Cosmic Microwave Background Radiation Polarization: The POLAR and COMPASS/SSPEX Experiments

1. Cosmic Microwave Background Radiation Polarization: The POLAR and COMPASS/SSPEX Experiments • G. Dall’Oglioa, P. Fareseb, T. Gaierc, J. Gundersend, B. Keatinge, S. Klawikowskif, L. Knoxg, A. Levyb, P. Lubinb, C. O’Dellf, L. Piccirilloh , J. Ruhlb, P. Timbief • Universita’ di Roma III, b. University of California at Santa Barbara, c. NASA/JPL, d. Princeton University, e. Stanford University, • f. University of Wisconsin-Madison, g. University of Chicago, h. Bartol Research Institute Introduction In the standard Big Bang model, the Cosmic Microwave Background Radiation (CMB) is the thermal relic of a hot, dense phase of the early Universe. When the temperature was greater than ~3000 K, the photon background was energetic enough to ionize hydrogen. These photons were intimately coupled to free electrons by Thomson scattering, forming a photon-baryon fluid. As the universe expanded and cooled, atomic hydrogen formed during recombination and the photon mean-free-path increased dramatically. These photons have a blackbody spectrum. In the standard recombination model these photons travel to us without further scattering, and carry the imprint of mass inhomogeneities at the “surface of last scattering”, located at a redshift of ~1000. The temperature variations, or anisotropies, in the CMB across the sky encode a bonanza of cosmological information. In particular, acoustic modes in the photon-baryon fluid give rise to a spectrum of temperature fluctuations at angular scales below a degree, the horizon size at the last-scattering surface. Measurement of this spectrum has driven experimentalists to map the sky at microwave and millimeter wavelengths to discover the values of the key parameters that define current cosmological models. An enormous amount of interest has developed recently in another property of the CMB, its polarization. All models predict the CMB to be polarized because the Thomson scattering that thermalizes the radiation has a polarization-dependent cross-section. Estimates for the size of the polarization signal are on the order of a few K,a factor of 10 to 50 below the temperature anisotropy signal, and a similar factor below current upper limits. Like the CMB temperature power spectrum, the polarization power spectrum contains information on all angular scales. An important feature of the polarization spectrum is that it directly probes the physical conditions at the epoch of last scattering, since it is only then that free electrons were present to Thomson scatter the radiation. In contrast, temperature anisotropy can evolve after the last scattering. In this sense, polarization observations can provide a “check” on the ongoing measurements of temperature anisotropy. Because of the symmetry of Thomson scattering, it turns out that only the quadrupole component of the radiation field that scatters from the free electrons produces polarization in the outgoing radiation. At the last scattering surface, the main source of quadrupole radiation is motion of the baryon-photon fluid, which causes a Doppler shift. Hence, the CMB polarization primarily probes the velocity of matter at the last scattering surface, while the temperature anisotropy measures the density variations. Because density and velocity are correlated in the early universe, so too will be the temperature and polarization power spectra we measure today. The polarization-temperature cross-correlation is always larger than the polarization autocorrelation, and is a robust cosmological probe in its own right. COMPASS/SSPEX: To measure polarization at sub-degree angular scales, the POLAR radiometer will be coupled to a 2.6 meter diameter on-axis Cassegrain telescope.This telescope is a copy of the OASI telescope, shown above, which is now deployed at Terra Nova Bay – Antarctica. The new telescope supports the secondary mirror using a molded Styrofoam cone to avoid scattering by the secondary support struts. Observations with COMPASS are planned for 2000 in Madison, WI. The system will be installed at Dome-C in 2002. A proposal to develop a new multi-pixel receiver using cooled bolometers for detectors has been submitted to NSF. This new system, called SSPEX, would extend the frequency coverage of COMPASS to 300GHz and improve the sensitivity dramatically. POLAR: This instrument is a polarimeter designed to detect CMB polarization on large angular scales. The system is capable of detecting extremely small signals while simultaneously rejecting terrestrial and astronomical foreground sources. Two similar millimeter-wave polaiimeters operate simultaneously in a shared cryostat (red cylinder at the top) which uses a mechanical cryocooler to maintain the horns and front-end amplifiers at 20 K. They are tuned to two different frequency bands,Ka (26-36 GHz) and W (85-105 GHz), to discriminate against galactic foreground sources of emission. Each polarimeter views the sky with a 7 FWHM corrugated feedhorn and the cryostat rotates about the vertical axis to allow synchronous detection of the Q and U Stokes parameters of the polarization signal. From Madison, WI, POLAR observes ~ 36 different pixels in a ring around the NCP at dec=43 for many months to reach the level of a few K per pixel. We are now making long-term observations with the Ka band polarimeter and anticipate adding the W-band polarimeter in 2000. This same two-channel polarimeter will be used with minor modifications for the COMPASS instrument for measurements of small angular scale polarization. This standard model may not tell the whole story; we know that the CMB photons may have scattered again by free electrons during a second, later, ionized phase of the Universe. In general, polarization of the CMB at large scales is always enhanced in models which predict early reionization, while the effect of reionization on the temperature anisotropy and polarization is primarily an attenuation of the signal at small scales. For reasonable non-standard models, the amplitude of polarization on large angular scales is on the level of 10% of the temperature anisotropy, while for the standard model of recombination the corresponding polarization level does not exceed 1%.The amplitude of CMB polarization on large scales tells us the epoch of reionization. Thepatterns formed by the polarization can be divided into ”E-modes” and ”B-modes” which are similar to E and B field lines respectively. E-modes are sensitive to the density perturbations mentioned above, while B-modes are produced by gravitational waves. Hence, polarization ultimately could be used to detect the presence of gravitational waves in the early universe. Systematic Effects: Reflection and emission of radiation from the mirror surfaces produce a small instrumental polarization. This figure is a simulation of polarized emission from an on-axis Cassegrain reflector antenna. In this model a polarimeter measuring the U Stokes parameter is positioned near the prime focus of an on-axis reflector antenna and views the reflector with a Gaussian beam pattern that tapers to –30dB at the edges of the mirror. The emission from various regions of the mirror is polarized in a manner that depends on its emissivity, curvature, and physical temperature. The colored lobes indicate the contribution from these regions of the mirror to the Stokes parameter measurement. These values are less than 2 mK. For a completely symmetric optical system, the contributions from the pattern on the mirror cancel, yielding U = 0. However, in this simulation the feedhorn is displaced in the vertical direction from the symmetry axis of the antenna by one horn diameter to illustrate the effects of an off-axis pixel in the focal plane. In this case, the integrated contribution of polarized emission from the mirror is ~ 10 uK. The mirror is taken to be f/1 with a 2.6 meter diameter mirror at 300 K, with an emissivity of 0.1% at 90 GHz, scaling as n1/2  to 30 GHz for this simulation. The results are similar for the Q parameter, with the pattern rotated by 45 . The scale on the frame of the figure is in meters. Polarization power spectra (for E modes) for a standard CDM cosmology with no reionization (reddash) and total reionization at redshift z = 100 (solid black). Cl is the polarization power spectrum amplitude for the multipole moment l. l = 200 corresponds to approximately 1 degree. The angular sensitivities (window functions) of the two measurements are shown: COMPASS 20 (FWHM) beam (dots) and the POLAR 7o (FWHM) beam (dot-dash) for comparison. Clearly, the small sale features of the power spectra are only probed with a beamsize < 20'. POLAR Radiometer. The schematic above shows the Ka-band (26-36 GHz) system, which is a correlation radiometer. RF signals enter the circular corrugated feedhorn and then an orthomode transducer, a waveguide device which splits the incoming signal into its two linear polarization components. These signals are then amplified by low-noise high electron mobility transistor (HEMT) amplifiers. At room temperature the signals are further amplified and then downconverted to the IF band (2-12 GHz). This band is then subdivided into three equal sub-bands. The signals from the two amplifier chains are then multiplied (correlated) in an analog mixer. An electronic phase switch modulates the relative phase of the LO signals reaching the two mixers by 180. As a result, the sign of the correlated output of the polarimeter is modulated at the switch frequency and is detected by a lock-in amplifier. The output of the polarimeter is proportional to Qcos(2) +Usin(2), where  is the rotation angle about the zenith and Q and U are the two Stokes parameters that define linear polarization. MEASURING CMB POLARIZATIONThe fundamental quantities measured by instruments described here are the Stokes parameters, Q and U, at each pixel which completely characterize the linear polarization of the CMB. The Q and U Stokes parameters are each the difference between two orthogonal linear polarizations. The U basis vectors are rotated by 45 with respect to those for Q. Foreground Radiation: Typical levels of anisotropy in the high galactic latitude polarized foreground emission at degree angular scales. The synchrotron spectrum is normalized to the rms brightness temperature of synchrotron at 19 GHz and assumes 30% polarization. The bremsstrahlung spectrum is normalized to 30 K at 10 GHz and assumes a 10% polarization. The spinning dust spectrum is proposed by Draine and Lazarian (1998) and assumes 3% polarization. The thermal dust spectrum assumes 5% polarization, a dust temperature of 18 K, an emissivity index of 1.8 and uses 3 K/MJy/sr to scale typical degree scale rms values (of 0.5 MJy/sr) at 100 microns to 90 GHz. The CMB E-polarization spectrum is assumed to be 10^-6 of the CMB brightness spectrum and the CMB B-polarization is assumed to be 0.1 of the E-polarization spectrum. The three frequency bands (Ka, W and D) of POLAR and COMPASS are shown above the spectra. Sensitivity: These plots display estimated error contours for the proposed SSPEX Dome C experiment. The l-space resolution is set as 90. The telescope drift scans while observing the zenith. The blue solid line represents error levels that can be obtained by averaging all 7 horns over one 21' pixel. The red dotted line corresponds to errors obtained using only the central pixel with 7'resolution. The green dashed line displays the errors expected on the COMPASS project observing from Wisconsin in 2000. Errors are calculated with Q and U measured simultaneously in each horn with assumed sensitivity to each Stokes parameter of respectively 250 and 800 μK s1/2 integrated for a total of 90 days. CMB power spectra are indicated by respectively, <TT> for temperature-temperature correlation, <TE> for temperature-polarization correlation and <EE> for polarization-polarization. <TT> is shown for comparison. Foreground Power Spectra: The angular power spectra of the foregrounds are expected to differ significantly from the CMB polarization. These figures show power spectra of CMB E polarization and foregrounds. The foreground model is from Knox 1999 and the parameters chosen are similar to those in the “MID” model of Tegmark et al. 1999. The foregrounds are dust (dots), point sources (short dashes), synchrotron (dot-dash), and spinning dust (long dashes). Observations at 90 and 150 GHz are expected to have substantially lower contamination than lower-frequency measurements.