Broadband interferometry-based searches for photon-axion conversion in vacuum

A novel experiment is introduced to detect photon-axion conversion independent of the dark-matter hypothesis in a broad mass-range called WISP Interferometer (WINTER). The setup consists of a free-space Mach-Zehnder-type interferometer incorporating an external magnetic field and vacuum in one of the arms, where photon-axion mixing occurs via the Primakoff effect and is detected through changes in amplitude. The expected axion-induced signal is then modulated by polarization changes. The experiment is designed to integrate a Fabry-Pérot cavity with a finesse of $10^{5}$ that will be operated in a vacuum environment, significantly enhancing the sensitivity. It is projected to reach the DFSZ theoretical line with photon-axion coupling sensitivities down to $g_{aγγ}\simeq 3.7\times10^{-14}$ $\text{GeV}^{-1}$ for axion masses up to 380 $μ$eV.

💡 Research Summary

The manuscript proposes a novel experimental concept, the WISP Interferometer (WINTER), to search for photon–axion conversion without relying on the dark‑matter hypothesis. The core of the setup is a free‑space Mach‑Zehnder interferometer (MZI) in which one arm (the “sensing arm”) is placed inside a strong dipole magnet (≈9 T, 10 m long) and operated under high vacuum. Inside this arm a Fabry‑Pérot cavity (FPC) of length 10 m and finesse ≈10⁵ is installed. The cavity dramatically increases the effective photon path length in the magnetic field, thereby enhancing the conversion probability P_{γ→a} ∝ (g_{aγγ} B L)².

The theoretical framework follows the standard Primakoff mixing of a pseudo‑scalar axion field a with the electric field in the presence of an external magnetic field B_ext. In the plane‑wave approximation the evolution equation can be written as (∂²_z + M²)ψ = 0, where ψ = (E, a)ᵀ and the mixing matrix M² contains the photon and axion momenta as well as the mixing term G = g_{aγγ} B_ext/2. For axion masses much smaller than the photon energy (m_a² ≪ ω_γ²) and for short interaction lengths (k_osc z ≪ 1) the conversion probability reduces to P_{γ→a} ≈ (G z)², i.e. quadratic in the interaction length. The FPC therefore boosts the signal by a factor (L_FPC)² ≈ (10 m)² while the high finesse ensures a circulating power enhancement of order F/π.

Detection relies on two orthogonal modulation schemes. First, an electro‑optic amplitude modulator (EOM‑AM) imposes a sinusoidal intensity modulation at ω_m ≈ 1 kHz. Second, a polarization modulator (EOM‑PC) switches the linear polarization between a state parallel to B_ext and a state orthogonal to it at a much lower frequency ω_sig ≈ 20 Hz. Because the conversion probability depends on the projection of the photon polarization onto the magnetic field, the axion‑induced loss appears only in the parallel state. By demodulating the dark‑port signal (PD2) first at ω_m (to lock the interferometer near a dark fringe) and then at ω_sig (to isolate the polarization‑dependent loss), the experiment extracts a slowly varying amplitude proportional to P_{γ→a}.

Stabilization of the interferometer and the cavity is achieved with standard feedback techniques. The MZI phase is locked using a piezo‑driven mirror (M4) driven by the error signal from the bright port (PD1) after demodulation at ω_m. The FPC length is locked via a Pound‑Drever‑Hall scheme: a phase‑modulation sideband at f_FPC = 7.5 MHz (half the free‑spectral range) is generated by EOM‑PM, reflected light is detected on PD3, and the piezo‑driven mirror (M3) is adjusted to keep the cavity resonant. The chosen modulation frequencies avoid cross‑talk, and the polarization‑modulation frequency is kept well below the cavity photon lifetime (τ_ph ≈ 0.7 ms) to allow the intracavity power to follow the modulation.

Noise analysis identifies the dominant contributions as the dark‑current noise of the InGaAs detector (NEP ≈ 0.7 fW · Hz⁻¹ᐟ²) and shot noise from the residual carrier leaking through the dark fringe (≈0.01 % of the input power). With an input laser power of ~100 W, modulation depth β_m ≈ 0.5, and the magnetic field and cavity parameters above, the expected signal power at the dark port is P_sig = ½ P_tot β_m P_{γ→a}. Solving for a signal‑to‑noise ratio of unity yields a sensitivity to the photon‑axion coupling of g_{aγγ} ≈ 3.7 × 10⁻¹⁴ GeV⁻¹. This reaches the DFSZ theoretical line and surpasses existing Light‑Shining‑Through‑Wall (LSW) experiments by 3–4 orders of magnitude.

The mass reach is limited by the oscillation length L_osc = π/k_osc. For the chosen magnetic field and photon energy (λ = 1064 nm), the condition L_osc ≥ L_FPC translates to m_a ≲ 0.5 meV (≈500 µeV). The authors therefore claim sensitivity up to 380 µeV, covering a substantial portion of the QCD axion window (50–1500 µeV) without assuming that axions constitute dark matter.

Practical challenges discussed include maintaining a 9 T field over 10 m while preserving ultra‑high vacuum, achieving and stabilizing a finesse of 10⁵ with a laser linewidth narrower than the cavity linewidth (≈0.1 kHz), and mitigating thermal lensing and diffraction losses on the cavity mirrors. The paper provides detailed calculations of mirror heating, diffraction clipping, and the required PID control loops for power‑matching the two polarization states.

In summary, WINTER combines a broadband interferometric readout with a high‑finesse resonant enhancement to achieve unprecedented sensitivity to photon‑axion conversion in vacuum. By exploiting polarization‑dependent modulation and dual‑stage demodulation, the experiment isolates the tiny axion‑induced amplitude loss from technical noise. If realized, WINTER would set the most stringent model‑independent limits on g_{aγγ} in the 0–380 µeV mass range, opening a new avenue for probing axion‑like particles beyond the reach of traditional haloscope or LSW techniques.