🔬 optics

Axion polarimetry with parametric readout: photon-number conversion and sensitivity limits

This paper proposes a bias-assisted active photon-number conversion scheme using a four-port optical ring cavity with parametric readout to linearize axion-induced polarization rotation signals, potentially extending low-mass axion sensitivity to the $10^{-18}$ – $10^{-16}$ eV range while quantifying specific noise advantages and stability requirements for future implementations.

Original author: Antonio Capolupo, Gabriele Pisacane, Raoul Serao

Published 2026-11-03

📖 5 min read🧠 Deep dive

CC BY 4.0

Original author: Antonio Capolupo, Gabriele Pisacane, Raoul Serao

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the author. For technical accuracy, refer to the original paper. Read full disclaimer

The universe is filled with a substance that we cannot see, touch, or directly detect, yet its gravitational pull holds galaxies together and shapes the cosmos. Astronomers call this invisible mass dark matter, and while its existence is supported by decades of observation, its true nature remains one of the greatest mysteries in physics. Among the many theoretical candidates proposed to explain this hidden substance, a particle called the axion stands out. If it exists, the axion would be incredibly light and would interact very weakly with ordinary light. This faint interaction is the key to finding it: if axions are present, they should cause a tiny, rhythmic wobble in the polarization of light beams passing through space, effectively rotating the direction in which the light waves vibrate. Detecting this minuscule rotation is the goal of many experiments, but the signal is so small that it is easily drowned out by the background noise of the instruments themselves.

A team of physicists has now proposed a new way to listen for this whisper, using a sophisticated optical setup designed to amplify the signal without amplifying the noise. Their work focuses on a specific method of reading out the data from a ring-shaped optical cavity, a device that traps light and makes it bounce around many times to increase its sensitivity. The researchers developed a scheme that uses a calibrated bias, a deliberate and precise adjustment applied to the light before it is measured, to convert a signal that would normally be too weak to see into a clear, linear response. In their simulations, this approach opens a window to a range of axion masses that has not been thoroughly explored, specifically between 10⁻¹⁸ and 10⁻¹⁶ electronvolts. This range is lower than the limits set by previous astronomical observations, such as those from the NuSTAR telescope, suggesting that if axions exist in this mass range, this new method could potentially probe a region not excluded by current constraints.

The core of the proposal involves a sequence of optical transformations that act like a filter for the axion signal. Light enters the system as a coherent beam, which is then amplified and sent into a ring cavity where it interacts with the hypothetical axion field. If axions are present, they cause the light's polarization to rotate slightly. The researchers then pass this light through a device called an optical parametric oscillator, which can squeeze the noise out of the light or amplify specific components. The critical innovation in this work is the placement of a static polarization bias just before the final stage of this process. Without this bias, the total number of photons detected would change in a way that depends on the square of the rotation angle, making the signal vanish for very small rotations. By applying the bias, the researchers shift the operating point so that the total number of photons changes linearly with the rotation. This linear relationship turns a quadratic, hard-to-detect effect into a first-order signal that is much easier to measure.

The team calculated the performance of this system using a detailed model of a four-mirror ring cavity, accounting for the finite coherence of the axion field and the specific way light travels through the device. They found that under ideal conditions, this setup could probe axion-photon couplings as low as 3.2 × 10⁻¹³ GeV⁻¹ in the specified mass range. This sensitivity is projected to be better than the current exclusion limits from NuSTAR, which have ruled out axions with stronger couplings in this mass region. The authors are careful to note that this is a noise-equivalent projection, meaning it represents the level of sensitivity the instrument would achieve if it were limited only by fundamental quantum noise and a modest allowance for technical imperfections. It is not a claim that axions have been found, nor is it a guarantee that the experiment will exclude all possibilities in this range, but rather a demonstration that the proposed architecture is theoretically capable of reaching into this unexplored territory.

In addition to the main polarimetric search, the paper explores a different way of reading the data: counting the total number of photons in both polarization directions rather than measuring their interference. This total-count method offers a secondary advantage in a higher-frequency band. The researchers found that using the optical parametric oscillator in this mode could reduce the noise-equivalent coupling by approximately 5% compared to a standard passive attenuation method. This improvement holds even when the system includes realistic imperfections, such as a 5% internal loss in the optical components and a small error in the pump phase. However, this benefit is modest and does not apply to the low-mass polarimetric search, where a different, optimized configuration remains more sensitive. The study identifies that the advantage comes from the active readout process itself, rather than the initial preparation of the light.

The researchers also took care to map out the stability requirements needed to make this experiment work in the real world. They determined that the system would need to maintain extremely low levels of noise in its polarization, pump phase, and light intensity to achieve the projected sensitivity. For instance, the noise in the pump phase must be kept below a specific threshold to prevent it from mimicking the axion signal. These requirements are stringent, but they provide a clear roadmap for future experimentalists. The work does not claim to have solved the dark matter problem or to have built a working device, but it establishes the operating principles and the stability targets for a future implementation. By identifying a prospective sensitivity window that extends below existing bounds and by detailing the specific technical challenges that must be overcome, the paper offers a concrete path forward for the search for axions in the low-mass regime.

Technical Summary: Axion Polarimetry with Parametric Readout

Problem Statement
The paper addresses the detection of ultralight axion dark matter via its electromagnetic coupling to photons. In the off-resonant regime, an axion field induces a small, oscillatory rotation in the linear polarization of light. While cavity-based experiments (e.g., DANCE, LIDA, ADBC) exploit this polarization conversion, the standard response of the summed output photon number to this rotation is quadratic (second-order) in the rotation angle. This quadratic dependence limits sensitivity, particularly for small rotation angles where the signal is weak. Furthermore, distinguishing the axion signal from technical noise and establishing sensitivity in the low-mass regime ( $10^{-18}$ – $10^{-16}$ eV) requires precise handling of cavity transfer functions, finite axion coherence, and noise budgets.

Methodology
The authors propose a bias-assisted active photon-number conversion scheme within a planar four-mirror optical ring cavity. The experimental sequence involves:

Input: A coherent $x$ -polarized seed and vacuum in the $y$ -polarization.
Amplification: A degenerate optical parametric oscillator (OPO) amplifies the carrier quadrature before the cavity.
Interaction: The axion field induces a passive, number-conserving rotation ( $\psi_a$ ) between the $x$ and $y$ polarizations.
Readout: An inverse OPO acts on the extracted $x$ polarization.
Bias and Detection:
- Polarimetric Readout: A downstream electro-optic modulator (EOM) introduces a reference modulation to create a linear interference term for the $y$ -polarization.
- Total-Count Readout (Novel): A calibrated, static polarization bias (implemented as a phase shift between circular polarizations or a $\pi/4$ rotation) is applied before the inverse OPO. This shifts the operating point away from the number-conversion null, converting the quadratic response of the summed photon number ( $N_x + N_y$ ) into a linear signal.

The authors derive the photon moments, extracted-field covariance, and noise spectra using a four-port optical transfer model that accounts for finite axion coherence (Maxwellian velocity distribution) and linewidth. They compare their scheme against passive selective attenuation and standard coherent polarimetry using the quantum Cramér–Rao bound and Fisher information analysis.

Key Contributions

Linearization of Total Count: The paper demonstrates that applying a calibrated polarization bias before the final inverse OPO transforms the summed photon number response from quadratic to linear with respect to the axion-induced rotation. This is described as "pump-assisted photon-number transduction" rather than photon creation by the axion interaction.
Four-Port Sensitivity Projection: The authors derive the sensitivity for a specific four-port geometry, retaining the full optical transfer and sideband structure. They calculate the noise-equivalent coupling ( $g_{a\gamma}$ ) under a stochastic-power criterion ( $A/\sigma_A = 1$ ).
Parametric Readout Advantage: A comparative analysis shows that in a favorable high-frequency band (above the bright-cavity linewidth), the OPO-based total-count readout reduces the noise-equivalent coupling by approximately 5% relative to a matched passive selective-attenuation comparator.
Robustness Analysis: The study quantifies the impact of internal optical loss and static pump-phase offsets on the OPO advantage, showing that the benefit persists even with moderate imperfections (e.g., 5% internal loss and 0.10 rad phase offset).

Results

Low-Mass Sensitivity Window: The projected sensitivity identifies a prospective window for axion masses between $10^{-18}$ and $10^{-16}$ eV. In this range, the projected noise-equivalent coupling ( $3.2 \times 10^{-13}$ to $6.0 \times 10^{-13}$ GeV $^{-1}$ ) lies below the published NuSTAR M82 exclusion limits (conservative and fiducial). The authors explicitly state this is a "noise-equivalent projection" and not a 95% confidence-level exclusion forecast.
Total-Count Performance: In the high-frequency regime, the two-OPO configuration with total counting achieves a coupling ratio of $\approx 0.95$ relative to the passive attenuator (a $\sim 5\%$ improvement). This advantage is attributed to the final active readout stage; a coherent source followed only by the final OPO retains this benefit, while the first OPO primarily prepares the probe covariance.
Baseline Comparison: In the low-mass polarimetric projection (the primary sensitivity curve), the modeled OPO configuration is slightly less sensitive than a matched passive comparator. The advantage of the active scheme is specific to the total-count observable in the high-frequency band.
Technical Requirements: To approach the theoretical optical limits, the system requires strict control of technical noise. Specifically, the pump-phase noise must be below $\sim 5.9 \times 10^{-9}$ rad/ $\sqrt{\text{Hz}}$ (at $10^{-11}$ eV) to maintain the OPO advantage, and the total technical noise power spectral density must be kept below $\sim 20\%$ of the shot noise.

Significance and Claims
The paper claims to establish the operating principle of active number conversion for axion detection and defines the stability requirements for future implementations.

Prospective Reach: The authors claim their scheme offers a potential path to probe a low-mass region of axion parameter space ( $10^{-18}$ – $10^{-16}$ eV) that is currently unexcluded by the specific constraints they compare against (notably NuSTAR M82). They emphasize that this is a conditional projection based on specific optical resources and noise assumptions, not a guaranteed exclusion limit.
Readout Innovation: The work highlights that the $\sim 5\%$ improvement in the high-frequency total-count channel is a distinct finding from the low-mass polarimetric sensitivity. It identifies the final active readout as the source of this improvement, distinguishing it from probe preparation effects.
Modesty: The authors are careful to note that their results do not constitute a confidence-level exclusion forecast. They do not claim to have solved the problem of technical noise but rather provide the theoretical framework and specific noise budgets (e.g., pump phase, intensity noise) that a future experiment must meet to realize the projected sensitivity. They also clarify that optimized coherent polarimetry remains more sensitive than the total-count method in the baseline low-mass scenario.

More like this