The Atacama Cosmology Telescope: Probing the baryon content of SDSS DR15 galaxies with the thermal and kinematic Sunyaev-Zel’dovich effects

We present measurements of the average thermal Sunyaev Zel’dovich (tSZ) effect from optically selected galaxy groups and clusters at high signal-to-noise (up to $12\sigma$) and estimate their baryon content within a ${2.1}^{\prime }$ radius aperture. Sources from the Sloan Digital Sky Survey Baryon Oscillation Spectroscopic Survey DR15 catalog overlap with 3,700 sq deg of sky observed by the Atacama Cosmology Telescope (ACT) from 2008 to 2018 at 150 and 98 GHz (ACT DR5), and 2,089 sq deg of internal linear combination component-separated maps combining ACT and Planck data (ACT DR4). The corresponding optical depths $\overline{\tau }$, which depend on the baryon content of the halos, are estimated using results from cosmological hydrodynamic simulations assuming an active galactic nuclei feedback radiative cooling model. We estimate the mean mass of the halos in multiple luminosity bins, and compare the tSZ-based $\overline{\tau }$ estimates to theoretical predictions of the baryon content for a Navarro-Frenk-White profile. We do the same for $\overline{\tau }$ estimates extracted from fits to pairwise baryon momentum measurements of the kinematic Sunyaev-Zel’dovich effect (kSZ) for the same dataset obtained in a companion paper. We find that the $\overline{\tau }$ estimates from the tSZ measurements in this work and the kSZ measurements in the companion paper agree within $1\sigma$ for two out of the three disjoint luminosity bins studied, while they differ by $2–3\sigma$ in the highest luminosity bin. The optical depth estimates account for one-third to all of the theoretically predicted baryon content in the halos across luminosity bins. Potential systematic uncertainties are discussed. The tSZ and kSZ measurements provide a step toward empirical Compton-$\overline{y}-\overline{\tau }$ relationships to provide new tests of cluster formation and evolution models.

Figure 1

Top: The $\mathrm{ACT}+\mathit{Planck}$ map used for the DR5 f150 analysis with the overlapping 343,647 SDSS DR15 selected sources plotted in blue over 3,700 sq deg, and the BN and D56 areas covered by the ILC maps plotted in green and orange, respectively. Bottom: The inverse white noise variance map associated with the DR5 f150 coadded $\mathrm{ACT}+\mathit{Planck}$ map highlighting regions representing a noise equivalence of 45 and $65\mu \mathrm{K}$ per pixel (with a 0.5 arcmin resolution plate Carré projection), which were used to cut the SDSS sample for the DR5 f150 analysis. The orange and yellow regions of higher noise overlapped with 27% of the DR15 sample. Results are shown for the more conservative $45\mu \mathrm{K}$ per pixel inverse white noise variance map cut shown in purple. We performed an equivalent cut for the DR5 f090 map and analysis.

Figure 2

The SDSS DR15 redshift distribution for the 602,461 total galaxy sample and selected 343,647 galaxy sample for analysis overlapping with the $\mathrm{ACT}+\mathit{Planck}$ DR5 map, as compared to the $\sim 9$ and $\sim 7$ times fewer DR11 galaxies overlapping with the ACT DR3 area and those used for the 2017 result (DB17), respectively.

Figure 3

Stacked raw submaps for the five cumulative (top) and four disjoint (bottom) luminosity bins as defined in Table 1, for the DR5 f150 map (top two rows, in differential units of CMB temperature which are calibrated to Planck [28]), the DR5 f090 map (middle two rows, in the same units as the top two rows), and the DR4 ILC Compton-$y$ maps (bottom two rows, in negative units of $y$ to better compare to the coadded maps). The submaps represent the weighted average submaps of the sources in a given bin, where the weight for each source is taken to be the average value inside the accompanying $R_1=2.1$ submap in the inverse white noise variance map. The sub-${0.5}^{\prime }$-scale structure in the submaps is an artifact of the subpixel interpolation and is not physical. The maps are normalized with the average value within the AP annulus, such that the mean of the pixels in the annulus in these maps is equal to zero. The apertures used for the tSZ and kSZ AP are drawn, where $R_1={2.1}^{\prime }$ (red) and $\sqrt{2}{R}_1$ (black). Radial averages of these submaps are plotted in Fig. 4. A central bright spot due to dust on approximately the beam scale can be seen across luminosity bins in the DR5 f150 submaps, but not the DR5 f090 submaps. Dust contamination of the DR4 ILC maps is more subtle, but can be noticed in radial average plots (Fig. 4).

Figure 4

Radial average of the stacked submaps, which have been repixelized to ${0.1}^{\prime }$ per pixel, normalized to the average annulus value, for each luminosity bin, for the DR5 f150 and DR5 f090 coadded maps as well as the DR4 ILC map shown for illustrative purposes only. The native units of the DR5 f150 and DR5 f090 maps are in $\mu \mathrm{K}$ and the DR4 ILC maps in $y$. Negative $y$ is plotted here to compare with the decrements present in temperature. The aperture photometry disk radius is plotted as a vertical black dashed line, and the annulus outer radius is plotted as a vertical solid black line. The beam radius is plotted as a blue vertical line for DR5 f150, an orange vertical line for DR5 f090, and a black vertical line for the effective DR4 ILC beam. A central bright spot is observed in nearly all but the most luminous bin, and is attributed to dust emission. Due to this effect, we studied the core-excised AP approach for the DR5 f150 and DR5 f090 analysis (Secs. 3b and 3c).

Figure 5

Aperture photometry tSZ signals in units of temperature rescaled by $f_{\mathrm{SZ}}$ [Eq. (2)] for disjoint luminosity bins and the full analysis sample, with $1\sigma$ jackknife estimated uncertainties, for the DR5 f150 and DR5 f090 maps. Raw results using the ${2.1}^{\prime }$ radius are compared to results after Herschel dust correction (Sec. 3c) and the removal of the pixels within a beam-scale radius (${0.7}^{\prime }$ for the DR5 f150 map and ${1.1}^{\prime }$ for the DR5 f090 map; Sec. 3b). The DR5 f090 results shown have been beam corrected (Sec. 3b). The Herschel dust correction has a small effect on the temperature signals and lowers the signal-to-noise for the DR5 f090 map due to the relatively large error bars on the dust estimates which are propagated into the uncertainties on the tSZ signal. The “core-excised” AP approach removes a noticeable amount of tSZ signal in the L116 bin, as there is less dust contamination apparent on the beam scale in the stacked submaps for this bin as compared to the lower luminosity bins (Fig. 4). For the most part, it has less than a $1\sigma$ effect on the tSZ signals, and is not included in the rest of this analysis. The dust-corrected (filled triangle) results are propagated through to our comparisons of optical depth estimates.

Figure 6

Average Compton-$y$ in the ${2.1}^{\prime }$ aperture for the five disjoint (highest luminosity bin L116 and bin labels with suffix “D") luminosity bins and the full source sample ($L>4.3\times {10}^{10}{L}_{\odot }$), with jackknife estimated uncertainties, for the DR5 f150 and DR5 f090 maps after dust and beam correction (Secs. 3b, 3c, and Table 2) and the DR4 ILC Compton-$y$ map after beam correction. The lower panel shows the signals divided by their associated uncertainties. The significance of the tSZ effect observed generally decreases for less luminous sources, as expected. The results from each of the three maps are consistent.

Figure 7

Fraction of the theoretically predicted optical depth ($f_{\mathrm{c}}$) for the full DR15 sample. The estimates are extracted from tSZ measurements (filled circles) from three different maps (DR5 f150, blue; DR5 f090, orange; ILC, black) and kSZ measurements from the same maps, as described in C21 (open circles). The tSZ measurements are converted to optical depth estimates using a scaling relationship from hydrodynamic simulations [3]. The tSZ jackknife uncertainties are plotted in color, and the systematic uncertainties from the simulation-based scaling relationship are plotted as gray bars. The plotted kSZ uncertainties are from bootstrap estimates. The kSZ and tSZ results agree within $1\sigma$ in the L61D and L43D bins, while in the highest signal-to-noise L79 bin they differ at $2-3\sigma$. This results in the $2-3\sigma$ difference observed in the full sample, L43. The difference between the kSZ and tSZ results is discussed in Sec. 4c. The kSZ results are lowest for the L79 bin.

Figure 8

Optical depth fits from kSZ signals versus average Compton-$y$ from the tSZ measurements for the three jointly analyzed disjoint luminosity bins with statistical error bars from jackknife estimates (bootstrap estimated uncertainties for the kSZ; see C21). The scaling relation between $\overline{\tau }$ and $\overline{y}$ from hydrodynamical simulations with aperture $\mathrm{\Theta }={1.8}^{\prime }$ (the closest scaling relation in [3] to our ${2.1}^{\prime }$ aperture) is plotted as the green model curve [Eq. (3)] with the $1\sigma$ uncertainty envelope shaded. The tSZ and kSZ results in the L43D and L61D luminosity bins are consistent with the model, while the kSZ results in the L79 bin fall below the model line.

Figure 9

Luminosity bin cuts for the SDSS DR15 catalog plotted over a histogram of the full sample (green), DR5 f150 and DR5 f090 selected analysis sample (blue), and DR4 ILC sample (black). The bottom three bins were selected to each have over 100,000 galaxies for the joint tSZ and kSZ analyses of the DR5 maps, while being roughly evenly spaced and overlapping with bin selection from the DB17 analysis. The top two bins were added for the tSZ analysis to study higher mass bins that have a strong tSZ signal, while also overlapping with DB17 bins.