Planck Constraints on Axion-Like Particles through Isotropic Cosmic Birefringence

TL;DR

This paper investigates isotropic cosmic birefringence induced by axion-like particles (ALPs) using Planck HFI polarization data. By solving the full Boltzmann equations for a time-dependent rotation β(η) and exploiting the spectral shape of the $EB$ power spectrum, the authors jointly constrain the ALP mass, initial rotation, instrumental miscalibration, and dust foregrounds within a multi-frequency likelihood. They find that certain ALP masses are excluded at more than $2\sigma$, while very light masses remain consistent with a near-constant rotation, suggesting ALPs could play a role in dynamical dark energy in the ultra-light regime. The study demonstrates the viability of using the $EB$ spectrum shape to distinguish ALP-induced birefringence from calibration systematics and foregrounds, and it outlines paths for improved constraints with future high-resolution CMB and low-redshift polarization data.

Abstract

We present constraints on isotropic cosmic birefringence induced by axion-like particles (ALPs), derived from the analysis of cosmic microwave background (CMB) polarization measurements obtained with the high-frequency channels of Planck. Recent measurements report a hint of isotropic cosmic birefringence, though its origin remains uncertain. The detailed dynamics of ALPs can leave characteristic imprints on the shape of the $EB$ angular power spectrum, which can be exploited to constrain specific models of cosmic birefringence. We first construct a multi-frequency likelihood that incorporates an intrinsic nonzero $EB$ power spectrum. We also show that the likelihood used in previous studies can be further simplified without loss of generality. Using this framework, we simultaneously constrain the ALP model parameters, the instrumental miscalibration angle, and the amplitudes of the $EB$ power spectrum of a Galactic dust foreground model. We find that, if ALPs are responsible for the observed cosmic birefringence, ALP masses at $\log_{10}m_φ[{\rm eV}]\simeq-27.8$, $-27.5$, $-27.3$, $-27.2$, $-27.1$, as well as $\log_{10}m_φ[{\rm eV}]\in[-27.0,-26.5]$, are excluded at more than $2\,σ$ statistical significance.

Figures (6)

• Figure 1: Marginalized posterior distribution of the logarithmic ALP mass $\mu_\phi=\log_{10}m_\phi [\mathrm{eV}]$ and the rescaled amplitude parameter $A_{\rm EB}$, which characterizes the overall strength of the birefringence-induced $EB$ power spectrum. The two-dimensional panel shows the distribution of MCMC samples along with the $2\,\sigma$ contour.
• Figure 2: Same as Fig. \ref{['fig:const:2params']}, but shown in the $\log_{10}m_\phi[\mathrm{eV}]$–$\ln|g\phi_{\rm ini}/2|$ plane, where $g\phi_{\rm ini}/2$ is expressed in degrees. The two-dimensional posterior is visualized with $1\,\sigma$ (cyan) and $2\,\sigma$ (blue) contours.
• Figure 3: Comparison of theoretical $EB$ power spectra for different ALP masses and amplitudes: $\mu_\phi=-27.822$, $A_{\rm EB}=1$ (blue solid), $\mu_\phi=-26.846$, $A_{\rm EB}=1$ (orange solid), and $\mu_\phi=-33$, $A_{\rm EB}=0.36/0.3$ (green solid). For reference, the spectrum from a constant rotation angle $\beta=0.36\,$deg is shown as a black dashed line. The black points represent the stacked, foreground-subtracted $EB$ power spectrum derived from the data using the best-fit foreground model.
• Figure 4: Same as Fig. \ref{['fig:const:2params']}, but without modeling the intrinsic dust-induced $EB$ foreground correlation. This comparison illustrates the impact of foreground modeling on the inferred ALP parameters.
• Figure 5: Constraints on the ALP mass, the overall rescaled amplitude of the $EB$ power spectrum, miscalibration angles, $\alpha_i$, and dust $EB$ amplitude, $A^{\rm dust}_b$.