# Constraints on direction-dependent cosmic birefringence from Planck polarization data

## TL;DR

This study uses Planck polarization data to set new constraints on anisotropic cosmic birefringence, finding no significant detection but identifying a potential dipolar excess possibly due to foreground contamination.

## Contribution

It introduces a novel pixel-based method for analyzing large-scale anisotropies in cosmic birefringence using Planck data, improving constraints on dipole and quadrupole amplitudes.

## Key findings

- No significant detection of anisotropic birefringence.
- Set upper limits on the amplitude of scale-invariant power spectrum.
- Detected an excess dipolar power possibly due to foreground contamination.

## Abstract

Cosmic birefringence is the process that rotates the plane of polarization by an amount, $\alpha$, as photons propagate through free space. Such an effect arises in parity-violating extensions to the electromagnetic sector, such as the Chern-Simons term common in axion models, quintessence models, or Lorentz-violating extensions to the standard model. Most studies consider the monopole of this rotation, but it is also possible for the effect to have spatial anisotropies. Paying particular attention to large scales, we implement a novel pixel-based method to extract the spherical harmonics for $L \le 30$ and a pseudo-$C_L$ method for $L > 30$. Our results are consistent with no detection and we set 95% upper limits on the amplitude of a scale-invariant power spectrum of $L(L+1)C_L/2\pi < [2.2\, (\mathrm{stat.})\, \pm 0.7\, (\mathrm{syst.})]\times10^{-5} = [0.07\, (\mathrm{stat.}) \pm 0.02\, (\mathrm{syst.})] \,\mathrm{deg}^2$, on par with previous constraints. This implies specific limits on the dipole and quadrupole amplitudes to be $\sqrt{C_1/4\pi} < 0.2^\circ$ and $\sqrt{C_2/4\pi} < 0.1^\circ$, at 95% CL, respectively, improving previous constraints by an order of magnitude. We further constrain a model independent $M=0$ quadrupole in an arbitrary direction to be $\alpha_{20} = 0.02^\circ \pm 0.21^\circ$, with an unconstrained direction. However, we find an excess of dipolar power with an amplitude $\sqrt{3C_1/4\pi} = 0.32^\circ \pm 0.10^\circ\, (\mathrm{stat.})\, \pm 0.08^\circ\, (\mathrm{syst.})$, in the direction $(l, b) = (295^\circ, 17^\circ) \pm (22^\circ, 17^\circ)\, (\mathrm{stat.})\, \pm (5^\circ, 16^\circ)\, (\mathrm{syst.})$, larger than 1.4% of simulations with no birefringence. We attribute part of this signal to the contamination of residual foregrounds not accounted for in our simulations, although this should be further investigated.