Quasar Feedback in the Ultraluminous Infrared Galaxy F11119+3257: Connecting the Accretion Disk Wind with the Large-Scale Molecular Outflow
S. Veilleux, A. Bolatto, F. Tombesi, M. Melendez, E. Sturm, E. Gonzalez-Alfonso, J. Fischer, D. S. N. Rupke
aa r X i v : . [ a s t r o - ph . GA ] J un D RAFT VERSION J ULY
9, 2018
Preprint typeset using L A TEX style emulateapj v. 12/16/11
QUASAR FEEDBACK IN THE ULTRALUMINOUS INFRARED GALAXY F11119 + S. V
EILLEUX , A. B
OLATTO , F. T
OMBESI , M. M
ELÉNDEZ , E. S
TURM , E. G ONZÁLEZ -A LFONSO , J. F ISCHER , & D. S.N. R UPKE Draft version July 9, 2018
ABSTRACTIn Tombesi et al. (2015), we reported the first direct evidence for a quasar accretion disk wind driving amassive molecular outflow. The target was F11119 + µ m transition profile observed with Herschel . Here, we independently confirm thepresence of the molecular outflow in F11119 + − −
0) line emission appears to be spatially extendedon a scale of at least ∼ − R − M ⊙ yr − , (1.5 − R − L AGN / c , and (0.15 − R − L AGN are inferred from these data,assuming a CO − to − H conversion factor appropriate for a ULIRG ( R is the radius of the outflow normalizedto 7 kpc and L AGN is the AGN luminosity). These rates are time-averaged over a flow time scale of 7 × yrs.They are similar to the OH-based rates time-averaged over a flow time scale of 4 × yrs, but about a factor 4smaller than the local (“instantaneous”; . yrs) OH-based estimates cited in Tombesi et al. The implicationsof these new results are discussed in the context of time-variable quasar-mode feedback and galaxy evolution.The need for an energy-conserving bubble to explain the molecular outflow is also re-examined. Subject headings: galaxies: active — galaxies: evolution — ISM: jets and outflows — quasars: general —quasars: individual (F11119 + INTRODUCTION
Rapidly accreting supermassive black holes (SMBHs) pro-duce tremendous amounts of radiative energy. The couplingof this energy with gas near the black hole or at larger scalesin the host galaxy produces outflows of gas. These “quasar-mode” outflows are distinct from “radio-mode” jets in thatthey are much less collimated, and therefore have the poten-tial to impact a much greater swath of a galaxy’s gas. Quasar-mode outflows are often invoked to play a fundamental role inthe evolution of both SMBHs and their host galaxies, quench-ing star formation and explaining the tight SMBH-galaxy re-lations (e.g., Veilleux et al. 2005; Fabian 2012). Recent ob-servations of large-scale neutral and molecular outflows in Department of Astronomy, University of Maryland, College Park, MD20742, USA; [email protected] Joint Space-Science Institute, University of Maryland, College Park,MD 20742, USA X-ray Astrophysics Laboratory, NASA Goddard Space Flight Center,Greenbelt, MD 20771, USA Department of Astronomy and CRESST, University of Maryland, Col-lege Park, MD 20742, USA Dipartimento di Fisica, Università di Roma Tor Vergata, Via dellaRicerca Scientifica 1, I-00133 Roma, Italy NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA Wyle Science, Technology and Engineering Group, 1290 Hercules Av-enue, Houston, TX 77058 USA Max-Planck-Institute for Extraterrestrial Physics (MPE), Giessen-bachstrasse 1, D-85748, Garching, Germany Departamento de Física y Matemáticas, Universidad de Alcalá, Cam-pus Universitario, E-28871 Alcalá de Henares, Madrid, Spain Naval Research Laboratory, Remote Sensing Division, 4555 Over-look Avenue SW, Washington, DC 20375, USA Department of Physics, Rhodes College, Memphis, TN 38112, USA (U)LIRGs have provided supporting evidence for this idea, asthey directly trace the gas out of which stars form (e.g., Fis-cher et al. 2010; Feruglio et al. 2010, 2015; Sturm et al. 2011;Alatalo et al. 2011, 2015; Rupke & Veilleux 2011, 2013a,2013b, 2015; Veilleux et al. 2013, hereafter V13; Aalto et al.2012, 2015; Cicone et al. 2014; García-Burillo et al. 2015;Lindberg et al. 2016; González-Alfonso et al. 2014, 2017).Theoretical models suggest an origin of these outflows asenergy-conserving flows driven by fast AGN accretion diskwinds (e.g., Zubovas & King 2012, 2014; Faucher-Giguère &Quataert 2012; Costa et al. 2014; Nims et al. 2015).Our previous claims of a connection between large-scalemolecular outflows and AGN activity in (U)LIRGs werebased on the fact that systems with quasar-like AGN luminosi-ties host the faster and more powerful outflows (V13; Rupke& Veilleux 2013a; Cicone et al. 2014). Until recently, theseclaims were incomplete because they were lacking the detec-tion of the putative inner wind. Conversely, studies of pow-erful AGN accretion disk winds to date had focused only onX-ray observations of local radio-quiet and radio-loud AGNand a few higher redshift quasars, but had ignored the impactof these winds on the galaxy host (e.g., Tombesi et al. 2010,2014; Nardini et al. 2015 and references therein).This situation changed with the publication of Tombesi etal. (2015, hereafter T15), where we showed the clear (6.5- σ ) detection of a powerful AGN accretion disk wind with amildly relativistic velocity of ∼ c in the X-ray spectrumof IRAS F11119 + z = 0.190; 1 ′′ = 3.19 kpc)optically classified type 1 ULIRG hosting a powerful molec-ular outflow with velocity V out , OH = 1000 ±
200 km s − . Thiswas the first direct evidence for a fast quasar accretion diskwind driving a large-scale molecular outflow. The energeticsof the accretion disk wind and molecular outflow derived fromour data are consistent with the energy-conserving mecha-nism (T15). In this scenario, the violent interaction of the fastinner AGN wind with the ISM of the host results in shockedwind gas that does not efficiently cool, but instead expandsadiabatically as a hot bubble (e.g., Zubovas & King 2012,2014; Faucher-Giguère & Quataert 2012; Costa et al. 2014;Nims et al. 2015). The adiabatically expanding shocked windsweeps up gas and drives an outer shock into the host ISM.The outflowing gas cools radiatively, and most of it “freezesout” into clumps of cold molecular material. This picture isalso able to explain the existence of a fast ( ∼ − )neutral-atomic (Na I D) outflow in this system (Rupke et al.2005b). A variant on this scenario is that pre-existing molecu-lar clouds from the host ISM are entrained in the adiabaticallyexpanding shocked wind, accelerated to the observed veloci-ties without being destroyed by the many erosive forces andinstabilities (e.g., Cooper et al. 2009; McCourt et al. 2015,2016; Scannapieco & Brüggen 2015; Banda-Barragán et al.2016; Tanner, Cecil, & Heitsch 2016; Thompson et al. 2015,2016; Scannapieco 2017)While the existence of the molecular outflow inF11119 + ˙ M out , OH = 800 + − M ⊙ yr − ,a momentum flux log ˙ P out , OH = 36.7 ± ˙ E out , OH = 44.4 ± − . Thelarge uncertainties stem mainly from the high optical depthof the OH 119 µ m line and the lack of higher excitation lineprofiles (e.g., OH 65 and 84 µ m), and to some degree fromthe fact that the OH outflow is not spatially resolved in the Herschel data. In T15, we had to compare the predictionsof our radiative transfer models (e.g., González-Alfonso etal. 2014) with the observed velocity-resolved profile of OH119 µ m to constrain the location (0.1 – 1.0 kpc) of the OHmolecules that produce the OH profile. The energetics ofthe OH outflow scale linearly with the OH abundance X OH =OH/H. It is also important to note that the energetics citedin T15 are the local (“instantaneous”) quantities estimated ata radius R out , OH = 300 pc. These values are time-averagedover ∆ R out , OH / V out , OH . yrs, the time the outflowingshell of material takes to cross the shell thickness ∆ R out , OH ∼
75 pc. The values time-averaged over the flow time scale R out , OH / V out , OH are smaller by a factor of 4.In the present paper, we take a complementary approachto constrain the energetics of the molecular outflow inF11119 + −
0) emission in F11119 + ∼ ±
36 km s − ), centered on redshift z CO = 0 . × K km s − pc , corresponding toa molecular gas mass of ∼ × M ⊙ for a Galactic con-version factor of α CO = 4.3 M ⊙ (K km s − pc ) − . Our new Since the publication of T15, Feruglio et al. (2015) has reported thetentative detection of a ∼ c X-ray wind in Mrk 231 at the 3.5- σ level (seealso Reynolds et al. 2017). Note, however, that the OH abundance adopted in T15, X OH = 2.5 × − , is, within a factor of ∼
3, consistent with the value inferred in the Galac-tic Sgr B2, the Orion KL outflow, and in buried galaxy nuclei, as well as withthe predictions of chemical models of dense photodissociation regions and ofcosmic-ray and X-ray dominated regions (see González-Alfonso et al. 2017).
ALMA data are considerably more sensitive than the IRAMdata and reveal faint broad wings in the CO(1 −
0) line emis-sion profile. Section 2 describes the ALMA observations ofF11119 + z = 0 . + H = 69.6km s − Mpc − , Ω M = 0.286, and Ω Λ = 0.714 from Bennett etal. 2014). ALMA OBSERVATIONS AND DATA REDUCTION
F11119 + ∼ . ′′ ∼ −
0) line emission at 96.9GHz (Band 3). A RMS (1-sigma) sensitivity of 140 µ Jy ineach 100 km s − channel was targeted, corresponding to 0.7%of the peak flux density measured by Xia et al. 2012 (4 mK ∼
20 mJy). The CO(1 −
0) wing-to-peak ratios in (U)LIRGs(e.g., Cicone et al. 2014) are typically ∼ ∼ σ for F11119 + µ m equivalent widthinto CO(1 −
0) line flux based on the average observed relationin the outflows of ULIRGs Mrk 231, IRAS F08572 + + ∼ ≤ −
200 km/s) of the OH line equivalent width (i.e. only thetruly outflowing component) is used for this calculation.The requested angular resolution ( ∼ ′′ ∼ ∼ − ′′ ) of these objects safely al-low complete CO(1 −
0) flux recovery from this galaxy (broadwings + bright core emission near systemic velocity; Xia etal. 2012). The correlator setup was optimized to simultane-ously observe CO(1 −
0) and the adjacent continuum emission.The central frequency of baseband-4 was adjusted so thatbaseband-4 was contiguous with baseband-3 and also covered“for free” the CN (1-0) complex at 113.15 and 113.50 GHz,and the SiO v =0 (3-2), v = 1 (3-2), and v = 2 (3-2) transitionsat 130.269, 129.363, and 128.459 GHz, respectively, possi-ble tracers of shocked molecular gas in this galaxy. Spec-tral smoothing by a factor of 4 was used to reduce the datarate while maintaining a reasonable velocity resolution of ∼ − . The pipeline-calibrated interferometric visibilities de-livered by ALMA were continuum-subtracted in the uv -planeusing a first-order polynomial, then imaged at 20 km s − res-olution using Briggs weighting with a robust parameter of 0.5and cleaned using a tight box around the source. The restoringbeam is 3 . ′′ × . ′′
21 FWHM with PA = 12 ◦ . RESULTS
Figure 1 shows the full continuum-subtracted upper side-band (USB) spectrum extracted from within a circular 3 ′′ -radius aperture centered on the source. The channels are 20km s − wide but Hanning velocity smoothing was carried outto result in a spectral resolution of ∼
40 km s − . The cube hasa noise of 0.28 mJy in the 20 km s − channels. The strongCO(1 −
0) line emission is detected with a signal-to-noise ra-tio (S/N) of ∼
65 at the peak. The hyperfine components ofCN (1-0) at 113.15 and 113.50 GHz are also detected. Onthe other hand, the SiO v =0, 1, 2 (3-2) transitions at 130.269,129.363, and 128.459 GHz are not visible in our band-4 data,so they are not discussed any further in the remainder of thepaper.Figure 2 zooms in on the CO(1 −
0) line emission withincircular apertures with radii of 3 ′′ and 5 ′′ centered onF11119 + ∼ ± − relative to systemic ( z = 0.190), remarkably similar to the ve-locity of the OH outflow reported in T15. Three methods areused to quantify the strength of this broad emission.First, we carry out a simultaneous fit for two Gaussians(one narrow, one broad) to these data. The broad Gaussian isshown as the yellow area in Figure 2. The residuals are gen-erally less than ± ∼ − ∼ ′′ -radius aperturespectrum (0.19) than in the 3 ′′ -radius aperture spectrum (0.14)suggests that the broad line emission extends out to a radiusof 5 ′′ , although the 5 ′′ spectrum is noticeably more noisy thanthe 3 ′′ spectrum. The fluxes of the broad components derivedfrom these fits are considered upper limits to the actual fluxfrom the outflowing material since they include CO line emis-sion at low velocities which may not be associated with theoutflow. We attempt to remove this low-velocity material us-ing a different strategy.Figure 3 reproduces the continuum-subtracted CO (1 − ′′ . The red line shows the original spectrum (cut off verti-cally to show the details in the wings of CO(1 − − channel. First, a two-dimensional Gaussian was fit to the image for each channel inthe region where a source is detected. The results were thenused to make a smooth source model with linearly changingposition as a function of velocity (to account for a possible ve-locity gradient; see below), Gaussian changing intensity, butconstant size and orientation. This smooth source model wasthen removed from each velocity slice to arrive at a “residu-als” cube. The high S/N of the detection allows us to centroidthe source in each channel with very good sub-beam preci-sion. The velocity gradient measured is +350 km s − kpc − inright ascension and −
200 km s − kpc − in declination (Figure4). This compares well with the direction and amplitude ofthe velocity gradient measured in an unpublished Keck laserguide star adaptive optics Pa α data cube of F11119 + ∼ × M ⊙ within ∼ + . ′′ ± . ′′
05 in right ascension and − . ′′ ± . ′′
10 in dec-lination from the USB band-4 continuum emission (shown inFig. 1b). A Gaussian fit to the high-velocity line emission im-age of Figure 3 finds a FWHM size of 5 . ′′ × . ′′ . ′′ ◦ fromthe North-South direction; this is significantly larger than the3 . ′′ × . ′′
21 FWHM beam.Aperture photometry on the high-velocity gas confirms thatit is indeed extended. In Figure 6, the enclosed high-velocityintegrated flux peaks around a radius of 5 ′′ ± . ′′ ∼ ± − . This isour conservative estimate for the flux from the molecular out-flow. The uncertainty on the enclosed flux is estimated fromthe amplitude of the fluctuations around the value of 0.4 Jykm s − observed in Figure 6. A radius of 5 ′′ ± . ′′ . ′′ + . − . , after correctingfor the beam size (3 . ′′ × . ′′
21 FWHM i.e. ∼ . ′′ . ′′ + . − . = 15 + − kpc, is our best estimate of themaximum extent of the CO outflow.As an independent check on the results from our analysis ofthese imaging data, we also derived the sizes and fluxes of thewing emission by fitting the data directly in the uv plane. Forthis exercise, we used both the CASA uv -Plane Model Fittingroutine uvmodelfit and uvmultifit , the library of Martí-Vidal etal. (2014). The results are summarized in Table 3. In contrastto uvmodelfit , uvmultifit could not deal with the sum of the redand blue wings, so they were fit separately. The signals wereintegrated between 96.9893 and 97.1315 GHz ( −
820 to − − ) for the blue wing, and 96.6081 to 96.7244 GHz forthe red wing ( +
440 to +
800 km s − ). As shown in Figure 3,these channels are not affected by rotation so we did not haveto remove a disk model in the uv -plane fitting. The resultsfrom the two uv fitters are consistent with each other and withthe results from the imaging methods (compare the entries inTable 3 with those of Tables 1 and 2).Taken at face value and keeping in mind the large uncer-tainties on these estimates, the CO outflow in F11119 + R out , CO is typically ∼ + R out , CO ≤ ≤ + ∼ − ) cloudwas recently detected by Janssen et al. (2017, in prep.) at ∼ + DISCUSSION
Energetics of the CO Outflow
The mass of molecular gas involved in the outflow can bederived from the integrated flux densities quoted in the pre-vious section (0.3 − − ), using equation 3 fromBolatto, Wolfire, & Leroy (2013a). A Galactic CO − to − H conversion factor X CO = 2 × cm − (K km s − ) − (or equiv-alently α CO = 4.3 M ⊙ (K km s − pc ) − ) would imply a CO-based molecular mass M out , CO = (3 − × M ⊙ , given aluminosity distance of 933 Mpc. A conservative lower limiton the outflowing molecular gas mass M out , CO ∼ (2 − × M ⊙ is derived if we use the ∼ × smaller optically thin X CO used by Bolatto et al. (2013b) to estimate the outflow-ing molecular gas mass in NGC 253. A compromise betweenthese two extremes is to adopt a ULIRG-like α CO of 0.8 M ⊙ (K km s − pc ) − as done by Cicone et al. (2014). This resultsin an outflowing molecular gas mass of ∼ (0.6 − × M ⊙ , which falls at the high mass end of the spectrum coveredby local ULIRGs (Cicone et al. 2014; González-Alfonso et al.2017). For comparison, the non-outflowing material emits 5 –6 Jy km s − in CO(1 − X CO as for the outflowing material, the amount of quiescent molec-ular gas in the host galaxy is M host , CO ∼ (7 − × M ⊙ ,i.e. in the top quartile of local ULIRGs and infrared quasars(e.g., Solomon et al. 1997; Evans et al. 2001, 2006; Scovilleet al. 2003; Xia et al. 2012), and about 5 − × the amountin the CO outflow.The next step is to derive the CO-based mass outflowrate ˙ M out , CO , momentum flux ˙ P out , CO , and mechanical power ˙ E out , CO of the molecular outflow of F11119 + ˙ M out , ˙ P out , and ˙ E out valuesbased on the second approach to characterize the outflow ofF11119 + ˙ M out , CO = M out , CO V out , CO R out , CO (1) ˙ P out , CO = ˙ M out , CO V out , CO (2) ˙ E out , CO = 12 ˙ M out , CO V , CO . (3)These correspond to the “time-averaged thin shell” valuesof Rupke et al. (2005a), time-averaged over the flow timescale R out , CO / V out , CO , and have been used extensively to de-scribe the energetics of the ionized and neutral phases ofoutflows (e.g., Rupke & Veilleux 2013a; Arav et al. 2013;Borguet et al. 2013; Heckman et al. 2015) as well as somemolecular outflows (González-Alfonso et al. 2017). They aremost appropriate for comparison with outflow models (e.g.,Faucher-Giguère & Quataert 2012; Thompson et al. 2015;Stern et al. 2016). In some studies (e.g., Feruglio et al. 2010,2015; Maiolino et al. 2012; Rodríguez Zaurín et al. 2013; Ci-cone et al. 2014; Harrison et al. 2014; García-Burillo et al.2015), a factor of 3 higher values have been used under theassumption that the emitting spherical (or multiconical) vol-ume is filled with uniform density. However, for a steadymass-conserving flow with constant velocity, we would ex-pect a density at the outer radius only 1/3 that of the aver-age, thus also yielding the expression in eq. (1). The quickdrop-off in the radial intensity profile of the outflow emission of F11119 + R out , CO and V out , CO will set the flow timescale ( R out , CO / V out , CO ) in these expressions and therefore hasto be done with care. If, for instance, we set R out , CO / V out , CO = R maxout , CO / V maxout , CO , where R maxout , CO is the maximum extent of theoutflow ( ∼
15 kpc; Sec. 3) and V maxout , CO is the maximum outflowvelocity ( ∼ − , ignoring projection effects), then theflow time scale R out , CO / V out , CO ≈ × yrs. This valuewould underestimate the actual flow time if all of the outflow-ing gas originated from the center and was uniformly acceler-ated from rest. If this were the case, the flow time scale wouldbe longer by a factor ∼ & R out , CO . R out , CO / V out , CO . × yrs. The actual value of ( R out , CO / V out , CO ) most likelylies between these two extremes. In the following discussion,we adopt a conservatively low value for the radius, R out , CO =7 kpc, and V out , CO = 1000 km s − (hence R out , CO / V out , CO = 7 × yrs), as nominal values of the size and velocity of theCO outflow, and a ULIRG-like CO − to − H conversion factor(we discuss the validity of this latter assumption in Section4.3). From eqs. (1) − (3), we get ˙ M out , CO = (80 − M ⊙ yr − , ˙ P out , CO = (6 − × dyne = (1.5 − L AGN / c , and ˙ E out , CO = (3 − × ergs s − = (0.15 − L AGN , where L AGN = 1.5 × ergs s − , derived from the infrared 15-to-30 µ m color and the prescription of Veilleux et al. (2009).These numbers need to be scaled up by a factor of 5.3 if theCO − to − H conversion factor is Galactic rather than ULIRG-like, or scaled down by a factor of 2.4 if CO(1 −
0) is opticallythin. The results are summarized in Table 4.
Comparisons with the
Herschel
OH Outflow
Table 4 compares the mass, momentum, and kinetic energyoutflow rates derived from the new ALMA CO(1 −
0) data cubewith the values derived from the spatially unresolved
Herschel
OH 119 µ m spectral feature (V13; T15). As noted earlier, themeasured velocity of the CO(1 −
0) outflow is remarkably sim-ilar to that of the OH outflow derived from the
Herschel data.Note, however, that the scales probed by the two data sets aresignificantly different: modeling of the
Herschel
OH profilesuggests a scale for the OH outflow of ∼ ∼
15 kpc. This differencein scale between the OH and CO outflows is not unexpected:OH absorption is produced by gas in front of the source of FIRcontinuum, which is compact in ULIRGs, but there is no suchrequirement for the detection of the CO line emission. More-over, CO(1 −
0) traces the more diffuse low-excitation molecu-lar gas, from which there may not be excited absorption. Thisdifference in scale is important since the dynamical parame-ters of the CO outflow listed in Table 4 are quantities that aretime-averaged over a flow time scale ( R out , CO / V out , CO ) ∼ × yrs, while the published OH-based mass outflow rate isa local (“instantaneous”) estimate at R out , OH ∼
300 pc, whichis valid for timescales ( ∆ R out , OH / V out , OH ) . yrs, where ∆ R out , OH ≈
75 pc, the thickness of outflow shell of molecu-lar material derived from the OH 119 µ m profile. The thirdrow in Table 4 lists the dynamical quantities time-averagedover the flow time scale R out , OH / V out , OH = 4 × yrs; thesequantities are R out , OH / ∆ R out , OH = 4 times smaller than the lo-cal quantities and comparable to the values derived from theCO outflow.Given the well-known short- and long-term variability ofF11119 + + × yrs. Wereturn to this issue in the next section. Comparisons with Published Models
In T15, we argued that the dynamics of the X-ray windand OH outflow were consistent with the models where theOH outflow is an energy-conserving flow driven by a fastAGN accretion disk wind (e.g., Zubovas & King 2012, 2014;Faucher-Giguère & Quataert 2012; Costa et al. 2014; Nims etal. 2015). If this is the case, we have by energy conservation ˙ P out = f ( V wind / V out ) ˙ P wind (4) ∼ f ( V wind / V out ) ( L AGN / c ) , (5)where the quantities with subscript “out” refer to the molec-ular outflow, while those with subscript “wind” refer to theinner X-ray wind. The last equality (eq. 5) is valid only ifthe inner wind is radiatively accelerated, i.e. ˙ P wind ∼ L AGN / c ,which appears to be the case in F11119 + f is defined as the fraction of the kinetic energyin the X-ray wind that goes into bulk motion of the swept-upmolecular material. In T15, an independent estimate of f wasderived from the ratio of the covering fraction of the OH out-flow ( C f , out , OH ) to that of the X-ray wind ( C f , wind ). In T15, wederived C f , wind > .
85 from the X-ray data and C f , out , OH = 0.20 ± Herschel data, so f = 0.22 ± + − to − H conversion factor of α CO = 0.8 M ⊙ (K km s − pc ) − . We therefore make the implicit assumption that thephysical state (e.g., density, temperature, metallicity, internalrandom/turbulent velocity, external radiation field, etc.) of theoutflowing molecular gas in F11119 + α CO tobe 2 − × smaller than the Galactic value (e.g., Bolatto et al.2013a). We also naively expect a decrease in the CO(1 −
0) op-tical depth due to the likelihood of highly turbulent conditionsin the emitting gas, but the detection of high-density molec-ular gas entrained in the outflow of NGC 253 favors a highcolumn density (Walter et al 2017) and seems to rule out theconservatively smaller optically thin α CO value of 0.34 M ⊙ (Kkm s − pc ) − used by Bolatto et al. (2013a). In the end, wefeel that using a ULIRG-like α CO is the most realistic valuefor the outflowing molecular gas, given our current knowl-edge of the conditions in the outflowing material, and also agood compromise solution between the 5.3 × higher valuesderived assuming Galactic α CO and the 2.4 × smaller valuesbased on optically thin α CO (Table 4).Using the molecular momentum outflow rate based on theULIRG-like α CO in eq. (4), we derive f CO = 0.02 – 0.03, con-siderably smaller than the value based on the OH outflow us-ing the local estimates of the energetics ( f OH = 0.2; T15),but comparable to the value we would derive if we use thetime-averaged quantities of Table 4 ( f OH = 0.05). In prin-ciple, an independent value of f CO may be derived from theratio of the covering fraction of the CO outflow ( C f , out , CO ) tothat of the X-ray wind ( C f , wind > .
85; T15). However, inpractice, the modest angular resolution of our ALMA data,taken in compact-array configuration, provides only an upperlimit on C f , out , CO since the extended emission from the out-flowing material seen in Figure 5 will likely break up intosmaller cloudlets when observed at higher angular resolution(e.g., F08572 + C f , out , CO . We derive C f , out , CO < f CO < α CO is only a few times larger than the radiation pres-sure, L AGN / c , exerted by the AGN in F11119 + + α AGN ) L BOL / c = [ (1 − α AGN ) α AGN ] L AGN / c ∼ L AGN / c ( L BOL is the bolometric luminosity; Veilleux et al.2009). A similar statement can be made when consideringthe time-averaged OH-based momentum outflow rate. Thus,the only time we need to invoke models of energy-conservingflows driven by accretion disk winds to explain the molec-ular outflow in F11119 + R − geometricdilution factor of the AGN radiation field.It is important to consider the CO and OH outflows togetherrather than independently. As discussed in Section 4.2, bothare likely related to one another but refer to significantly dif-ferent physical scales ( ∼ vs ∼ × yrs vs ∼ × yrs). The CO-based momen-tum outflow rate listed in Table 4 is a quantity that has beentime-averaged over a ∼ × longer time scale than the localOH-based momentum outflow rate, so one has to use cautionwhen making direct comparisons between the two molecu-lar outflows and with the present properties of the AGN. In-deed, the CO-based quantities are in much closer agreementwith the OH-based quantities that are time-averaged over theflow time scale ( R / V ). This general agreement between theenergetics of the OH and CO outflows suggests that the ef-ficiency of the quasar to drive the large-scale molecular out-flow in F11119 + × yrs. This is not to say that the luminosity ofthe AGN in F11119 + × yrs. With thisin mind, it is important to use methods that are insensitive toshort-term AGN variability when estimating the AGN lumi-nosity. Our use of the global 15-to-30 µ m color (Veilleux etal. 2009) to estimate the fraction of the bolometric luminosityof F11119 + . − yrs) AGN variability. CONCLUSIONS
We report the results of our analysis of deep new ALMACO(1 −
0) data on F11119 + ∼ . ′′ • The CO(1 −
0) spectrum shows the presence of broadwings extending ∼ ± − relative to systemicvelocity, indicative of a fast CO outflow with velocitiescomparable to those measured from the Herschel
OH119 µ m line profile. • Careful photometric and uv -plane analyses of theALMA data indicate that the broad-wing CO(1 − ∼ • The mass of molecular gas involved in the CO outflowis (0.6 − × M ⊙ , assuming a ULIRG-like α CO of 0.8 M ⊙ (K km s − pc ) − . This represents ∼ R out , CO / V out , CO ) of this large COoutflow is ∼ × yrs. The molecular mass, mo-mentum, and energy outflow rates time-averaged overthe flow time scale are (80 − M ⊙ yr − , (6 − × dyne = (1.5 − L AGN / c , and (3 − × ergs s − = (0.15 − L AGN , respectively ( L AGN =1.5 × ergs s − is the AGN luminosity derived fromthe infrared 15-to-30 µ m color and the prescription ofVeilleux et al. 2009). • At face value, the CO-based momentum outflow rate isnot inconsistent with the scenario where the CO out-flow is momentum-conserving and driven by the AGNradiation pressure. This is a different picture than thatproposed by Tombesi et al. (2015), who used the lo-cal (“instantaneous”) value of the OH-based momen-tum outflow rate estimated at R ∼
300 pc and valid for timescales ∆ R out , OH / R out , OH . yrs i.e. nearly twoorders of magnitude shorter than the flow time scale ofthe CO outflow ( ∆ R out , OH is the thickness of the out-flowing shell of molecular material). In contrast, theOH-based dynamical quantities time-averaged over theflow time scale R out , OH / V out , OH are R out , OH / ∆ R out , OH = 4times smaller than the local quantities and thus compa-rable to the values derived from the CO outflow. Theseresults suggest that the efficiency of the quasar to drivethe large-scale molecular outflow in F11119 + × yrs.The modest angular resolution of the ALMA data set is amajor limitation of our analysis. It will be important to re-visit F11119 + + X-Ray Astronomy Recovery Mission (XARM) , the replacementto
Hitomi (ASTRO-H) , will change the landscape and allowus to search for X-ray winds in the X-ray brightest ULIRGswith known molecular outflows as well as some high-redshiftquasars. F11119 + Facilities: ALMA, Herschel, Suzaku
REFERENCES
Aalto, S., Garcia-Burillo, S., Muller, S., et al. 2012, A&A, 537, A44Aalto, S., Garcia-Burillo, S., Muller, S., et al. 2015, A&A, 574, A85Alatalo, K., Blitz, L., Young, L. M., et al. 2011, ApJ, 735, 88Alatalo, K., Lacy, M., Lanz, L., et al. 2015, ApJ, 798, 31Banda-Barragán, W. E., Parkin, E. R., Federrath, C., Crocker, R. M., &Bicknell, G. V. 2016, MNRAS, 455, 1309Bennett, C. L., Larson, D., Weiland, J. L., & Hinshaw, G. 2014, ApJ, 794,135Bolatto, A. D., Wolfire, M., & Leroy, A. K. 2013, ARAA, 51, 206Bolatto, A. D., Warren, S. R., Leroy, A. K.. et al. 2013, ApJ, 768, 17Cicone, C., Maiolino, R., Sturm, E., et al. 2014, A&A, 562, A21Cooper, J. L., Bicknell, G. V., Sutherland, R. S., & Bland-Hawthorn, J. 2009,ApJ, 703, 330Costa, T., Sijacki, D., & Haehnelt, M. G. 2014, MNRAS, 444, 2355Evans, A. S., Solomon, P. M., Tacconi, L. J., Vavilkin, T., & Downes, D.2006, AJ, 132, 2398Evans, A. S., Frayer, D. T., Surace, J. A., & Sanders, D. B. 2001, AJ, 121,1893Fabian, A. C. 2012, ARAA, 50, 455Faucher-Giguère, C.-A., & Quataert, E. 2012, MNRAS, 425, 605Feruglio, C., Maiolino, R., Piconcelli, E., et al. 2010, A&A, 518, L155Feruglio, C., Fiore, F., Carniani, S., et al. 2015, A&A, 583, 99Fischer, J., Sturm, E., González-Alfonso, E., et al. 2010, A&A, 518, L41García-Burillo, S., Combes, F., Usero, A., et al. 2015, A&A, 580, A35González-Alfonso, E., Fischer, J., Graciá-Carpio, J., et al. 2014, A&A, 561,A27González-Alfonso, E., Fischer, J., Spoon, H., et al. 2017, ApJ, 836, 11González-Alfonso, E., Fischer, J., Sturm, E., et al. 2015, ApJ, 800, 69Harrison, C. M., Alexander, D. M., Mullaney, J. R., & Swinbank, A. M. 2014,MNRAS, 441, 3306Kawakatu, N., Imanishi, M., & Nagao, T. 2007, ApJ, 661, 660Keel, W. C., Chojnowski, S. D., Bennert, V. N., et al. 2012a, MNRAS, 420,878Keel, W. C., Lintott, C. J., Schawinski, K., et al. 2012b, AJ, 144, 66Keel, W. C., Maksym, W. P., Bennert, V. N., et al. 2015, AJ, 149, 155Keel, W. C., Lintott, C. J., Maksym, W. P., et al. 2017, ApJ, 835, 256Kim, D.-C., Veilleux, S., & Sanders, D. B. 2002, ApJS, 143, 315Lindberg, J. E., Aalto, S., Muller, S., et al. 2016, A&A, 587, 15Maiolino, R., Gallerani, S., Neri, R., et al. 2012, MNRAS, 425, 66McCourt, M., Oh, S. P., O’Leary, R. M., & Madigan, A.-M. 2016, preprint(arXiv:1610.01164)McCourt, M., O’Leary, R. M., Madigan, A.-M., & Quataert, E. 2015,MNRAS, 449, 2 Nardini, E., Reeves, J. N., Gofford, J., et al. 2015, Science, 347, 860Nims, J., Quataert, E., & Faucher-Giguère, C.-A. 2015, MNRAS, 447, 3612Reynolds, C., Punsly, B., Miniutti, G., O’Dea, C. P., & Hurley-Walker, N.2017, ApJ, 836, 155Rodríguez Zaurín, J., Tadhunter, C. N., Rose, M., & Holt, J. 2013, MNRAS,432, 138Rupke, D. S. N., & Veilleux, S. 2011, ApJ, 729, L27Rupke, D. S. N., & Veilleux, S. 2013a, ApJ, 768, 75Rupke, D. S. N., & Veilleux, S. 2013b, ApJl, 775, L15Rupke, D. S. N., & Veilleux, S. 2015, ApJ, 801, 126Rupke, D. S., Veilleux, S., & Sanders, D. B. 2005a, ApJS, 160, 115Rupke, D. S., Veilleux, S., & Sanders, D. B. 2005b, ApJ, 632, 751Rupke, D. S., Veilleux, S., & Baker, A. J. 2008, ApJ, 674, 172Scannapieco, E. 2017, ApJ, 837, 28Scannapieco, E., & Brüggen, M. 2015, ApJ, 805, 158Schawinski, K., Evans, D. A., Virani, S., et al. 2010, ApJ, 724, L31Schawinski, K., Koss, M., Berney, S., & Sartori, L. F. 2015, MNRAS, 451,2517Scoville, N. Z., Frayer, D. T., Schinnerer, E., & Christopher, M. 2003, ApJ,585, L105Solomon, P. M., Downes, D., Radford, S. J. E., & Barrett, J. W. 1997, ApJ,478, 144Stern, J., Faucher-Giguère, C.-A., Zakamska, N. L., & Hennawi, J. F 2016,ApJ, 819, 130Sturm, E., González-Alfonso, E., Veilleux, S., et al. 2011, ApJ, 733, L16Tanner, R., Cecil, G. & Heitsch, F. 2016, preprint (arXiv:1608.05342)Tombesi, F., Cappi, M., Reeves, J. N., et al. 2010, ApJ, 521, 57Tombesi, F., Meléndez, M., Veilleux, S., et al. 2015, Nature, 519, 436 (T15)Tombesi, F., Tazaki, F., Mushotzky, R. F., et al. 2014, MNRAS, 443, 2154Thompson, T. A., Fabian, A. C., Quataert, E., & Murray, N. 2015, MNRAS,449, 147Thompson, T. A., Quataert, E., Zhang, D., & Weinberg, D. H. 2016, MNRAS,455, 1830Turner, B. E., & Gammon, R. H. 1975, ApJ, 198, 71Veilleux, S., Cecil, G., & Bland-Hawthorn, J. 2005, ARAA, 43, 769Veilleux, S., Kim, D.-C., & Sanders, D. B. 2002, ApJS, 143, 315Veilleux, S., Rupke, D. S. N., Kim, D.-C., et al. 2009, ApJS, 182, 628Veilleux, S., Meléndez, M., Sturm, E., et al. 2013, ApJ, 776, 27 (V13)Xia, X. Y., Gao, Y., Hao, C.-N., et al. 2012, ApJ, 750, 92Zubovas, K., & King, A. 2012, ApJ, 745, L34Zubovas, K., & King, A. 2014, MNRAS, 439, 400
TABLE 1M
EASURED Q UANTITIES FROM T WO -G AUSSIAN F ITS OF THE I NTEGRATED
CO(1 −
0) E
MISSION
Component V central
FWHM Integrated Flux Peak Flux(km s − ) (km s − ) (Jy km s − ) (mJy)(1) (2) (3) (4) (5)3 ′′ -radius apertureNarrow Gaussian − ± ± ± ± − ±
53 1113 ±
171 0.66 ± ± ′′ − radius apertureNarrow Gaussian − ± ± ± ± ±
48 1068 ±
168 0.98 ± ± EASURED Q UANTITIES FROM R ESIDUAL M AP AFTER R EMOVAL OF THE R OTATING D ISK
Component Velocity Range Integrated Flux Size (FWHM) ( a ) RA Offset DEC Offset(km s − ) (Jy km s − ) (arcsec) (arcsec) (arcsec)(1) (2) (3) (4) (5) (6)Blue + red wings [ − − ± × ± ± − ± OTE . — ( a ) Not corrected for the beam size (3 . ′′ × . ′′
21 FWHM) TABLE 3M
EASURED Q UANTITIES FROM UV P LANE F ITTING
Component Velocity Range Integrated Flux Size (FWHM) RA Offset DEC Offset(km s − ) (Jy km s − ) (arcsec) (arcsec) (arcsec)(1) (2) (3) (4) (5) (6)Fitter: uvmodelfit Blue + red wings [ − − ± ± ± − ± uvmultifit Blue wing [ − − ± ± − ± − ± ± ± ± − ± TABLE 4D
ERIVED P ROPERTIES OF THE S MALL - AND L ARGE -S CALE O UTFLOWS IN
F11119 + ˙ M ˙ P ˙ E Type Velocity (lower limit) (upper limit) Fraction [ M ⊙ yr − ] [ L AGN /c] [ L AGN ](1) (2) (3) (4) (5) (6) (7) (8)Accretion Disk Wind ( a ) ± r s r s > − ( b ) − ( c ) (6 − ( d ) OH Outflow (local) ( e ) ±
200 km s − ± − ( f ) − ( g ) (0.5 − ( h ) OH Outflow (average) ( i ) ±
200 km s − ± − ( j ) − ( g ) (0.1 − ( h ) CO Outflow (ULIRG-like) ( k ) ±
200 km s − < < ( l ) − ( m ) − ( n ) (0.15 − ( o ) CO Outflow (Galactic) ( p ) ±
200 km s − < < ( l ) − ( m ) − ( n ) (0.80 − ( o ) CO Outflow (optically thin) ( q ) ±
200 km s − < < ( l ) − ( m ) − ( n ) (0.06 − ( o ) N OTE . — Boldfaced entries indicate favored estimates. Column (1): This table lists the physical properties of three different outflows: (i) the X-ray windon the scale of the accretion disk first reported in T15, (ii) the
Herschel -detected OH outflow first reported in V13, and (iii) the ALMA-detected CO outflowreported in the present paper. Column (2): Estimate of the outflow velocity. Column (3): Lower limit on the size of the outflow. Column (4): Upper limit on thesize of the outflow. Column (5): Estimate of the fraction of the sky covered by the outflowing material. Column (6): Mass outflow rate in M ⊙ yr − . Column (7):Momentum flux normalized to the radiation pressure, L AGN / c . Column (8): Mechanical power normalized to the AGN luminosity, L AGN = 1.5 × ergs s − ,derived from the infrared 15-to-30 µ m color and the prescription of Veilleux et al. (2009). ( a ) Quantities derived from the
Suzaku data of T15. ( b ) ˙ M wind = 1.5 × ( r wind /15 r s ) ( N H /6.4 × cm − ) × ( C f , wind /1.0) × ( V wind /0.255c) M ⊙ yr − ,where r wind is the wind radius, r s is the Schwarzschild radius of the SMBH in F11119 + M BH = 1 . × M ⊙ ; Kawakatu et al. 2007), N H is columndensity of the (fully ionized) wind, and C f , wind is the wind covering fraction. ( c ) ˙ P wind = ˙ M wind × V wind . ( d ) ˙ E wind = ˙ M wind × V . ( e ) Local (“instantaneous”)quantities derived by T15 from the
Herschel
OH 119 µ m presented in V13. ( f ) ˙ M out , OH = 800 × ( R out , OH /300 pc) × ( n H /100 cm − ) × ( C f , out , OH /0.2) × ( V out , OH /1000 km s − ) M ⊙ yr − = M out , OH V out , OH ∆ R − , OH , where R out , OH is the radius of the OH outflow, n H is the Hydrogen number density, C f , out , OH is theOH outflow covering fraction, M out , OH is the total outflowing mass of molecular gas, and ∆ R out , OH is the thickness of the outflowing shell (= 75 pc). ( g ) ˙ P out , OH = ˙ M out , OH × V out , OH . ( h ) ˙ E out , OH = ˙ M out , OH × V , OH . ( i ) Time-averaged quantities derived from the
Herschel
OH 119 µ m presented in V13. ( j ) ˙ M out , OH = 200 × ( R out , OH /300 pc) × ( N H /2.3 × cm − ) × ( C f , out , OH /0.2) × ( V out , OH /1000 km s − ) M ⊙ yr − = M out , OH V out , OH R − , OH , where R out , OH is the radius of the OHoutflow, N H is the Hydrogen column density, C f , out , OH is the OH outflow covering fraction, M out , OH is the total outflowing mass of molecular gas. ( k ) Quantitiesderived from the ALMA CO(1 −
0) data using a ULIRG-like α CO of 0.8 M ⊙ (K km s − pc ) − . ( l ) The covering fraction of the CO outflow is estimated from thepatchiness of the high-velocity CO emission on large scale in Figure 5 (see Section 4.3 for more detail). ( m ) ˙ M out , CO = 140 ( M out , CO /1 × M ⊙ ) × ( R out , CO /7kpc) ( V out , CO /1000 km s − ) − M ⊙ yr − . ( n ) ˙ P out , CO = ˙ M out , CO × V out , CO . ( o ) ˙ E out , CO = ˙ M out , CO × V , CO . ( p ) Quantities derived from the ALMA CO(1 − α CO = 4.3 M ⊙ (K km s − pc ) − . ( q ) Quantities derived from the ALMA CO(1 −
0) data using an optically thin α CO = 0.34 M ⊙ (K km s − pc ) − . frequency (GHz) -202468101214161820 F l u x D en s i t y ( m Jy )
107 107.5 108 108.5 109 109.5 110 frequency (GHz)
CN (1−0)CO (1−0)(a) (b) v=2 v=1 v=0
SiO (3−2) F IG . 1.— Full continuum-subtracted USB spectrum integrated inside a 3 ′′ -radius circular aperture: (a) 94.4 − − − wide but Hanning velocity smoothing was carried out to result in a spectral resolution of ∼
40 km s − . The vertical black lines in (a) show the expectedpositions for the CN (1-0) hyperfine components, with the relative intensities observed in Orion (Turner & Gammon 1975). The SiO v = 0 (3-2), v = 1 (3-2), and v = 2 (3-2) transitions are not detected in (b). -2000 -1500 -1000 -500 0 500 1000 1500 2000 V src (km s -1 ) -1-0.500.511.522.53 F l u x D en s i t y i n s i de R = " ( m Jy ) Spectrum2-Component fitBroad Component -2000 -1500 -1000 -500 0 500 1000 1500 2000 V src (km s -1 ) -1-0.500.511.522.53 F l u x D en s i t y i n s i de R = " ( m Jy ) Spectrum2-Component fitBroad Component (a)(b) F IG . 2.— Simultaneous two-Gaussian fit to the CO(1 −
0) line emission within (a) a 3 ′′ -radius circular aperture centered on F11119 + ′′ -radiuscircular aperture centered on F11119 + -2000 -1500 -1000 -500 0 500 1000 1500 2000 V src (km s -1 ) -1-0.500.511.522.53 F l u x D en s i t y ( m Jy ) F IG . 3.— Same as Figure 2a, but here the blue line shows the residuals after fitting and removing a two-dimensional Gaussian source model to each 20 km s − channel, representative of the gas in pure rotation. See text in Section 2 for more detail on the removal method. The yellow region shows the high-velocityemission in CO, from −
820 to −
400 km s − and +
280 to +
800 km s − , which cannot be accounted for by the gas in pure rotation. -0.4-0.3-0.2-0.100.10.20.3 R.A. Offset (") -0.100.10.20.30.4 D e c . O ff s e t ( " ) -150-100-50050 k m s - F IG . 4.— Velocity gradient due to rotation in the host galaxy derived from the ALMA CO (1 −
0) line emission. The linear scale is 3.19 kpc per arcsecond. SeeSection 2 for more detail on the derivation. Blue+Red Wings -10-50510
R.A. Offset (") -10-8-6-4-20246810 D e c . O ff s e t ( " ) Jy k m s - Red Wing -10-50510
R.A. Offset (") -10-8-6-4-20246810 D e c . O ff s e t ( " ) Jy k m s - Galaxy -10-50510
R.A. Offset (") -10-8-6-4-20246810 D e c . O ff s e t ( " ) Jy k m s - Blue Wing -10-50510
R.A. Offset (") -10-8-6-4-20246810 D e c . O ff s e t ( " ) Jy k m s - Galaxy residual -10-50510
R.A. Offset (") -10-8-6-4-20246810 D e c . O ff s e t ( " ) Jy k m s - (b)(a) (d) (e)(c) F IG . 5.— Maps of the CO(1 −
0) emission from the various kinematic components of F11119 + −
400 and +
280 km s − after subtraction of the rotatingmaterial, (c) blue + red wings, i.e. the “residual” channels between −
820 and −
380 km s − and between +
280 and +
800 km s − , (d) blue wing only, i.e. between −
820 and −
380 km s − , (e) red wing only, i.e.between +
280 and +
800 km s − . The linear scale is 3.19 kpc per arcsecond. For each panel, the color scale on the right indicates the flux level (note that the panels are on different scales). The white contoursindicate − −
2, and − × the rms noise (= 0.033, 0.033, 0.04, 0.026, and 0.029 Jy km s − for panels a, b, c, d, and e, respectively). The black contours show the USB continuum emission (0.1, 0.25, and 0.5 mJy).The beam size is shown in the lower left corner of each panel, and the 3 ′′ - and 5 ′′ -radius circular apertures centered on the CO peak are shown as red dashed circles. Note that the emission from the high-velocitygas is extended and offset from the continuum emission and the rotating disk. Aperture Radius (") E n c l o s ed F l u x f o r H i gh - V e l o c i t y C O ( Jy k m s - ) F IG . 6.— Aperture photometry on the high-velocity CO (1 −
0) emission shown in Figures 3 and 5. The integrated high-velocity CO line flux is plotted in blueas a function of the radius of the circular aperture. For comparison, the integrated continuum flux, which is unresolved, is shown in red. The flux peaks around R = (5 ± ′′ and then stays around ∼ ± − . A radius of (5 ± ′′ measured on the image corresponds to an actual radius of 4 . ′′ + . − . = 15 + −8