Outflows, Bubbles, and the Role of the Radio Jet: Direct Evidence for AGN Feedback at z~2
aa r X i v : . [ a s t r o - ph ] O c t SF2A 2008
C. Charbonnel, F. Combes and R. Samadi (eds)
OUTFLOWS, BUBBLES, AND THE ROLE OF THE RADIO JET: DIRECT EVIDENCEFOR AGN FEEDBACK AT Z ∼ Nesvadba, N., P., H. , and Lehnert, M. D. Abstract.
To accommodate the seemingly ”anti-hierarchical” properties of galaxies near the upper end ofthe mass function within our hierarchical paradigm, current models of galaxy evolution postulate a phase ofvigorous AGN feedback at high redshift, which effectively terminates star formation by quenching the supplyof cold gas. Using the SINFONI IFU on the VLT, we identified kpc-sized outflows of ionized gas in z ∼ − × M ⊙ in ionized gas with velocities near the escape velocity of a massive galaxy. Kinetic energies are equivalent to ∼ .
2% of the rest mass of the supermassive black hole. We discuss the results of this on-going study andthe global impact of the observed outflows.
AGN feedback is now a critical element of state-of-the-art models of galaxy evolution tailored to solve some ofthe outstanding issues at the upper end of the galaxy mass function. Observationally, a picture emerges whereAGN feedback is most likely related to the mechanical energy output of the synchrotron emitting, relativisticplasma ejected during the radio-loud phases of AGN activity: Giant cavities in the hot, X-ray emitting halosof massive galaxy clusters filled with radio plasma are robust evidence for AGN feedback heating the gas onscales of massive galaxy clusters (e.g., McNamara & Nulsen, 2007). Best et al. (2006) analyzed a large sampleof early-type galaxies from the SDSS catalog with FIRST and NVSS radio data and found that heating by theradio source may well balance gas cooling over 2 orders of magnitude in radio power and in stellar mass.However, since most of the growth of massive galaxies was completed during the first few Gyrs after the BigBang, observations at low redshift can only provide evidence that AGN feedback is able to maintain the hot,hydrostatic halos of massive early-type galaxies ( “maintenance mode” ). If we want to observe directly whetherAGN feedback indeed quenched star formation and terminated galaxy growth in the early universe ( “quenchingmode” ), we have to search at high redshift. With this goal, we started a detailed analysis of the rest-frameoptical line emission in powerful, z ∼ − ∼ −
3: Dying starbursts in the most massive galaxies?
The observed properties of HzRGs suggest they may be ideal candidates to search for strong, AGN-driven winds:They have large stellar (Seymour et al. 2007) and dynamical (Nesvadba et al. 2007a) masses of ∼ − M ⊙ and reside in significant overdensities of galaxies suggesting particularly massive underlying dark-matter halos(e.g., Venemans et al. 2007). Large molecular gas masses in some sources (e.g., Papadopoulos et al. 2000) andsubmillimeter observations suggest that some HzRGs at redshifts z ≥ − >
50% at z > ≤
15% at z < GEPI, Observatoire de Paris, CNRS, Universite Denis Diderot; 5, Place Jules Janssen, 92190 Meudon, France Marie-Curie Fellow c (cid:13)
Soci´et´e Francaise d’Astronomie et d’Astrophysique (SF2A) 2008
38 SF2A 2008
Fig. 1. (left to right:) [OIII] λ ∼ − To directly investigate whether HzRGs may be the sites of powerful, AGN driven winds, we collected a sampleof HzRGs at redshifts z ∼ − <
10 kpc in radius. For details see Nesvadba et al.(2006, 2007a, 2008).
Using an integral-field spectrograph, we were able to extract continuum-free line images as well as line-freecontinuum images from our three-dimensional data cubes (Fig. 1). We find that in all cases, the continuumemission is relatively compact, but spatially resolved in some cases, with half-light radii ≤ ∼ α longslit spectroscopy (e.g., Villar-Martin 2003). Overall, differentemission lines in the same galaxy show similar morphologies. In the galaxies with compact radio sources, theline emission appears also compact. This may suggest a causal relationship between the advance of the jet andthe extent of the high surface brightness emission line gas. We fitted spectra extracted from individual spatial resolution elements to construct two-dimensional maps ofthe relative velocities and line widths (Fig. 2). Typically, the velocity maps show two bubbles with relativelyhomogeneous internal velocity, and projected velocities relative to each other of 700 − − , reminiscentof back-to-back outflows extending from near the radio core. MRC1138-262 has a more complex structurewith at least 3 bubbles. Line widths are generally large, indicating strong turbulence, with typical FWHMs ∼ − − . Areas with wider lines may be due to partial overlap between bubbles.ive a shorter title using \ runningtitle Fig. 2. top, left to right:
Maps of relative velocities (in km s − ) for MRC0316-257 at z=3.1, MRC0406-244 at z=2.4,and TXS0828+193 at z=2.6. bottom, left to right : Maps of FWHMs (in km s − ) for the same galaxies. Contours showthe jet morphologies. For TXS0828+193, the lobes are outside of the area shown. Filamentary morphologies and low gas filling factors suggest that the UV-optical line emission may originatefrom clouds of cold gas that are being swept up by an expanding hot medium, most likely related to theoverpressurized ’cocoon’ of gas heated by the radio jet. In such a scenario the velocity of the clouds may yieldan estimate of the kinetic energy injection rate necessary to accelerate the gas to the observed velocities of upto ∼ erg s − (Nesvadba et al. 2006). The size and velocities of the outflow suggest dynamical timescales offew × yrs. Maintaining the observed outflows over such timescales requires total energy injections of ∼ erg. This is in the range of what is observed for AGN driven bubbles in massive clusters at low redshift (e.g.,McNamara & Nulsen, 2006, and references therein). The observed velocities and kinetic energies are also in therange of escape velocities and binding energies expected for galaxies with masses of few × M ⊙ (Nesvadbaet al. 2006). This may suggest that much of the gas participating in the outflows may ultimately be unboundfrom the underlying gravitational potential. Having measured H α line fluxes, we are able to roughly estimate ionized gas masses assuming case B recombi-nation (see Nesvadba et al. 2008 for details). For galaxies where we also measured H β , we correct for extinctionof A V ∼ − × M ⊙ . (Without the correction, estimates arefew × M ⊙ , Nesvadba et al. 2008.) This exceeds the amount of ionized gas found in any other high-redshiftgalaxy population by several orders of magnitudes, including galaxies with starburst-driven winds. Nesvadbaet al. (2007b) investigated a spatially-resolved, starburst driven wind in a strongly star-forming submillimeter-selected galaxy at z ∼ × M ⊙ in ionized gas in the wind. Compact radio galaxies havelower entrained gas masses, but in the range of what would be expected for less evolved outflows with similarentrainment rates as the galaxies with large radio lobes (Nesvadba et al. 2007a).Molecular gas masses in strongly star-forming galaxies at high redshift are also typically in the range offew × M ⊙ (e.g., Neri et al. 2003), and are a necessary prerequisite to fuel the observed starbursts withstar formation rates of few 100 M ⊙ yr − . However, not all HzRGs have been detected in CO. TXS0828+193specifically, which is part of our sample, appears to have less than ∼ M ⊙ in molecular gas (Nesvadba et al.,40 SF2A 2008in prep.). This illustrates that the AGN winds may affect a significant fraction of the overall interstellar mediumof strongly star-forming, massive galaxies in the early universe. Since the velocities are near the expected escapevelocity of a massive galaxy and underlying dark-matter halo ( § Four out of four HzRGs with extended radio jets show evidence for outflows with with kinetic energies of upto 10 erg over dynamical timescales of 10 yrs, and the preliminary analysis of our full sample suggests thatthis is far from being unusual. Nesvadba et al. (2006, 2008) estimate that the outflow energies correspond to ∼
10% of the jet kinetic luminosity. If this coupling efficiency between jet and interstellar medium is typicalfor HzRGs with similarly powerful radio sources, then the redshift-dependent luminosity function of Willott etal. (2001) suggests that at redshifts z ∼ −