1. Star Formation in High Redshift Galaxies Mauro Giavalisco Space Telescope Science Institute and the GOODS Team

2. GOODS: Great Observatories Origins Deep Survey

3. GOODS: Great Observatories Origins Deep Survey

4. GOODS: Great Observatories Origins Deep Survey Finding high-redshift galaxies: color selection B 435 V 606 z 850 Unattenuated Spectrum Spectrum Attenuated by IGM B 435 V 606 i 775 z 850 z~4 1.Color selection is very efficient in finding galaxies with specific spectral types in a pre-assigned redshift range 2.Wide variety of methods available, targeting a range of redshifts, galaxies’ SEDs: Lyman and Balmer break (Steidel, Adelberger, MG) DRG (Franx, Labbe et al.) BzK (Daddi et al.) Photo-z (Mobasher et al) Here, the case of “Lyman-break galaxies” GOODS yielded the deepest and largest quality samples of LBGs at z~4 to ~6 (7?)

5. GOODS: Great Observatories Origins Deep Survey Color selection at z>3 B-band dropouts: 3.5<z<4.5 Vanzella et al. 2006

6. GOODS: Great Observatories Origins Deep Survey Color selection at z>3 V-band dropouts: 4.5<z<5.5

7. GOODS: Great Observatories Origins Deep Survey Color selection at z>3 i-band dropouts: 5.5<z<6.5

8. GOODS: Great Observatories Origins Deep Survey Color selection at z>3 z-band dropouts: 6.5<z<7.5

9. GOODS: Great Observatories Origins Deep Survey The Redshift Distribution #183 #27 LBGs at z>3 are targets of the ongoing GOODS spectroscopic time with the ESO VLT and Keck Vanzella et al. 2006, 2005, 2006 in prep. Stern et al. 2006 in prep.

10. GOODS: Great Observatories Origins Deep Survey z~4 spectroscopy Variety of spectral “types” Very similar to the z~3 galaxies Emission of Lya observed together with weak interstellar absorption lines Stronger absorption lines are present when Lya is obsered in absorption Effect of geometry of ISM? Vanzella et al., in prep.

11. GOODS: Great Observatories Origins Deep Survey z~3 spectroscopy Popesso et al., Vanzella et al. in prep.

12. GOODS: Great Observatories Origins Deep Survey z~4 spectroscopy Popesso et al, in prep.

13. GOODS: Great Observatories Origins Deep Survey Exploring the geometry of the ISM Abs. Em. No obvious correlation of spectral “types” with UV color or ellipticity of the galaxies Whatever causes the absorption does not know about the geometry of the UV-luminous galaxy Outer ISM phase surrounding the UV-emitting regions whose spatial geometry DOES NOT correlate?

14. GOODS: Great Observatories Origins Deep Survey z~5 spectroscopy At z~5 and 6 selection effects make “emission” galaxies easier to confirm spectroscopically Vanzella et al. in prep.

15. GOODS: Great Observatories Origins Deep Survey Composite spectrum of i-band dropouts The spectral properties of “observed” LBGs at z~6 are very similar to some LBGs observed at z~3. At z~6 it is very hard to obtain spectra of those LBGs with no Lya. Selection effect! Vanzella et al., Giavalisco et al 2006, in prep.

16. GOODS: Great Observatories Origins Deep Survey LBG luminosity function Relatively mild evolution of the UV luminosity function at 2.5<z<5.5 Giavalisco et al. 2006 in prep.

17. GOODS: Great Observatories Origins Deep Survey The history of the cosmic star formation activity: This plot spans 94% of the cosmic time! We find that at z~6 the cosmic star formation activity was nearly as vigorous as it was at its peak, between z~2 and z~3. Giavalisco et al. 2004 Giavalisco et al. 2006, in prep.  =-1.6 assumed

18. GOODS: Great Observatories Origins Deep Survey Star formation rates Derive from far-UV continuum luminosity Dust obscuration correction: Calzetti starburst obscuration law Some rates are low, like z~0 spirals; other are prodigiously high But, does “corrected UV” trace SF well? Quite likely in these systems (Kennicutt et al., Calzetti et al 2006; also Dickinson’s talk) z~4 B-band dropouts

19. GOODS: Great Observatories Origins Deep Survey The morphology of the LBGs Giavalisco et al. 1994, 1996, 1998 Steidel, Giavalisco, Dickinson & Adelberger 1996; Lowenthal et al. 1997; Dickinson 1998; Giavalisco 1998; Papovich, Giavalisco, Dickinson, Conselice & Ferguson 2003 Papovich, Dickinson, Giavalisco, Conselice & Ferguson 2004 Smaller Regulars, Irregulars, Merging, Spheroids? Disks? No Hubble Seq. No -dependence Rest-UV light Rest-optical light Morphology does not depend much on wavelength: young systems

20. GOODS: Great Observatories Origins Deep Survey Galaxies get smaller at high redshift… Standard ruler R~H(z) -2/3 R~H(z) -1  First measures at these redshifts  Testing key tenets of the theory  Galaxies appear to grow hierarchically Ferguson et al. 2004

21. GOODS: Great Observatories Origins Deep Survey Surface Brightness Profile Analysis: allows convolution by the point spread function better handle on flux in the galaxy wings where S/N drops at low surface brightness levels Measurement biases minimized - 2-D modelling using a single Sérsic function: Exponential disks: n = 1 R 1/4 spheroids : n = 4 Quality control: low chi 2, small errors on parameters, m fit = m auto ±0.5 [Ravindranath et al. 2006] GALFIT

22. GOODS: Great Observatories Origins Deep Survey B-dropout with n > 3.0 ( spheroid-like )

23. GOODS: Great Observatories Origins Deep Survey B-dropout with n~ 0.8 ( disk-like )

24. GOODS: Great Observatories Origins Deep Survey B-dropout with n ≥ 5 ( centrally concentrated )  3"  100 x 100 pixels  3" 

25. GOODS: Great Observatories Origins Deep Survey B-dropout with n<0.5 ( mergers, multiple cores )

26. GOODS: Great Observatories Origins Deep Survey Profile Distribution of LBGs and z=1.2 starbursts (all M<0.5M UV *) LBGs at z > 2.5: ~ 40% exponential disks ~ 30% spheroid-like ~ 30% mergers, multiple cores Star - forming galaxies at z = 1.2: ~ 26% exponential disks ~16% spheroid-like ~ 58% mergers, irregulars? Similar conclusions from non-parametric study based on GINI, M20 and CAS coefficients Lotz, Madau, Giavalisco, Primack & Ferguson 2005

27. GOODS: Great Observatories Origins Deep Survey Probing the Intrinsic Shapes Through Ellipticity Distribution Observed peak in the  = (1- b/a), and skewed distribution  Not only spheroids and circular disks seen at random orientations Intrinsically elongated galaxies Peak  is lower at lower z

28. GOODS: Great Observatories Origins Deep Survey Ellipticity distribution for different LBG profile types…….

29. GOODS: Great Observatories Origins Deep Survey Possible explanations for the excess of “Elongated” morphologies among LBGs !  Rotation-dominated disks? Edge-on projections and selection effects  Star forming clumps along gas-rich filaments of cold gas infall in DM halos  High-z bars at early epochs of galaxy formation?

30. GOODS: Great Observatories Origins Deep Survey Star-formation in filaments of cold gas in DM halos ? Ravindranath et al. 2006 35 kpc (180 comoving)

31. GOODS: Great Observatories Origins Deep Survey z=4 M=3x10 11 T vir =1.2x10 6 R vir =34 kpc Hydro Simulation: ~Massive M=3x10 11 Kravtsov et al. virial shock Dekel & Birnboim 06

32. GOODS: Great Observatories Origins Deep Survey Cold, dense filaments and clumps (50%) riding on dark-matter filaments and sub-halos Birnboim, Zinger, Dekel, Kravtsov

33. GOODS: Great Observatories Origins Deep Survey Observing the first gas-rich bars among LBGs at z > 2.5? Classic bar morphology in the first few billion years! Bar in DGs encompasses the whole galaxy; ~2-3 kpc scalelength Ravindranath et al. 2006

34. GOODS: Great Observatories Origins Deep Survey More bar signatures among LBGs at z > 2.5 Spiral arms from bar ends?

35. GOODS: Great Observatories Origins Deep Survey More possible bars among LBGs at z > 2.5 Star formation at bar ends?

36. GOODS: Great Observatories Origins Deep Survey The mass of LBGs: spatial clustering Galaxies at high redshifts have “strong” spatial clustering, i.e. they are more clustered than the z~0 halos “de-evolved back” at their redshift. –High-redshift galaxies are biased, I.e. they occupy only the most massive portion of the mass spectrum. –Today, the bias of the mix is b~1. Idea is to test key tenets of the gravitational instability paradigm –evolution of galaxy clustering contains information on how the mass spectrum gets populated with galaxies as the cosmic time goes on. –Clustering of star-forming galaxies at a given redshift contains information on relationship between mass and star formation activity

37. GOODS: Great Observatories Origins Deep Survey The mass of LBGs: spatial clustering Giavalisco et al. 1998 Steidel et al. 2003 Adelberger et al. 1998 r 0 =3.3+/- 0.3 Mpc h -1  = -1.8 +/- 0.15

38. GOODS: Great Observatories Origins Deep Survey Strong clustering, massive halos Porciani & Giavalisco 2002 Adelberger et al. 2004  =1.55 r 0 =3.6 Mpc h -1

39. GOODS: Great Observatories Origins Deep Survey Clustering strength depends on UV luminosity: mass drives L UV (SFR) Adelberger et al. (2004) GOODS Ground Lee et al. 2006 Giavalisco & Dickinson (2001)

40. GOODS: Great Observatories Origins Deep Survey Clustering segregation at z~4 and 5 Lee et al. 2006 See also Ouchi et al. 2004, 2006 Clustering segregation is detected in the GOODS ACS sample at z~4

41. GOODS: Great Observatories Origins Deep Survey Halos and Galaxies at z~3-5: Evidence of Evolution? Clustering scaling in good agreement with hierarchical theory Implied halo mass: >5x10 10 M O (faint samples) >10 12 M O (bright samples) 1-σ scatter between mass and SFR ~smaller that 100% LBG halos at z ~ 5 are less Massive. Specific star formation higher at higher redshift. Up-sizing! Giavalisco & Dickinson 2001 Porciani & Giavalisco 2002 Adelberger et al. 2004; Lee et al. 2006

42. GOODS: Great Observatories Origins Deep Survey Implications Halo mass, I.e. local gravity, is a key parameter to control star fomation Relationship between mass and star formation is tight Possible to reconstruct the L UV (M H ) distribution function (e.g. CLF) Giavalisco & Dickinson 2002; Lee et al. 2006 in prep.

43. GOODS: Great Observatories Origins Deep Survey Halo sub-structure at z~4 ACS depth made possible to observe structure within the halo. Break observed at ~10 arcsec Note: 10 arcsec at z~4 is about ~350 kpc, about the size of the virial radius for M~10 12 Mo. Lee et al. 2006; see also Ouchi et al. 2006

44. GOODS: Great Observatories Origins Deep Survey HOD at z~5 Lee et al. 2006

45. GOODS: Great Observatories Origins Deep Survey The Halo Occupation Distribution at z~4 =(M/M 1 )  M>M min Major improvement from COSMOS (Lee et al. PhD Thesis) Lee et al. 2006

46. GOODS: Great Observatories Origins Deep Survey Halos and Galaxies at z~4 Lee et al. 2006 Halo substructure: we observe an excess of faint galaxies around bright ones. massive halos contain more than one LBG “Bright Centers”: z_ 850 <24.0 “Faint centers”: 24.0< z_ 850 <24.7 “Satellites”: z_ 850 >25.0 Substructure is observed with good S/N at faint luminosity L<L * /2

47. GOODS: Great Observatories Origins Deep Survey Inside the halo at z~4: are we seeing dwarf galaxies?

48. GOODS: Great Observatories Origins Deep Survey Inside the halo at z~4: are we seeing dwarf galaxies?

49. GOODS: Great Observatories Origins Deep Survey Conclusions With large samples of high-z galaxies it is possible to test key ideas on star formation and galaxy evolution LBGs at z>4 have mix of spectroscopic properties –Tracing geometry of ISM Relatively high SFR; mild evolution of the UV lum. density at high z Mix of UV morphology –Spheroid and disk-like systems observed –Higher fraction of irregular systems at z~1.5 than at z>3 – Intrinsic excess of elongated systems that disappear at lower redshifts Evidence of cold accretion in filaments? Large-scale bars? Size evolution consistent with hierarchical growth Detected halo sub-structure at z~4 (thanks to ACS sensitivity) –Proving key prediction of theory

50. GOODS: Great Observatories Origins Deep Survey Color selection at z~2 Distant Red Galaxies (DRGs): J-K>2.3 F(24  m) & z -> L IR using Chary & Elbaz 2001 templates X-ray detected GTO 24  m 50% completeness UV-IR SEDs span range of Hubble sequence or dusty galaxies, (Forster-Schreiber et al.) 50% detected with F(24  m)>60  Jy. SEDs consistent with either AGN or starbursts. 24  m-detected DRGs are typically ULIRGs (L IR >10 12 L o ) Papovich et al. 2005

51. GOODS: Great Observatories Origins Deep Survey DRGs at z~2 Galaxies selected from near-IR photometry [(J-K)>2.3] Most would NOT be selected by LBG criteria (UV selection) However, overlap with LBG not quantified And certainly significant (see Adelberger Et al. 2004). They appear in general more evolved, I.e. more massive (larger clustering), with larger stellar mass, more metal rich, and more dust obscured) than LBGs. Occurrence of AGN also seems higher. At z~3 these galaxies have about 50% of the volume density of LBGs (highly uncertaint). However; they possibly contribute about up to 100% of the LBG stellar mass density, because they have higher M/L ratios Van Dokkum et al. 2004

52. GOODS: Great Observatories Origins Deep Survey IRX-  for Distant Red Galaxies UV spectral slope  measured from ACS colors. DRGs typically have redder  than LBGs: = 3.1 mag L IR for DRGs typically exceeds expectation from L UV and  by factors of 10- 100x DRG IR excess larger than that for less luminous (typically more UV-bright) HDFN 24  m sources. SFR~10 to 1000 M o /yr

53. GOODS: Great Observatories Origins Deep Survey Stellar population modeling

54. GOODS: Great Observatories Origins Deep Survey Stellar population modeling

55. GOODS: Great Observatories Origins Deep Survey Stellar masses & properties of GOODS-S DRGs Typical DRG stellar masses ~few x 10 11 M o, (cf. FIRES work). GOODS-S sample is roughly complete at >10 11 M o for 2 < z < 3 2-component models frequently (but not always) give better fits to the photometry. Masses increase, but not as much as for blue, lower-mass HDFN LBGs. Loosely dividing by reddening: Heavily obscured: E B-V > 0.35: = 1.7 L IR ~ expected from L UV,  Lightly obscured: E B-V < 0.35: = 2.5 L IR >> expected from L UV,  (for 24  m- detected objects)

56. GOODS: Great Observatories Origins Deep Survey Specific star formation rates (SSFRs) Low-z comparison samples from COMBO-17: z ~ 0.4 and z ~ 0.7 Stellar masses estimated from COMBO-17 photometry SFRs from GTO MIPS 24  m data z 10 11 M o tend to be forming stars at low SSFRs. z > 1: Galaxies over a broad range of masses tend to span a broad range of SSFRs, with many DRGs forming stars prodigeously.

57. GOODS: Great Observatories Origins Deep Survey “Downsizing” of star formation in massive galaxies Treating COMBO-17 and GOODS DRG samples as representative for M > 10 11 M o : z~2.3 DRGs forming stars with SSFR > cosmic average z < 1 massive galaxies forming stars more slowly than the global average Further evidence that 1.5 < z < 3 was a key era for the rapid growth of stellar mass in the most massive galaxies. Global average from co-moving  SFR (z)

58. GOODS: Great Observatories Origins Deep Survey Color selection at z~2: BzK galaxies BzK selection well suited for 24  m MIPS studies: Selected range 1.4 < z < 2.5 places strong mid-IR features in 24  m band Color selection includes objects with red UV continuum, e.g., from extinction K-band selection suitable for relatively massive galaxies (Daddi et al. 2005) BzK selection: 1.4<z<2.5

59. GOODS: Great Observatories Origins Deep Survey

60. GOODS: Great Observatories Origins Deep Survey BzK samples in GOODS- N&S

61. GOODS: Great Observatories Origins Deep Survey 24  m detection of BzK galaxies 245 BzKs with K < 20.6 169 BzKs with K < 20 At present, spectroscopic redshifts available for only a few; Keck LRIS+DEIMOS runs ongoing. 36/169 detected in hard X-rays (mostly AGN; not considered for now) 109/133 (82%) for non-Xray BzKs detected at 24  m (undetected fraction consistent with expected number of “passive” BzKs) Median = 110  Jy Fainter K-band --> fainter 24  m

62. GOODS: Great Observatories Origins Deep Survey Multi-wavelength measures of SFR MIPS: =125  Jy, =1.9, and CE01 templates: = 1.7e12 L o, ~ 300 M o /yr UV continuum + reddening: ~ 220 M o /yr Radio: stacked VLA data = 17  Jy = 2e12 L o, ~ 340 M o /yr Sub-mm: stacked = 1.0 mJy (5  ) = 1.0e12 L o, ~ 170 M o /yr X-ray: stacked 8.5  soft-band detection, no significant hard-band. Far below expected AGN level. = 100 - 500 M o /yr (Persic 2004, Ranalli 2003 conversions) On average, multi-wavelength SFR tracers agree reasonably well with expectations from low-z correlations, templates & analogs.

63. GOODS: Great Observatories Origins Deep Survey UV vs. IR SFRs: BzK- selected galaxies at z ~ 2 B-band samples ~1500A UV continuum at z~2; B-z measures UV continuum slope. f(24  m) / f(B) correlates strongly with B-z color, as expected if UV continuum slope results from dust reddening. Log scatter is a factor of ~3 (including effects of the broad BzK z-range). Brighter/more luminous mid-IR sources (L IR > 10 12 L o ) tend to exceed expected IRX- , while less luminous sources match or fall below it (possibly including “passive” BzKs. Measure of mass in progress.

64. GOODS: Great Observatories Origins Deep Survey Star formation at z~1.5 – 2.5 Typical BzK and DRG galaxies appear to be both massive (~10 11 M o ) and rapidly star forming (L IR ~10 12 L o, ~ 200 M o /yr), with space density ~1000x larger than present-day ULIRGs 10-20% may be AGN; X-ray stacking favors star formation for the majority. ~ 80% MIPS detection rate for BzKs implies that most massive galaxies at 1.4 < z < 2.5 are forming stars prodigiously: –Implies high duty cycle for SF –Substantial mass build-up over this redshift range BzKs should form  * >~ 5x10 7 M o /Mpc 3 over ~2 Gyr, comparable to local stellar mass density in galaxies with M * > 2x10 11 M o Specific star formation rate (SFR/M * ) for massive (>10 11 M o ) galaxies at 2 downsizing.

65. GOODS: Great Observatories Origins Deep Survey VIMOS LBGs U B V 25 MR (Rwfi<24.5) - i