Measurement of the CMB Polarization at 95 GHz from QUIET

TL;DR

This paper reports on the measurement of CMB polarization at 95 GHz using the QUIET instrument, aiming to detect or constrain B-mode polarization predicted by inflation, through detailed instrument design, calibration, and data analysis of extensive observational data.

Contribution

The study introduces a new microwave polarimeter with optimized design and analysis techniques to measure CMB polarization, providing constraints on inflationary B-modes.

Findings

Collected over 5300 hours of CMB data

Developed a blind-analysis strategy with null tests

Characterized systematic errors and noise models

Abstract

(Abridged) Despite the great success of precision cosmology, cosmologists cannot fully explain the initial conditions of the Universe. Inflation, an exponential expansion in the first 10^-36s, is a promising potential explanation. A generic prediction of inflation is odd-parity (B-mode) polarization in the cosmic microwave background (CMB). The Q/U Imaging ExperimenT (QUIET) aimed to limit or detect this polarization. We built a coherent pseudo-correlation microwave polarimeter. An array of mass-produced modules populated the focal plane of a 1.4m telescope. Each module had a sensitivity to polarization of 756muK sqrt{s} with a bandwidth of 10.7+/-1.1 GHz centered at 94.5+/-0.8 GHz; the combined sensitivity was 87+/-7muK sqrt{s}. We incorporated deck rotation, an absorbing ground screen, a new time-stream double-demodulation technique, and optimized optics into the design to reduce instrumental polarization. We observed with this instrument at the Atacama Plateau in Chile between August 2009 and December 2010. We collected 5336.9 hours of CMB observation and 1090 hours of astronomical calibration. This thesis describes the analysis and results of these data. We characterized the instrument using the astronomical calibration data as well as purpose-built artificial sources. We developed noise modeling, filtering, and data selection following a blind-analysis strategy. Central to this strategy was a suite of 32 null tests, each motivated by a possible instrumental problem or systematic effect. We also evaluated the systematic errors in the blind stage of the analysis before the result was known. We then calculated the CMB power spectra using a pseudo-Cl cross-correlation technique that suppressed contamination and made the result insensitive to noise bias.