Molecular outflows and feedback in the local universe Eckhard Sturm (MPE) for the SHINING Team *) AGN winds 2017 *) A. Contursi, R. Davies, R. Genzel, E. Gonzalez-Alfonso, J. Gracia-Carpio, J. Fischer, A. Janssen, D. Lutz, R. Maiolino , A. Poglitsch, A. Sternberg, L. Tacconi, S. Veilleux and Collaborators

Molecular outflows and feedback in the local universe Tracers (OH, CO, CII) Evidence, statistics, simple correlations Outflow masses and energetics (masses, outflow rates, momentum, luminosities) Does it have an effect? (mass loading, depletion times) Source and mechanism (AGN and/or Star formation, momentum and/or energy) Positive feedback (= star formation IN the outflow) ? Eckhard Sturm (MPE) for the SHINING Team *) *) A. Contursi, R. Davies, R. Genzel, E. Gonzalez-Alfonso, J. Gracia-Carpio, J. Fischer, A. Janssen, D. Lutz, R. Maiolino , A. Poglitsch, A. Sternberg, L. Tacconi, S. Veilleux and Collaborators

I) OH OH P-Cygni 1200 km/s wavelength (mm) Fischer+2010, Sturm+2011

The SHINING outflow sample spans a broad range: 20 cool (f25/f60<0.2) pre-merger ULIRGs 18 warm (f25/f60>0.2) late-stage ULIRGs 5 „classic“ IR-faint QSOs 52 hard X-ray selected BAT AGN Fischer+2010 (Mrk 231), Sturm+2011 (5 ULIRGs + NGC253) Veilleux + 2013 (38 ULIRGs + 5 PG QSOs) Stone+ 2016 (+52 BAT AGN) See also Spoon + 2013 (24 (fainter) ULIRGs)

Massive molecular outflows detected in 70% of ULIRGs 9% of BAT AGNs Inflow: 11% of ULIRGs 17% of BAT AGNs Outflows ubiquitous in ULIRGs  not a pencil beam (radio jet) effect (see also EWs) Fischer+2010 (Mrk 231), Sturm+2011 (5 ULIRGs + NGC235) Veilleux + 2013 (38 ULIRGs + 5 PG QSOs) Stone+ 2016 (+52 BAT AGN) See also Spoon + 2013 (24 (fainter) ULIRGs)

outflow velocity v84 [km/s] Velocities, and their correlation with host galaxy properties (SFR, M*, LAGN, …) Sturm+2011, Veilleux+2013, Stone + 2016: 43 ULIRGs and PG QSOs, 52 BAT AGN (see also Spoon+ 2013) outflow velocity v84 [km/s] 400 -400 -800 9 10 11 12 13 log (LAGN/L) BAT AGN ULIRG NGC7479 Outflow velocity LAGN ~

Modeling and energetics OH119 OH79 OH84 OH65 OH119 OH79 OH84 OH65 Radiative Transfer Modelling:  Radius and covering factor r, f outflow mass Mout outflow rate Ṁout = Mout v/r Mass loading η = Ṁout / SFR Momentum flux: Ṗ = Ṁv Mechanical luminosity: E = 0.5 Ṁv2 * . González-Alfonso + 2017

* Modeling results, and: Does it matter? Outflow masses: M = (100 – 2900) 106M Mass outflow rate Ṁout = 200 – 1500 M /yr Mass loading: η = Ṁout / SFR = 1 – 10 Gas consumption time tcon = M(H2) / SFR Gas depletion time tdep = M(H2)/ Ṁout * tcon / tdep = 1.5 – 15 *) *) assuming continuous flow and no replenishment

AGN- or Starburst-driven? Momentum-conserving or energy-conserving? Starburst99 (Leitherer + 1999): starbursts supply a maximum momentum of ~3.5 L*/c (including ram pressure of winds and radiation pressure on dust grains) (Heckman+2015) AGN may supply a maximum momentum of ~2 LAGN/c

AGN- or Starburst-driven? Momentum-conserving or energy-conserving? Starburst99 (Leitherer + 1999): starbursts supply a maximum momentum of ~3.5 L*/c (including ram pressure of winds and radiation pressure on dust grains) (Heckman+2015) AGN may supply a maximum momentum of ~2 LAGN/c

AGN- or Starburst-driven? Momentum-conserving or energy-conserving? L=LAGN L=L* < 0.025 LAGN < 0.005 L* Supernovae and stellar winds can provide a mechanical luminosity of up to ~1.8% of L* (Leitherer + 1999, Veilleux+2005, Harrison+ 2014) of which less than ¼ will go into bulk motion of the ISM  energy-conserving winds from the starburst unable to drive the observed molecular outflows, at least in the strong outflow cases. Energy-conserving bubbles created by AGN winds supply up to ~5% of LAGN (e.g. King&Pounds 2015) with ½ going into bulk motion of the ISM (Faucher-Giguère & Quataert 2012)

II) CO

Mrk 231 OH P-Cygni 1200 km/s wavelength (mm) Feruglio+ 2010

IRAS F08572+3915 H-band image (Scoville et al. 2000). OH119 IRAS F08572+3915 H-band image (Scoville et al. 2000). Janssen + 2016 b (PhD Thesis)

Janssen + 2016 b (PhD Thesis) Main outflow: biconical outflow with a large opening angle, inclined w.r.t. line-of-sight. vmax close to the maximum observed velocity in the outflow: 1200 km s-1. The second redshifted outflow matches the description of an individual cloud. H2: Rupke&Veilleux 2013a OSIRIS / Keck Hα and Na I D: Rupke & Veilleux 2013b GMOS/Gemini Janssen + 2016 b (PhD Thesis)

Masses derived from OH, CO and CII agree within a factor ~2 (αCO=0 Masses derived from OH, CO and CII agree within a factor ~2 (αCO=0.8 M(K km s-1 pc2)-1) SFR = 61 M /yr , η = 6 Janssen + 2016 b (PhD Thesis)

5.9kpc Momentum rate: 20 LAGN/c Kinetic power: 3.6% of LAGN Momentum boost of a factor 20 suggests that the outflow is (mainly) energy-driven. Depletion time: tdep = 3.6 Myr (cp. OH: 2 Myr). But: episodic Red blob: M = 5 * 107 M flow time: 6 * 106 yr 5.9kpc Cp. observational evidence for AGN variability (e.g. Hickox+ 2014) AGN models predict variability on time scales of 105 - 108 years Janssen + 2016 b (PhD Thesis)

Cicone, Maiolino, Sturm + 2014 log Ṁ (M/yr) Cicone, Maiolino, Sturm + 2014

Comparison CO - OH González-Alfonso + 2017

III) [CII]

M82, velocity dispersion Contursi + 2012

CII as tracer of (molecular) outflows IRAS10565 CII FWHM=856km/s 22 SHINING ULIRGs 15/22 exhibit broad [CII] components All ULIRGs with broad [CII] also show an OH outflow: 13/15, one is an inflow, one OH non-detection due to S/N 13/16 OH outflow objects also show broad [CII]. Non-detections of [CII] mostly for objects with low outflow velocities Janssen+ 2016

Outflow masses – OH vs. [CII] 1:1 Janssen+ 2016.

SDSS J1148+5152, z=6.4189 [CII]158μm Maiolino + 2012 Cicone + 2015 IRAM [CII]158μm Capak+2015, Gallerani+2016, Pallottini+ 2016, See also Riechers+2014 (z=5.3 SMG)

IV) Positive feedback Several recent models1) predict that such massive outflows may ignite star formation within the outflow itself. This star-formation mode, in which stars form with high radial velocities, could contribute to the morphological evolution of galaxies the evolution in size and velocity dispersion of the spheroidal component of galaxies the population of high-velocity stars, which could even escape the galaxy. Such star formation could provide in situ chemical enrichment of the circumgalactic and intergalactic medium (through supernova explosions of young stars on large orbits), and some models also predict it to contribute substantially to the star-formation rate observed in distant galaxies. 1) e.g. Ishibashi, Fabian+ 2012, 2013; Zubovas+ 2013, 2014; Silk 2013, ESO science release 1710: https://www.eso.org/public/news/eso1710/

within the outflow itself. Several recent models1) predict that such massive outflows may ignite star formation within the outflow itself. This star-formation mode, in which stars form with high radial velocities, could contribute to the morphological evolution of galaxies the evolution in size and velocity dispersion of the spheroidal component of galaxies the population of high-velocity stars, which could even escape the galaxy. Such star formation could provide in situ chemical enrichment of the circumgalactic and intergalactic medium (through supernova explosions of young stars on large orbits), and some models also predict it to contribute substantially to the star-formation rate observed in distant galaxies. 1) e.g. Ishibashi, Fabian+ 2012, 2013; Zubovas+ 2013, 2014; Silk 2013,

Star formation inside a galactic outflow Maiolino+ 2017 IRAS F23128−5919, ULIRG merger at z= 0.0448, southern nucleus hosts an AGN approaching outflow (weak) receding

X-shooter spectra: detailed optical and near-IR diagnostics approaching outflow (weak) receding X-shooter spectra: detailed optical and near-IR diagnostics Maiolino+ 2017

decomposition between narrow and broad components, as well as the Subsections of continuum-subtracted X-shooter spectra, extracted from the central region, around some of the relevant emission lines, showing the decomposition between narrow and broad components, as well as the non-detection of coronal lines Maiolino+ 2017

Maiolino+ 2017

Maiolino+ 2017

They are inconsistent with schock or AGN excitation The outflow line ratios are consistent with those observed in star-forming galaxies They are inconsistent with schock or AGN excitation The presence of young stars is clearly revealed in the ultraviolet Hubble Space Telescope (HST) images. BUT: from imaging alone it is not possible to disentangle putative young stars in the outflow from young stars in the galaxy disks, whose ultraviolet radiation field can potentially ionize the gas in the outflow (externally) and produce the line ratios observed in the outflow. Maiolino+ 2017

Even better evidence for stars formed in the outflow is the direct detection of a young stellar population with the kinematical fingerprints of formation inside the outflow Maiolino+ 2017

Kinematics Together with the nebular diagnostics quiescent gas (disk) kinematics of stars formed in the outflow (falling back) outflowing gas gaseous outflow kinematics of stars formed in the outflow (escaping) trajectories of stars formed in the outflow host galaxy disk+bulge Together with the nebular diagnostics provide further evidence to the scenario of SF in outflows

Hα (outflow) SFR(outflow) > (2x) 15 M compare: SFR(total) = 115 M Chabrier IMF Hα (outflow) SFR(outflow) > (2x) 15 M compare: SFR(total) = 115 M Maiolino+ 2017

SUMMARY - In a sample of ~50 ULIRGs and ~50 AGN: detections of massive molecular outflows in 70% of ULIRGs and 9% of BAT AGNs; inflows in 11% of ULIRGs and 17% of BAT AGNs - Outflows ubiquitous in ULIRGs  not a pencil beam effect - Outflow velocity LAGN, vmax > 1000km/s, - Outflow dominated by AGN, at least for luminous AGN - Combined momentum from AGN and starburst can drive the outflows in weak and moderate cases. - The strongest outflows require energy-driven AGN mechanism - tcon / tdep = 1.5 – 15, Mass loading η = 1 - 10 - Momentum rates of about 20 LAGN/c are common among the AGN-dominated sources in our sample. - CO results agree very well with OH results - OH outflow velocities and [CII] FWHM are correlated - Outflowing masses derived from [CII] and from OH are similar and show the same trends - Strong evidence found for star formation IN the outflow of IRAS F23128−5919 Fischer+2010, Sturm+2011, Veilleux + 2013, Cicone+ 2014, Stone+ 2016, Janssen+ 2016, 2017, Maiolino+ 2017 ~