SPIDER: A Balloon-borne Large-scale CMB Polarimeter

TL;DR

Spider tackles measuring CMB polarization to probe the early Universe by combining a balloon-borne platform with six monochromatic refracting telescopes, a rotating half-wave plate for systematic control, and large antenna-coupled TES bolometer arrays read out by SQUID multiplexers. The multi-band (100/150 GHz primary; 225/275 GHz for foregrounds) design enables robust separation of Galactic dust and CMB polarization, while aiming to constrain the reionization optical depth $\tau$ and to search for B-mode signals from primordial gravitational waves characterized by $r$. The 2–6 day initial flight from Alice Springs will map ~50% of the sky at large scales to tighten $\tau$ constraints, and longer-duration flights are planned to reach inflationary B-mode targets. The work demonstrates the viability of large-format TES arrays and in-flight systematics control on a sub-orbital platform, informing future space missions such as CMBpol.

Abstract

Spider is a balloon-borne experiment that will measure the polarization of the Cosmic Microwave Background over a large fraction of a sky at 1 degree resolution. Six monochromatic refracting millimeter-wave telescopes with large arrays of antenna-coupled transition-edge superconducting bolometers will provide system sensitivities of 4.2 and 3.1 micro K_cmb rt s at 100 and 150 GHz, respectively. A rotating half-wave plate will modulate the polarization sensitivity of each telescope, controlling systematics. Bolometer arrays operating at 225 GHz and 275 GHz will allow removal of polarized galactic foregrounds. In a 2-6 day first flight from Alice Springs, Australia in 2010, Spider will map 50% of the sky to a depth necessary to improve our knowledge of the reionization optical depth by a large factor.

Figures (6)

• Figure 1: Simulated recovery of the temperature and polarization angular power spectra from a 4-day first flight of Spider, with 100 and 145 GHz as in Table \ref{['tbl:detectors']}. The simulations of time-ordered data include 1/f noise with a 100 mHz knee frequency. The half-wave plate is stepped each hour during the daytime scan mode and each night during nighttime spin mode. The sky signal is reconstructed from the time ordered data with a naively binned map-maker, power spectra are estimated with xfasterxfaster. The input model (solid curve) is the best fit $\Lambda$CDM + tensors of WMAP3. The dashed curve is our own best fit spectrum with cosmomc software 2002PhRvD..66j3511L - see Figure \ref{['fig:params']}. These are expected to be conservative estimates of Spider' s sensitivity; we expect to reduce the error in the lowest EE bin by $\sim$ 30% with an optimal map maker.
• Figure 2: Marginalized likelihood contours for WMAP 5-year and Spider' s first flight. $r$ is the ratio of tensor perturbations to scalar perturbations, $\tau$ is the optical depth through reionization, and $n_s$ is the spectral index of scalar perturbations. The $+$ represents the input value of the parameters; in this case $n_s=0.984$, $r=0.1$, $\tau=0.09$. Spider' improvement comes from more sensitive polarimetery on large scales.
• Figure 3: The Spider payload. Six independent monochromatic telescopes are housed in a single long hold time cryostat. Each telescope is fully baffled from radiation from the ground and balloon. The gondola scans in azimuth with a reaction wheel and a motorized pivot. The cryostat, mounted on bearings, can be adjusted in elevation. Solar arrays provide power.
• Figure 4: Measurements of the transmission of a polarized signal as a function of frequency and rotation angle for a single birefringent sapphire plate appropriate for our 150GHz band, which will be used in Spider' s half-wave plate with an anti-reflection (AR) coating. The data are well-described by a birefringence model that gives an index difference of $\Delta n = 0.3350 \pm 0.0003$ between the fast and slow axes. The model fit also recovers the angle of the crystal axes to 0.1$^\circ$ precision. The addition of an AR-coat greatly reduces the channel spectra evident in band.
• Figure 5: Left: A prototype Spider focal plane unit, consisting of four detector wafers. Each wafer comprises 64 spatial pixels, sensing two polarizations each, and read out by 128 detectors. The smaller chips on the periphery are NIST column multiplexer chips, each reading out 32 detectors. Light enters through a square aperture under the tiles, back illuminating the detectors. A niobium plate covers the entire assembly and serves as a superconducting magnetic shield as well as a backshort surface for the antennae. Right: the passband of a 145 GHz device. This passband is of the device alone, the Spider optical train is not yet mated to the detector.