Polarization-shape alignment of IllustrisTNG star-forming galaxies

TL;DR

This study uses high-resolution IllustrisTNG50 simulations to quantify polarization–shape alignment in star-forming disk galaxies from redshift 0 to 2. By modeling synchrotron emission, sub-grid magnetic-field depolarization, and internal Faraday rotation, it derives a non-Gaussian distribution for the misalignment angle Δθ and computes the effective per-galaxy uncertainty σ_{α,eff} on a uniform cosmic birefringence rotation α via Fisher information. At 4.8 GHz, σ_{α,eff} ≈ 18°, 23°, and 33° for z = 0, 1, and 2 respectively, with higher source-frame frequencies mitigating depolarization and redshift evolution. The work shows SKA-scale surveys could yield competitive constraints on cosmic birefringence by leveraging the large samples of polarization-aligned disk galaxies, while also highlighting the need to account for AGN contamination and to pursue pilot observations to empirically calibrate polarization–shape alignment.

Abstract

In star-forming disk galaxies, the radio continuum emission ($1$-$10\,$GHz) powered by star formation has an integrated polarization direction imperfectly aligned with the apparent disk minor axis. This polarization-shape alignment effect was previously observed in a small sample of local spirals. If this is prevalent for disk galaxies out to cosmological redshifts, novel measurements of cosmic birefringence and cosmic shear will be enabled by leveraging radio continuum surveys synergized with galaxy shape measurements. We calculate the polarization-shape misalignment angle for star-forming galaxies in the \texttt{IllustrisTNG50} simulation at $0 < z < 2$, assuming that additional polarized radio emission from an AGN is negligible. The alignment found for $z=0$ is consistent with local spiral data, but significantly deteriorates as redshift increases. Moreover, it degrades toward lower frequencies due to internal Faraday depolarization. Thanks to cosmic redshifting, observing higher-$z$ galaxies at a fixed frequency greatly mitigates degradation due to reduced Faraday depolarization at the source-frame frequency. We present analytic fits to the non-Gaussian misalignment angle distribution, and evaluate Fisher information per galaxy for measuring cosmic birefringence. For observation at 4.8 GHz, the effective RMS misalignment angle $σ_{α,{\rm eff}}$ is $18^\circ$, $23^\circ$ and $33^\circ$ at $z=0$, $1$ and $2$, respectively. Analyzing $N$ independent galaxies reduces the uncertainty on an isotropic cosmic birefringence signal to $σ_{α,{\rm eff}}/\sqrt{N}$, providing competitive sensitivity once large samples are available. Our results motivate pilot observations to empirically characterize polarization-shape alignment, facilitate forecasts of cosmology and fundamental physics applications that exploit this effect.

Figures (10)

• Figure 1: Illustration of the polarization-shape alignment effect. We show the V-band stellar surface brightness profile of a simulated $z=0$ galaxy (IllustrisTNG-50 Subhalo ID: 19) with its inclined disk. The yellow dashed line marks the apparent semi-minor axis of the galaxy, which is determined from optical photometric moments. White segments represent synchrotron emission at $4.8\,$GHz near major sites of star formation. The length of the segment is proportional to the local intensity of the polarized emission $\sqrt{Q^2+U^2}$, and the orientation indicates the direction of polarization. For clarity, only 1 in 20 of the segments is shown. The white dashed line marks the polarization direction of synchrotron emission integrated over the entire galaxy. This galaxy shows a misalignment angle $\Delta\theta=20.96^\circ$ between polarization and shape.
• Figure 2: Distribution of inclination $i$ inferred from photometric moments (colored histograms) compared to the theoretical isotropic distribution (black histogram). Inclinations are estimated from the axial ratio $b/a$ (see Appendix \ref{['app:galaxy_shape']}), assuming all galaxies are thin disks without bulges. The discrepancy with the isotropic distribution at very high $i$ may arise from neglecting the bulge in our thin-disk modeling of galaxy shapes.
• Figure 3: Modification of polarization fraction due to unresolved random magnetic field components. Here $r$ is the ratio between the random component and the regular component of the magnetic field. Requiring $\langle B_{\rm reg}/B_{\rm tot} \rangle$ to be $0.6$--$0.7$ from Milky-Way ISM observations Beck2004role, the ratio $r=1.35$--$1.44$. This leads to a constant reduction factor $\Pi_{p}(r)/\Pi_{p}(0)$ ranging from $0.327$ to $0.353$ for polarization fraction from all simulation cells, where $\Pi_{p}(r)$ is the polarization fraction in the presence of a random magnetic field component, and $r \equiv B_{\rm rand} / B_{\rm reg}$ is the ratio of the random to regular magnetic field strengths.
• Figure 4: Radio galaxy luminosity function (RLF) at $1.4\,$GHz for each of the six Illustris redshift snapshots we analyze. In each panel, the dash-dotted histograms show our selected IllustrisTNG50 star-forming galaxies with a halo mass $M_{\rm halo}=10^{11}$--$10^{13}\,M_\odot$ and star formation rate $0.2$--$10\,M_\odot\,{\rm yr}^{-1}$, while the solid histogram includes all IllustrisTNG50 galaxies. The luminosities include only star formation contribution for the black (solid and dash-dotted) histograms. All luminosities are estimated using the analytic recipes of Hansen2024SHARKmodel. For comparison, the blue curve shows the star-formation RLF taken from Hansen2024SHARKmodel, and should be compared to the solid black histogram. The red curve shows the AGN RLF from the same work. These results are available and shown for $z=0$, $1$ and $2$. The blue dotted curve is an analytic fit to the RLF, consisting of a power-law at low luminosities and a Schechter function at high luminosities. The grey shaded region indicates the luminosity range where AGN-dominated radio galaxies are expected to outnumber those dominated by star formation Hansen2024SHARKmodel. The majority of the detectable radio galaxies at $z<1.5$ will be star-formation dominated.
• Figure 5: Computed polarization fraction $\Pi_{p}$ for our selected IllustrisTNG50 galaxies at $z=0$, at three different frequencies and with internal Faraday depolarization (FE) accounted for. The distributions show fair agreement with independent model predictions from stil2009integrated (red histograms). Internal Faraday depolarization significantly reduces the polarization fraction at $1.4\,$GHz and $4.8\,$GHz, but is unimportant at $8.4\,$GHz.