We report the results from a systematic search for molecular (OH 119 µm) outflows with Herschel-PACS 1 in a sample of 43 nearby (z < 0.3) galaxy mergers, mostly ultraluminous infrared galaxies (ULIRGs) and QSOs. We find that the character of the OH feature (strength of the absorption relative to the emission) correlates with that of the 9.7-µm silicate feature, a measure of obscuration in ULIRGs. Unambiguous evidence for molecular outflows, based on the detection of OH absorption profiles with median velocities more blueshifted than −50 km

Powerful winds driven by active galactic nuclei are often thought to affect the evolution of both supermassive black holes and their host galaxies, quenching star formation and explaining the close relationship between black holes and galaxies. Recent observations of large-scale molecular outflows in ultraluminous infrared galaxies support this quasar-feedback idea, because they directly trace the gas from which stars form. Theoretical models suggest that these outflows originate as energy-conserving flows driven by fast accretion-disk winds. Proposed connections between large-scale molecular outflows and accretion-disk activity in ultraluminous galaxies were incomplete because no accretion-disk wind had been detected. Conversely, studies of powerful accretion-disk winds have until now focused only on X-ray observations of local Seyfert galaxies and a few higher-redshift quasars. Here we report observations of a powerful accretion-disk wind with a mildly relativistic velocity (a quarter that of light) in the X-ray spectrum of IRAS F11119+3257, a nearby (redshift 0.189) optically classified type 1 ultraluminous infrared galaxy hosting a powerful molecular outflow. The active galactic nucleus is responsible for about 80 per cent of the emission, with a quasar-like luminosity of 1.5 × 10(46) ergs per second. The energetics of these two types of wide-angle outflows is consistent with the energy-conserving mechanism that is the basis of the quasar feedback in active galactic nuclei that lack powerful radio jets (such jets are an alternative way to drive molecular outflows).

We report on the energetics of molecular outflows in 14 local Ultraluminous Infrared Galaxies (ULIRGs) that show unambiguous outflow signatures (P-Cygni profiles or high-velocity absorption wings) in the far-infrared lines of OH measured with the Herschel/PACS spectrometer. All sample galaxies are gas-rich mergers at various stages of the merging process. Detection of both ground-state (at 119 and 79 µm) and one or more radiatively-excited (at 65 and 84 µm) lines allows us to model the nuclear gas ( 300 pc) as well as the more extended components using spherically symmetric radiative transfer models. Reliable models and the corresponding energetics are found in 12 of the 14 sources. The highest molecular outflow velocities are found in buried sources, in which slower but massive expansion of the nuclear gas is also observed. With the exception of a few outliers, the outflows have momentum fluxes of (2 − 5) × L IR /c and mechanical luminosities of (0.1 − 0.3)% of L IR . The moderate momentum boosts in these sources ( 3) suggest that the outflows are mostly momentum-driven by the combined effects of AGN and nuclear starbursts, as a result of radiation pressure, winds, and supernovae remnants. In some sources (∼ 20%), however, powerful (10 10.5−11 L ⊙ ) AGN feedback and (partially) energy-conserving phases are required, with momentum boosts in the range 3 − 20. These outflows appear to be stochastic, strong-AGN feedback events that occur throughout the merging process. In a few sources, the outflow activity in the innermost regions has subsided in the last ∼ 1 Myr. While OH traces the molecular outflows at sub-kpc scales, comparison of the masses traced by OH with those previously inferred from tracers of more extended outflowing gas suggests that most mass is loaded (with loading factors ofṀ /SFR = 1 − 10) from the central galactic cores (a few × 100 pc), qualitatively consistent with an ongoing inside-out quenching of star formation. Outflow depletion timescales are < 10 8 yr, shorter than the gas consumption timescales by factors of 1.1 − 15, and are anti-correlated with the AGN luminosity.