A Measurement of the CMB <EE> Spectrum from the 2003 Flight of BOOMERANG

Abstract

We report measurements of the CMB polarization power spectra from the January 2003 Antarctic flight of BOOMERANG. The primary results come from six days of observation of a patch covering 0.22% of the sky centered near R.A. = 82.5 deg., Dec= -45 deg. The observations were made using four pairs of polarization sensitive bolometers operating in bands centered at 145 GHz. Using two independent analysis pipelines, we measure a non-zero <EE> signal in the range 100< l <1000 with a significance 4.8-sigma, a 2-sigma upper limit of 8.6 uK^2 for any <BB> contribution, and a 2-sigma upper limit of 7.0 uK^2 for the <EB> spectrum. Estimates of foreground intensity fluctuations and the non-detection of <BB> and <EB> signals rule out any significant contribution from galactic foregrounds. The results are consistent with a Lambda-CDM cosmology seeded by adiabatic perturbations. We note that this is the first detection of CMB polarization with bolometric detectors.

Figures (6)

• Figure 1: Sky coverage from the 2003 Boom-erang flight. The top panel shows the large region covered during the first part of the flight (the shallow region) and the bottom panel is the smaller region covered during the second half of the flight (the deep region). In the top panel, the outer set of black lines shows the sky cut used in the shallow mask. The inner outline shows the outline of the deep region sky cut. The shallow scans covered 3.0% of the sky, and sky cut used for the CMB analysis covers 1.8%. The deep region observations covered 0.28% of the sky and the outlined region covering 0.22% of the sky was used for the CMB analysis. The integration time per pixel for the deep observations is roughly 20 times longer than integration time per pixel during shallow observations. In both panels, the small circles represent regions of map which are excised due to the presence of known point sources.
• Figure 2: Angular Power Spectra Results. From top to bottom the panels show the $\left<EE\right>$, $\left<BB\right>$ and $\left<EB\right>$ power spectrum results from the NA (blue circles) and IT (red squares) pipelines. The solid line in the $\left<EE\right>$ plot is the best fit $\Lambda$CDM model to the WMAP $\left<TT\right>$ results spergel03a and the dashed line in all plots represent zero-signal. From these plots and the statistical tests in Table \ref{['tab:lambda_multi']}, it is clear the $\left<BB\right>$ and $\left<EB\right>$ are consistent with zero signal while $\left<EE\right>$ is consistent with $\Lambda$CDM, but inconsistent with zero signal. The $\Lambda$CDM $\left<EE\right>$ model predicted by the WMAP $\left<TT\right>$ results is nearly identical the best fit $\left<EE\right>$ model predicted from recent $\left<TT\right>$ results including B03 cmt.
• Figure 3: Window functions from the NA wide band results in Tables \ref{['tab:wideband']} and \ref{['tab:widelam']}. In the top panel the solid blue line is the $\left<EE\right>$ window function for the band $201\ge \ell \le 1000$. The dashed red line characterizes the leakage of E-modes into B-modes ($E \rightarrow B$) which has a maximum value of $\sim 0.01$. The $\left<BB\right>$ window function and $B \rightarrow E$ leakage window function are similar to those plotted here. The low amplitude of the $E \rightarrow B$ and $B \rightarrow E$ shows that B03 is able to separate E-mode and B-mode polarization. In the bottom panel, the window functions are shown for the different power spectrum parameterizations (i.e. $C_{\ell}^{(S)}$) used in the bandpower estimation. The shape of the $\left<EE\right>$ window function indicates the effective weight applied to each multipole moment. For all cases used in the NA analysis, the window function is significantly different than the flat band used in the IT wide band analysis. This is due to an effective Wiener filter which weights each multipole by $C_{\ell}^{(S)}/(C_{\ell}+N_{\ell})^2$ where $N_{\ell}$ is the noise at a given multipole, $C_{\ell}^{(S)}$ is the shape function and $C_{\ell}$ is the expected signal. In a given band the expected signal depends on the form of $C_{\ell}^{(S)}$.
• Figure 4: Results of Jackknife Tests. The left side shows the results for the $(h1-h2)/2$ test and the right side shows the results for the (WX-YZ)/2 test. The blue circles are results from the NA pipeline and the red squares are results from the IT pipeline. Table \ref{['tab:jack']} shows the $\chi^2$ and $PTE$ calculated from these results. For both tests, all three spectra are consistent with zero signal.
• Figure 5: Propagation of measurement errors in instrumental parameters to $\left<EE\right>$ error. The hatched bands show the upper edges of the $\left<EE\right>$ error (due to noise) for the bands in the multi-bin NA results (Figure \ref{['fig:full']}). The other symbols show errors in relative calibration (stars), bolometer time constant (upward triangles) and polarization angle (downward triangles). Because errors in absolute calibration and polarization efficiency are multiplicative factors which act identically on each bin, their effect is left off this plot and reported instead in Table \ref{['tab:sysparm']}.