Dark Photons in the Radio Sky: II. Resonant Conversions in the Intergalactic Medium

TL;DR

This paper develops a comprehensive framework to forecast SKA’s sensitivity to resonant gamma-to-A' conversions in the intergalactic medium, across three environments (halos, EoR IGM, and late-time IGM). It constructs a realistic mock-observation pipeline, incorporating foregrounds and a needlet ILC, and analyzes auto- and cross-correlations with galaxy surveys to extract the signal. The study finds that SKA, especially when cross-correlated with low-redshift galaxies, could probe $m_{A'}$ in the $10^{-14}$–$10^{-14}$ eV range with $\epsilon$ down to $10^{-7}$–$10^{-8}$, surpassing some Planck limits, and that 21-cm global signals may yield competitive sensitivity post-reionization. Power-spectrum searches in 21-cm interferometers face significant challenges due to foreground wedges, making global-signal and cross-correlation approaches particularly promising for constraining ultralight dark photons. The work highlights the strong potential of upcoming radio experiments to explore BSM physics beyond the reach of traditional laboratory tests.

Abstract

This is the second part in a pair of papers forecasting the sensitivity of the Square Kilometre Array (SKA) to dark photons, a highly motivated, simple extension of the Standard Model. Through a kinetic mixing term, visible photons from the cosmic microwave background can resonantly convert into dark photons, generating new temperature anisotropies in the sky. In this work, we detail the entire analysis pipeline that we use to compute SKA's sensitivity, focusing on resonant conversions that occur in the intergalactic medium. We also discuss the sensitivity of 21-cm experiments to dark photons. Our results show that both SKA in combination with galaxy surveys and 21-cm experiments could discover dark photons with masses between $5\times 10^{-15}$ and $5\times 10^{-12}$ eV, and kinetic mixing parameter $ε$ as low as $10^{-8}$.

Figures (11)

• Figure 1: (Left): The auto-power spectra of the map obtained using our ILC procedure $C_\ell^{\rm TT, obs}$ (red), compared to the environment A dark photon signal $C_\ell^{\rm A, TT}$, shown here for $m_{A'} = \qty{5.6e-13}{\eV}$ with $\epsilon = 9\times10^{-8}$ (black, solid) and $\epsilon = 2 \times 10^{-7}$ (black, dashed), with $\omega_0 / 2 \pi = \qty{410}{\mega\hertz}$. The post-ILC power spectra here are normalized to a signal-only map at $\omega_0 / 2 \pi = \qty{410}{\mega\hertz}$. (Right): The cross-power spectrum of the ILC map with the mock low-redshift galaxy survey $C_\ell^{\rm Tg, obs}$ (red). The predicted signal in environment A is shown for a dark photon with $m_{A'}=\qty{5.6e-13}{\eV}$ and $\omega_0/2\pi = \qty{410}{\mega\hertz}$ for $\epsilon = 10^{-8}$ (black, solid), and $\epsilon=5\times 10^{-8}$ (black, dashed). In each case, the error bars are the square root of the diagonal entries of the covariance matrix. https://github.com/bakerem/dark_photons_radio_sky/tree/main/notebooks_for_paper/Cl_Comparison.ipynb
• Figure 2: A 2D cross-section of the angular lightcone of $m_{\gamma}^2$ that we use in this work, as a function of $z$ and one sky angular coordinate. Lines of sight are drawn horizontally through cells in the lightcone. In each cell, we compute $m_{\gamma, i}^2$ and compare it to $m_{A'}^2$ to determine the probability of conversion in that cell. https://github.com/bakerem/dark_photons_radio_sky/tree/main/notebooks_for_paper/21cm_Lightcone.ipynb
• Figure 3: Two-point angular correlation functions from our mock observations, computed in 20 logarithmically spaced angle bins. The error bars shown are the square root of the diagonal elements of the covariance matrix, obtained from our mock observations. These are to be compared to the signal from conversions in environment B. (Left): $\xi^{\rm TT, obs}$ (red) computed from the post-ILC map, compared to the signal $\xi^{\rm B, TT}$, which is computed from the 21cmFAST simulation for $m_{A'}=\qty{3e-14}{\eV}$ at $\omega_0/2\pi = \qty{410}{\mega\hertz}$ for $\epsilon = 8 \times 10^{-7}$ (black, solid) and $\epsilon = 10^{-6}$ (black, dashed). (Right): $\xi^{\rm Tg, obs}$ (red) computed from the cross-correlation of the post-ILC map and the mock Roman galaxy catalog. The signal $\xi^{\rm B, Tg}$ is shown for $m_{A'} = \qty{3e-14}{\eV}$, with $\epsilon = 5 \times 10^{-7}$ (black, solid) and $\epsilon = 8 \times 10^{-7}$ (black, dashed). https://github.com/bakerem/dark_photons_radio_sky/tree/main/notebooks_for_paper/21cmFAST_corr_funcs.ipynb
• Figure 4: $T_{\gamma,0} P_{\gamma\to A'}$ is shown for $m_{A'}=\qty{3e-14}{\eV}$, $\epsilon=10^{-6}$, and $\omega_0/2\pi =\qty{410}{\mega\hertz}$. Overlaid in black points is a random subset of 10,000 galaxies from the mock Roman galaxy survey. $\gamma \to A'$ conversions for dark photons of this mass preferentially occur in regions of the sky with more galaxies. https://github.com/bakerem/dark_photons_radio_sky/tree/main/notebooks_for_paper/Conversions_Galaxy_Map.ipynb
• Figure 5: Two point angular correlation functions from our mock observations, computed in 20 logarithmically spaced angle bins. The error bars are the square root of the diagonal elements of the covariance matrix, obtained from our mock observation. In black, the theoretical prediction $\xi^{\rm C, TT}$ is plotted for $m_{A'}=\qty{3e-14}{\eV}$ at $\omega_0/2\pi = \qty{410}{\mega\hertz}$ for $\epsilon=5\times 10^{-7}$ (black, solid) and $\epsilon=8\times 10^{-7}$ (black, dashed). This correlation function is compared to the $\xi^{\rm TT, obs}$ from the post-ILC map (red). $\xi^{\rm C, TT}$ is greatest at small angular scales. https://github.com/bakerem/dark_photons_radio_sky/tree/main/notebooks_for_paper/Analytic_corr_funcs.ipynb