Anton Koekemoer

1

• EXOs Candidate AGN at z 6
• and intermediate-z evolved populations

• Anton Koekemoer
• (Space Telescope Science Institute)
• GOODS (Alexander, Bauer, Brandt, Chary,
  Conselice, Cristiani, Daddi, Dickinson, Elbaz,
  Grogin, Mainieri, Tozzi,..)
• COSMOS-XMM (Brusa, Comastri, Elvis, Finoguenov,
  Fiore, Gilli, Hasinger, Impey, Mainieri, Salvato,
  ..)

2

• Large X-ray/optical/IR surveys are opening up new
  regions of parameter space
• Depth probes faint end of AGN LF to high z
• Area probes high end of AGN LF at high z
• Combined optical X-ray depth allows wider
  exploration of FX/FOpt and new populations

• Relevant surveys
• GOODS/CDFNS (15)
• E-CDFS (30)
• XMM-LH (30)
• EGS (10x 60)
• COSMOS (1.4o x 1.4o)
• ..

3

• FX/FOpt parameter space
• AGN typically 1-2 dex
• SB are lower Fx/Fopt
• Highest FX/FOpt
• EXOs - Extreme X-ray /Optical sources
• Fx/Fopt gt 100
• Only revealed byextending optical depthbelow i
  or z 24 - 25
• Appear to have no comparable analogs in the local
  universe
• Why do we only start seeing them at faint
  magnitudes?
• What produces the high FX/FOpt - these sources
  likely consist of two sub-populations
• Balmer break z 2 - 3 evolved or dusty hosts
  (DRGs, EROs)
• Lyman break candidate z gt 6-7 agn

4

• Fx/Fopt vs Colour
• Sources with low FX/FOptare generally blue
  low-zstar-forming galaxies
• Some blue sources alsohave FX/FOpt typical
  ofunobscured AGN- quasars
• Moderately red galaxies(z - K gt 2) all have
  higherFX/FOpt, obscured AGN

• Reddest z - K sources
• for z - K gt 4, FX/FOpt increases to 10 - 100x
  the value for even typical obscured AGN
• Not a selection effect
• high FX/FOpt should not necessarily imply bright
  IR flux
• Thus red opt/IR colour is intrinsic
  characteristic of EXOs

5

• Previous studies of optically faint X-ray
  sources
• Initial Deep Chandra/XMM fields revealed that
  20-30 ofX-ray sources are optically faint, R
  gt 24(Koekemoer et al. 2002, Tozzi et al. 2002)
• Most optically faint sources are also X-ray
  faint, ie have fairly normal FX/FOpt typical of
  obscured AGN at z 1-3 (Brusa et al. 2003,
  Mainieri et al. 2004, Fiore et al 2005)
• Some optically faint sources are EROs (z
  1-1.5) - but also have normal FX/FOpt (Stevens et
  al. 2003, Yan et al. 2003, Rigby et al. 2005)
• EXOs
• Optically faint sources with anomalously high
  FX/FOpt gt100
• No apparent local analogs at brighter magnitudes
• Typically have redder z-K than EROs (Koekemoer
  et al. 2004, 2006)

6

• Nature of EXOs
• Need further IR constraints on SED to determine
  which sub-population a given EXO belongs to based
  on its red opt/IR
• Balmer break (z 2 - 3 evolved or dusty
  populations)
• Lyman break (z gt6 AGN)
• NOTE expect EXOs to contain both z2-3 and zgt6
  sources
• Spitzer data
• currently based purely on GOODS CDFS HDFN
  (Dickinson)
• all EXOs are detected in IRAC data
• red K - IRAC colour
• across IRAC, have a mixture of SEDs (red, flat or
  blue)
• MIPS detections/non-detections are consistent
  with IRAC properties
• flat/red IRAC -gt MIPS detections, usually
  brighter than IRAC
• blue IRAC colours -gt generally undetected in MIPS

7
(No Transcript)
8
(No Transcript)
9

• SED fitting
• Parameters SSP constant SFR, reddening, IMF,
  dust
• Models
• initally used CB2003
• will also now include Maraston
• Results
• Most EXOs arewell fit by
• z 2 - 3
• evolved
• 1010-11 Mo
• Some requirereddening AV 1
• A few EXOs arenot well fit byz 2 - 3 models

10

• Using EXOs to trace high-z AGN evolution
• AGN regulate galaxy growth / SFR via feedback
• may trace hierarchical dark matter halos
• possibly different types of accretion mechanisms
• luminous AGN may trace major mergers
• lower-luminosity AGN may trace more minor
  interactions and accretion events (e.g. Merloni
  et al. 2004)
• Hasinger etal. luminosity-dependent density
  evolution LDDE
• High-lum AGN grow earlier in universe, peak at z
  2
• Lower-lum AGN peak much later, z1, decline by
  10x to z0
• Questions
• how does the faint end of the AGN LF evolve
  beyond z 6?
• does obscured/unobscured AGN ratio increase
  beyond z 6?
• does more rapid evolution of high-lum AGN trace
  merging history of spheroid formation? (e.g,
  Franceschini et al 1999)

11

• Using EXOs to count high-z AGN in GOODS
• Use XLF to estimate expected number of optically
  unidentified sources as a function of redshift
• Most of the optically unidentified AGN are
  evolved interlopers at intermediate z gt 2
• Compare with observed number of undetected
  sources
• use existing X-ray detection limits
• apply optical detection cut-off (z(AB) 27.5 for
  ACS)
• Integrate over X-ray luminosities at each
  redshift bin
• assume Type 1/2 ratio found in GOODS by Treister
  et al
• Use the difference to calculate cumulative number
  N(gt6)
• Compare with N(gt6) from XLF

12

• Predict optically unidentified sources in each
  redshift bin using Hasinger et al. LDDE
  description, extrapolating upto z 7
• Apply to GOODS X-rayselection, including
  theoptical detection limits, tocalculate EXOs
  expected.

• Number of optically unIDd sources N(z) based
  onz(AB)27.5 limit, for currentChandra
  catalogs
• LDDE predicts 9 - 16 EXOs in GOODS (out of 607
  X-ray sources)
• 8-13 should be at z 2 - 5
• 1-3 should be at z 6 - 7
• Actually detect 13 EXOs
• SED modelling confirmsthat 10 are at z 2 - 4
• 3 are indeterminate

13

• Conclusions
• Number of EXOs found in GOODS agrees well with
  that expected based on LDDE (13 vs 9-16,
  respectively)
• Number of intermediate-z interlopers among EXOs
  also agrees with expected from LDDE (10/13 vs
  8-13/9-16)
• Remaining number of 3 EXOs in GOODS with possible
  high redshifts (ie gt 6) is consistent with the
  1-3 EXOs expected at z gt 6 based on extending
  LDDE to z 6 7
• Therefore LDDE appears to extend up to at least z
  6 - 7
• This suggests that AGN growth/accretion
  mechanisms continue to track galaxy growth into
  reionization
• AGN feedback regulating star formation up to
  early epochs
• black holes tracing dark matter halos since at
  least z 7
• Future
• extend to COSMOS to improve LF constraints
  (whenever Spitzer catalogs are available for SED
  fitting..)