1. Current and Future SZ Surveys Sunil Golwala California Institute of Technology July 7, 2001

2. Current and Future SZ Surveys Sunil Golwala/2001/07/07 2 Overview The Sunyaev-Zeldovich effect in galaxy clusters Science with blind SZE surveys Interferometers and bolometer arrays Calculating expected sensitivities Laundry list of current and future instruments: specifications and sensitivities Summary for the near future Thanks to all the instrument teams for specs and numbers!

3. Current and Future SZ Surveys Sunil Golwala/2001/07/07 3 The Sunyaev-Zeldovich Effect in Galaxy Clusters Thermal SZE is the Compton up-scattering of CMB photons by electrons in hot, intracluster plasma last scattering surface z ~ 1100 CMB photons T = (1 + z) 2.725K galaxy cluster with hot ICM z ~ 0 - 3 scattered photons (hotter) observer z = 0 ∆T CMB /T CMB depends only on cluster y ~ line-of-sight integral of n e T e. Both ∆T CMB and T CMB are redshifted similarly  ratio unchanged as photons propagate and independent of cluster distance thermal SZE causes nonthermal change in spectrum. CMB looks colder to left of peak, hotter to right Sunyaev & Zeldovich (1980)

4. Current and Future SZ Surveys Sunil Golwala/2001/07/07 4 Current SZE Data Beautiful images of the SZE in clusters at a large range of redshifts from Carlstrom group using 1 cm (30 GHz) receivers at BIMA and OVRO But sensitivity of this and other instruments too poor for blind surveys Carlstrom et al, Phys. Scr., T: 148 (2000) SZE only, ~15 - 40 µK/beam rms CL0016+16, SZE + X-ray (ROSAT PSPC)

5. Current and Future SZ Surveys Sunil Golwala/2001/07/07 5 The Sunyaev-Zeldovich Effect in Galaxy Clusters Proportional to line integral of electron pressure: Fractional effect is independent of cluster redshift Thermal SZE causes unique spectral distortion of CMB  “hole in the sky” to left of peak Simplifies in Rayleigh-Jeans limit  But same spectrum as CMB in this limit

6. Current and Future SZ Surveys Sunil Golwala/2001/07/07 6 Secondary CMB Anisotropy The thermal SZE is the dominant contributor to CMB secondary anisotropy (beyond the damping tail) = thermal SZE from LSS at low z Probes baryon pressure distribution, early energy injection Spectrally separable from primary anisotropy Other effects (kinematic/Ostriker-Vishniac, patchy reionization) at much lower level, same spectrum as primary Predictions for secondary anisotropy: Springel et al, Ap. J.,549: 681 (2000) Seljak et al, PRD, 63: 063001 (2001) Limits (95%CL): ATCA: Subrahmanyan et al, MNRAS, 315: 808 (2000). BIMA: Dawson et al, Ap. J., 553: L1 (2001) Ryle: Jones et al, Proc. PPEU (1997). Springel Seljak BIMA Ryle ATCA

7. Current and Future SZ Surveys Sunil Golwala/2001/07/07 7 Unbiased Cluster Detection via the SZE Central decrement is bad observable because of dependence on core characteristics Integrated SZE over cluster face more robust and provides largely z- independent mass limit (Barbosa et al (1996), Holder et al (2000), etc.)  M 200 is virial mass (inside R 200 ), equal to volume integral of n e /f ICM   T e  n is electron-density weighted electron temperature  Under “fair sample” assumption, f ICM given by BBN value  d A 2 factor arises from integration  weak z-dependence arises from fortuitous cancellation: d A 2 factor tends to reduce flux as z increases (1/r 2 law) But for a given mass, a cluster at high redshift has smaller R 200 and hence higher  T e  n : M 200 ~  (R 200 )3,  increases with z, so R 200 must decrease to get same M 200, and T ~ M 200 /R 200

8. Current and Future SZ Surveys Sunil Golwala/2001/07/07 8 Unbiased Cluster Detection via the SZE Holder, Mohr, et al (2000) modeled the mass limit of an interferometric SZE survey (synth. beam ~ 3’) using simulations Bears out expectation of weak dependence of mass limit on z: SZE provides an essentially z- independent selection function; it allows detection of all clusters above a given mass limit v. different selection function from optical/x-ray surveys For any survey, careful modelling will be required to determine this precisely, understand uncertainties Holder et al, Ap. J., 544:629 (2000) limiting mass vs. z for an interferometric survey for different cosmologies

9. Current and Future SZ Surveys Sunil Golwala/2001/07/07 9 Science with Blind SZE Surveys Galaxy clusters  largest virialized objects  so large that formation not severely affected by “messy” astrophysics – star formation, gas dynamics  mass, temperature, radius understood within simple spherical tophat collapse model  good probe of cosmological quantities:  power spectrum amplitude (  8 )  total matter density (  m )  volume element (  tot )  growth function (  m,   )  with higher statistics, equation of state p = w , dependence of w on z  (see talks by Holder, Kamionkowski) Non-Gaussianity: clusters are high-significance excursions, sensitive to non-Gaussian tails

10. Current and Future SZ Surveys Sunil Golwala/2001/07/07 10 Science with Blind SZE Surveys Constraining cosmological parameters  Best done with redshift distribution  Separation at high redshift between OCDM and  CDM due to different growth functions, volume element (more high-z volume in open universe)  Normalization of redshift distribution v. sensitive to  8 (= power spectrum normalization) Reichardt, Benson, and Kamionkowski, in preparation

11. Current and Future SZ Surveys Sunil Golwala/2001/07/07 11 Science with Blind SZE Surveys Looking for non-Gaussianity  assume a cosmology  non-Gaussianity changes z- distribution: if tail is longer, get more clusters at high z Reichardt, Benson, and Kamionkowski, in preparation

12. Current and Future SZ Surveys Sunil Golwala/2001/07/07 12 Where is it possible to do high-l measurements (from the ground)? Rayleigh-Jeans tail (10s of GHz)  atmosphere not a big problem  HEMT receivers provide good sensitivity  have to contend with radio pt srces, but subtraction demonstrated (DASI, CBI, BIMA, ATCA) near the peak (100-300 GHz)  least point source contamination  have to contend with sky noise  bolometric instruments provide best sensitivities in this band shorter wavelengths  sky noise horrendous  IR point sources difficult (impossible?) to observe and subtract SZE Instrument Parameter Space point sources (radio, IR) dust free-free and synchrotron

13. Current and Future SZ Surveys Sunil Golwala/2001/07/07 13 Techniques Interferometers pros:  many systematics and noises do not correlate (rcvr gains, sky emission)  phase switching and the celestial fringe rate can be used to reject offsets, 1/f noise, non-celestial signals (if not comounted)  individual dish pointing requirements not as stringent as for single dish (if not comounted)  HEMT rcvrs, no sub-K cryogenics cons:  not natural choice for brightness sensitivity – must make array look like single dish to achieve  operating frequency, BW limited by rcvr technology, correlator cost Bolometers pros:  sensitivity, bandwidth  simplicity of readout chain  scalability (big FOV arrays) cons:  sub-K cryogenics  standard single-dish problems: spillover, sky noise, etc.  requires chopping or az scan to push signal out of 1/f noise

14. Current and Future SZ Surveys Sunil Golwala/2001/07/07 14 Noise sources: specified as noise-equivalent power (NEP), power incident on detector that can be detected at 1  in 1 sec, units of W√sec  detector noise: Johnson noise of thermistor, phonon noise, amplifier noise, etc.  BLIP noise: shot noise on DC optical load; present even if sky is perfectly quiet  sky noise: variations in sky loading These yield noise-equivalent flux density (NEFD): flux density (Jy) that can be detected at 1  in 1 sec; units of Jy√sec Beam size gives noise-equivalent surface brightness (NESB): units of (Jy/arcmin 2 )√sec Can then calculate noise-equivalent temperature (NET CMB ), units of (µK CMB /beam)√sec and finally, noise-equivalent y parameter (NEy), units of (1/beam)√sec Instantaneous Bolometer Sensitivity

15. Current and Future SZ Surveys Sunil Golwala/2001/07/07 15 Single-antenna noise sources: summed to give T sys  T rcvr (receiver noise): ~ bolometer detector noise, like a NEP  T sky (optical loading): due to DC optical load, but, unlike bolometers, this NEP scales with T sky, not as √T sky  √P sky because coherent receiver  sky noise: nonexistent unless imaging the atmosphere T sys and number of baselines n yield NEFD As for bolometers, calculate NET [(µK/beam)√sec] from NEFD  must assume well-filled aperture (uv) plane so it is valid to use simple Ω beam should include correction for central hole in uv plane, ignore for this  In RJ limit, simplifies greatly:  And using antenna theorem finally, NEy, units of (1/beam)√sec Instantaneous Interferometer Sensitivity

16. Current and Future SZ Surveys Sunil Golwala/2001/07/07 16 Mapping Speed Straightforward to calculate a mapping speed for a bolometer array Also pretty trivial for an interferometer and in RJ limit counterintuitive? Increased FOV hurts unless beam size also increased: fixed sensitivity spread over larger sky area Comparing mapping speeds: must be careful about beam size. Affects NET and FOV, though in different ways for bolometers and interferometers. Point source mapping speed? Only appropriate for large-beam experiments, hard to compare bec. SZ flux strong function of frequency.

17. Current and Future SZ Surveys Sunil Golwala/2001/07/07 17 New SZE Survey Instruments Now: ACBAR, BOLOCAM Soon (2003/2004): SZA, AMI, AMiBA Not so soon (>2004?): ACT, SP Bolo Array Telescope, etc. Numbers:  all numbers calculated with true CMB spectrum; i.e., not in RJ limit  For thermal SZ, best to compare y/beam sensitivity, since this can be compared at different frequencies. Could also use Y = area integral of y. NEY is like NEFD, except corrected for SZ spectrum. Using y assumes a beam-filling source, using Y assumes an unresolved source. y favors large-beam experiments, Y favors small-beam experiments, both impressions are artificial  Mass limits are those provided by each experiment, or in the literature. They are not consistent with each other! Further comment later.

18. Current and Future SZ Surveys Sunil Golwala/2001/07/07 18 ACBAR Instrument Specs Arcminute Cosmology Bolometer Array Receiver UCB, UCSB, Caltech/JPL, CMU 2m VIPER dish at South Pole spider-web bolometers at 240 mK 4 horns each at 150, 220, 270, 350 GHz 4.5’ beams at 150 GHz BW ~ 25 GHz N det Ω beam ~ 64 arcmin 2 chopping tertiary, 3 deg chop, raster scan in dec Unique multifrequency coverage: promises separation of thermal SZE and primary CMB 250mK filt & lens Corrugated feeds 4K filters & lenses Bolometers Thermal gap 274 219 150 345 GHz

19. Current and Future SZ Surveys Sunil Golwala/2001/07/07 19 ACBAR Sensitivity Achieved (2001), dominated by 4x150  NET = 440 µK CMB √sec (per row)  NEy = 150 x 10 -6 √sec (per row)  M T = 34 deg 2 (10 µK/beam) -2 month -1  M y = 2.8 deg 2 (10 -6 /beam) -2 month -1  M Y = 0.55 deg 2 (10 -5 arcmin 2 ) -2 month -1 Map 10 deg 2 in ~ 200 hrs (live) to  T rms ~ 10 µK/beam  y rms ~ 4 x 10 -6 /beam  Y rms ~ 8 x 10 -5 arcmin 2 2002: 4x150 + 12x280 focal plane  NEy = 95 x 10 -6 √sec (per row)  M y = 7.2 (10 -6 /beam) -2 month -1  M Y = 1.4 (10 -5 arcmin 2 ) -2 month -1  significant improvement in NEy from better-matched multifrequency coverage  Possibly ~2X better sensitivity if optical loading problem fixed Mapping speeds benefit from large beams, though also gives high mass limit (few x 10 14 M sun )

20. Current and Future SZ Surveys Sunil Golwala/2001/07/07 20 BOLOCAM Instrument Specs Caltech/JPL, Colorado, Cardiff 10.4m CSO on Mauna Kea Spider-web bolometer array at 300 mK 144 horns at 150, 220, 270 GHz (not simultaneous) 1’ beams at 150 GHz BW ~ 20 GHz N det Ω beam ~ 160 arcmin 2 drift scan + raster in dec, possible az. scan, raster in ZA Large number of pixels at high resolution – unique for SZ Multifrequency coverage, but at poorer sensitivity in other bands and delayed in time

21. Current and Future SZ Surveys Sunil Golwala/2001/07/07 21 BOLOCAM Sensitivity Expected, based on extrapolation from SuZIE 1.5:  NET = 1300 µK CMB √sec  NEy = 470 x 10 -6 √sec  M T = 6.8 deg 2 (10 µK/beam) -2 month -1  M y = 0.53 (10 -6 /beam) -2 month -1  M Y = 42 (10 -5 arcmin 2 ) -2 month -1 Map 1 deg 2 in ~ 100 hrs (live)  T rms ~ 10-15 µK/beam  y rms ~ 4 x 10 -6 /beam  Y rms ~ 0.4 x 10 -5 arcmin 2 Expectations consistent with achieved sensitivity in engineering run at 220 GHz Mapping speed degraded by small beams; but small beams yield low mass limit (~ 2-3 x 10 14 M sun )

22. Current and Future SZ Surveys Sunil Golwala/2001/07/07 22 SZ Array Instrument Specs SZ Array Chicago (Carlstrom), MSFC (Joy), et al 8 x 3.5m at 30 GHz NRAO HEMT receivers, ~10K noise, ~21K system noise 8 GHz digital correlator (in conjunction with OVRO) FOV FWHM ~ 10.5’, Beam FWHM ~ 2.25’? (unable to get definite number for beam, so scale from AMI) 1-year survey of 12 deg 2, part of time in heterogeneous mode later upgrade to 90 GHz

23. Current and Future SZ Surveys Sunil Golwala/2001/07/07 23 SZ Array Sensitivity Sensitivity and mapping speed for 8x3.5m array assuming 2.25’ beam:  NET = 730 (mK CMB /beam)√s  NEy = 140 (10 -6 /beam)√s  M T = 17 deg 2 (10 µK/beam) -2 month -1  M y = 4.7 (10 -6 /beam) -2 month -1  M Y = 15 (10 -5 arcmin 2 ) -2 month -1 Map 12 deg 2 in 1 yr at 75% eff.:  T rms ~ 2.8 µK/beam  y rms ~ 0.5 x 10 -6 /beam  Y rms ~ 0.3 x 10 -5 arcmin 2 HETEROGENEOUS BASELINES HAVE NOT BEEN INCLUDED HERE! They improve sensitivity to low masses (counteract beam dilution) Mass limit ~ 10 14 M sun, found by Monte Carlo in visibility space pt. src. subtraction – won’t need continuous monitoring, intermittent monitoring sufficient SZA + OVRO

24. Current and Future SZ Surveys Sunil Golwala/2001/07/07 24 AMI Instrument Specs Arcminute Microkelvin Imager MRAO/Cavendish/Cambridge group 10 x 3.7m at 15 GHz NRAO HEMT receivers, ~13K noise, ~25K system noise 6 GHz analog correlator FOV FWHM ~ 21’, Beam FWHM ~ 4.5’ concurrent point source monitoring by Ryle Telescope (8 x 13m), no heterogeneous correlation Expect to upgrade receivers to InP HEMTs, ~6K rcvr noise, ~18K system noise

25. Current and Future SZ Surveys Sunil Golwala/2001/07/07 25 AMI Expected Sensitivity Sensitivity and mapping speed:  NET = 470 (mK CMB /beam)√s  NEy = 90 (10 -6 /beam)√s  M T = 160 deg 2 (10 µK/beam) -2 month -1  M y = 47 deg 2 (10 -6 /beam) -2 month -1  M Y = 9.2 (10 -5 arcmin 2 ) -2 month -1 Map 36 deg 2 in 6 months at 75% eff.:  T rms ~ 2 µK/beam  y rms ~ 4 x 10 -6 /beam  Y rms ~ 1 x 10 -5 arcmin 2 Map 2 deg 2 in 6 months at 75% eff.:  T rms ~ 0.5 µK/beam  y rms ~ 0.1 x 10 -6 /beam  Y rms ~ 0.2 x 10 -5 arcmin 2 Mass limit ~ 10 14 M sun in deep survey As with ACBAR, mapping speed greatly helped by large beam, but also yields high mass limit (or long integration time and small area coverage for low mass limit) clusters detectable in simulated observations; note how redsfhit range increases as Y is lowered

26. Current and Future SZ Surveys Sunil Golwala/2001/07/07 26 AMiBA Instrument Specs Array for Microwave Background Anisotropy ASIAA + ATNF + CMU 19 x 1.2m at 90 GHz MMIC HEMT receivers under development in Taiwan, ~45K noise expected, ~75K system noise 20 GHz analog correlator FOV FWHM ~ 11’, Beam FWHM ~ 2.6’ Also: 19 x 0.3m for CMB polarization

27. Current and Future SZ Surveys Sunil Golwala/2001/07/07 27 AMiBA Expected Sensitivity Sensitivity and mapping speed:  NET = 590 (µK CMB /beam)√s  NET = 140 (10 -6 /beam)√s  M T = 28 deg 2 (10 µK/beam) -2 month -1  M y = 5 deg 2 (10 -6 /beam) -2 month -1  M Y = 8.9 (10 -5 arcmin 2 ) -2 month -1 3 different surveys (eff. = 50%):  deep: 3 deg 2 in 6 months to T rms ~ 0.2 µK/beam, y rms ~ 0.4 x 10 -6 /beam,Y rms ~ 0.3 x 10 -5 arcmin 2  med.: 70 deg 2 in 12 months to T rms ~ 0.6 µK/beam, y rms ~ 1.5 x 10 -6 /beam,Y rms ~ 1.1 x 10 -5 arcmin 2  wide: 175 deg 2 in 6 months to T rms ~ 1.4 µK/beam, y rms ~ 3.4 x 10 -6 /beam,Y rms ~ 2.6 x 10 -5 arcmin 2 Mass limits: 2, 4.5, 6.5 x 10 14 M sun pt. src. confusion much less at 90 GHz; will do survey to check src. counts, but expect confusion from low-flux clusters will be more important

28. Current and Future SZ Surveys Sunil Golwala/2001/07/07 28 ACT Instrument Specs Atacama Cosmology Telescope Princeton/Penn (Page, Devlin, Staggs) 6m off-axis dish with ground screen, near ALMA site 3 x 32x32 arrays of TES-based pop-up bolometers with multiplexed SQUID readout 150, 220, 265 GHz bands 1.7’, 1.1’, 0.9’ beam sizes 22’ x 22’ FOV azimuth scan of entire telescope l-space coverage from l ~ 200 to 10 4 Expected NETs: 300, 500, 700 µK CMB √s detector/BLIP limited T sky = 20K assumed sky noise expected to be negligible at l > 1000 in Chile

29. Current and Future SZ Surveys Sunil Golwala/2001/07/07 29 ACT Sensitivity Sensitivity and mapping speed:  NET = 300 (µK CMB /beam)√s  NEy = 115 (10 -6 /beam)√s  M T = 2600 deg 2 (10 µK/beam) -2 month -1  M y = 180 deg 2 (10 -6 /beam) -2 month -1  M Y = 1700 (10 -5 arcmin 2 ) -2 month -1 Huge mapping speed because of good sensitivity and large FOV: 100 deg 2 in 4 months at eff. = 25% to  T rms ~ 2 µK/beam  y rms ~ 0.7 x 10 -6 /beam  Y rms ~ 0.2 x 10 -5 arcmin 2  Will actually do significantly better because of multi-frequency coverage (not accounted for in above) Expected mass limit ~ 4 x 10 14 M sun (seems overly conservative!) Multi-frequency coverage promises excellent separation of thermal SZE and CMB-like secondary effects **Proposed, not yet funded**

30. Current and Future SZ Surveys Sunil Golwala/2001/07/07 30 South Pole Bolometer Array Telescope Chicago (Carlstrom et al, Meyer), UCB (Holzapfel, Lee), UCSB (Ruhl), CfA (Stark), UIUC (Mohr) 32x32 bolometer array, ~90% at 150 GHz, ~10% at 220 GHz 1.3’ beam at 150 GHz FOV: telescope: 1 deg array: ~ 17’ x 17’?

31. Current and Future SZ Surveys Sunil Golwala/2001/07/07 31 South Pole Bolometer Array Telescope Sensitivity and mapping speed:  NET = 250 (µK CMB /beam)√s  NEy = 90 (10 -6 /beam)√s  M T = 2000 deg 2 (10 µK/beam) -2 month -1  M y = 160 deg 2 (10 -6 /beam) -2 month -1  M Y = 4700 (10 -5 arcmin 2 ) -2 month -1 4000 deg 2 in 2 months live to  T rms ~ 10 µK/beam  y rms ~ 3.5 x 10 -6 /beam  Y rms ~ 0.7 x 10 -5 arcmin 2  multi-frequency coverage not so good, so has little effect Expected mass limit ~ 3.5 x 10 14 M sun Proposal into NSF-OPP

32. Current and Future SZ Surveys Sunil Golwala/2001/07/07 32 Summary and Scalings Plot of mapping speeds vs. beam FWHM for y and Y = area integral of y Overresolution: can correct for this by coadding adjacent pixels. Corrects y and Y mapping speeds by  src 2 and  src -2, respectively. Note: ratio of mapping speeds for two experiments scaled to same  src is independent of whether y or Y is used. Beam dilution: for y mapping speed, beam-filling source is assumed. If not, apparent y in beam is degraded by (  beam /  src ) 2, mapping speed by (  src /  beam ) 4 x-axis is  src for scaling lines,  beam for experiments (  src /  beam ) 4 (  src /  beam ) 2 (  src /  beam ) -2

33. Current and Future SZ Surveys Sunil Golwala/2001/07/07 33 Random Parting Thoughts Calculation of mass limits seems still to be highly scientist-dependent  Would be nice to have agreed-upon estimation method  Of course, some disagreement is inevetible and indicative of our ignorance Expecting µK/beam maps with v. small pixels over large areas  What kind of instrumental junk is going to turn up?  Do we really not expect to run into diffuse backgrounds?  When does point-source subtraction begin to fail?  When do mm-wave instruments become point-source confused? Interferometers vs. Bolometers  Will interferometers ever be competitive near the null?  What about interferometers with multi-pixel receivers to increase FOV? Large telescopes with bolometer arrays getting too large for small groups (manpower + $$). Heading out of the small experiment regime. You don’t get a new measurement technique for free!

34. Current and Future SZ Surveys Sunil Golwala/2001/07/07 34 Conclusion First blind cluster surveys using SZE underway or beginning soon New instrumentson the horizon with remarkable raw sensitivities and mapping speeds Exciting new science coming in the next few years  New, independent measures of  8,  m,    Prospect for new measure of equation of state