AGN and Star Formation at High Redshift

1. AGN and Star Formation at High Redshift Anton Koekemoer (Space Telescope Science Institute) + CDFS/GOODS-AGN(Alexander, Brandt, Bergeron, Conselice, Chary, Cristiani, Dickinson, Elbaz, Grogin, Mainieri, Schreier, Treister, Urry, Wiklind...) +COSMOS-AGN (Brusa, Carilli, Capak, Civano, Comastri, Daddi, Elvis, Fiore, Hasinger, Ilbert, Impey, Jahnke, Mainieri, Martinez-Sansigre, Mobasher, Murayama, Nagao, Oyama, Salvato, Sanders, Sasaki, Schinnerer, Scoville, Smolcic, Treister, Taniguchi, Trump, Urry, Zamorani)

2. Black holes in context: • Likely tracer of hierarchical dark matter halos • Provide hard ionizing continuum, relevant to low- reionization • May regulate galaxy growth / SFR via feedback • M-s relation suggests intimate connection between BH/galaxy formation and growth • To date: • At z ~ 6 - 6.5, already havesupermassive BHs to ~ 109 Mo(Fan et al 2001-06; Willott et al. 2003;Vestergaard et al. 2004) • AGN LF evolves stronglyfrom z~6 to z~2; PLE isruled out (Fan et al. 2001+;Richards et al. 2006) • Faint end of AGN LF steepensat z~6 (Jiang et al 2009, Fan et al. • (Richards et al. 2006)

3. Instead of PLE, AGN appear to follow “anti-hierarchical evol” -- luminosity-dependent density evolution “LDDE” at least for XLF (Ueda et al. 2003, Hasinger et al. 2005; Gilli, Comastri, Hasinger 2006), QSO LF (Richards et al. 2006); also Merloni et al. (2004): • luminous AGN peak earlier (z~2-3) • fainter AGN peak more recently (z~1) • (XLF; Hasinger et al. 2005) • (QSO LF, Richards et al. 2006)

4. Below LX ~ 1044 erg s-1, density evolution is drastically different from higher-lum sources, peaking at lower z • higher-lum sources show essentially pure density evolution • suggests possible difference in accretion / galaxy evolution as a function of luminosity (Hasinger et al. 2005)

6. Questions: • How do the most massive BHs form within < 1 Gyr? • How does BH growth influence the M-s relation? • What is the ionizing budget of AGN integrated over the LFbeyond z ~ 6, and its contribution to reionization? • (How) does obscured/unobsc. AGN ratio evolve at high z? • Our knowledge has been limited by the following: • only the top of the AGN LF has been studied at z ~ 6 • no AGN previously confirmed at z > 7 • Approach: • Understand AGN / SFR evolution at z ~ 2 - 5 • Set out to quantify the faint end of AGN LF at z ~ 6 • Search for more luminous AGN at z > ~ 6 - 7

7. AGN log FX (erg s-1cm-2) FOpt (mag) • Require Wide+Deep X-ray / Optical / IR Surveys: • Depth probes faint/moderate-lum AGN to high z • Area probes high-lum AGN at high z • (Hard) X-rays penetrate obscuring torus, IR probes rest-frame optical emission from AGN + host galaxy • New part of parameter space: • Combined optical + X-ray depth allows wider exploration of FX/FOpt:

8. Some current relevant surveys

9. COSMOS COSMOS GOODS SXDF AEGIS SXDF GOODS AEGIS ECDFS ECDFS GOODS COSMOS COSMOS SXDF ECDFS GOODS AEGIS ECDFS AEGIS SXDF

10. A Brief Selection of Extragalactic Survey Fields • UDF: • 3’x3’ HST/ACS (B,V,i,z) + NICMOS (JH) • Other coverage same as CDFS • GOODS/CDFS+CDFN: • 2x 10’x16’ HST/ACS (B,V,i,z) • Chandra 1Ms, Spitzer (3-70 mm), ATCA, VLA, VLT, KPNO • GEMS/ECDFS: • 30’x30’ HST/ACS (i,z) • Chandra 1Ms+4x 250ks, Spitzer, ATCA, CTIO • EGS: • 10’ x 1o HST/ACS (i,z) • Chandra 7x50ks, CFHT (UBVRIz) • COSMOS: • 1.4o x 1.4o HST/ACS (z) + NICMOS (J,H) • Subaru (UBVRiz), XMM, Spitzer, VLA

11. CDFS/GOODS-S+N survey: • powerful combination of wide area and depth, opt/Xray: CDFS + CDFN have 1 & 2 Msec Chandra depth respectively • X-ray depth sufficient for AGN LF faint end (LX~1043-44 erg s-1) up to z ≥ 6 - 7 • area sufficient (0.1 sq deg) to provide number statistics on AGN LF at these redshifts • More than 800 AGN from Chandra in GOODS-N & S (> 600 covered by HST/ACS and Spitzer) • Extensive optical spectroscopic coverage • Deep multi-band optical/NIR/Spitzer coverage: • JHK + Spitzer/IRAC 3.6 – 8 mm observations trace host stellar mass for z > 1-2 • Spitzer/MIPS 24 mm data helps constrain thermal dust emission

12. GOODS = GreatObservatoriesOriginsDeepSurvey(Hubble, Chandra, Spitzer) • Centered on CDFS, CDFN • Supporting observations: • VLT, Gemini, Keck, Subaru, KPNO, CTIO, VLA, ATCA, JCMT • Goals: • Extensive, deepmulti-l coverage • Large enougharea to probeboth galaxy andAGN evolution • Two fields reduceimpact of cosmicvariance

13. Advantages of X-ray selection: • avoid obscuration effects in optical/IR • at z > 2-3, hard X-rays are redshifted into soft observed X-ray bandpass, allows a more complete selection of obscured AGN (except for Compton thick sources) • allows selection of more AGN than radio selection. • Other wavelengths needed: • deep optical to ensure good identification of dropouts: • at least R,I, preferably also z • deep near-IR to provide detections relative to optical: • at least K, preferably also J,H • Spitzer data to allow SED fitting: • at least IRAC 3.6 - 8 micron, preferably also MIPS 24 micron

14. (Brusa et al. 2009) • Starformation in obscured AGN: IRAC colours • Significant fraction (~50%) of z~2-4 obscured AGN show strong SFR signatures (Brusa et al. 2009): • S(8.0)/S(4.5) < 2 classified as SFR-dominated (Pope et al. 2008) Green points: 24m selected field galaxies Blue points: AGN LX < 1043 erg s-1 Red points: AGN LX > 1043 erg s-1 Purple, yellow: other samples of high-z obscured AGN (Pozzi et al 2008, Pope et al 2008)

15. (Brusa et al. 2009) (Brusa et al. 2009) • Starformation in obscured AGN: SFR • Comparison with field galaxies: • For 65% of the obscured AGN, find significant SFR (>10M/yr) • Result is robust for both optically and MIR-selected samples Upper panel: optically-selected field galaxies, AGN Lower panel: MIPS 24m-selected field galaxies, AGN

16. Starformation in obscured AGN: SSFR • Significant fraction of AGN still have active SSFR w.r.t mass (Alonso-Herrero et al. 2008, Busa et al. 2009) • Red colours likely due to dust (not stellar population age)

17. Increased warm dust in z~1-2 obscured AGN: • Lanzuisi et al (2009) sample: SWIRE sources, F24/Fopt > 1000 • X-ray spectral properties used to compare with AGN in the local universe (yellow shaded region) • Result: majority of the AGN have significant MIR excess wrt intrinsic Lx: requires excess dust component relative to local

18. LFs: Della Ceca et al (2009) Yencho et al (2009) Normalized to Treister et al Treister et al. (2009) • Compton-Thick AGN Redshift Evolution Red: Lx > 1043 erg/s Blue: Lx > 1044 erg/s Steep rise above z~2 relative to non-CT AGN

19. Expected contributions of AGN & starbursts to LIR • Examine triggering of AGN and starbursts in mergers (Hopkins et al 2009) - how significant is the contribution from AGN to the MIR LF? • Find that, for log(LIR)>12-13, both starburst & AGN dominate LF To a similar extent, up to z~3

20. HST Morphologies of MIR Excess AGN

21. CDFS (Brandt et al, Koekemoer et al, Treister et al) FX/Fopt = 0.1 FX/Fopt = 10 u u u u u u u • High-z AGN Candidates: “EXOs” • Based on SED change with z: • Drop-outs in z850lp (>27) • Anomalous Fx/Fopt (>100) • Red z850lp - K (>4) • Expect mostly obscuredsources, but unobscuredAGN are not excluded • Highest FX/FOpt: • found in several studies so far (Koekemoer et al 2002,2004; Brusa et al. 2004; Treister et al 2006; Koekemoer et al 2009) • EXO’s - ExtremeX-ray/Optical sources • Only revealed by extending optical depth below ~27 • Optically faint sources with anomalously high FX/FOpt >100 • Typically have extremely red z-K > 4-6 • Appear to have no comparable analogs in the local universe

22. X-ray properties: • Well-detected by Chandra(~10-16-10-15 erg s-1cm-2) • FX/FOpt is a lower limit,and is > ~100x abovethe average for AGN • Generally have soft andhard X-ray emission (excludes z<2 obscured AGN) • Redder z-K colour: • most AGN with FX/FOpt ~ 0.1 - 10 have fairly tight z-K ~1-2, with some slight scatter: • z-K ~ -1 to 2 for quasars/Seyferts • z-K ~ 2 to 4 for ERO’s • However, the EXO high-z candidates generally have z-K >4 Candidate high-z AGN: X-z-K selection:

23. CDFS EXOs

24. EXO/high-z candidates from GOODS N+S HST/ACSVLT+NOAOSPITZER/IRAC MIPS

25. EXO Close-up (contours = Chandra 0.5-8 keV)

26. Overall parameterization: • Use CB03, Maraston06, vary IMF, metallicity • h - stellar mass formed as SSP / CSP • t - reddening (Fall & Charlot; Calzetti; SMC; LMC) • Fits driven by two observational features: • red opt/NIR - IRAC • colours within IRAC • General results: • Use the fits to identify plausible z~2-3 interlopers • Down-select to the remainder that are not well fit by z~2-3: • Follow these up with NIR spectroscopy (Subaru/Gemini/VLT) • Use their number counts to directly constrain high-z logN-logS SED Fitting

27. total ~2 million galaxies, ~1300 AGN (XMM: Hasinger et al, Brusa et al 2007) + radio (Schinnerer et al 2007) ~1.4°

28. COSMOS X-Ray Observations (Comastri, Brusa 2007): • XMM dataset: 53 pointings, ~1.6 sq deg (Brusa et al) • Chandra dataset (1.8 Msec) ~1.4 sq deg, deeper than XMM • Total of ~1300 sources in XMM, ~1800 in Chandra • Limiting fluxes: • F(0.5-2 keV) ~ 1 x 10-15 erg cm-2 s-1 • F(2-10 keV) ~ 5 x 10-15 erg cm-2 s-1 • Full survey: • Combined XMM + Chandra ~2500 sources to date • identify with ACS sources • remove ambiguous IDs • Finally, produce sample of EXOs: • detect a total of 35 EXOs (cf ~7 for GOODS/CDFS) • Likely a mixture of z~2-4 sources and z>~6 candidates, similar to GOODS

29. EXO/high-z candidates from COSMOS ACS-iKSPITZER/IRAC1,2,3,4

30. (cont’d) ACS-iKSPITZER/IRAC1,2,3,4

31. Constraining high-z AGN LF: • Use hard X-ray 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 > 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-rayluminosities at eachredshift bin: • Use the difference tocalculate cumulativenumber N(>6) • Compare with N(>6)from XLF

32. Number of optically unID’d sources N(z) based onz(AB)=27.5 limit, for current Chandra catalogs, using: • X-ray sensitivity • Optical flux limits • Spitzer flux limits • LDDE predicts total of ~5 AGN at z>7 in GOODS+COSMOS (out of all the X-ray sources with ACS coverage): • After removing low-z interlopers, currently have 4 likely z~7 candidates in GOODS+COSMOS - rough agreement with LDDE

33. New results - MOIRCS Spectroscopy (special thanks to Yoichi Oyama!) • NIR spectroscopy of remaining high-z candidates • Subaru - MOIRCS: ~4 hours/mask; objects faint (K~23+) • Data reduction currently in progress • One example EXO spectrum (no lines detected so far..):

34. Summary: • New results by several groups on z~2-4 AGN: • Many AGN are now found to have strong MIR SFR signatures • May be seeing “pre/post-blowout” QSO phase in action • Overall number of high-z candidate AGN found in GOODS+COSMOS supports steepening of LF at z~6-7: • Intermediate-z interlopers successfully accounted for, and removed from dropout samples • Not clear yet whether faint-end slope needs to change as well or whether evolution can be accounted for by change in L* • Next steps: • larger/deeper coverage at MIR with Spitzer and Herschel to better constrain rest-frame opt/IR and dust SEDs • MAIN QUESTION: • What observational signatures to look for in determining the direction of causality for BH / bulge growth?