Evidence for CO shock excitation in NGC 6240 from Herschel SPIRE spectroscopy
R. Meijerink, L. E. Kristensen, A. Weiss, P. P. van der Werf, F. Walter, M. Spaans, A. F. Loenen, J. Fischer, F. P. Israel, K. Isaak, P. P. Papadopoulos, S. Aalto, L. Armus, V. Charmandaris, K. M. Dasyra, T. Diaz-Santos, A. Evans, Y. Gao, E. Gonzalez-Alfonso, R. Guesten, C. Henkel, C. Kramer, S. Lord, J. Martin-Pintado, D. Naylor, D. B. Sanders, H. Smith, L. Spinoglio, G. Stacey, S. Veilleux, M. C. Wiedner
aa r X i v : . [ a s t r o - ph . C O ] N ov Draft version June 22, 2018
Preprint typeset using L A TEX style emulateapj v. 5/2/11
EVIDENCE FOR CO SHOCK EXCITATION IN NGC 6240 FROM HERSCHEL SPIRE SPECTROSCOPY
R. Meijerink , , L.E. Kristensen , A. Wei ß , P.P. van der Werf , F. Walter , M. Spaans , A.F. Loenen , J.Fischer , F.P. Israel , K. Isaak , P.P. Papadopoulos , S. Aalto , L. Armus , V. Charmandaris , K.M. Dasyra ,T. Diaz-Santos , A. Evans , , Y. Gao , E. Gonz´alez-Alfonso , R. G¨usten , C. Henkel , , C. Kramer , S.Lord , J. Mart´ın-Pintado , D. Naylor , D.B. Sanders , H. Smith , L. Spinoglio , G. Stacey , S.Veilleux , and M.C. Wiedner Kapteyn Astronomical Institute, PO Box 800, 9700 AV Groningen, The Netherlands Leiden Observatory, Leiden University, P.O. Box 9513, NL-2300 RA Leiden, Netherlands Max-Planck-Institut f¨ur Radioastronomie, Auf dem H¨ugel 16, Bonn, D-53121, Germany Max-Planck-Institut f¨ur Astronomie, K¨onigstuhl 17, Heidelberg, D-69117, Germany Naval Research Laboratory, Remote Sensing Division, Washington, DC 20375, USA ESA Astrophysics Missions Division, ESTEC, PO Box 299, 2200 AG Noordwijk, The Netherlands Department of Radio and Space Science, Onsala Observatory, Chalmers University of Technology, 43992 Onsala, Sweden Spitzer Science Center, California Institute of Technology, MS 220-6, Pasadena, CA 91125, USA University of Crete, Department of Physics, 71003 Heraklion, Greece Observatoire de Paris, LERMA (CNRS:UMR8112), 61 Av. de l’Observatoire, F-75014, Paris, France Department of Astronomy, University of Virginia, 530 McCormick Road, Charlottesville, VA 22904, USA National Radio Astronomy Observatory, 520 Edgemont Road, Charlottesville, VA 22903, USA Purple Mountain Observatory, Chinese Academy of Sciences, 2 West Beijing Road, Nanjing 210008, PR China Universidad de Alcal´a Henares, Departamente de F´ısica, Campus Universitario, 28871 Alcal´a de Henares, Madrid, Spain Astron. Dept., King Abdulaziz University, P.O. Box 80203, Jeddah, Saudi Arabia Instituto Radioastronomie Millimetrica (IRAM), Av. Divina Pastora 7, Nucleo Central, 18012 Granada, Spain NASA Herschel Science Center, California Institute of Technology, M.S. 10022, Pasadena, CA 91125, USA Departamento de Astrofisica Molecular e Infrarroja-Instituto de Estructura de la Materia-CSIC, Calle Serrano 121, 28006 Madrid,Spain Department of Physics, University of Lethbridge, 4401 University Drive, Lethbridge, Alberta, T1J 1B1, Canada University of Hawaii, Institute for Astronomy, 2680 Woodlawn Drive, Honolulu, HI 96822, USA Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA Istituto di Astrofisica e Planetologia Spaziali, INAF, Via Fosso del Cavaliere 100, I-00133, Roma, Italy Department of Astronomy, Cornell University, Ithaca, NY 14853, USA Department of Astronomy, University of Maryland, College Park, MD 20742, USA and Observatoire de Paris, LERMA, CNRS, 61 Av. de l’Observatoire, 75014 Paris, France
Draft version June 22, 2018
ABSTRACTWe present Herschel SPIRE FTS spectroscopy of the nearby luminous infrared galaxy NGC 6240.In total 20 lines are detected, including CO J = 4 − J = 13 −
12, 6 H O rotational lines,and [C i ] and [N ii ] fine-structure lines. The CO to continuum luminosity ratio is 10 times higher inNGC 6240 than Mrk 231. Although the CO ladders of NGC 6240 and Mrk 231 are very similar, UVand/or X-ray irradiation are unlikely to be responsible for the excitation of the gas in NGC 6240. Weapplied both C and J shock models to the H v = 1 − S (1) and v = 2 − S (1) lines and the COrotational ladder. The CO ladder is best reproduced by a model with shock velocity v s = 10 km s − and a pre-shock density n H = 5 × cm − . We find that the solution best fitting the H linesis degenerate: The shock velocities and number densities range between v s = 17 −
47 km s − and n H = 10 − × cm − , respectively. The H lines thus need a much more powerful shock thanthe CO lines. We deduce that most of the gas is currently moderately stirred up by slow (10 km s − )shocks while only a small fraction ( . Subject headings: galaxies: individual (NGC 6240) — galaxies: active — galaxies: nuclei — galaxies:starburst — infrared: galaxies INTRODUCTION
We present
Herschel SPIRE FTS (Griffin et al. 2010)observations of the nearby luminous infrared galaxyNGC 6240 (IRAS 16504+0228, UGC 10592). With aredshift of z = 0 . H = 70 .
50 km s − Mpc − , Ω matter = 0 . vacuum = 0 .
73) NGC 6240 is at a luminosity distance of D L = 107 Mpc, with 1 ′′ corresponding to 492 pc. The de- Herschel is an ESA space observatory with science instrumentsprovided by European-led Principal Investigator consortia and withimportant participation from NASA rived 8 − µ m luminosity of the merger galaxy NGC6240 is L IR = 7 . × L ⊙ . Different power sources aresuggested for this infrared luminosity, which complicatethe interpretation of observations. The source is a strongX-ray emitter. Based on modeling of Chandra between0 . − L X ,N = 1 . × erg s − and L X ,S = 0 . × erg s − for the northern and southernAGN cores in the energy range 0 . −
10 keV. However,using BeppoSAX observations, Vignati et al. (1999) findthat the intrinsic hard X-ray luminosity shows up above9 −
10 keV, and their models of the higher energy yields L (2 −
10 keV) ≈ . × erg s − . Hydrogen re-combination lines (Rieke et al. 1985; Depoy et al. 1986;Elston & Maloney 1990; van der Werf et al. 1993) andluminous PAH emission (Armus et al. 2006) in the twonuclei indicate recent star formation. The H v =1 − S (1) 2.12 µ m line, presented by van der Werf et al.(1993), shows a peak in the overlap region between thesetwo nuclei, which are 2 ′′ apart. This H emission extendsover several kpc and shows a complex morphology. Theauthors conclude that the bright H emission between thenuclei is generated in shocks resulting from the collisionof the interstellar media (ISM) of the merging galaxies.More recently, Engel et al. (2010) observed the same lineat high spatial (0.5 ′′ ) and spectral ( ∼
90 km s − ) reso-lution. The line profiles also indicate multiple compo-nents, and the dispersion map suggests that the gas ishighly disturbed and turbulent. Interferometric observa-tions of several CO, HCN, and HCO + lines show a high-density peak between the two nuclei (Tacconi et al. 1999;Nakanishi et al. 2005; Iono et al. 2007). Tacconi et al.(1999) conclude that NGC 6240 is in an earlier mergingstage than Arp 220, the prototypical ultra-luminous in-frared galaxy, and state that the gas is in the process ofsettling between the two nuclei, and dissipating angularmomentum rapidly. Engel et al. (2010) also presented a CO J = 2 − and CO emission are coextensive, but do not co-incide with the stellar emission distribution.This paper is ordered as follows. In Section 2, the ob-servations, data reduction and line luminosities are dis-cussed. In Section 3, we discuss the different excitationcomponents (PDR, XDR, shocks) in combination withthe geometry of NGC 6240. Consequently, we composea CO ladder using both the available SPIRE FTS andground based CO data from Papadopoulos et al. (2011)and compare this system to the ULIRG Mrk 231 (pre-viously studied by us with Herschel , van der Werf et al.2010). Then we analyze this with shock models fromKristensen et al. (2007). In Section 4, we conclude witha discussion and a summary of the results. OBSERVATIONS, DATA REDUCTION, AND RESULTS
NGC 6240 was observed (Observations ID 1342214831)in staring mode with the SPIRE FTS on February 27,2011, as part of the
Herschel
Open Time Key ProgramHerCULES (P.I. Van der Werf). The high spectral res-olution mode was used, yielding a resolution of 1 . ν = 447 −
989 GHz ( λ = 671 − µ m) andthe high frequency band covering ν = 958 − λ = 313 − µ m). In total 97 repetitions (194 FTSscans) were carried out, yielding an on source integrationtime of 12920 s (3.6 hrs) for each band. A reference mea-surement was used to subtract the combined emissionfrom the sky, telescope and instrument. The data wereprocessed and calibrated using HIPE version 6.0. Theextent of the CO J = 3 − ′′ (Wilson et al.2008), while the SPIRE beam varies from 17 ′′ to 42 ′′ overour spectrum. Therefore, a point source calibration pro-cedure was adopted, and no corrections for wavelengthdependent beam coupling factors were necessary.The full SPIRE FTS spectrum of NGC 6240 is shownin Fig. 1. In the overlap region between the two fre- quency bands (958 −
989 GHz), the noisy parts of thetwo spectrometer bands were clipped and plotted on topof each other. A total of 20 lines were detected of whichone line at the observed frequency ν rest = 1481 . J = 4 − J = 13 −
12, 6 H O lines, [C i ] 370 and 609 µ m, and [N ii ]205 µ m. A well-constrained upper limit was obtained forthe CO J = 6 − ± ± ± ± . transitions (Tecza et al. 2000). From thesefour H lines, Tecza et al. (2000) estimated an unatten-uated power of L H ≈ × L ⊙ . ANALYSIS
We will focus on the excitation of the CO ladder, andcompare CO excitation conditions to those of H . Amulticomponent Large Velocity Gradient (LVG) analy-sis of the CO ladder, considering also constraints fromthe HCN and [CI] lines will be presented in anotherpaper (Papadopoulos et al., in prep.). Preliminarymass estimates from this study are consistent with thosefound by Papadopoulos et al. (2012), Greve et al. (2009),Tacconi et al. (1999), and Engel et al. (2010). Estimatesrange between M total = 3 × and 4 × M ⊙ , de-pending on whether low and/or high density tracers areused.The total luminosity in the CO lines listed in Table1 is L CO ≈ × L ⊙ . In the absence of J up >
13 mea-surements this is a lower limit to the total luminosity inthe CO lines. Although the line luminosity of the CO J = 8 − J up >
13. The shape ofthe CO ladder of NGC 6240 is similar to that of Mrk 231(see Fig. 2) and can be fitted by two photon-dominatedregion (PDR) models and an X-ray dominated region(XDR) model (see van der Werf et al. 2010). The physi-cal and geometrical properties of NGC 6240 are differentfrom those of Mrk 231. We argue below why the fit usedfor Mrk 231 is not appropriate for NGC 6240, and whyshocks must be responsible for the CO excitation in NGC6240:
Absence of OH + and H O + : The NGC 6240 spec-
Fig. 1.—
Full SPIRE FTS spectrum of NGC 6240 and zoom-in on line blends, with on the x-axis the frequency and on the y-axis the fluxdensity in Jy. The detected lines are marked: CO (red), [C i ] and [N ii ] (green), H O (orange), and unidentified (brown). In the overlapregion between the two frequency bands (958 −
989 GHz), the noisy parts were clipped and plotted on top of each other.
Fig. 2.—
Comparison of the CO ladder of NGC 6240 to thoseobtained for Mrk 231 (van der Werf et al. 2010). The CO ladderof Mrk 231 is normalized to the CO J = 8 − trum does not show emission line features of theionic species OH + and H O + , observed in Mrk 231(van der Werf et al. 2010). Large OH + and H O + abun-dances are only sustained in gas clouds with high ion-ization fractions ( x e > − ), which are produced byelevated cosmic ray or X-ray fluxes (cf., Meijerink et al.2011). Their absence hints that the bulk of the gas is notexposed to high ionization rates resulting from AGN orstarburst/supervae activity. Line-to-continuum ratio:
The CO luminosity (up tothe J = 13 −
12 transition) to infrared (measured between8 − µ m) luminosity ratio is L CO /L IR = 7 × − in NGC 6240. This is exceptionally high, and approx-imately an order of magnitude higher than the ratio found for Mrk 231 (van der Werf et al. 2010) and Arp220 (Rangwala et al. 2011). An exceptionally high lineto FIR continuum ratio is also found for the H linesof NGC 6240 (van der Werf et al. 1993). Our PDR andXDR models (Meijerink & Spaans 2005; Meijerink et al.2007) give a maximum L CO /L IR ratio of ∼ − (as-suming I (FIR) = 2 . × − G erg s − cm − sr − , seeKaufman et al. 1999), where the XDR ratios are high-est. Most of the absorbed photons in a PDR will heatthe dust. An AGN (creating an XDR) generates a UVcontinuum which contains approximately ten times moreenergy than the X-ray field, and also heats the dust effi-ciently. In shocks, on the contrary, the gas is compressedand heated to higher temperatures, while the dust is notaffected (except for shock velocities and densities that areorders of magnitude higher). Assuming that shocks arenot heating the dust and that all the far-infrared lumi-nosity is reprocessed radiation from the AGN, we obtaina maximum AGN contribution of 10 −
15 percent. So,a shock dominated ISM can yield a much larger line-to-continuum ratio than PDRs and XDRs and this is exactlywhat we see in NGC 6240.
Geometry of NGC 6240:
The bulk of the gas mass doesnot coincide with the two AGN nuclei or with star for-mation. Engel et al. (2010) relate gas masses, as tracedby the CO J = 2 − ′′ . Our FTS beam ( > ′′ )is larger than the galaxy, and traces the total CO emis-sion. We determined a FWHM of 450 ±
40 km s − for theCO J = 13 −
12 emission with a gaussian × sinc profile fit(larger than the instrumental resolution of 245 km s − TABLE 1Derived fluxes from the SPIRE FTS observationssupplemented with ground-based observations
Line ν obs λ rest S line L line [Ghz] [ µ m] [Jy km s − ] [L ⊙ ]CO J = 1 − a ±
29 4 . × ,d CO J = 2 − a ±
250 4 . × CO J = 3 − a ±
640 1 . × CO J = 4 − ±
370 2 . × CO J = 5 − ±
150 3 . × CO J = 6 − ±
82 4 . × CO J = 7 − ±
60 5 . × CO J = 8 − ±
89 6 . × CO J = 9 − ±
82 5 . × CO J = 10 − ±
67 5 . × CO J = 11 −
10 1236.9 236.613 3160 ±
75 4 . × CO J = 12 −
11 1349.1 216.927 2590 ±
60 4 . × CO J = 13 −
12 1461.2 200.272 2080 ±
60 3 . × CO J = 6 − b < ±
67 8 . × [C i ] P − P ±
220 1 . × [C i ] P − P ±
52 3 . × [N ii ] P − P ±
77 5 . × H O 2 − ±
52 4 . × H O 2 − ±
75 9 . × H O 3 − ±
60 7 . × H O 1 − ±
52 3 . × H O 3 − ±
60 8 . × H O 2 − ±
60 5 . × UID line 1472.7 198.689 277 ±
60 4 . × H v = 1 − S (1) c . × H v = 1 − S (0) c . × H v = 2 − S (1) c . × H v = 2 − S (2) c . × a Papadopoulos et al. (2011) b upper limit c Tecza et al. (2000), extinction corrected values, luminosities ad-justed to adopted distance d relative uncertainties for L line are the same as for S line at this frequency range). This is similar to the FWHMof the CO J = 2 − J = 13 −
12 emis-sion is also located in this region. The projected dis-tance between these nuclei on the sky is ∼
750 pc, andthe true distance between them is estimated at 1.4 kpc(Tecza et al. 2000). These authors assume that the twonuclei are on a circular orbit, of which the position angleand inclination are the same as that of the CO-disk be-tween the two nuclei (Tacconi et al. 1999), and that thevelocity difference between the two nuclei is 150 km s − .The AGN X-ray luminosities are not enough to powerthe CO excitation of the gas that is residing betweenthe two nuclei, since they are geometrically diluted andabsorbed: The luminosities derived by Komossa et al.(2003) give an X-ray flux at 250 pc from the AGN of < − s − . Vignati et al. (1999) derive a muchlarger luminosity by an obscured AGN, with an absorb-ing screen of N H ∼ cm − which also reduces theX-ray fluxes below < − s − . This leaves roomfor an XDR component located near the AGN nuclei,but even the most optimistic estimate of the X-ray lu-minosity is not enough to explain the combined H andCO luminosity of L = 2 . × L ⊙ = 9 . × erg s − .This would require a strong coupling of the X-rays to the Fig. 3.—
C-shock model (solid line) with density n H = 5 × cm − and shock velocity v s = 10 km s − overlaid on theobserved CO line fluxes (diamonds) for NGC 6240. molecular gas. Shock modeling
Given our arguments discussed above and the fact thatvarious papers (cf., van der Werf et al. 1993; Tecza et al.2000) have argued that C type shocks are exciting theH lines in NGC 6240, we analyze the CO ladder withshock models. Two types of shocks are used: a mag-netic continuous ( C -type) and a non-magnetic jump ( J -type) shock model. In both types, ro-vibrational lev-els of H are excited through high temperature H − H ,H − H , and He-H collisions. As the shock develops,the temperature of the gas becomes such that the ex-cited vibrational states become populated. We use theFlower & Pineau des Forˆets (2003) shock code to modelthe H v = 1 − S (1) /v = 2 − S (1) ratio and theCO ladder. Kristensen et al. (2007) used this model tocalculate a grid, spanning hydrogen number densities be-tween n H = 10 and 10 cm − , velocities between v = 10and 50 km s − , and transverse magnetic field densities b × n / µ G, with b=1 and 5. The magnetic field den-sity relation implies values between 100 and 3000 µ G forthe adopted densities, which is within the range of valuesobserved for galactic molecular clouds (Crutcher 1999).Using a rotational diagram and assuming that the COlines are optically thin, we find T rot = 66 K for theCO rotational transitions between CO J = 5 − J = 7 −
6. This rotational temperature increases to val-ues T rot ∼
150 K at the highest CO transitions. We notethat for the highest transitions the rotational temper-ature is a lower limit to the kinetic temperature, sincethese transitions are slightly sub-thermally excited. Us-ing a chi-square fit and including the errors provided inTable 1, the best-fitting C-type shock model has a hydro-gen number density of n H = 5 × cm − and velocityof v = 10 km s − (see Fig. 3). The density is well con-strained, and chi-square values increase by an order ofmagnitude when going to densities n = 10 or 10 cm − .The uncertainty in the shock velocity is a few km s − .Downstream in this particular model, the CO gas hasbeen compressed by a factor of 7.2 and the post-shockdensity of the CO emitting gas is 3 . × cm − .The v = 1 − S (1) and v = 2 − S (1) H lines have up-per level energies of E = 6500 and 12500 K, respectively,and are therefore only excited when the gas temperature T & v = 1 − S (1) /v = 2 − S (1) ratio of ∼ n H = 5 × cm − and a shock velocity v s = 47 km s − to a pre-shockdensity of n H = 10 cm − combined with a velocity of v s = 16 km s − . DISCUSSION AND CONCLUSIONS
Shock modeling:
Combining the model results for theH and CO emission, we find that the H v = 1 − S (1)to CO J = 10 − ∼ / CO line luminosity ratio isapproximately ∼ .
5. The low density, low velocity shockmodel fitting the CO lines, has a low temperature anddoes not produce H emission. From this we concludethat only a very small fraction of the gas mass is currentlyexposed to very powerful shocks (with either very highdensities, n H = 10 cm − and moderate shock velocities v s = 16 km s − or moderate densities, n H = 4 × cm − ,combined with a high shock velocity v s ∼
50 km s − ).Most of the shocked gas is settling and equilibrating withthe ambient ISM, which is in agreement with the fastdissipation timescale derived below. Dissipation timescales:
If we assume that all the gasis colliding at a shock velocity v s = 50 km s − (which isthe highest velocity allowed by the models reproducingthe H lines), the total amount of energy available (foran adopted gas mass of M = 1 . × M ⊙ , in themiddle of the LVG masses derived by various authors) is E = 0 . M v s = 4 . × erg, which is very similar tothe value derived by Tacconi et al. (1999). Assuming noadditional energy input and that CO traces the bulk ofthe gas mass, this would imply that the CO shock energyis dissipated away within 6.6 million years, well withinthe orbital timescale of the two nuclei of 30 million years(Tacconi et al. 1999). Excitation by UV/X-rays vs. Shocks:
The observed H v = 1 − S (1) / v = 2 − S (1) ratio falls within therange of ratios that are produced by X-ray dominatedregions (see, e.g., Maloney et al. 1996, Fig. 6). Also, theCO ladder of NGC 6240 resembles the one observed forMrk 231 (van der Werf et al. 2010). However, as men-tioned before, NGC 6240 is lacking the bright OH + andH O + lines, which are associated with gas clouds thatare exposed to extremely high cosmic ray or X-ray ion-ization rates (Meijerink et al. 2011), and the available X-ray photons are not sufficient to dominate the chemistryand thermal balance of the bulk of the gas (see Sect. 3.1). The H O lines are less luminous than in Mrk 231. Thisimplies either lower water abundances or a less efficientmode of excitation. A full analysis is beyond the scope ofthis paper. These results and the similarity to the high- J CO line distribution in Mrk 231 suggest that shocks pos-sibly due to the massive molecular outflow (Fischer et al.2010; Feruglio et al. 2010; Sturm et al. 2011) may alsocontribute to the CO line emission in Mrk 231. Indeed,based on millimeter interferometric observations of Mrk231, Cicone et al. (2012) note that in the inner region(
R < . − −
0) ratio is slightlyhigher, indicative of a shock contribution.
Observing strategies for ALMA and other sub-millimeter facilities:
Although the observed CO laddersfor Mrk 231 and NGC 6240 are practically indistinguish-able, we argue that shock excitation and not X-rays areresponsible for the excitation of the CO ladder, based onour knowledge of the geometry of the NGC 6240. Suchan analysis will not be possible in the study of the ISMin galaxies at high redshift. At those distances, we areunable to resolve the offset between the two radio nu-clei and the location of the CO emitting gas, and X-rayfluxes cannot be determined. Therefore, we have to relyon dust continuum and line emission at far-infrared andsub-millimeter wavelengths.