Measurement of Cosmic Microwave Background Polarization Power Spectra from Two Years of BICEP Data

TL;DR

This work presents initial CMB polarization results from the BICEP experiment using two years of South Pole observations. The authors deploy two independent data-processing pipelines to produce temperature and polarization maps, estimate power spectra, and perform extensive tests to control systematics, noise, and foregrounds. They detect the $E$-mode polarization with a pronounced peak near $\ell \sim 140$ and find no evidence for $B$-mode polarization, deriving a constraint on the tensor-to-scalar ratio of $r < 0.72$ (95% CL) with a best-fit near $r = 0.02^{+0.31}_{-0.26}$, validating the $\Lambda$CDM framework while setting the first direct polarization-based limit on the inflationary gravitational wave background. The results demonstrate that polarization data from BICEP are robust against instrumental systematics and provide a solid foundation for future, deeper constraints on $r$ as the full data set are analyzed and combined with other experiments.

Abstract

Background Imaging of Cosmic Extragalactic Polarization (BICEP) is a bolometric polarimeter designed to measure the inflationary B-mode polarization of the cosmic microwave background (CMB) at degree angular scales. During three seasons of observing at the South Pole (2006 through 2008), BICEP mapped ~2% of the sky chosen to be uniquely clean of polarized foreground emission. Here we present initial results derived from a subset of the data acquired during the first two years. We present maps of temperature, Stokes Q and U, E and B modes, and associated angular power spectra. We demonstrate that the polarization data are self-consistent by performing a series of jackknife tests. We study potential systematic errors in detail and show that they are sub-dominant to the statistical errors. We measure the E-mode angular power spectrum with high precision at 21 < ell < 335, detecting for the first time the peak expected at ell ~ 140. The measured E-mode spectrum is consistent with expectations from a LCDM model, and the B-mode spectrum is consistent with zero. The tensor-to-scalar ratio derived from the B-mode spectrum is r = 0.03+0.31-0.26, or r < 0.72 at 95% confidence, the first meaningful constraint on the inflationary gravitational wave background to come directly from CMB B-mode polarization.

Figures (15)

• Figure 1: Bicep's CMB and Galactic fields are outlined on the 150-GHz FDS Model 8 prediction of dust emission finkbeiner99, plotted here in equatorial coordinates.
• Figure 2: Bicep$T$, $Q$, $U$, and coverage maps. The resolution is about 0.9$^\circ$ and 0.6$^\circ$ at 100 and 150 GHz, respectively, and no smoothing or apodizing has been applied to the maps. The noise per square degree in the central region of the $Q$ and $U$ maps is 0.81 $\mu$K at 100 GHz and 0.64 $\mu$K at 150 GHz. Note that the color scales of the temperature and polarization maps differ by a factor of 10.
• Figure 3: Data from Bicep's 100-GHz and 150-GHz channels are combined to form temperature, $E$, and $B$ signal and jackknife maps. The $E$ and $B$ maps are apodized to downweight noise-dominated edge pixels. The temperature anisotropies are measured with high S/N, and the $E$ signal map shows resolved degree-scale structure. The $B$ signal map and the $E$ and $B$ jackknife maps are consistent with noise.
• Figure 4: Raw power spectra, uncorrected for noise and filter bias, are shown here for 500 signal-plus-noise simulations (gray lines) and Bicep 150 GHz data (black points). The dashed red lines indicate the level of noise bias, $\ell(\ell+1){\hat{N}}_\ell/(2\pi)$, and the black lines show the $\Lambda$CDM spectra used as inputs for the signal simulations. Noise dominates much of the $EE$ spectrum and all of the $BB$ spectrum. The effect of filter bias is visible in the low-$\ell$ portion of the $TT$ spectrum, where the simulated spectra fall below the model curve.
• Figure 5: Signal-only simulations are used to evaluate the filter function $F^X_b$, shown here for 150 GHz ($F^X_b$ for other frequency combinations look similar). The beam--pixel window functions ${\cal B}_\ell^2 = B_\ell^2 H_\ell^2$ are shown for 100 and 150 GHz.