Mass dependence of halo baryon fractions from the kinetic Sunyaev-Zeldovich effect

TL;DR

This work detects the kinetic SZ imprint from DESI Legacy Survey galaxy groups using ACT CMB maps and forwards-models based on AbacusSummit simulations to constrain halo gas content. By reconstructing a velocity field and modeling halo gas with a two-parameter Gaussian profile ($f_\mathrm{gas}$, $\alpha$), the authors quantify how the gas fraction within the virial radius depends on halo mass and reveal depletion in low-mass halos, possibly due to energetic feedback. A momentum-weighted stacking approach reconciles a near-universal baryon fraction in massive halos with extended gas distributions in smaller ones, implying most baryons in low-mass halos lie outside the virial radius. The results are consistent with X-ray, lensing, and hydrodynamical simulations, highlighting the importance of probing gas at large radii to fully recover the universal baryon content and informing models of galaxy formation and feedback.

Abstract

We detect the kinetic Sunyaev-Zeldovich imprint of peculiar motions of galaxy groups and clusters, using the photometric DESI Legacy Survey together with cosmic microwave background (CMB) maps from the Atacama Cosmology Telescope (ACT). We develop a comprehensive forward model based on the AbacusSummit cosmological simulations: mock galaxy group catalogues and synthetic kSZ maps are generated, together with a reconstructed peculiar velocity field that allows for photo-$z$ errors, redshift-space distortions, and survey masks. We investigate possible contamination from the cosmic infrared background (CIB), finding that CIB effects are subdominant to the kSZ signal in the relevant ACT frequency channel. We then predict the kSZ signal expected when stacking CMB temperature maps around groups, taking account of their estimated radial velocity. Comparing the model with observations, we are able to constrain the total baryon fraction within haloes, as well as their internal gas profiles. We find evidence for mass dependence of the halo baryon fraction within the virial radius. The gas fraction in massive groups is consistent with the universal baryon fraction, but low-mass groups ($10^{12.5} \lesssim M\,/h^{-1}\mathrm{M}_\odot \lesssim 10^{14}$) are depleted to $0.21 \pm 0.06$ times the universal baryon fraction. We find this low virial baryon fraction to be consistent with an extended gas profile, for which the total baryon content reaches the universal value well beyond the virial radius. This conclusion is consistent with previous analyses using X-ray, kSZ, and weak lensing, and plausibly reflects energetic feedback processes from the galaxies in these haloes.

Figures (7)

• Figure 1: Two-dimensional histogram showing groups in the DESI Legacy catalogue in redshift and mass. The full catalogue is shown in greyscale and the catalogue after redshift and mass cuts is shown in colour. Both grey and colour scales are logarithmic. Minor panels show marginalised redshift (top) and mass (right) distributions. Grey and colour show full and cut DESI-LS catalogues. Note that the cuts in redshift have almost no effect on the marginalised mass distribution.
• Figure 2: Reconstructed grid LoS velocities against true grid LoS velocities for the AbacusSummit halo catalogue, with RSDs and photo-$z$ errors applied to the mock data. Blue shading shows density of haloes in this space, with the contours containing 68 and 95 per cent of data. Red points and error bars show mean and standard deviation of a Gaussian fitting in bins, with the red line showing a least-squares linear fit. The black line shows $v_{\parallel,\mathrm{recon}}=v_{\parallel,\mathrm{true}}$. The halo catalogue is treated identically to the DESI group catalogue, with the same sky masking and redshift distribution. The velocities are reconstructed as described in Sec. \ref{['sec:vel_rec']}, and identical grid resolution and smoothing are applied to the true and reconstructed velocities.
• Figure 3: A comparison of CMB temperature fluctuations, $\delta T$, over the same region of kSZ templates with different radial scaling parameters, $\alpha$. These maps are convolved with an isotropic Gaussian with FWHM of 1.0 arcmin, in order to match the beam of AdvACT. The mock kSZ templates are generated from AbacusSummit halo light-cone as described in Sec. \ref{['sec:kSZ_template']}. Lower values of $\alpha$ lead to sharply peaked halo profiles with smaller angular extent compared to the extended and shallower profiles produced for higher $\alpha$. But effective point sources of this kind are smeared identically by the beam, so that the predicted profile is less sensitive to any change in $\alpha$. These effects, expected to occur in real observations, are reproduced in our kSZ template. Note that the colour scale is not shared between panels for visual clarity. The full templates cover around 40 per cent of the sky over $0.1<z<0.9$, equivalent to the DESI-LS catalogue (see Sec. \ref{['sec:kSZ_template']} for details).
• Figure 4: Stacked simulated and observed temperature fluctuations, $\delta T$, showing absolute LoS momentum-weighted averages of regions around haloes. Panels (a--c) show stacks of a mock kSZ template (without superimposed intrinsic CMB fluctuations), generated using a halo light-cone catalogue from the AbacusSummit simulations (with $\alpha=0.6$; see Sec. \ref{['sec:kSZ_template']}). Panels (d--f) show stacks of a filtered $220\,\mathrm{GHz}$ ACT CMB temperature map over the positions of DESI-LS galaxy groups, smoothed with a $0.5\,\mathrm{arcmin}$ FWHM Gaussian kernel to remove visual noise. Panels (a) & (d), (b) & (e) show stacks over only the receding and approaching haloes, respectively. Observational stacks are both positive due to possible CIB emission of groups. Panels (c) & (f) show the receding stack minus the approaching stack, thereby cancelling out all temperature fluctuations that are not halo velocity-dependent. The dominant remaining effect being kSZ. Panel (g) & (h) show radial means of difference stacks with absolute LoS momentum weighting and equal weighting, respectively. Note that the radius extends further than shown in the left panels. The black curve in panel (g) corresponds to observation [i.e. panel (f) with no smoothing] with shading showing one standard error from the diagonal components of the bootstrapping covariance; the grey vertical line denotes the absolute LoS momentum-weighted average angular halo virial radius, $\sigma_\mathrm{R}$; the blue dashed curve shows predictions from the mock kSZ template assuming gas fractions equal to the universal baryon fraction found by PlanckCollaboration_18_VI; and the red solid curve shows best-fit profile from stacking of kSZ templates with $f_\mathrm{gas} = (1.2 \pm 0.4) (\Omega_\mathrm{b}/\Omega_\mathrm{m})$ and $\alpha = 0.59 \pm 0.12$. Panel (h) is similar to panel (g), but with no weighting. With each group given equal weight, we instead obtain $f_\mathrm{gas} = (0.2 \pm 0.2) (\Omega_\mathrm{b}/\Omega_\mathrm{m})$ and $\alpha = 0.4 \pm 0.2$. See Sec. \ref{['sec:fgas']} for details.
• Figure 5: Stacked temperature profiles of the mock kSZ templates, with universal baryon content and varying radial scaling parameter, $\alpha$, where $\sigma_\mathrm{R}=\alpha R / D_\mathrm{A}$. Note that the beam angle, $\sigma_\mathrm{B}$ is included in the total Gaussian radius $\Sigma = (\sigma_\mathrm{R}^2 + \sigma_\mathrm{B}^2)^{1/2}$ (see Sec. \ref{['ssec:tau_profiles']} for details). The total spatially integrated optical depth is unchanged by $\alpha$, meaning $\int \delta T \, \theta \, \mathrm{d}\theta$ is conserved, causing sharp slopes and strong temperature decrements for low $\alpha$.