1. Probing the formation and evolution of massive galaxies with near- to mid-infrared surveys Scuola Nazionale di Astrofisica VIII Ciclo (2005-2006): I Corso S. Agata sui due golfi, 8-13 Maggio 2005 LECTURE 2 Andrea Cimatti (INAF-Arcetri)

2. Lecture 2 - Outline The K20 and other recent near-IR surveys: - Sample definition and observations - Redshift distribution - Nature of K-selected galaxies - Morphology evolution - New selection and segregation techniques based on colors - Evolution of the luminosity density and function - Evolution of the stellar mass density and function - Evolution of the clustering - Comparison with model predictions First Spitzer results The GMASS project Future steps

3. Near-infrared surveys … First “seminal” work: Hawaii Deep Survey, Ks<20, 26 arcmin 2, zspec, Keck telescopes (Cowie et al. 94,96) Main players in 2000+ : FIRES (Franx et al.): zphot + zspec, very deep, small area K20 (Cimatti et al.): zspec GDDS (Abraham et al.): zspec GOODS (public survey): zphot+zspec, depth and area MUNICS (Drory et al.): zphot+zspec, large area, big sample

4. The K20 survey 20 ESO/VLT nights (visitor mode) Deep optical and (sometimes) near-IR spectroscopy 546 objects with Ks<20 (no other selection criteria) 2 fields (0055-269 + fraction of GOODS-South/CDFS) = 52 arcmin 2 UBVRIzJHKs imaging -> SEDs and photometric redshifts K20/CDFS/GOODS-South field: multi-wavelength public data 92% complete (zspec), 100% (zspec+zphot) 0 < z(spec) < 2.5 Highest z(spec) completeness to date for a faint K-selected sample

5. GOODS-South HST+ACS+F850LP Giavalisco et al. 04

6. Observations Optical: FORS1 & FORS2 1” slit width, 5”-20” slit lengths Spectral resolution ~ 260, 380, 660 Integration times: 30 min – 8 hours Near-IR: ISAAC 1” slit width, 1-2 targets per slit Spectral resolution ~ 500 Integration times: 1-2 hours Good (≤ 1”) and stable seeing is essential for faint galaxy spectroscopy

7. VLT+ISAAC Ks-band mosaic of the GOODS-S field

8. Example of a spectroscopic mask design FORS2 + MXU

9. Example of multi-object spectroscopic raw data

10. The (big) problem of CCD fringing Fringing is caused by multiple reflections inside the CCD. At longer wavelengths, where thinned chips start to become transparent, light can penetrate through and be reflected from the rear surface. It then interferes with light entering for the first time. This can give rise to constructive and destructive interference and a series of fringes where there are minor differences in the chip thickness. For spectroscopic applications, fringing can render some thinned CCDs unusable, even those that have quite respectable QEs in the red. Thicker deep depletion CCDs, which have a much lower degree of internal reflection and much lower fringing are preferred by astronomers for spectroscopy.

11. Removing fringing with “dithered” observations ABAB slit galaxy Spectrum + sky + fringing (A) Spectrum + sky + fringing (B) A-B B-A (A-B)+(B-A) S

12. Example of dithered data

13. Optical spectra Early-type z=1.096 Early+emission z=0.735 Emission lines z=1.367 Absorption lines z-=1.725 FORS2 Cimatti et al. 2002

14. ISAAC 2D spectrum H-band Low-resolution H  z=1.729

15. Near-IR spectra ISAAC z=1.729 z=1.715 Cimatti et al. 2002

16. Photometric redshifts =0.012  =0.089 Cimatti et al. 2002

17. The importance of N(z) A test proposed by Kauffmann & Charlot (1998) Predicted N(z) for K<20

18. The observed N(z)

19. Cumulative number of galaxies No galaxies expected with z>1.7 Cimatti et al. 2002

20. The nature of K-selected galaxies (I): Extremely Red Objects (EROs)

21. Luminous red galaxies at z~1 EROs Pozzetti et al. 2003 0.75<z<1.30 observed (K20 survey) predicted (Kauffmann et al. 1999)

22. The importance of Extremely Red Objects (EROs) R-K>5 Instantaneous burst (SSP) or SFR(t)=SFR(0)exp(-t/τ) R-Ks>5 is the observed color expected at z > 1 in case of high formation redshift followed by passive evolution or rapidly declining SFR  EROs trace the oldest envelope of galaxies at z>1 However, very red colors (such as R-Ks > 5) are also expected in case of galaxies with ongoing star formation and strong dust reddening

23. Sometimes they are really very red… Dickinson et al. 2000 Hubble Deep Field images

24. The nature of K-selected galaxies (II): old EROs (fossil, quiescent, passive galaxies)

25. Old passive spheroidal EROs at z~1 “Reddest envelope” galaxies Passively evolving 0.9<z<1.3 2 – 4 Gyr old z(SF onset) > 2 M(stellar) > 10 11 Msun r 0 ~ 10 Mpc comoving N ~ N(PLE) (2σ) Strongly underpredicted by hierarchical merging models Cimatti et al. 1999, 2002, 2003 Daddi et al. 2000a, 2000b, 2001

26. Comparison with SDDS spectra Pozzetti et al. 2005 VLT+FORS2, 3Å resolution, 11h integ. z~1 spectra similar to z~0 ones, i.e. over a time interval of ~8 Gyr !! Elliptical at z=1.096 (0.7<z<1.2)

27. Weak spectral evolution to z~1 0<z<0.6 0.6<z<0.75 0.75<z<1.25 (0.6<z<0.75) – (0.75<z<1.25) (0.6<z<0.75) – (0<z<0.6) Mignoli et al. 2005

28. More surprises…

29. Searching for old galaxies at z>1.5 The region around MgII 2800 Å rest-frame is useful for redshift estimates at z>1.5 (when D4000 exits from optical spectra) and it is strong for old stellar populations Daddi et al. 2005

30. Old massive spheroids to z~2 ! Cimatti et al. 2004, Nature 25-60% of total stellar mass density Absent in hierarchical models Cosmic variance result ? No ! HST+ACS (GOODS-S data) R-Ks>6, ~0.2/□‘, 1.6<z<2.0 1-2 Gyr old, passive, z(SF onset)>2.5-4 Spheroidal morphology Massive: M*>10 11 Msun

31. Comparison with other spectra SSP synthetic spectra (BC03) Solar metallicity, Salpeter IMF From top: 3.0, 1.1, 0.5 Gyr Old EROs at z~1 Dusty star-forming EROs at z~1 Early-type galaxies at z~0.5 (SDSS) F5 V star F2 V star Cimatti et al. 2004 (Nature)

32. More cases Saracco et al. 2005 TNG + NICS near-IR spectroscopy TESIS project McCarthy et al. 05 Glazebrook et al. GDDS project 1.5<z<2.0

33. Old ellipticals at z>1.4 found with HST+ACS What is the number density of high-z old spheroids ? Comoving density at =1.7 is formally 1/3 of the local (z=0) density of galaxies with M>1e11 Msun However, taking into account the cosmic variance due to clustering, such fraction can range from 10% to 80% of the local density value (at 1σ confidence level) Daddi et al. 2005

34. The nature of K-selected galaxies (III): dusty star-forming EROs

35. Dusty star-forming EROs 0.7<z<1.7 E(B-V)~0.5 - 1.0 ~ 30 Msun/yr SFR up to ≥ 100 Msun/yr Weakly clustered (r 0 <3 Mpc) ? Probably unrelated to spheroids Strongly under-predicted by hierarchical merging models Cimatti et al. 2002,2003 HST + ACS (F850LP filter)

36. X-ray, submm, radio emission from dusty EROs 9” x 9” Chandra X-ray full-band stacked images of dusty (left) and old (right) EROs (Brusa et al. 2002) Dusty EROs are (weak) X-ray sources The individual detections indicate Type 2 AGN activity Old EROs are below the detection threshold SFR ~ 5-44 Msun/yr (from SFR – L(hard X) relation) SFR ~ 30-60 Msun/yr (from SFR – L(radio) relation) SFR ~ 3 Msun/yr (from SFR – L([OII]) relation) SFR < 1 Msun/yr (from SFR – L(UV) relation) SFRD(dEROs)/SFRD(total) > 30% at z~1 [Msun/yr/Mpc 3 ] L(FIR) typically < 10 12 Lsun (i.e. most dEROs are not ULIGs)

37. The nature of K-selected galaxies (IV): Progenitors ? (dusty massive starbursts at z~2 and beyond)

38. Massive starburst galaxies at 1.7<z<2.7 E(B-V)~0.3-0.5, SFR > 100 Msun/yr Stellar masses ~ 10 11 Msun (confirmed with Spitzer + IRAC data) Merging-like morphology in rest-frame UV “Compact” morphology in rest-frame optical Strongly clustered (r 0 ~ 10 Mpc) Daddi et al. 2004; De Mello et al 2004 B ACS Z ACS Massive spheroids in formation ? B Z Ks

39. … More progenitor candidates… Submm/mm, mid- and near-infrared surveys are also finding progenitor candidates at z~2-4: Large masses + dust + strong SFRs Number density still poorly known (e.g. Totani et al. 2001, Chapman et al. 2003; Franx et al. 2003; Saracco et al. 2004; Chen et al. 04; Yan et al. 2004)

40. … J-K>2.3 – selected galaxies J-Ks>2.3 (Vega system) allows to select galaxies at z>2 (Franx et al. 03) These distant red galaxies (DRGs) are significantly redder than optically-selected galaxies in the same redshift range

41. … Complementarity of optical and K/J-K>2.3 selections Selection region for optically-selected galaxies at z~2 (Adelberger et al. 03) DRGs do not fall in the optical color selection region van Dokkum et al. 2004

42. … Properties of DRGs DRGs are physically distinct from optically-selected galaxies (van Dokkum et al. 2004)

43. J-K>2.3 vs. “pure” K-selection K20 data Blue = zphot Red = zspec Galaxies at z>2 with J-K<2.3 Dusty galaxies at z 2.3 J-K>2.3 is contaminated by dusty galaxies at z<2 J-K>2.3 does not select the whole galaxy population at z>2: there are galaxies at z>2 with J-K<2.3

44. New segregation and selection techniques

45. “BZK”: a new selection for 1.4<z<2.5 BZK = (Z-K)-(B-Z) > -0.2 BZK 2.5 Daddi et al. 2004 K20/CDFS field Also verified with simulated SED tracks Complementary to optical selection reddening

46. The diversity of EROs Is there a simple and “cheap” photometric criterion to segregate different types of EROs ? Cimatti et al. 2003

47. R-K vs. J-K diagram Pozzetti & Mannucci 2000 Cimatti et al. 2003 EROs dominated by old stellar populations Dust-reddened EROs

48. Summary of selection techniques

49. Luminosity evolution

50. The evolution of the K-band luminosity function Also : Cohen 2000 Drory et al. 2001 Kashikawa et al. 2003 Feulner et al. 2003 Chen et al. 2003 Inconsistent with hierarchical merging Pozzetti et al. 2003

51. Morphology evolution

52. The morphology evolution in the rest-frame B-band HST+ACS + K20/GOODS-S data Four ACS bands (bviz) Galaxies divided into 4 redshift bins to overcome morpho k-correction and sample the rest-frame B-band (up to z~1.2) Cassata et al. 2005 Galaxies at z ~ 1 z-band v-band

53. Morphological classes and CAS parameters Cassata et al. 2005

54. Morphology and merger evolution to z~1 Morphological fractions rather constant to z~1 Merger fraction increases rapidly Pair: companion with i<24.5 at <20 kpc (corrected for projection contamination) Cassata et al. 2005

55. Stellar mass evolution

56. K20 stellar masses from SED fitting UBVRIzJHKs bands + spectroscopic redshifts for 92% of the sample ! M(BF) = stellar masses derived from the SED best-fit M(MA) = stellar masses assuming all galaxies form at z=20 and requiring an exponentially declining SFR which reproduces the R-Ks color of each galaxy. This maximizes the stellar mass of each galaxy. Salpeter IMF is assumed Stellar masses decrease by ~1.5-2x if a two-slope IMF (e.g. Kroupa) is used Fontana et al. 2004

57. K20 stellar masses from SED fitting Fontana et al. 2004

58. The evolution of stellar mass function and density Also: Brinchmann & Ellis 2000, Cohen 2002, Drory et al. 2001,2004, Fontana et al. 2003;Dickinson et al. 2003; Rudnick et al. 2003, Glazebrook et al. 2004 Fontana et al. 2004 20-30% decrease of MF to z~1 30-40% of present-day massive systems in place at 1.5<z<2 Old/passive and star-forming systems contribute ~equally to mass density at 1.5<z<2 Local MF

59. The stellar mass function by spectral type The large mass tail is dominated by early-type galaxies up to z~1 Fontana et al. 2004

60. Contribution of morphological types to stellar mass density evolution Bundy et al. 2005

61. Other results: Drory et al. (2005) Comparing different stellar mass functions

62. Other results: Drory et al. (2005) The evolution of the stellar mass density: a comparison between different surveys

63. Clustering evolution

64. Large scale structure & clustering evolution Simulation of large scale structure Real galaxy distribution projected on the sky W(θ) -> ξ(r)=(r/r 0 ) -γ

65. Clustering evolution of massive galaxies Predictions: evolution of DM halos from Mo & White 2002 M>10 11 Msun red/old galaxies consistently hosted by >10 13 Msun halos at all epochs  Decrease in number density consistent with that of same halos  Picture in good agreement with hierarchical clustering scenario  K L* red galaxies SCUBA

66. Comparison with models

67. What doesn’t work Observation Agreement with models N(z) No (no luminous/massive galaxies z~1-2) LF(z) No (strong density evolution) Colors No (too blue) Old EROs /spheroids No (under-predicted or absent) Dusty EROs No (under-predicted or absent) GSMF(z) and mass density No (faster evolution) Models by: Durham (Cole et al.), Munich (Kauffmann et al.), Santa Cruz (Somerville et al.), Rome (Menci et al.) What works Non-violation of DM halo number density Clustering evolution

68. GSMF(z): comparison with models “Classical” semi-analytical models tend to under-predict massive galaxies at high-z (Cole et al. 00, Somerville et al. 01, Menci et al. 02) Models with strong AGN feedback are in better agreement with the data (Nagamine et al. 01-04, Granato et al. 04) Fontana et al. 2004

69. The ΛCDM “violation” test Theoretical galaxy baryonic mass function (GBMF): Press-Schechter distribution + + M(baryon)=M(DM) (Ω baryons /Ω DM ) Ω baryons /Ω DM = 0.045/0.23 (WMAP) Observed massive galaxy number density does not exceed the GBMF Fontana et al. 2004

70. Upgrading models Fly-by galaxy interactions, merging starbursts, feedback from AGN (gas heating and expelling to stop star formation in massive halos), chemical and dust evolution (Menci et al. 04, Somerville et al. 04, Granato et al. 04, Nagamine et al. 2004). Earlier and faster SF in most massive halos. “Anti-hierarchical” behaviour mimicking old-fashioned “monolithic” collapse scenario Better agreement with global/integrated properties, but not yet with individual galaxy properties (colors, ages, SFRs, masses; cf Cimatti et al. 04) Apparently OK, but too many passive old galaxies… Granato et al. 04 K-band LF at z=1.5

71. Spitzer

72. DRGs (J-Ks>2.3) with Spitzer 70% star-forming + 30% passive. Large masses confirmed. Most massive formed at z>5 (Labbè et al. 2005)

73. Dusty star-forming EROs 50% of ALL EROs detected at 24 μm (Ks<20.2) L(8-1000μm) ~ 3e11–3e12 Lsun SFR ~ 50-170 Msun/yr (Yan et al. 2004)

74. IRAC EROs (IEROs) Selected at 3.6μm F ν (3.6μm)/F ν (z-band) > 20 SEDs need old stars + star formation 1.6<z<2.9, some very massive (Yan et al. 2004)

75. Spitzer spectroscopy Yan et al. 2005 This opens the unprecendented possibility to derive redshifts and study the nature of extremely dust-obscured galaxies for which optical/NIR spectroscopy is impossible due to their faintness

76. GMASS Galaxy Mass Assembly ultradeep Spectroscopic Survey ESO VLT + FORS2 Large Program (145h; PI Cimatti) Collaboration with GOODS Team (Dickinson et al.) 4.5µm-selected galaxies (m<23 AB) (Spitzer+IRAC) (rest-frame NIR covered to z~3, accurate stellar masses) 51 arcmin 2 in the GOODS-S/K20/HUDF region N~190 galaxies with z(phot)>1.4 Blue or red spectra (B<26.5, I<26.0 AB) 12h-30h integration per mask (60-90h for repeated targets) Spitzer VLT Arcetri

77. GMASS: why 4.5µm selection ? Best compromise among IRAC bands (sensitivity, PSF, IQ, blending) Rest-frame near-infrared coverage up to z~3 Reduction of 2-3x in stellar mass estimate uncertainties Less dust extinction

78. GOODS-South SST+IRAC 4.5µm 1.4 < z < 4.0 candidates GMASS

79. First red spectra (12 hours, grism 300I)

80. The emerging picture  LF and stellar MF: mild luminosity evolution to z~1  Old, massive, passive E/S0 galaxies already present up to z~2 require a “formation” epoch at z~2-3  LF(z) of E/S0 is still poorly constrained even at z=1  Substantial population of dusty, luminous, massive, K-selected starbursts at z~2-3: E/S0 progenitors ?  Massive galaxies at 1<z<3 are strongly clustered  Current hierarchical merging models match the clustering evolution, but not the galaxy properties  The discrepancy is likely due to incorrect treatment of SF history/timescales, feedback, dust evolution.  Significant discrepancies are still present even with the newest upgraded hierarchical merging models

81. Future (near and far)

82. Ongoing deep public surveys PI M. Dickinson (USA) 150 + 150 arcmin 2 Deep multiwalength coverage HST Ground-based optical & NIR Spitzer (all bands) X-ray Radio UV (GALEX) Optical spectroscopy Submm Far-IR (Herschel) www.stsci.edu/science/goods

83. COSMOS PI N. Scoville Public survey 2 square degrees ! HST+ACS (b,i) Multiwalength coverage X-ray UV (GALEX) Optical & NIR Optical spectroscopy (VLT+VIMOS) Radio Spitzer Submm/mm ? Far-IR ? www.astro.caltech.edu/~cosmos

84. Future possibilities Spitzer ALMA JWST ESO OWL-ELT Observations from Antarctica