Sunyaev Zel`dovich Effect Experiments

Transcript

Sunyaev-Zel’dovich Effect Experiments and CMB polarization Experiments : differential techniques at work Silvia Masi Dipartimento di Fisica, Sapienza Università di Roma L’ Aquila, apr.25th, 2014. ISSS course “Observing the Universe  with the Cosmic Microwave Background” Plan of the talk : – The Cosmic Microwave Background (CMB) is a 3K  blackbody. Detour on differential measurements of  the CMB and the FTS. – Basics of the Sunyaev‐Zeldovich (Inverse Compton)  effect. – Current status of the experiments – The future : SZ spectroscopy ‐ WHY – The new instruments (differential FTS) we are  developing: • OLIMPO • Millimetron – CMB polarization measurement techniques The spectrum of the CMB is a precise blackbody. Mather et al. 1994 Nobel prize in 2006 150GHz 300GHz wavenumber σ (cm-1) COBE – FIRAS : A differential Fourier-transform spectrometer: measures the difference between the brightness collected at two input ports. One port collects radiation from the sky. The other collects radiation from an internal reference blackbody. Differential measurements are good measurements • • • Imagine you want to obtain an accurate measurement of the brightness of the CMB. In the RJ region B is proportional to T. Imagine you want to measure T with an accuracy of 1 mK. If you have a regular photometer, producing a signal proportional to the temperature, you measure an output voltage : V=RTCMB where the constant R is the calibration constant, or gain, of the photometer. You observe the CMB, measure V, and infer TCMB • • • • • V = R , with ΔTCMB ΔV ΔR = + TCMB V R The first contribution to the error is due to the noise in your instrument (i.e. depends on the sensitivity of the detector). The second term is due to the calibration accuracy of the instrument. Both need to be < 1mK/2.7K , i.e. < 0.04% (!!!). While it is certainly possible to have low noise detectors, and so the condition above is easily met for the first term, it is extremely difficult to calibrate the photometer ( i.e. measure R ) with such a precision. This difficulty is typical of absolute measurements. The solution is to convert the measurement in a differential measurement. Differential measurements are good measurements • • • • Imagine you have a differential photometer, i.e. an instrument with two input ports, producing a signal proportional to the difference in the brightness (temperature) present at the two input ports. You focus the CMB on the first port, and focus radiation from a reference blackbody on the second port. You measure an output voltage proportional to the temperature difference: V=R(TCMB-TREF) where the constant R is the calibration constant of the differential photometer. You observe the CMB, measure V, and infer TCMB − TREF = • Now you have a second handle : you can change TREF as you like. So you can try to change it until V=0, within the noise, say +ΔV. In this case TCMB = TREF • • V R ΔV ± R the only contribution to the error is due to the noise ΔV. If ΔV/R is small with respect to T, the exact value of the calibration constant is irrelevant. This solves the main difficulty of absolute measurements. There are additional advantages in differential measurements. Differential measurements are good measurements • Your photometer is emitting radiation, with a birghtness corresponding to a temperature Tp. If it is a regular photometer, what you really measure is V=R(TCMB +Tp ) • • You need to know precisely Tp to be able to subtract it and find TCMB . In a differential photometer the emission of the spectrometer is a common mode, i.e. is present in both ports, so what you measure is the difference : V=R[(TCMB+Tp )–( TREF+Tp)] =R(TCMB-TREF) • • The common-mode emission cancels completely in an ideal differential photometer, and you are back to the previous case. A room-temperature instrument can emit much more than the 3K blackbody you are measuring, but you can do the measurement anyway. In a real differential instrument, the rejection of the common mode signal is not perfect, because for example the two ports of the instrument are slightly different. In that case you will measure something like V=R(TCMB-TREF)+ cmrr R(TCMB+TREF) • where cmrr is the common-mode rejection ratio, a small number in good differential instruments (usually 10-5 or even less) Example of the power of differential measurements: CMB anisotropy measurements • You want to measure ΔT of the order of 100μK with an instrument and an atmosphere emitting like room-temperature blackbodies, with say few % emissivity (i.e. 10K signals), plus the 2.7K CMB monopole itself. • It seems hopeless, but this large signal is approximately independent of the direction observed (the temperature of the instrument is stable during the scan of the sky; atmospheric emission is approximately stable at least for constant elevation scans, CMB monopole is stable by definition). • So you use a differential photometer, and compare the signals from two contiguous sky regions. The common mode emission is rejected (as long as it is stable and really common-mode) and only the tiny anisotropy signal ΔT is measured, despite of the overwhelming emission from the instrument, atmosphere and CMB monopole. How can we implement this technique in spectral measurements ? • The only solution is to use a Fourier-transform spectrometer. • This is an efficient spectrometer based on the measurement of the auto-correlation function of the radiation under analysis. • In the MPI (Martin-Puplett Interferometer) configuration, the instrument becomes differential. Elementary theory of the FTS • Imagine one of the two arms is longer than the other by x: the optical path difference is 2x. • The two electrical fields E on the detector are E(t ) = Eo (σ )RT (σ ) cos(2πσct) + + Eo (σ )RT (σ ) cos(2πσct + 4πσx) • Here σ is the wavenumber (1/λ), RT(σ) is the efficiency of the beamsplitter (reflectivity times transmission) ZPD OPD Michelson Interferometer Elementary theory of the FTS E(t ) = Eo (σ )RT (σ ) cos(2πσct) + Eo (σ )RT (σ ) cos(2πσct + 4πσx) • The detected power is proportional to the square of the total Electrical field: I ( x ) = E ( t ) 2 = E ( t ) E * (t ) = [ (σ )rt (σ )[1 + e ][ + 1] = E ] Eo2 (σ )rt (σ ) e i 2πσct + e i 2πσct e i 4πσx e −i 2πσct + e −i 2πσct e −i 4πσx = Eo2 − i 4πσ x + e i 4πσx 2 o (σ )rt (σ )2(1 + cos 4πσx) • So we have an interferogram I ( x) − I ∝ rt(σ ) cos(4πσx) • For the non monochromatic case we sum all the components, each with amplitude S(σ) (spectrum): ∞ I ( x ) − I ∝ ∫ S (σ )rt (σ ) cos( 4πσ x ) d σ 0 Spectrum and interferogram are a Fourier pair. ∞ I (x) − I ∝ ∫ S (σ )rt (σ ) {1 + cos [4 πσ x ]}d σ 0 ∞ = ∫ 0 S (σ ) 8 R (1 − R ) 2 (1 − cos [4 π nd cos θ ' σ (1 − 2 R cos [4 π nd cos θ ' σ ] + R 2 Performing the inverse Fourier transform of the interferogram one gets the product of the spectrum of the radiation times the efficiency of the beamsplitter. For a dielectric beamsplitter interference occurs reducing the useful signal at selected frequencies, depending on the refraction index and on the thickness of the beamsplitter. Low frequencies have an efficiency problem. CMB is important at wavenumbers lower than 20 cm-1 (!) = ]){1 + cos [4 πσ x ]}d σ 2 ) Alternative beam-splitters: metallic wire-grids • Radiation polarized with electrical filed parallel to the wires generates an electrical current in the wires, which Main axis in turn produces a reflected electromagnetic wave. Very little is transmitted. • Radiation polarized with electrical field orthogonal to the wires cannot generate electrical currents in the wires, so it is transmitted with very little attenuation. Main axis • If the incoming radiation is not polarized, this is very close to an ideal intensity beamsplitter (50% transmitted, 50% reflected). E E • Depending on the distance between the wires and on the wavelength, one can have perfect beamsplitters. • G25 transmits perfectly Eperp and does not transmit Epar in the frequency range 10-1000 GHz of interest here. I θ I’ Martin Puplett Interferometer • An instrument with two input ports and two output ports. • Consider two unpolarized sources and two intensity (unpolarized) detectors • Let’s study the components. Martin Puplett Interferometer • The input polarizer A relects radiation from source S1’ and transmits radiation from S1 • The polarizer wires are horizontal (parallel to the plane of the drawing) • So beam 3 has one vertical component from S1 and a horizontal component from S1’ . Martin Puplett Interferometer • The input polarizer A relects radiation from source S1’ and transmits radiation from S1 • The polarizer wires are horizontal (parallel to the plane of the drawing) • So beam 3 has one vertical component from S1 and a horizontal component from S1’ . Martin Puplett Interferometer • The beamsplitter polarizer B is rotated so that its wires are seen by beam 3 with and angle of 45o with respect to the plane of the drawing. • In this way, B reflects a fraction of the verical component of beam 3 (coming from S1 ) and the same fraction of the horizontal component of beam 3 (coming from S1’ ). • So beam 4 is polarized along the direction of the wires of B and contains radiation from both S1 and S1’ in equal parts. 4 how beamsplitter B reflects radiation from both sources 3 Martin Puplett Interferometer • Moreover, B transmits both a fraction of the vertical component of beam 3 (coming from S1 ) and the same fraction of the horizontal component of beam 3, coming from S1’ . • So beam 4’ is polarized along the direction orthogonal to the wires of B and contains radiation from both S1 and S1’ in equal quantities. 4’ beamsplitter B transmits radiation from both sources, and reflects radiation from both sources. 4 3 Martin Puplett Interferometer • The roof mirror CD reflects beam 4 towards B (beam 6). • Using a roof mirror, the polarization plane is rotated by 90o. • So beam 6, which had been reflected (as beam 4) by B, is now transmitted and contributes to beam 8. The roofmirror rotates the incoming polarization by 90o. 5 4 6 Martin Puplett Interferometer • Analogously for the other arm, roofmirror C’D’ reflects beam 4’ back towards B, as beam 6’ • The polarization plane is rotated by 90° so that beam 6’, which had been transmitted by B (as 4’) is now reflected and contributes to beam 8. • • • • • Martin Puplett Interferometer The output polarizer G is rotated so that its wires are horizontal (as in the input polarizer). Since reflected and transmitted rays from B have i raggi provenienti dal beamsplitter are both at 45°, are split in the same way towards both detectors. So detectors receive radiation from both sources, that traveled both arms of the interferometer, exactly as in a Michelson interferometer. However, there is a difference: the radiation from S1’ has been reflected 1 times more than radiation from S1, so has 180° phase difference. This is why the instrument is sensitive to the difference between the intensities from the two sources. Jones calculus • To treat linearly polarized radiation (Jones 1941) • The interaction of the electrical field of the EM wave with an optical component is described by a 2x2 matrix. ⎛ E x ,OUT ⎞ ⎛ J11 ⎜ ⎟ = ⎜⎜ ⎟ ⎜E y , OUT ⎝ ⎠ ⎝ J 21 J12 ⎞⎛ E x , IN ⎞ ⎟ ⎟⎟⎜⎜ J 22 ⎠⎝ E y , IN ⎟⎠ • This calculus works only if light is fully polarized. Partially polarized light can be described by Muller calculus, but phase information is lost. E field • Described by 2-components vectors. if E is the amplitude: • Linear polarization along x ⎛1⎞ E ⎜⎜ ⎟⎟ ⎝0⎠ • Linear polarization along y • 45° from x • -45° from x • Right-end circular polarization • Left-hand circular polarization E ⎛1⎞ ⎜⎜ ⎟⎟ 2 ⎝1⎠ E ⎛1⎞ ⎜⎜ ⎟⎟ 2 ⎝−i⎠ ⎛0⎞ E ⎜⎜ ⎟⎟ ⎝1⎠ E ⎛1⎞ ⎜⎜ ⎟⎟ 2 ⎝ −1⎠ E ⎛1⎞ ⎜⎜ ⎟⎟ 2 ⎝i⎠ Reference system specchio • Comoving with the light ray : Mirrors y y z x x z • Ideal mirror orthogonal to xz plane: ⎛1 0 ⎞ ⎟⎟ M = ⎜⎜ ⎝ 0 − 1⎠ • Ideal roof mirror orthogonal ⎛ 1 RM = ⎜⎜ to plane xz: 0 ⎞⎛ 1 0 ⎞ ⎛ 1 0⎞ ⎟⎟⎜⎜ ⎟⎟ = ⎜⎜ ⎟⎟ ⎝ 0 −1⎠⎝ 0 −1⎠ ⎝ 0 1 ⎠ y Ein PA Linear polarizer • Transmission: ⎛1 0⎞ ⎟⎟ Polarizer with horizontal ⎜⎜ ⎝ 0 0⎠ principal axis • Transmission: Polarizer with vertical principal axis • Transmission: Polarizer with principal axis tilted by φ from horizontal ⎛ 0 0 ⎞ Eout, R ⎜⎜ ⎟⎟ ⎝0 1⎠ Eout, T φ x ⎛ cos 2 ϕ Pt (ϕ ) = ⎜⎜ ⎝ cos ϕ sin ϕ ⎛ sin 2 ϕ • Reflection: Polarizer with Pr (ϕ ) = ⎜ ⎜ cos ϕ sin ϕ principal axis tilted by φ ⎝ from horizontal. cos ϕ sin ϕ ⎞ ⎟ 2 sin ϕ ⎟⎠ − cos ϕ sin ϕ ⎞ ⎟ 2 − cos ϕ ⎟⎠ Delay • The last component needet to describe the MPI is the delay introduced by an optical path difference between the two arms δ=4πσx : this is the same for the two polarizations, so is described by the matrix D: ⎛ e iδ D (δ ) = ⎜⎜ ⎝ 0 0 ⎞ ⎟ e iδ ⎟⎠ • Imagine to have two sources S1 and S’1 , whose E fileds are described by Jones vectors ⎛ Ax ⎞ ⎛ Bx ⎞ ' S1 = ⎜⎜ ⎟⎟ S1 = ⎜⎜ ⎟⎟ ⎝ Ay ⎠ ⎝ By ⎠ • beam 3, aftyer the input polarizer (with horizontal principal axis) is simply ⎛ Ax ⎞ ⎟ S3 = Pt (0) S1 + Pr (0) S = ⎜⎜ ⎟ − B y ⎝ ⎠ ' 1 • beam 4, reflected by the beamsplitter, and beam 4’ , transmitted by the dal beamsplitter are: 1 ⎛1 − 1⎞⎛ Ax ⎞ 1 ⎛ Ax + B y ⎞ ⎟= ⎜ ⎟ ⎟⎟⎜⎜ S 4 = Pr (π / 4) S3 = ⎜⎜ 2 ⎝1 − 1⎠⎝ − B y ⎟⎠ 2 ⎜⎝ Ax + B y ⎟⎠ 1 ⎛1 1⎞⎛ Ax ⎞ 1 ⎛ Ax − B y ⎞ ' ⎟= ⎜ ⎟ ⎟⎟⎜⎜ S 4 = Pt (π / 4) S3 = ⎜⎜ ⎟ ⎜ 2 ⎝1 1⎠⎝ − B y ⎠ 2 ⎝ Ax − B y ⎟⎠ • Since roof mirrors are represented by Identity matrices, 1 ⎛ Ax + B y ⎞ ⎟ S 6 = S 4 = ⎜⎜ 2 ⎝ Ax + B y ⎟⎠ iδ − A B ⎛ ⎞ − A B e ( ) ⎛ ⎞ 1 x 1⎜ x y y ' ' ⎟ ⎟= S 6 = DS 4 = D ⎜⎜ i δ 2 ⎝ Ax − B y ⎟⎠ 2 ⎜⎝ ( Ax − B y )e ⎟⎠ • The former is transmitted by the beamplitter, while the latter is reflected, so that 1 ⎛ 1 1⎞ 1⎛ 1 1 ⎞ ' ' ⎟⎟ S 6 + ⎜⎜ ⎟⎟ S 6 S8 = Pt (π / 4) S 6 + Pr (3π / 4) S 6 = ⎜⎜ 2 ⎝ 1 1⎠ 2 ⎝ − 1 − 1⎠ 1 ⎛ 1 1⎞ 1⎛ 1 1 ⎞ ' ⎟⎟ S 6 + ⎜⎜ ⎟⎟ S 6 S8 = Pt (π / 4) S 6 + Pr (3π / 4) S = ⎜⎜ 2 ⎝ 1 1⎠ 2 ⎝ − 1 − 1⎠ • So we obtain ' 6 iδ iδ ⎛ + − − 1) ⎞ ( 1 ) ( A e B e 1⎜ x y ⎟ S8 = ⎜ iδ iδ 2 ⎝ − Ax (e − 1) + B y (e + 1) ⎟⎠ • And including the effect of the output polarizer: iδ iδ ⎛ A e B e ( + 1 ) − ( − 1) ⎞ 1 x y ⎟ S9 = Pt (0) S8 = ⎜⎜ ⎟ 2⎝ 0 ⎠ 0 ⎞ 1⎛ ' ⎟ S9 = Pr (0) S8 = ⎜⎜ iδ iδ 2 ⎝ Ax (e − 1) − B y (e + 1) ⎟⎠ • We are now able to compute the intensity for each of the detectors at the two outputs: I = E x E x* + E y E *y → Ax2 + By2 Ax2 − By2 1 2 I 9 = [ Ax (1 + cosδ ) + By2 (1 − cosδ )] = + cosδ 2 2 2 2 2 2 2 A + B A − B 1 2 x y x y ' 2 I 9 = [ Ax (1 − cosδ ) + By (1 + cosδ )] = − cosδ 2 2 2 • We are now able to compute the intensity for each of the detectors at the two outputs: I = E x E x* + E y E *y → Ax2 + By2 Ax2 − By2 1 2 I 9 = [ Ax (1 + cosδ ) + By2 (1 − cosδ )] = + cosδ 2 2 2 2 2 2 2 A + B A − B 1 2 x y x y ' 2 I 9 = [ Ax (1 − cosδ ) + By (1 + cosδ )] = − cosδ 2 2 2 • We are now able to compute the intensity for each of the detectors at the two outputs: I = E x E x* + E y E *y → Brightness DIFFERENCE ! Ax2 + By2 Ax2 − By2 1 2 I 9 = [ Ax (1 + cosδ ) + By2 (1 − cosδ )] = + cosδ 2 2 2 2 2 2 2 A + B A − B 1 2 x y x y ' 2 I 9 = [ Ax (1 − cosδ ) + By (1 + cosδ )] = − cosδ 2 2 2 I9 = Ax2 + By2 I = ' 9 2 + Ax2 + By2 2 Ax2 − By2 − 2 Ax2 − By2 2 cosδ cosδ COBE-FIRAS • COBE-FIRAS was a cryogenic MartinPuplett FourierTransform Spectrometer with composite bolometers. It was placed in a 400 km orbit. • A zero instrument comparing the specific sky brightness to the brightness of a cryogenic Blackbody ∞ I SKY ( x) = C ∫ [SSKY (σ ) − SREF (σ )]rt(σ ){1+ cos[4πσx]}dσ 0 ∞ ICAL( x) = C ∫ [SCAL(σ ) − SREF (σ )]rt(σ ){1+ cos[4πσx]}dσ 0 Why from space ? • Compare sky emission to internal reference blackbody emission. • In ground-based experiments the CMB is observed through the earth atmosphere, which is a powerful emitter of millimeter waves. • The differential nature of the instrument does not help here, since the path through atmospheric gases is present only for the photons from the CMB, while is basically absent for the photons from the internal reference blackbody. • So, for a ground-based intrument, the quantity measured is ( V = ℜ BCMB e −τ + Batm (1 − e −τ ) − Bref ) • where Batm is much larger than BCMB : • In space, instead V = ℜ(BCMB − Bref ) • If the internal reference is a blackbody, V can be zero-ed if and only if the CMB brightness is a blackbody with the same temperature of the internal blackbody. 90 150 210 270 330 390 GHz FIRAS • The FIRAS guys were able to change the temperature of the internal blackbody until the interferograms were null within the noise. • This is a null measurement, which is much more sensitive than an absolute one: (one can boost the gain of the instrument without saturating it !). • This means that the difference between the spectrum of the sky and the spectrum of a blackbody is 0, i.e. the spectrum of the sky is a blackbody with that temperature. • This also means that the internal blackbody is a real blackbody: it is unlikely that the sky can have the same deviation from the Planck law characteristic of the source built in the lab. σ (cm-1) wavenumber 0 5 brightness temperature of the sky (K) at 150 GHz • The CMB dominates the sky brightness at mm wavelengths • And is very much isotropic: the early universe was very homogeneous • The most boring picture of the sky ever ! Boosting the contrast of the image and removing the foregrounds … Planck 2013 Planck 2013 Comological parameters base‐model results Planck 2013 A different way to use the CMB • Compared to the sensitivity of current surveys, the CMB  is a bright background light.  • Most of the matter in the Universe is transparent for CMB photons.  • Important interaction only with molecules (resonant scattering & absorption) and ionized gas (Thomson scattering & Inverse Compton scattering) – Molecules form very late and mainly in galaxies, so occupy a  very small solid angle. Their interaction with the CMB is masked by dominant galactic emission.  – Ionized gas, instead, is present at all scales, including the  largest structures: clusters, superclusters, filaments: i.e. LSS.  Missing baryons !  • So we expect to be able to map the ionized matter in the  universe observing the fine‐scale anisotropy of the CMB  induced by the ionized gas. Horologium-Reticulum Shapley Sloan great wall Pisces-Cetus Galaxies are clustered, and form filaments and foils, surrounding large voids. Clusters of galaxies are at filaments/foils crossings. The picture above represents two slices of the universe, where the position of about 220000 galaxies are marked as blue dots. A few superclusters are also labeled. Clusters of galaxies • Gravitationally bound structures • Composed of  – dark matter – ionized gas  – Galaxies/Star with: • Mtot ~ 1015 Msun • MDM ~ 10 Mgas • Mgas ~ 10 Mstars A2218 Dark matter in clusters A1689 Hot gas in clusters • The potential well of the cluster is so deep that gas  falling into the  well heats up to millions of K :  • kT ~ 10 keV • For this reason galaxy clusters are powerful X‐ ray emitters.  The center of the Phoenix Cluster, in X‐rays, and in visible light • • • • S-Z Inverse Compton Effect for CMB photons against charged particles in the hot gas of clusters Cluster optical depth: τ=nσl Incoming CMB photons Incoming CMB photons l = a few Mpc = 1025 cm n < 10-3 cm-3 σ = 6.65x10-25 cm2 So τ = nσl < 0.01 : there is a 1% likelihood that a cluster CMB photon crossing the cluster is scattered by an electron Eelectron >> Ephoton, so the electron transfers energy to the photon. To first order, the energy gain of the photon is Δν 5keV kTe = ≈ = 0.01 2 500keV ν mec • The resulting CMB temperature anisotropy is ΔT Δν ≈τ ≈ 0.01 × 0.01 = 10−4 T ν Sunyaev R., Zeldovich Y.B., 1972, Comm. Astrophys. Space Phys., 4, 173 Birkinshaw M., 1999, Physics Reports, 310, 97-195 Scattered CMB photons Brightness Sunyaev‐Zeldovich Effect A decrement at low frequencies ( <217GHz ) All photons increase their energy,  but the number is conserved.   The result is a shift of the spectrum of the CMB in the direction of rich clusters of galaxies An increment at high  frequencies ( > 217GHz ) frequency Relevant reviews : • Sunyaev, R. A.; Ya. B. Zel'dovich (1970). "Small-Scale Fluctuations of Relic Radiation". Astrophysics and Space Science 7: 3. • Rephaeli, Y. (1995). "Comptonization Of The Cosmic Microwave Background: The Sunyaev-Zeldovich Effect". Annual Review of Astronomy and Astrophysics 33 (1): 541–580 • Birkinshaw, Mark (1999). "The Sunyaev Zel'dovich Effect". Physics Reports 310 (2–3): 97. arXiv:astroph/9808050. • John E. Carlstrom, Gilbert P. Holder, Erik D. Reese, Cosmology with the Sunyaev-Zel'dovich Effect, Ann.Rev.Astron.Astrophys. 40:643-680, (2002) arXiv:astro-ph/0208192 thermal SZ hν x= kTCMB Very characteristic spectral behaviour amplitude Integral of gas pressure along the line of sight through the cluster • The boost in the energy of each scattered photon is small, and the fraction of boosted photons is also small, so the shift of the spectrum is tiny. • We need differential instruments to measure it ! ISD Δ emission, 18K 6 kJy/sr @ 150 GHz Thermal SZ Kinematic SZ Non-Thermal SZ Thermal SZ The Sunyaev‐Zeldovich Effect • Being produced by scatterings, the S‐Z signal amplitude does not depend on the distance (redshift) of the cluster • Depends linearly on the density of the gas  • The X‐ray brightness, instead, decreases significantly with distance and gas density (depends on the density  squared) X‐ray S‐Z X‐ray S‐Z X‐ray S‐Z Angular Size of a cluster • However, the apparent size of a cluster does depend on the distance. θ= L DA and, for a flat geometry 1 c 1 daˆ DA = 1 (1 + z ) Ho ∫(1+ z ) aˆ 2 [Ω Mo aˆ −3 + Ω Λ ]1/ 2 • Distant clusters have arcmin sizes. Angular resolution is required. At 150 GHz, FWHM(arcmin) = 8.4 / d(m), so several-meters-diameter telescopes are needed. Simulations:  e.g. Da Silva et al. astro‐ph/0011187 ΩΛ=0.7 ΩΛ=0.0 Differential measurements are badly needed ! 0.1 ΔI = I1 − I 2 DIFFERENTIAL SIGNAL I1 -0.1 I2 COMMON-MODE SIGNALS • The instrument should compare a LOS through the cluster to a LOS outside the cluster, and should reject common-mode signals many orders of magnitude larger than the SZ brightness gradient. Anisotropy measurementsI • Old-fashioned method: beamswitch, modulation & synchronous demodulation 1 I2 wobbling mirror detector common-mode only on-cluster Measured signal AC-filtered signal off-cluster SZ signal, recovered via synchronous demodulation Anisotropy measurements • I(t3) Modern method: scanning telescopes (with low-noise fast detectors, AC-couple and sample fast): BOOMERanG, Archeops, …, Planck, BICEP… B150 • I(t2) Sky-scanning detector Brightness reaching the detector during the scan I(t1) SZ common mode • • recorded signal: commonmode is gone, but data are filtered and do not follow exactly the incoming brightness. For bolometers, complex transfer function, deconvolution needed (Piacentini’s talk) 0 scan angle V 0 SZ effect of filter time Tenerife (1990) : beam switch BOOMERanG (2000) WMAP (2003) Sc a nn in g Planck (2013) te le sc op es SZ measurements OVRO • Single‐beam radio‐telescopes and radio‐ interferometers (70s) • Pronaos (balloon, first detection of positive  effect) • Single‐beam mm‐wave telescopes (MITO@TG,  SUZIE@CSO, multiband single‐pixel  bolometers) • Targeted observations of rich clusters • Now: Bolometer Arrays at large telescopes: – – – – South Pole Telescope APEX ACT Green Banks (Mustang) • Systematic blind‐surveys of large areas • WMAP & Planck (satellites, multiband) • Forthcoming: SZ‐Spectrometers Pronaos MITO CSO SZ measurements, today • Large telescopes (10m class) – Atacama Cosmology Telescope – APEX‐SZ – South‐Pole Telescope λ 1.2 × 2 mm FWHM ≈ 1.2 = ≈ 1' D 8000 mm • Arrays of TES Bolometers (cheap to replicate, low power dissipation in the  cryostat) – order of 1000  pixels ACT APEX-SZ SPT Sky coverage of the SPT SZ survey (astro‐ph/1210.7231) 2500 sq. deg = 9Mpixels ! Discovery (!) of new clusters • SPT [~1.1' @150GHz] – 244 SZ sources – Williamson+2010, Reichardt+2013 • ACT [~1.4' @148GHz] – 91 SZ sources – Marriage+2010, Menanteau+2011, Hasselfield+2013 0.5 mm PWV ISD Δ emission, 18K 6 kJy/sr @ 150 GHz Thermal SZ Kinematic SZ Non-Thermal SZ Thermal SZ 0.5 mm PWV ISD Δ emission, 18K 6 kJy/sr @ 150 GHz Thermal SZ Kinematic SZ Non-Thermal SZ Thermal SZ Windows available for gorund-based surveys (only in excellent sites !) Planck is a very ambitious experiment. It carries a  complex CMB  experiment (the  state of the art, a  few years ago) all the way to L2,  improving the  sensitivity wrt WMAP by at least a factor 10, extending the  frequency coverage towards high frequencies by a factor about 10 z NEPb = 15 aW/Hz1/2 -> 70 μK/Hz1/2 Total NET (bolo+photon) = 85 μK/Hz1/2 Planck LFI + HFI: 7 bands, including high frequencies A2319 As seen by Planck All-sky Sunyaev-Zeldovich clusters • Planck multiband observations of SZ clusters over the full sky: 1227 cluster and candidates detected ~200 newly discovered 683 known clusters ~344 candidate clusters • It is a shallow survey, most sensitive to massive clusters, mostly at moderate redshifts • Shallow full-sky observations from Planck and deeper observations on selected regions from SPT, ACT complement each-other quite nicely. Planck early results VIII, Planck 2013 results XXIX • • • Half of the baryons known to be present in the universe (from nucleosynthesis estimates) are  missing (i.e. have not been detected in emission, nor in  absorbtion). A possible physical state of  baryons escaping detection in the  radio, IR, visible, X rays domains, is a ionized medium with low density  or warm temperature.   In principle, this is detectable in  the microwaves, because it produces a SZ effect. The Planck survey, observing the  whole sky, has also observed couples of galaxy clusters, and  there is evidence for filaments of  gas connecting two of the couples (astro‐ph/1208.5911). This is hot  gas, probably heated by shocks,  but the density is low, so X‐ray emission is also very low.  It might be one step towards the solution of the missing baryons problem.  ROSAT cts/s • SZ 106Y Filaments X rays SZ-X scaling relation Cosmology from SZ clusters counts • Well characterised sample (189 clusters @S/N≥7 on 65% cleanest part of sky) • Completeness function of filter size & position on the sky • Tension at ~3σ level on σ8 from SZ counts vs CMB • Agreement can be recovered if – Missing ~half of the massive clusters – (1-b)~0.55, i.e. true mass higher by ~50% – More complex bias dependence – massive neutrino Σmν~0.2ev – Combination (calibration, high b, Σmν>0) How to improve • Better spectral coverage • Better angular resolution • Possibly together … • Photometric observations of the SZ can be significantly biased, when there are less spectral channels than free parameters. • Components, LOS through a rich cluster: ThSZ NThSZ pmin , Amp KSZ CMB ISD Td , τd ….(β) • Photometric observations of the SZ can be significantly biased, when there are less spectral channels than free parameters. • Components, LOS through a rich cluster: ThSZ NThSZ pmin , Amp KSZ At least, 8 independent parameters ! CMB ISD Td , τd ….(β) The final solution: spectroscopic measurements of the SZ • Requirements: – Wide spectral coverage (in principle 100 to 1000 GHz) – Modest spectral resolution (λ/Δλ = 100 to 1000) – Differential input, high rejection of common mode signal (CMB is common mode and is 2750000 μK, cluster signal is differential and can be as low as 10 μK). – Imaging instrument, resolution ad high frequency comparable to SPT 150 GHz (1 arcmin). – Wide field of view to image the whole cluster and have a clean reference area to compare OLIMPO (PI S. Masi, La Sapienza, Roma) Test specchio primario 2.6m ‐ f/0.5  0.3K cryostat (made in Sapienza) 65L superfluid 4He 70L liquid N 40LSTP 3He refrigerator 50L experimental volume Hold time – 15 days @ 0.3K 4He plate 4He shield 5th mirror 348GHz 483GHz 148GHz 215GHz tertiary 4th mirror,  Lyot‐stop & cal‐lamp From telescope window filters OLIMPO: Cold Optics and Arrays dychroics OLIMPO: Low‐frequency arrays (140 GHz & 220 GHz) OLIMPO: Low‐frequency arrays (140 GHz & 220 GHz) Dati da presentazione gualtieri The instrument is based on a double Martin Pupplett Interferometer configuration to avoid the loss of half of the signal. Olimpo Telescope A wedge mirror splits the sky image in two halves IA and IB, used as input signals for both inputs of the two FTS’s. outgoing fields : E FTSII E FTSI ⎛ Bx cos(δ / 2) + i Ay sin(δ / 2) ⎞ ⎟⎟ = ⎜⎜ 0 ⎝ ⎠ 0 ⎞ ⎛ ⎟ ⎜ =⎜ ⎟ B δ i A δ cos( / 2 ) + sin( / 2 ) y x ⎠ ⎝ Olimpo Cryostat Optical optimization has been performed using ZEMAXTM software optimizing the optical quality in the full FOV of OLIMPO. Cold Optics FTS The instrument was designed to fit the available room in between the primary mirror and the cryostat, a 75x75x30 cm3 box.(A.Schillaci, astro-ph/1402.4091 ) Expected performance of OLIMPO for the first flight (1 out of 40 clusters to be observed) Ideal ground-based 4-bands photometer OLIMPO spectrometer 0.75m Telescope / primary mirror DFTS cryostat / detectors arrays Main components of OLIMPO integrated for the first time, as of yesterday @ 8 PM …. • • • NASA‐CSBF has flown balloons around the  south pole for many years.  We have flown long duration stratospheric balloons around the North Pole launching from Longyearbyen (Svalbard) both in the  summer (heavy litf payloads) and in winter (pathfinders) [see Peterzen, S., Masi, S., et  al., Mem. S. A. It., 79, 792‐798 (2008)]  In this way CMB experiments can access most of the northern sky in a single flight,  – without contamination from the sun in the  sidelobes – within a cold and very stable environment – Accumulating more than 10 days of  integration at float (38 km altitude). Top: Ground path of a flight performed in June 2007. Bottom right: Ground path of a small pathfinder test flight performed in January 2011, in the middle of the polar night. The eastward trajectory is evident. Bottom left: Launch of a heavy-lift balloon from the Longyearbyen airport (Svalbard Islands, latitude 78oN). Polar flights June 2014 ! • Larger telescopes • Wider frequency coverage РадиоАстрон Millimetron ASC Moscow ROSCOSMOS • Antenna diameter: 10 m • Range of wavelengths: 0.01 – 20 mm • Bolometric sensitivity (λ0.3mm, 1h integration): 5x10-9 Jy • Interferometry sensitivity (λ0.5mm, 300s integration, 16GHz bw) : 10-4 Jy • Interferometer beam: 10-9 arcsec MILLIMETRON            We have been assigned a  large sector of  the  focal plane to insert a  low.resolution differential spectrometer.  ASI Phase‐A study MILLIMETRON            Dr. Alessandro Schillaci ASI Phase‐A study 21/09/2012           MILLIMETRON            ASI Phase‐A study Oscillating pantograph (negligible dissipation)  for cryogenic delay lines. Dr. Alessandro Schillaci 21/09/2012           3 hours of observations of a rich cluster with a DFTS on Millimetron Absolutely outstanding. USING A PHOTON NOISE LIMITED BOLOMETER IN THE COLD ENVIRONMENT OF L2 WITH 4K TELESCOPE 3h integration on the same LOS through a rich cluster P. de Bernardis, et al., Astronomy and Astrophysics, 538, A86 (2012) Oppure ammassi speciali … 1ES0657-556 7.5 ’ Isolating SZDM (at 223 GHz) Mχ = 20 GeV Mχ = 40 GeV Mχ = 80 GeV The SZE from the hot gas disappears at x0,th (∼ 220-223 GHz) while the SZDM expected at the locations of the two DM clumps remains negative and with an amplitude and spectrum which depend on Mχ. [Colafrancesco, de Bernardis, Masi, Polenta & Ullio 2006] Perseus Cluster In X‐rays: Hot gas  with Cavities, Shocks … 4.7’ • Very useful to study the internal structure of clusters (shocks, cavities, cooling flows …) The 100 m Green Bank Telescope (USA) has a W‐band array (Mustang) We have the 64m Sardinia Radio Telescope, and we are considering to install a  DFTS for the W‐band at the focus.  λ 1.2 × 3 mm FWHM ≈ 1.2 = ≈ 11" D 64000 mm 64 m • • XXXL telescopes & SZ 0m 0 1 GBT Measurements of CMB polarization CMB Polarization – Why ? • Linear Polarization of CMB photons is induced via Thomson scattering by quadrupole anisotropy at recombination (z=1100, t =1.2x1013s). + - - • In turn, quadrupole anisotropy is induced by – Density perturbations (scalar relics of inflation) producing a curl-free polarization vectors field (E-modes) – Gravitational waves (tensor relics of inflation) producing both curl-free and curl polarization fields (B-modes) + E-modes • No other sources for a curl polarization field of the CMB at large angular scales: • B-modes are a clear signature of inflation. B-modes - • 0.1 μK is extremely small, and is embedded in a common mode signal >107 times larger ! • Which is the best way to measure such a small signal ? • Differential measurements ! THE SIMPLEST BOLOMETRIC POLARIMETER ΔV (t ) = V⊥ (t ) − V// (t ) = = ℜ⊥ I ⊥ [θ (t )] − ℜ// I // [θ (t )] PSD[V⊥ (t )] PSD[ΔV (t )] B2K, Planck and BICEP2 work this way Jones et al. A&A 470, 771–785 (2007) Polarization modulators Pisano, G., Ng, M.W., Haynes, V., Maffei, B., “A Broadband Metal-Mesh Half-Wave Plate for Millimetre Wave Linear Polarisation Rotation”, Submitted to PIER JEMWA (2012) Linear Polarimeter source polarizer θ Intensity detector • A polarimeter is a device able to detect polarized light and measure its polarization characteristics. • The simplest polarimeter we can imagine is a linear polarimeter, which can be built with a rotating polarizer in front of an intensity detector. • An intensity detector is represented by a Stokes vector D=(1,0,0,0). The power detected by the detector from an optical beam with Stokes vector S is simply w=DS=So • If we put a polarizer in front of the detector, the polarizer is called analyzer, and the power detected will be w(θ) =DMP(θ)S Linear Polarimeter source Polarizer (analyzer) θ Intensity detector c2 Δ s2 Δ ⎛ Σ ⎜ 2 2 1 ⎜ c 2 Δ c2 Σ + s 2 X s 2 c 2 ( Σ − X ) w = DM P (θ ) S = (1,0,0,0) ⎜ 2 s2 Δ s2 c2 (Σ − X ) s22 Σ + c22 X ⎜ ⎜ 0 0 0 ⎝ w = 1 2 (Σ S o + Δ S 1 cos 2 θ + Δ S 2 sin 2 θ 0 ⎞⎛ S o ⎞ ⎟⎜ ⎟ 0 ⎟⎜ S1 ⎟ ⇒ ⎜ ⎟ ⎟ 0 S2 ⎟⎜ ⎟ X ⎟⎠⎜⎝ S3 ⎟⎠ ) This polarimeter is not sensitive to circular polarization (no S3). It is sensitive to linear polarization (S1 and S2) and to unpolarized light (So). If the polarizer is ideal: Δ = 1 ; Σ = 1 ; X = 0 w= 1 2 (S o + S 1 cos 2θ + S 2 sin 2θ ) Linear Polarimeter • If we are interested to the linear polarized component only, we can rotate continuously the polarizer: θ=ωt and look only for the AC signal at frequency 2ω. • This allows to reject the unpolarized component, even if it is dominant, and to remove all the noise components at frequencies different than 2ω (synchronous demodulation). source ω Rotating analyzer Intensity detector (Σ S o + Δ S 1 cos 2ω t + Δ S 2 sin 2ω t ) V (t ) = Rw (t ) + N (t ) = 12 R [Σ S o + Δ (S1 cos 2ω t + S 2 sin 2ω t )] + N (t ) w= 1 2 detector responsivity constant signal (DC) modulated signal (AC) noise (AC) Linear Polarimeter source Log P(ω) <…>T ω Detector Rotating analyzer Rw+N Ref(2ω) C A x A(Rw+N)AC A[Rw(2ω)+N(Δω)] Demodulated signal noise signal σ2 = ∫ P( ω)dω Δω Δω-=1/T 1/RC 2ω Log ω R How do we separate S1 and S2 V ( t ) = Rw ( t ) + N ( t ) = 1 2 R [Σ S o + Δ (S 1 cos 2ω t + S 2 sin 2ω t )] + N ( t ) • Neglecting the stochastic effect of noise (we integrate enough that N becomes negligible) and of the constant term (which we remove with the AC decoupling) V (t ) = Rw (t ) = 12 R [Δ (S1 cos 2ω t + S 2 sin 2ω t )] • We measure V and we want to estimate S1 and S2. We can use two reference signals, out of phase by T/8 and synchronously demodulate with them: How do we separate S1 and S2 T T ⎡ ⎤ 1 1 RΔ X = T ∫ V (t ) sin 2ωtdt = 2 T ⎢S1 ∫ cos 2ωt sin 2ωtdt + S2 ∫ sin 2ωt sin 2ωtdt⎥ 0 0 ⎦ ⎣ 0 T T T ⎡ ⎤ 1 1 RΔ Y = T ∫ V (t ) cos 2ωtdt = 2 T ⎢S1 ∫ cos 2ωt cos 2ωtdt + S2 ∫ sin 2ωt cos 2ωtdt⎥ 0 0 ⎣ 0 ⎦ T X = Y = 1 8 1 8 RΔS2 R Δ S1 • So the double linear polarimeter is insensitive to So and it is easy to calibrate. • Is this a troubleless instrument ? No ! • It is inefficient (factor 1/8 from modulation and demodulation) • It can be microphonic. • And, as all polarimeters, needs a telescope. Other polarimeters • Wave plates can be used to build polarimeters sensitive to linear or to circular polarization. ⎛1 0 0 0 ⎞ ⎜ ⎟ o • Half-wave plate (φ=180 ) ⎜0 1 0 0 ⎟ HWP = ⎜ 0 0 −1 0 ⎟ ⎜ ⎟ ⎜ 0 0 0 − 1⎟ ⎝ ⎠ • Quarter-wave plate (φ=90o) ⎛1 ⎜ ⎜0 QWP = ⎜ 0 ⎜ ⎜0 ⎝ 0 1 0 0 0 0 0 1 0 ⎞ ⎟ 0 ⎟ − 1⎟ ⎟ 0 ⎟⎠ Polarimeter with HWP HWP source θ Polarizer Intensity detector W = D × {PH × R(− θ )× HWP × R(θ )}× S ⎛1 ⎜ ⎜0 HWP = ⎜ 0 ⎜ ⎜0 ⎝ ⎛I⎞ ⎜ ⎟ ⎜Q⎟ S =⎜ ⎟ U ⎜ ⎟ ⎜V ⎟ ⎝ ⎠ 0 0 1 0 0 −1 0 0 0⎞ ⎟ 0⎟ 0⎟ ⎟ −1⎟⎠ ⎛1 ⎜ ⎜1 PH = ⎜ 0 ⎜ ⎜0 ⎝ D = (1 0 0 0) 1 0 1 0 0 0 0 0 0⎞ ⎟ 0⎟ 0⎟ ⎟ 0 ⎟⎠ ⎛1 0 ⎜ ⎜ 0 c2 R(θ ) = ⎜ 0 − s2 ⎜ ⎜0 0 ⎝ ⎛1 0 ⎜ ⎜ 0 c2 R(− θ ) = ⎜ 0 s2 ⎜ ⎜0 0 ⎝ 0 0⎞ ⎟ 0⎟ 0⎟ ⎟ 1 ⎟⎠ 0 − s2 c2 0 0⎞ ⎟ 0⎟ 0⎟ ⎟ 1 ⎟⎠ 0 s2 c2 Polarimeter with HWP HWP source Polarizer θ Intensity detector W = D × {PH × R(− θ )× HWP × R(θ )}× S ⎛1 ⎜ ⎜0 ⎜0 ⎜ ⎜0 ⎝ 0 1 0 0 0 0 −1 0 0 ⎞⎛1 ⎟⎜ 0 ⎟⎜ 0 0 ⎟⎜ 0 ⎟⎜ −1⎟⎠⎜⎝ 0 ⎛1 ⎜ ⎜0 ⎜0 ⎜ ⎜0 ⎝ 0 0 c2 − s2 s2 c2 0 0 0⎞⎛1 ⎟⎜ 0⎟⎜0 0⎟⎜0 ⎟⎜ 1⎟⎠⎜⎝0 0⎞ ⎛1 ⎟ ⎜ 0⎟ ⎜ 0 =⎜ ⎟ 0 0 ⎟ ⎜ 1⎟⎠ ⎜⎝ 0 0 0 0⎞ ⎟ c2 0 0 ⎟ 0 − c2 0 ⎟ ⎟ 0 0 −1⎟⎠ 0 0 0 ⎞ ⎛1 ⎟ ⎜ c2 0 0 ⎟ ⎜0 =⎜ ⎟ 0 − c2 0 0 ⎟ ⎜ 0 0 −1⎟⎠ ⎜⎝0 0 0 0⎞ ⎟ 2 c2 0 0 ⎟ 0 − c22 0 ⎟ ⎟ 0 0 −1⎟⎠ 0 0 c2 − s2 s2 c2 0 0 Polarimeter with HWP HWP source Polarizer θ Intensity detector W = D × {PH × R(− θ )× HWP × R(θ )}× S ⎛1 ⎜ ⎜1 ⎜0 ⎜ ⎜0 ⎝ 1 0 0⎞⎛ 1 ⎟⎜ 1 0 0⎟⎜ 0 0 0 0⎟⎜ 0 ⎟⎜ 0 0 0⎟⎠⎜⎝ 0 0 c22 0 0 0 − c22 0 0 0 ⎞ ⎛1 ⎟ ⎜ 0 ⎟ ⎜1 =⎜ ⎟ 0 0 ⎟ ⎜ −1⎟⎠ ⎜⎝ 0 c22 0 0⎞ ⎟ 2 c2 0 0⎟ ⎟ 0 0 0⎟ 0 0 0⎟⎠ ⎛ I + c22Q⎞ ⎟ ⎜ 2 ⎜ I + c2 Q⎟ 1 2 W = 12 (1 0 0 0)⎜ = + I Qc 2 ⎟ 2 0 ⎟ ⎜ ⎜ 0 ⎟ ⎠ ⎝ ( ) ⎛1 ⎜ ⎜1 ⎜ ⎜0 ⎜0 ⎝ c22 0 0⎞⎛ I ⎞ ⎛ I + c22Q⎞ ⎟ ⎟⎜ ⎟ ⎜ 2 2 c2 0 0⎟⎜ Q ⎟ ⎜ I + c2 Q⎟ ⎟ ⎟⎜U ⎟ = ⎜ 0 0 0⎟ 0 ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎜ ⎟ 0 0 0⎠⎝V ⎠ ⎝ 0 ⎟⎠ ( W = 12 I + Q cos2 2θ f = 4 fo ) QUBIC Bolometric Interferometer in Dome-C (Antarctica) 2016 Rotating waveplate modulator 6 modules, 140 GHz, >1000 det. APC Paris & international collaborators; important Italian contribution (PNRA, S. Masi) Target r=0.01 See Piat et al. J Low Temp Phys (2012) 167:872–878 EBEX EBEX Focal Plane 738 element array 141 element hexagon Single TES Lee, UCB 250 150 420 150 150 150 3 mm 250 5 cm • Total of 1476 detectors • Maintained at 0.27 K • 3 frequency bands/focal plane Slide: Hanany • • • • G=15-30 pWatt/K NEP = 1.4e-17 (150 GHz) NEQ = 156 μK*rt(sec) (150 GHz) τ = 3 msec, William Jones Princeton University for the Spider Collaboration Suborbital Polarimeter for Inflation Dust and the Epoch of Reionization The Path to CMBpol June 31, 2009 Spider: A Balloon Borne CMB Polarimeter Suborbital Polarimeter for Inflation Dust and the Epoch of Reionization • Long duration (~30 day cryogenic hold time) balloon borne polarimeter • Surveys 60% of the sky each day of the flight, with ~0.5 degree resolution • Broad frequency coverage to aid in foreground separation • Will extract nearly all the information from the CMB E-modes • Will probe B-modes on scales where lensing does not dominate • Technical Pathfinder: solutions appropriate for a space mission Carbon Fiber Gondola Six single freq. telescopes 30 day, 1850 lb, 4K / 1.4 K cryostat Attitude Control • flywheel • magnetometer • rate gyros • sun sensor Pointing Reconstruction • 2 pointed cameras • boresight camera • rate gyros Flight Computers/ACS • 1 TB for turnaround • 5 TB for LDB SWIPE STRIP The Large Scale Polarization Explorer LSPE in a nutshell • The Large-Scale Polarization Explorer is – – – – • • • • • a spinning stratospheric balloon payload flying long-duration, in the polar night aiming at CMB polarization at large angular scales using polarization modulators to achieve high stability Frequency coverage: 40 – 250 GHz (5 channels) Angular resolution: 1.5 – 2.3 deg FWHM Sky coverage: 20-25% of the sky per flight Combined sensitivity: 10 μK arcmin per flight Two instruments: STRIP (hear Daniele tomorrow) and SWIPE SWIPE • • • The Short Wavelength Instrument for the Polarization Explorer Uses overmoded bolometers, trading angular resolution for sensitivity Sensitivity of photon-noise limited bolometers vs # of modes: Number of modes actually coupling to the bolometer absorber LSPE - SWIPE eff = 0.25 Bolometric D lens = 0.4 m Instrument F= 0.8 m N modes (geom) = 15 25 40 f (GHz) 90 145 220 λ (mm) 3.3 2.1 1.4 N det = 37 58 83 FWHM (deg) = 2.4 1.9 1.6 NET (μK/sqrt(Hz) ) = 15 25 30 NET Focal Plane (μK/sqrt(Hz))= 2.5 3.3 3.2 Target : r = 0.01 in a single long-duration flight in the polar night Covering large angular scales Cryogenic Rotating HWP (Salatino et al. 2012) Very much complementary to SPIDER, EBEX Flight in 2016 Life is hard…. • … and for CMB polarization experimentalists is even harder …. Winter balloon flights demonstrated from Longyearbyen (78N)