1. der Paul van der Werf Leiden Observatory Star formation and molecular gas in (U)LIRGs Sant Cugat April 17, 2012

2. Credits Kate Isaak (ESTEC) Padelis Papadopoulos (MPI für Radioastronomie) Marco Spaans (Kapteyn Astronomical Institute) Eduardo González-Alfonso (Henares) Rowin Meijerink (Leiden Observatory) Edo Loenen (Leiden Observatory) Alicia Berciano Alba (Leiden Observatory/ASTRON) Axel Weiß (MPI für Radioastronomie) + the HerCULES team 2 Molecular gas in (U)LIRGs

3. Conditions in ULIRGs  Starbursts cannot be simply scaled up. be simply scaled up.  More intense starbursts are also more efficient are also more efficient with their fuel. with their fuel. (Gao & Solomon 2001) L IR  SFR L IR / L CO  SFR/ M H 2 SFE 3Molecular gas in (U)LIRGs

4. (U)LIRGs from low to high z   LIRGs dominate cosmic star formation at high redshift 4 Molecular gas in (U)LIRGs (Magnelli et al. 2011)

5. ISM in luminous high-z galaxies 5  Even in ALMA era, limited spatial resolution on high-z galaxies.  For unresolved galaxies, multi-line spectroscopy will be a key diagnostic (Weiß et al. 2007) Molecular gas in (U)LIRGs (Danielson et al. 2010)

6. Outline   CO as a probe of the energy source un (U)LIRGs   H 2 O emission in (U)LIRGs   The X-factor in (U)LIRGs 6 Molecular gas in (U)LIRGs

7. HerCULES 7 Herschel Comprehensive (U)LIRG EmissionSurvey Open Time Key Program on the Herschel satellite Molecular gas in (U)LIRGs

8. Who is HerCULES? Paul van der Werf (Leiden; PI) Susanne Aalto (Onsala) Lee Armus (Spitzer SC) Vassilis Charmandaris (Crete) Kalliopi Dasyra (CEA) Aaron Evans (Charlottesville) Jackie Fischer (NRL) Yu Gao (Purple Mountain) Eduardo González-Alfonso (Henares) Thomas Greve (Copenhagen) Rolf G ü sten (MPIfR) Andy Harris (U Maryland) Chris Henkel (MPIfR) Kate Isaak (ESA) Frank Israel (Leiden) Carsten Kramer (IRAM) Edo Loenen (Leiden) Steve Lord (NASA Herschel SC) Jesus Martín-Pintado (Madrid) Joe Mazzarella (IPAC) Rowin Meijerink (Leiden) David Naylor (Lethbridge) Padelis Papadopoulos (Bonn) Dave Sanders (U Hawaii) Giorgio Savini (Cardiff/UCL) Howard Smith (CfA) Marco Spaans (Groningen) Luigi Spinoglio (Rome) Gordon Stacey (Cornell) Sylvain Veilleux (U Maryland) Cat Vlahakis (Leiden/Santiago) Fabian Walter (MPIA) Axel Wei ß (MPIfR) Martina Wiedner (Paris) Manolis Xilouris (Athens) 8 Molecular gas in (U)LIRGs

9. HerCULES in a nutshell   HerCULES has uniformly and statistically measured the neutral gas cooling lines in a flux-limited sample of 29 (U)LIRGs.   Sample:   all IRAS RBGS ULIRGs with S 60 > 12.19 Jy (6 sources)   all IRAS RBGS LIRGs with S 60 > 16.8 Jy (23 sources)   Observations:   SPIRE/FTS full high-resolution scans: 200 to 670  m at R ≈ 600, covering CO 4 — 3 to 13 — 12 and [CI] + any other bright lines   PACS line scans of [CII] and both [OI] lines   All targets observed to same (expected) S/N   Extended sources observed at several positions 9 Molecular gas in (U)LIRGs

10. HerCULES sample Target log(L IR /L  ) Mrk 231 12.51 IRAS F17207—0014 12.39 IRAS 13120—5453 12.26 Arp 220 12.21 Mrk 273 12.14 IRAS F05189—2524 12.11 Arp 299 11.88 NGC 6240 11.85 IRAS F18293—3413 11.81 Arp 193 11.67 IC 1623 11.65 NGC 1614 11.60 NGC 7469 11.59 NGC 3256 11.56 Target log(L IR /L  ) IC 4687/4686 11.55 NGC 2623 11.54 NGC 34 11.44 MCG+12—02—001 11.44 Mrk 331 11.41 IRAS 13242—5713 11.34 NGC 7771 11.34 Zw 049.057 11.27 NGC 1068 11.27 NGC 5135 11.17 IRAS F11506—3851 11.10 NGC 4418 11.08 NGC 2146 11.07 NGC 7552 11.03 NGC 1365 11.00 10 Molecular gas in (U)LIRGs

11. Warning: may contain... 11  quiescent molecular (and atomic) gas  star-forming molecular gas (PDRs)  AGN (X-ray) excited gas (XDRs)  cosmic ray heated gas  shocks  mechanically (dissipation of turbulence) heated gas  warm very obcured gas (hot cores) Molecular gas in (U)LIRGs

12. PDRs vs. XDRs Four differences:  X-rays penetrate much larger column densities than UV photons  Gas heating efficiency in XDRs is ≈10 — 50%, compared to <3% in PDRs  Dust heating much more efficient in PDRs than in XDRs  High ionization levels in XDRs drive ion-molecule chemistry over large column density 12 Molecular gas in (U)LIRGs

13. PDRs vs. XDRs: CO lines  XDRs produce larger column densities of warmer gas  Identical incident energy densities give very different CO spectra  Very high J CO lines are excellent XDR tracers  Need good coverage of CO ladder (Spaans & Meijerink 2008) 13 Molecular gas in (U)LIRGs

14. 14 Mrk231 HST/ACS (Evans et al., 2008)  At z=0.042, one of the closest QSOs (D L =192 Mpc)  With L IR = 4  10 12 L , the most luminous ULIRG in the IRAS Revised bright Galaxy Sample  “Warm” infrared colours  Star-forming disk (~500 pc radius) + absorbed X-ray nucleus  Face-on molecular disk, M H 2 ~ 5  10 9 M  Molecular gas in (U)LIRGs

15. Mrk231 SPIRE FTS 15 (Van der Werf et al., 2010) Molecular gas in (U)LIRGs

16. CO excitation 2 PDRs + XDR 6.4:1:4.0 n=10 4.2, F X =28 * n=10 3.5, G 0 =10 2.0 n=10 5.0, G 0 =10 3.5 * 28 erg cm -2 s -1  G 0 =10 4.2 16 Molecular gas in (U)LIRGs

17. CO excitation 3 PDRs 6.4:1:0.03 n=10 6.5, G 0 =10 5.0 n=10 3.5, G 0 =10 2.0 n=10 5.0, G 0 =10 3.5 17 Molecular gas in (U)LIRGs

18. High-J lines: PDR or XDR?   High-J CO lines can also be produced by PDR with n=10 6.5 cm —3 and G 0 =10 5, containing half the molecular gas mass.   Does this work?   G 0 =10 5 only out to 0.3 pc from O5 star; then we must have half of the molecular gas and dust in 0.7% of volume.   With G 0 =10 5, 50% of the dust mass would be at 170K, which is ruled out by the Spectral Energy Distribution   [OH + ] and [H 2 O + ] > 10 —9 in dense gas requires efficient and penetrative source of ionization; PDR abundances factor 100— 1000 lower 18 Molecular gas in (U)LIRGs

19. High-J CO lines and AGNs Highly excited CO ladders are found in all high luminosity/compact sources with an energetically dominant AGN (and only in those sources). 19 Molecular gas in (U)LIRGs

20. Outline   CO as a probe of the energy source in (U)LIRGs   H 2 O emission in (U)LIRGs   The X-factor in (U)LIRGs 20 Molecular gas in (U)LIRGs

21. Water in molecular clouds  H 2 O ice abundant in molecular clouds  Can be released into the gas phase by UV photons, X-rays, cosmic rays, shocks,...  Can be formed directly in the gas phase in warm molecular gas  Abundant, many strong transitions  expected to be major coolant of warm, dense molecular gas 21 Molecular gas in (U)LIRGs Herschel image of (part of) the Rosetta Molecular Cloud

22. Molecular gas in (U)LIRGs 22 Low-z H 2 O: M82 vs. Mrk231  At D = 3.9 Mpc, one of the closest starburst galaxies  With L IR = 3  10 10 L , a very moderate starburst  At z=0.042, one of the closest QSOs (D L =192 Mpc)  With L IR = 4  10 12 L , the most luminous ULIRG in the IRAS Revised bright Galaxy Sample M82Mrk231

23. H 2 O lines in M82  Faint lines, complex profiles  Only lines of low excitation 23 Molecular gas in (U)LIRGs (Weiß et al., 2010)(Panuzzo et al., 2010)

24. Mrk231: strong H 2 O lines, high excitation 24 Molecular gas in (U)LIRGs (Van der Werf et al., González- Alfonso et al., 2010)

25. H 2 O lines in Mrk231  Low lines: pumping by cool component + some collisional excitation  High lines: pumping by warm component  Radiative pumping dominates and reveals an infrared-opaque (  100  m ~ 1) disk. (González-Alfonso et al., 2010) 25 Molecular gas in (U)LIRGs

26. High-z connection: H 2 O at z=3.9 26Molecular gas in (U)LIRGs  Line ratios similar to Mrk231  FIR pumping dominates, implies 100  m-opaque disk  Radiation pressure dominates, Eddington-limited Van der Werf et al., 2011

27. H 2 O in HerCULES Target log(L IR /L  ) Mrk 231 12.51 IRAS F17207—0014 12.39 IRAS 13120—5453 12.26 Arp 220 12.21 Mrk 273 12.14 IRAS F05189—2524 12.11 Arp 299 11.88 NGC 6240 11.85 IRAS F18293—3413 11.81 Arp 193 11.67 IC 1623 11.65 NGC 1614 11.60 NGC 7469 11.59 NGC 3256 11.56 Target log(L IR /L  ) IC 4687/4686 11.55 NGC 2623 11.54 NGC 34 11.44 MCG+12—02—001 11.44 Mrk 331 11.41 IRAS 13242—5713 11.34 NGC 7771 11.34 Zw 049.057 11.27 NGC 5135 11.17 IRAS F11506—3851 11.10 NGC 4418 11.08 NGC 2146 11.07 NGC 7552 11.03 NGC 1365 11.00 27 Molecular gas in (U)LIRGs red = wet

28. Lessons from H 2 O (1 + 2 + 3 + 4) 28Molecular gas in (U)LIRGs 1) In spite of high luminosities, H 2 O lines are unimportant for cooling the warm molecular gas. 2) Radiatively H 2 O lines reveal extended infrared-opaque circumnuclear gas disks. 3) Extinction and radiative pumping of highest CO lines.4) Detection of H 2 O lines implies high FIR radiation field, but not the presence of an AGN.

29. Lessons from H 2 O (5) Radiation pressure from the strong IR radiation field: 29Molecular gas in (U)LIRGs Since both  100 and T d are high, radiation pressure dominates the gas dynamics in the circumnuclear disk. 5) Conditions in the circumnuclear molecular disk are Eddington-limited.

30. Mechanical feedback   Radiation pressure can drive the observed molecular outflows (e.g., Murray et al., 2005)   Aalto et al., 2012: flow prominent in HCN  dense gas   Key process in linking ULIRGs and QSOs?   Shocks probably of minor importance in Mrk231 30 Molecular gas in (U)LIRGs (Feruglio et al., 2010) (Fischer et al., 2010)

31. Possible consequences   Eddington-limited conditions account for L FIR -L HCN relation within factor 2 for standard dust/gas ratio (Andrews & Thompson 2011)   If accreting towards nuclear SMBH, can account for M * /M  relation (Thompson et al., 2005)   Radiation pressure can expel nuclear fuel and produce Faber- Jackson relation (Murray et al., 2005) 31 Molecular gas in (U)LIRGs (Andrews & Thompson, 2011)

32. Outline   CO as a probe of the energy source in (U)LIRGs   H 2 O emission in (U)LIRGs   The X-factor in (U)LIRGs 32 Molecular gas in (U)LIRGs

33. The X-factor   Converting CO flux (luminosity) into H 2 column density (mass): (MW:  =4; ULIRGs:  =0.8) Warning: discussions of the X- factor have been the death-blow for many conferences. 33 Molecular gas in (U)LIRGs (Papadopoulos, Van der Werf, Isaak & Xilouris, astro-ph/1202.1803) (U)LIRG sample

34. The X-factor and optical depth Highly turbulent motions  low optical depths (  high 12 CO/ 13 CO line ratios)  low X-factor (e.g., ULIRGs: X=0.8,Downes & Solomon 1997) 34 Molecular gas in (U)LIRGs NB: T line is T b of the line, not kinetic temperature

35. Calculating X-factors   Sample of 70 (U)LIRGs, 12 CO 1  0, 2  1, 3  2, (4  3, 6  5) and 13 CO 1  0, (2  1) lines   2 component model: high excitation and low excitation component   X-factor explicitly calculated using non-LTE excitation model and finite optical depths 35 Molecular gas in (U)LIRGs

36. X-factors based on low CO lines only Assumptions:   One gas component   Based on low-J CO lines only Results:   X-factors cluster between 0.5 and 1 (cf., Downes/Solomon value of 0.8 for ULIRGs) with tail up to and exceeding Milky Way values   Subthermal excitation But:   Higher CO lines and density tracing lines reveal a substantial dense gas component 36 Molecular gas in (U)LIRGs (Papadopoulos, Van der Werf, Isaak & Xilouris, astro-ph/1202.1803) (U)LIRG sample

37. X-factors in a 2-component model Assumptions:   Orion-like high excitation component (probably not needed with denser coverage of CO ladder  Herschel)   Radiation pressure supported disk  L FIR /M(star forming gas) = constant (observed: L FIR /L HCN = constant; e.g., Gao & Solomon, Wu et al.) Results:   X-factors in (U)LIRGs dominated by dense, star-forming gas   X-factors have values of the order 2  4 (i.e., significantly higher than the Downes/Solomon value of 0.8) 37 Molecular gas in (U)LIRGs

38. Summary: getting at X   Derive X from 12 CO 1  0  6  5 + 13 CO 1  0 (or 2  1); ideally, use HCN (or HCO + or CS) lines as well   Can be used at high z too, but 13 CO is challenging   Aside: why did Downes & Solomon get it right (for the low-J lines) without 13 CO?   High quality data showing turbulent velocity field   Correctly produced low optical depth   Getting the optical depth right is key 38 Molecular gas in (U)LIRGs

39. Summary 39 Molecular gas in (U)LIRGs  CO-H 2 conversion factor can be derived from multi-line CO data (up to J=6) – see Papadopoulos et al., 2012ab (astro-ph/1109.4176 and 1202.1803)  Multi-line CO data (up to at least J=11) can separate star formation and AGN accretion as power sources of unresolved galaxies (vdW et al., 2010)  Luminous H 2 O lines trace infrared-opaque nuclear disks and reveal Eddington- limited circumnuclear conditions (vdW et al., 2011)  Strong shocks are only a minor contributor to CO excitation, barring exceptional cases (in preparation).  OH + emission requires a high electron abundance and forms another way to reveal XDRs (in preparation).