A measurement of secondary cosmic microwave background anisotropies with two years of South Pole Telescope observations

Abstract

We present the first three-frequency South Pole Telescope (SPT) cosmic microwave background (CMB) power spectra. The band powers presented here cover angular scales 2000 < ell < 9400 in frequency bands centered at 95, 150, and 220 GHz. At these frequencies and angular scales, a combination of the primary CMB anisotropy, thermal and kinetic Sunyaev-Zel'dovich (SZ) effects, radio galaxies, and cosmic infrared background (CIB) contributes to the signal. We combine Planck and SPT data at 220 GHz to constrain the amplitude and shape of the CIB power spectrum and find strong evidence for non-linear clustering. We explore the SZ results using a variety of cosmological models for the CMB and CIB anisotropies and find them to be robust with one exception: allowing for spatial correlations between the thermal SZ effect and CIB significantly degrades the SZ constraints. Neglecting this potential correlation, we find the thermal SZ power at 150 GHz and ell = 3000 to be 3.65 +/- 0.69 muK^2, and set an upper limit on the kinetic SZ power to be less than 2.8 muK^2 at 95% confidence. When a correlation between the thermal SZ and CIB is allowed, we constrain a linear combination of thermal and kinetic SZ power: D_{3000}^{tSZ} + 0.5 D_{3000}^{kSZ} = 4.60 +/- 0.63 muK^2, consistent with earlier measurements. We use the measured thermal SZ power and an analytic, thermal SZ model calibrated with simulations to determine sigma8 = 0.807 +/- 0.016. Modeling uncertainties involving the astrophysics of the intracluster medium rather than the statistical uncertainty in the measured band powers are the dominant source of uncertainty on sigma8 . We also place an upper limit on the kinetic SZ power produced by patchy reionization; a companion paper uses these limits to constrain the reionization history of the Universe.

Figures (14)

• Figure 1: The results of the jackknife tests applied to the SPT data. Red squares denote the 95 GHz bandpowers, black circles mark the 150 GHz bandpowers, and the 220 GHz bandpowers are plotted with blue diamonds. Note that the 150 GHz uncertainties are smaller than the symbols. The top panel displays the results of the "first half - second half" jackknife. The PTE to exceed the measured $\chi^2$ is 0.90. The "left - right" jackknife is shown in the second panel and has a PTE of 0.21. The "azimuth-split" jackknife is plotted in the third panel and has a PTE of 0.84. Finally, the auto-spectra of the three frequencies are plotted in the bottom panel for comparison.
• Figure 2: The WMAP7 and SPT bandpowers. The SPT bandpowers at $\ell \le 2000$ are taken from K11 and are at 150 GHz only. At $\ell \ge 2000$, we show the bandpowers at 95, 150, and 220 GHz measured with the SPT in this work. Below $\ell = 2000$, the primary CMB anisotropy is dominant at all frequencies. On smaller scales, the CIB, radio sources, and secondary CMB anisotropies contribute to the signal. With the SPT source masking, the CIB is the largest source of power on sub-arcminute scales at 150 and 220 GHz. Due to the relative spectral behavior of the CIB and synchrotron emission, the 95 GHz bandpowers also have a significant contribution from radio sources.
• Figure 3: The six auto- and cross-spectra measured with the 3-frequency SPT data. Overplotted on the bandpowers is the best-fit model for the fiducial set of model parameters. The bandpowers have not been corrected by the best-fit calibration or beam uncertainties in the MCMC chains; for reference, the best-fit temperature calibration factors at 95, 150, and 220 GHz are 0.999, 0.997, and 1.003 respectively. In addition to the complete model ( black lines), each individual model component is shown. The tSZ effect is marked with the blue solid line. The best-fit kSZ power is near-zero and off-scale. The Poisson power from DSFGs and radio galaxies are shown by solid orange and green lines respectively. The clustered component to the DSFGs is shown with a orange dot-dash line.
• Figure 4: 2D likelihood surface for the tSZ and kSZ power in the SPT 150 GHz band at $\ell=3000$. 1, 2, and 3-$\sigma$ constraints are shown in red, yellow, and blue respectively. The left panel shows the constraints for the modified BB CIB frequency scaling. The right panel displays the likelihood surfaces for the modified BB CIB frequency scaling with the introduction of free parameter, $\xi$, describing the correlations between the tSZ and CIB power spectra. The tSZ-CIB correlation is degenerate with the kSZ effect, thereby weakening the kSZ limits.
• Figure 5: 2D Likelihood curves for the correlation between the tSZ and CIB spectra versus the kSZ power (Left panel) or tSZ power (Right panel) in $\mu {\rm K}^2$. The filled contours show the 1, 2, and $3\,\sigma$ constraints. The kSZ template consists of the sum of CSF homogeneous kSZ model and patchy kSZ contribution for a model where reionization begins at $z = 11$ and ends at $z=8$. Increasing anti-correlation increases the allowed kSZ power and reduces the tSZ power. The data prefer anti-correlation suggesting that DSFGs are over-dense in galaxy clusters.