum limit. To scan a band, the cavity is mechanically tuned (movable rod, piezo-driven plunger) over weeks. The two leading platforms are dark-photon-specific dish/cavity searches (DM-Radio, Dark E-Field) below 1 µeV and reused axion haloscope cavities (ADMX, HAYSTAC) for masses in the µeV–tens-of-µeV range.
In plain terms
Imagine standing in a perfectly soundproof room with a tuning fork, listening for a faint hum that isn't coming from anything you can see. If a different kind of light is filling the universe — light invisible to your eyes — your tuning fork would still pick it up, just barely. A dark-photon haloscope is a metal can listening for that faint hum at a chosen pitch.
Also called: dark-photon haloscope, kinetic-mixing cavity search, dark SRF detector.
Why it matters
- One of the cleanest tests of 'hidden sector' models that arise generically in string compactifications and SUSY extensions.
- Cavity techniques developed here directly improve quantum sensors and primary-frequency standards.
- Same hardware constrains axion–photon coupling — a single experiment chases two distinct dark matter candidates.
- Reaches couplings six to ten orders of magnitude weaker than any direct accelerator probe of dark photons.
Where it lives
- ADMX (Axion Dark Matter eXperiment, University of Washington)
- HAYSTAC (Yale/Berkeley, with squeezed-state read-out)
- DM-Radio (Stanford SLAC, sub-µeV mass range)
- Dark SRF (Fermilab, superconducting RF cavity dark-photon search)
Key numbers
- Mass range explored: 0.1 µeV – 100 µeV (≈ 25 MHz – 25 GHz)
- Cavity Q: 10⁴–10⁵ (Loaded Q at mK)
- Operating temperature: 20 mK – 100 mK
- Magnetic field (axion-search variants): 7–9 T
- Sensitivity floor on ε: ~10⁻¹³ (ADMX run 1B-class limits)
Surprising facts
- These experiments routinely measure power signals smaller than 10⁻²³ W — about a million times less than a single solar photon arriving at Earth.
- ADMX operates at ~100 mK in superconducting magnetic fields of 7–8 T, the same regime as superconducting quantum computers, but tuned to listen rather than compute.
- Modern haloscopes use Josephson parametric amplifiers that approach the standard quantum limit — they add no more than half a quantum of noise per measurement bandwidth.
- A 4.7σ excess at 22 GHz reported by a Taiwan-based dark-photon search in 2025 is intriguing but unconfirmed: the result is a single-experiment hint and has not been reproduced elsewhere.
Open questions
- Reported 4.7σ excess at 22 GHz (2025): A 2025 Taiwan-based dark-photon search reported a 4.7σ candidate signal at 22 GHz consistent with kinetic-mixing parameter ε ~ 10⁻¹³. The result is single-experiment and unconfirmed; ADMX, DM-Radio and other haloscopes are scheduled to attempt direct reproduction. Until then, the community treats the excess as a hint, not evidence of dark photons.
- Unexplained narrow-band noise spurs: Cavity haloscopes routinely encounter narrow-band noise spurs that move with cryostat temperature or magnet ramping. Most are eventually traced to mechanical resonances or RF ingress, but a handful in published ADMX and HAYSTAC data have no firmly identified origin and are simply masked.
A short history
Pierre Sikivie proposed the cavity haloscope in 1983 as an axion search; the same technology was rederived for dark-photon kinetic mixing by Arias et al. (2012) and Chaudhuri et al. (2015). ADMX achieved DFSZ-axion sensitivity in 2018, the first axion search to probe theoretically motivated couplings. Dark-photon-specific designs (DM-Radio, Dark SRF, BREAD) followed in the early 2020s. In 2025 a Taiwan-based search reported a 4.7σ excess at 22 GHz consistent with a dark photon — the result remains single-experiment, unconfirmed, and is treated by the community as a candidate, not a discovery.
Sources
- Sikivie, PRL (1983) — cavity-haloscope proposal (https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.51.1415)
- Arias et al., JCAP (2012) — dark-photon CDM searches (https://iopscience.iop.org/article/10.1088/1475-7516/2012/06/013)
- ADMX collaboration — recent results (https://depts.washington.edu/admx/)
- Snowmass dark-photon report (2022) (https://arxiv.org/abs/2203.14923)