Spider Optimization: Probing the Systematics of a Large Scale B-Mode Experiment

Abstract

Spider is a long-duration, balloon-borne polarimeter designed to measure large scale Cosmic Microwave Background (CMB) polarization with very high sensitivity and control of systematics. The instrument will map over half the sky with degree angular resolution in I, Q and U Stokes parameters, in four frequency bands from 96 to 275 GHz. Spider's ultimate goal is to detect the primordial gravity wave signal imprinted on the CMB B-mode polarization. One of the challenges in achieving this goal is the minimization of the contamination of B-modes by systematic effects. This paper explores a number of instrument systematics and observing strategies in order to optimize B-mode sensitivity. This is done by injecting realistic-amplitude, time-varying systematics in a set of simulated time-streams. Tests of the impact of detector noise characteristics, pointing jitter, payload pendulations, polarization angle offsets, beam systematics and receiver gain drifts are shown. Spider's default observing strategy is to spin continuously in azimuth, with polarization modulation achieved by either a rapidly spinning half-wave plate or a rapidly spinning gondola and a slowly stepped half-wave plate. Although the latter is more susceptible to systematics, results shown here indicate that either mode of operation can be used by Spider.

Figures (12)

• Figure 1: Left: 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. Power is supplied by solar arrays. The baseline observing strategy is to spin the payload in azimuth at fixed elevation. Spider is designed to obtain maximum sky coverage during a 20-30 day, mid-latitude, around-the-world flight. Right:Spider optical train. The telescope yields a flat and telecentric focal plane. The apodized Lyot-stop, which is fixed with regard to the instrument, is maintained at 4 Kelvin. All dimensions are in millimeters.
• Figure 2: Left: A single pixel of a 145 GHz antenna-coupled bolometer, comprising a 288 element phased array of dual-polarization slot antennas coupled to a matched load by a superconducting microstrip network. Microstrip filters, which determine the spectral response, and TES detectors, which measure the power dissipated in the load are visible at bottom. Right: The measured beam pattern of the dual-polarization antenna. The upper limit on differential beam ellipticity of is 1%, limited by the testbed. The polarization efficiency is greater than 98%. It is important to note that the beam pattern here is the feed beam pattern. The beam on the sky is influenced by the telescope. While the Spider telescope edge taper is modest, the beam on the sky will be more symmetric than the feed pattern shown here. In particular, the visibly large and asymmetric lobes above will not propagate to the sky.
• Figure 3: A schematic representation of the simulation pipeline. During the time-stream generation the (stepped or spinning) half-wave plate polarization angle is added to the intrinsic polarization angle of the individual detectors. In addition polarization angle systematics, beam offsets and gain drift are also applied during time-stream generation. Pointing jitter and pendulation systematics are added to the pointing time-streams during flight simulation. For the case of multiple beam distortions, multiple pointing time-streams are produced and the full-sky map is observed by each beam/pointing.
• Figure 4: Left: The Spider mask for 36 dps simulations. The dark band represents a galactic cut at $\pm 10^{\circ}$ in galactic latitude and the white region represents sun flagging. With these regions flagged the fraction of the sky covered for this observing strategy is $\sim 60$%. Pixel values are the number of observations in the pixel divided by the maximum hits value. The most obvious features are constant declination lines where scan circles on the sky for each detector overlap and the coverage is deepest. The pixel weighting is applied in order to reduce the effect of badly sampled pixels at the edge of the map. Top Right:Spider coverage projected into equatorial coordinates. Bottom Right: The IRAS 100 $\mu$m map SFD:1998 of Galactic dust is shown for comparison.
• Figure 5: Maps of the residuals in the Q Stokes parameters for the two polarization modulation strategies. The top panel shows the residuals for the continuous half-wave plate rotation case. For this case a "naive", or zero-iteration, map is shown. The middle panel shows the naive map for the stepped half-wave plate mode with the gondola spinning at 36 dps. In this case significant striping is present due to the loss of low frequency modes. This is caused by the inverse noise filtering of the time-streams during the map-making phase which uses a noise kernel with a realistic 100 mHz 1/f knee. For the stepped case the polarization modulation is not sufficient. However, iterated map-making reduces the impact of the striping as shown in the bottom panel and 10 iterations of the map-maker are sufficient to recover most of the lost modes.