aa r X i v : . [ a s t r o - ph . C O ] S e p Structure and Dynamics of Disk GalaxiesASP Conference Series, Vol. **Volume Number**M. Seigar and P.Treuthardt, eds. c (cid:13) Gas accretion in disk galaxies
Francoise Combes
Observatoire de Paris, LERMA, CNRS, 61 Av de l’Observatoire, 75014 Paris,France
Abstract.
Gas accretion is necessary to maintain star formation, spiral and bar struc-ture, and secular evolution in galaxies. This can occur through tidal interaction, or massaccretion from cosmic filaments. Di ff erent processes will be reviewed to drive gas to-wards galaxy centers and trigger starbursts and AGN. The e ffi ciency of these dynamicalprocesses can be estimated through simulations and checked by observations at di ff erentredshift, across the Hubble time. Large progress has been made on galaxies at moderateand high redshifts, allowing to interpret the star formation history and star formatione ffi ciency as a function of gas content, dynamical state and galaxy evolution.
1. Introduction
In the last decade, cosmological simulations have emphasized the importance of coldgas accretion onto galaxies in the mass assembly, in particular at high redshift (e.g.Keres et al. 2005, Dekel et al. 2009, Devriendt et al. 2010). This mode of accretionis thought to be about one order of magnitude more important than galaxy mergers inmass assembly, unlike what was assumed in the hierarchical scenario.In parallel, simulations of isolated galaxies show how important is the gas accre-tion to maintain star formation at a constant level, as is observed in spiral galaxies, toexplain abundance gradients, and also to maintain the spiral structure. In the following,the impact of gas accretion is first described, and then we will review the evidence ofcircumgalactic gas inflow.
2. Role of gas accretion: secular evolution, bars
Secular evolution involves mainly the disk galaxies of the Hubble sequence: all spiralsand irregulars. As for the ellipticals, their formation scenario is still heavily relyingon mergers, either a few major mergers, or more likely a series of minor mergers, toheat the stellar component, destroy disks progressively, and cancel out any angularmomentum, explaining its low values in this class.In disk galaxies, the main motor of evolution is non-axisymmetries and bars.When the disk is abundant in gas, as are most high redshift disky objects, then thebars are not long-lived, but weakened and destroyed by the accumulation of mass in thecenters, and by the exchange of angular momentum between the gas and the stars ofthe bar (Friedli & Benz 1993, Berentzen et al. 1998, Bournaud & Combes 2002). Aweakened bar would transiently look like a lens inside its inner ring (e.g. Laurikainenet al. 2009). 1 F.CombesThe frequency of bars in disk galaxies as a function of mass shows an interest-ing bimodality (Nair & Abraham 2010), with two maxima in the blue cloud and in thered sequence. This was also found by Masters et al (2011) with the Galaxy Zoo, al-though not by the S4G consortium (see K. Sheth, this meeting), but this could be dueto selection e ff ects.Bars can act in conjunction with spirals, to redistribute the angular-momentumacross galaxies, and to modify significantly stellar radial profiles. The non-linear inter-actions at resonances overlap multiply the e ff ects (Minchev et al 2011). In less then 3Gyrs, the e ff ective sizes of galaxy disks may be multiplied by 3, and the correspondingradial migration brings high-velocity dispersion stars in the outer parts. Disk thickeningis also substantial (Minchev et al 2012). When gas accretion is considered, the strengthof bars can be re-boosted, and new stars formed out of the accreted gas continuouslyre-shape the radial stellar profiles. It is then possible to obtain the three observed types:Type I as a single exponential disk, Type II as a truncated one, when star formation isnewly forming at the break, and beyond the break the star formation threshold is notyet reached, and Type III, the anti-truncated profile can be obtained in case of stronggas accretion in the outer parts (cf Fig 1). The Type III morphology appears relativelytransient, and able to evolve into Type II or Type I. Its presence could be a tracer ofaccretion events. There is some correlation of these Type II with weakened bars, asexpected from strong gas accretion (Bournaud & Combes 2002, Combes 2011).When the gas accretion occurs essentially from non-aligned cosmic filaments,characteristic signatures may occur, such as inclined and warped rings (Roskar et al2010), or even polar rings, when the accretion is near polar. Brook et al (2008) haveshown how a lenticular system is first formed from matter accretion, which suddenlystops when the birth filament is consumed out. The next filament in the perpendiculardirection then fuels a relatively stable polar ring system. Gas accretion may mimickgalaxy interactions, since it can produce asymmetries, lopsidedness, clumpiness, andsfatbursts. Even if the accretion is globally symmetric and isotropic, it may be tem-porarily on one side only.Gas accretion replenishes the extended gas reservoirs present around most spiralgalaxies. This gas slowly spirals in, when in quiescent state. However, the first tidal in-teraction may drive the gas violently towards the center, and strongly a ff ects abundancegradients, that can even be reversed (Montuori et al 2010): low-metallicity gas flowsinto the center and dilutes the abundance of the central gas, in a time-scale shorter thanthe time required for this gas to re-enrich through the triggered nuclear starburst. Suchgradient reversals have been observed at high redshift in the MASSIV survey (Queyrelet al 2012).
3. Inside-out disk formation, inflow / outflow, metallicity Gas accretion occurs in the outer parts of disks, and is a way to explain inside-outdisk formation. There are now multiple evidence of this progressive mode of diskformation. Through observations of galaxies as a function of redshift, it is possible totrack the evolution directly. However, finding the progenitors of today galaxies is noteasy. Statistically, this is solved by matching the galaxies at a given cumulative numberdensity, for instance 1.4 10 − Mpc − . Plotting the mass of these matched galaxiesat a given redshift, Patel et al (2013) follow the mass increase of galaxies over z = Figure 1. Simulations of smooth, in-plane gas accretion, for two spiral galaxies,giant Sa (left) and giant Sb (right). The various colored lines correspond to di ff erentepochs. The first row shows the evolution of stellar surface density, and the twoothers the radial and vertical velocity dispersions. From Minchev et al (2012). while the mass inside 2kpc is stable in all the same galaxies. This fact is related to theobservations that the normalised galaxy radius, at a given mass, increases by a factor 3from z = α in the 3D-HST project in 57 galaxies at z ∼
1, Nelsonet al (2012) find that the e ff ective radii of galaxies in ionised gas and new stars isabout 1.3 larger than the e ff ective radius of the rest-frame R-band stellar continuum,representing older stars. The e ff et is larger for massive galaxies, that certainly are thefirst to form inside-out. F.CombesIn this inside-out scenario, with gas accretion, the radial migration will produce atypical reversal of stellar ages in the outer parts: the gas is accreted at the radius of thebreak, which is where the new stars are formed, accentuating the negative age gradientfrom the center. After the break, only old stars migrated from the center are expected,and there is now a positive gradient (Roskar et al 2008). This age gradient reversal hasbeen observed in M33 by Williams et al (2009). Some other galaxies do not show anybreak, but a flat age gradient in the outer parts (Vlajic et al 2011).
4. Evidence of gas accretion: HVC, warps, QSO absorption
Most of the gas from cosmic filaments is accreted at large scales, settles down to thedisk, and spirals in progressively. Since the alignment process occurs through preces-sion and dissipation, with a time-scale of the order of a few dynamical times at theselarge radii, this can take some Gyrs, during which galaxies appear warped or perturbedin the outer parts. Warps and polar rings are therefore the best tracers of external ac-cretion, and indeed most spiral galaxies are observed to be warped (e.g. Briggs 1990,Binney 1992, Reshetnikov & Combes 1998). Also the frequency of asymmetries andlopsidedness in spiral galaxies cannot be explained but with external accretion (Jog &Combes 2009).Searches have been done of extra-planar gas in the halo of spiral galaxies (Fra-ternali et al. 2002, Heald et al 2011, Gentile et al. 2013) and the quantities found arerelatively small, NGC 891 being the most remarkable for its gas entension. The ori-gin of this gas is multiple. Some gas can be ejected into the halo by stellar feedback(fountain e ff ect), or through tidal disruption of satellites. In the Milky Way, the HighVelocity Clouds (HVC) and the Magellanic stream are good examples. This gas is ac-creted progressively, with an interface of multiphase gas, but at a rate lower than thestar formation rate (0.4 M ⊙ / yr in the Galaxy, Putman et al. 2012).The way external hot gas is accreted might be complex (Fraternali & Binney2008). The fountain e ff ect ejects ionised gas into the halo, where it encounters thehot coronal gas. Merging with this assumed non-rotating gas, it looses angular mo-mentum, and cold gas condensates in the shock. Finally more gas is infalling down,that was uplifted by star formation feedback. The metallicity of the infalling gas isrelatively high, since it is a mixture of gas from very di ff erent origins. The same kindof processes is invoked in cool core clusters, as schematically shown in Figure 2. Thehot X-ray gas in the cluster halo is dense enough at the center to cool down and fuel acentral AGN. But the radio jets of the AGN re-heat the medium, in creating two cav-ities of plasma, pushing the X-ray gas out at the cavity boundaries. There the gas iscooling down to low temperatures, and molecular clouds are observed through theirCO emission (Salome et al 2006, 2008). In parallel, the AGN jets drag some moleculargas previously settled in the central galaxy, and this uplifts some high-metallicity gasat 20-30kpc, which explains the more e ffi cient cooling in filaments far from the center,and the gas abundance su ffi cient to produce CO emission.High velocity clouds have been found around M31 and M33 in the local group,and also along a gas bridge between M31 and M33 (Lockman et al. 2012). About 121HVC have been detected around the isolated galaxy NGC 6946 (Boomsma et al. 2008).In nearly face-on galaxies, these HVC re traced by the numerous HI holes they createdin the disk plane, as demonstrated in M101 (Kamphuis et al. 1991). Some of thesehighly perturbed galaxies are aslo lopsided, like UGC7989 (Noordermeer et al. 2005).asaccretion in disk galaxies 5 Figure 2.
Left:
Schematical view of the fountain gas in the hot corona of theMilky Way, from Fraternali et al (2013). Ionised gas is elevated to the hot halo,compresses and provokes the cooling of the hot gas, which comes back to the plane,as high (or intermediate) velocity clouds. This is a way to accrete external gas, sincemore gas cools down than was elevated by the star formation feedback.
Right:
Ananalogous feedback mechanism occurs in cool core clusters. Here we see in thePerseus cluster, the two radio jets filling two cavities in the hot X-ray gas (whitecontours and bubbles), where the heated X-ray gas does not cool. At the boundaryof the cavities, gas can cool (enhancement of X-ray emission), and even very coldgas forms and is visible in CO emission (dark contours and arcs). In this scenario,gas is not cooling towards the center, as previously expected, but at ∼ More generally, 30% of galaxies show an asymmetry larger than 10%, as quantified bythe Fourier decomposition of the density. In poor environments, low surface brightness(LSB) galaxies are dwarfs particularly rich in HI gas, and revealing spectacular warps,like NGC 2915 (Meurer et al. 1006), NGC 5055 (Battaglia et al. 2006), or NGC 3741(Begum et al. 2005, Gentile et al. 2007).Another important way to discover the circum-galactic gas around galaxies isthrough absorption measurements in front of remote quasars. The various systemscan be sorted by their column density, from the Ly α forest at very low columns, tothe damped Ly α (DLA) at NHI > cm − , passing through the Lyman limit systems(LLS). The abundance of the systems varies as a power law (Prochaska et al. 2010).Fumagalli et al. (2011) have derived from cosmological simulations the probabilityto observe absorptions in front of backgound sources, at z = M ⊙ at z = ffi cult, for the small filling fators, but also due to the confusionwith the galaxy host, when the line-of-sights are too close, or the low metallicity of thecircum-galactic gas, if carbon lines are used. F.CombesAnother possibility would be to detect Ly α photons emitted when the gas is in-falling in dark matter haloes, in some way when it re-radiates its gravitational energy. Ithas been proposed that the frequently observed Ly α blobs at high redshift are preciselydue to this radiation. But the simulations from Faucher-Gigu`ere et al. (2010), takinginto account the proper self-shileding, etc., have shown that the gravitational emissionin these blobs is negligible. Ly α blobs are powered by star formation or AGN. Now,several hundreds of Ly α blobs have been discoverd (Matsuda et al. 2006, 2011). TheLy α line is resonant and requires radiative transfer to better understand the shapes ofthe profiles observed. Verhamme et al. (2006) have showned that in case of outflowingmaterial, a P-cygni profile is expected, with emission on the red side, and absorptionon the blue side. The reverse is expected for inflowing gas. However, in all Ly α blobsobserved until now, there are always P-cygni profiles with emission in the red, thereforeonly outflows have been detected.Another way to detect the filaments could then be through fluorescence of Ly α photons emitted by a starburst or a luminous quasar. This has been done in a fewcases (Rauch et al. 2011, Cantalupo et al. 2012), but this clever technique should bedeveloped more. A powerful quasar can illuminate dark gas, or dark galaxies, up to100kpc distance.Finally, when absorption lines are detected in front of quasars, what are the meth-ods to distringuish inflows from outflows in the circumgalactic medium (CGM)? Twomethods have been used, and are illustrated in Figure 3.Using an H I-selected sample of 28 Lyman limit systems (LLS) at z <
1, observedin absorption with the HST-COS spectrograph, Lehner et al. (2013) are able to deter-mine their metallicity from weakly ionized metal species (e.g., O II, Si II, Mg II) andfind that the metallicity distribution of the LLS is bimodal with metal-poor and metal-rich branches peaking at about 2.5% and 50% solar metallicities. Both branches havecomparable number of absorbers. The metal-rich branch likely traces winds, recycledoutflows, and tidally stripped gas, while the metal-poor branch has properties consistentwith cold accretion streamsWhen the galaxy host is detected in the proximity of the absorbant lines of sight,it is possible to determine the azimuthal orientation of the gas: Is it towards the minoraxis? more likely to be an outflow, or the major axis? then it is probably an inflow.Bouch´e et al. (2012, 2013) have shown that the number of absorbants as the functionof angle with respect to the major-axis, reveals a bimodal distribution also. These areMgII absorbants at z ∼ / s,i.e. of the order of the circular velocity, and smaller than the escape velocity by a factorof ∼
2. The outflow rates are typically two to three times the instantaneous SFRs.
5. Evolution with redshift of gas content
It is now well established that the global SFR in the Universe evolves towards a max-imum at about z = =
0, and then regularly decreases withincreasing z. There was a recent debate about a possible plateau of the sSFR after z = Figure 3.
Left:
Bimodal metallicity distribution of the Lyman Limit Systems(LLS) at z <
1. The hashed histograms correspond to upper limits. The peak of thetw maxima are marked by vertical dotted lines. The low metallicity LLS could cor-respond to inflows, while the high metallicity ones to outflows. From Lehner et al(2013).
Right:
Bimodal distribution of the MgII absorbers at z ∼ spectroscopy, it was not possible to disentangle the continuum from the lines, and thecontinuum was over-estimated. With the recent re-normalisation, the sSFR increases atall redshift, which corresponds better to the simulations predictions.Why galaxies in the main sequence of star formation are more e ffi cient to formstars at high redshift? According to our PHIBSS survey, detecting about 52 galaxies atz = = = = dep = + z)-1.05 Gyr. We have shown that the gas fraction was strongly correlatedto the sSFR, and explored the whole range from 10 to 90%. In the Kennicutt-Schmidtdiagram, the surface density of gas and the SFR surface density follows the main se-quence branch. With respect to the gas surface density divided by the dynamical time,all objects (main sequence and starbursts) align on the same almost linear curve. Thisis due to the smaller dynamical time-scale of nuclear starbursts.When only ULIRGS and starbursts are considered, the strong increase of star for-mation e ffi ciency rate is explained at high redshift by two factors with comparableweight: first the gas fraction is higher by a factor 3 at z = =
0, and thestar formation e ffi ciency (SFR per unit gas mass), is also higher by a factor 3 (Combeset al 2013).
6. Conclusion
External gas supply is fundamental for secular evolution: bars drive gas towards thevery center, and accretion is necessary to replenish the disk, to reform a bar, or for thegalaxy to stay on the blue cloud, instead of directly move on the red sequence. F.Combes
Figure 4. Specific star formation rate sSFR = SFR / M ∗ as a function of redshift.The large blue filled circles denote the sSFR inferred from the average CO gas frac-tions, using t dep = Mmol gas / SFR = + z)-1.05 Gyr inferred from the COLDGASSand PHIGS surveys (1.5 Gyr at z = = / infrared imaging surveysin the literature, as compiled in Weinmann et al. (2011), Sargent et al. (2010) andGonzalez et al (2012). The gas accretion is slow, progressive, from gas reservoirs in the outer parts ofgalaxies. Matter and angular momentum is redistributed by non-axisymmetries andbars, including radial migration. These processes explain the inside-out disk formation,that is now currently observed at all redshifts. The evolution of the disk sizes at agiven mass can also be explained by slow matter accretion and dry minor mergers. Inparticular secular evolution can multiply the e ff ective radii by up to a factor 3.Gas accretion was more intense in the past. It could be the explanation, togetherwith galaxy interactions and fly-bys, to abundance gradients reversal in galaxy disks.Tidal forces drive the gas reservoir into the center, and this nearly primordial gas dilutesthe metallicity of the central gas, on a time-scale shorter, than the enrichment time-scaledue to the triggered starburst.Warps and polar rings might be the best evidence of gas accretion, since theirsetlling takes a few Gyr, and they reflect non-aligned accretion of matter. Other evi-dence of accretion can be seen in high velocity clouds, high-column density absorbantsin front of quasars. There is a bimodality between the gas inflowing and outflowing,with almost equal weights, through the metallicity and the angle of accretion.asaccretion in disk galaxies 9There is a strong redshift evolution of gas fraction in galaxies, which can explainthe SFR history, peaking at z = ff erent from what isobserved locally, since the high gas fraction makes disks highly unstable and turbulent. Acknowledgments.
Thanks to Mark Seigar and the organisers for such an inter-esting and nicely located meeting.
References
Battaglia G., Fraternali, F., Oosterloo, T., Sancisi, R.: 2006, A&A 447, 49Begum A., Chengalur, J. N., Karachentsev, I. D.: 2005, A&A 433, L1Berentzen I., Heller, C. H., Shlosman, I., Fricke, K. J.: 1998, MNRAS 300, 49Binney J.: 1992, ARAA 30, 51Boomsma R., Oosterloo, T. A., Fraternali, F. et al.: 2008 A&A 490, 555Bouch´e N., Hohensee, W., Vargas, R. et al. 2012, MNRAS 426, 801Bouch´e N., Murphy, M. T., Kacprzak, G. G. et al. 2013, Science 341, 50Bournaud F., Combes F.: 2002 A&A, 392, 83Bouwens, R., Bradley, L., Zitrin, A. et al. 2012 arXiv-1211.2230Briggs F.: 1990, ApJ 352, 15Brook C. B., Governato, F., Quinn, T. et al.: 2008 ApJ 689, 678Buta R., Combes F. 1996, Fundamentals of Cosmic Physics 17, 95Cantalupo S., Lilly, S. J., Haehnelt, M. G.: 2012, MNRAS 425, 1992Combes F.: 2011, MSAIS 18, 53Combes F., Garca-Burillo, S., Braine, J. et al.: 2013, A&A 550, A41Dekel A., Birnboim, Y., Engel, G. et al.: 2009, Nature 457, 451Devriendt J., Rimes, C.; Pichon, C. et al.: 2010, MNRAS 403, L84Faucher-Gigu`ere C.A., KereÅ, D., Dijkstra, M. et al.: 2010, ApJ 725, 633Faucher-Gigu`ere C.A., Keres D.: 2011 MNRAS 412, L118Fraternali, F., van Moorsel, G., Sancisi, R., Oosterloo, T. 2002, AJ, 123, 3124Fraternali F., Binney J.: 2008, MNRAS 386, 935Fraternali F., Marasco A., Marinacci F., Binney J.: 2013, ApJ 764, L21Friedli D., Benz W.: 1993, A&A 268, 65Fumagalli M., Prochaska, J. X., Kasen, D. et al. 2011, MNRAS 418, 1796Gentile G., Salucci, P., Klein, U. et al. 2007, MNRAS 375, 199Gentile G., Jozsa G.I.G., Serra P. et al 2013: A&A 554, A125Gonzalez, V., Bouwens, R., Illingworth, G., et al.: 2012, ApJ 755, 148Heald, G., Jozsa, G., Serra, P., et al. 2011, A&A, 526, A118Jog C., Combes F.: 2009, Physics Reports, Volume 471, Issue 2, p. 75-111Kamphuis J., Sancisi, R., van der Hulst, T.: 1991, A&A 244, L29Keres D., Katz, N., Weinberg, D. H., Dav´e, R.: 2005, MNRAS 363, 2Kimm T., Slyz, A., Devriendt, J., Pichon, C.: 2011, MNRAS 413, L51Laurikainen E., Salo H., Buta R., Knapen J.: 2009, ApJ 692, L34Lehner, N., Howk, J. C., Tripp, T. M. et al. 2013, ApJ 770, 138Lockman, F. J., Free, N. L.,Shields, J. C.: 2012, AJ 144, 52Masters K., L., Nichol, R. C., Hoyle, B. et al.: 2011, MNRAS 411, 2026Matsuda Y., Yamada, T., Hayashino, T. et al.: 2006, ApJ 640, L123Matsuda Y., Yamada, T.; Hayashino, T.; et al.: 2011 MNRAS 410, L13Meurer G. R., Carignan, C., Beaulieu, S. F., Freeman, K.C.: 1996, AJ 111, 1551Minchev I., Famaey, B., Combes, F. et al.: 2011, A&A 527, A147Minchev I., Famaey, B., Quillen, A. C. et al.: 2012, A&A 548, A126Montuori M., Di Matteo, P., Lehnert, M. D., Combes, F., Semelin, B.: 2010, A&A 518, A56Nair P.B., Abraham R.G.: 2010 ApJ 714, L260Nelson, E. J., van Dokkum, P. G., Brammer, G. et al: 2012 ApJ 747, L28Newman A.B., Ellis R.S., Bundy K., Treu T.: 2012, ApJ 746, 162Noordermeer E., van der Hulst, J. M., Sancisi, R. et al.: 2005 A&A 442, 137
Patel S. G., van Dokkum, P. G.; Franx, M. et al. 2013, ApJ 766, 15Prochaska J. X., O’Meara, J. M., Worseck, G.: 2010, ApJ 718, 392Putman M. E., Peek, J. E. G., Joung, M. R.: 2012, ARAA 50, 491Queyrel J., Contini, T., Kissler-Patig, M. et al.: 2012, A&A 439, A93Rauch, M., Becker, G. D., Haehnelt, M. G. et al.: 2011, MNRAS 418, 1115Reshetnikov V., Combes F.: 1998, A&A 337, 9Roskar R., Debattista, V. P., Stinson, G. S. et al.: 2008 ApJ 675, L65Roskar R., Debattista, V. P., Brooks, A. M. et al.:2010 MNRAS 408, 783Salom, P., Combes, F., Edge, A. C. et al. 2006, A&A 454, 437Salom, P., Combes, F., Revaz Y. et al. 2008, A&A 484, 317Sargent, M. T., Schinnerer, E., Murphy, E. et al. 2010, ApJS 186, 341Smit, R., Bouwens, R. J., Labbe, I. et al.: 2013, arXiv1307.5847Tacconi, L. J., Genzel, R., Neri, R. et al.: 2010, Nature 463, 781Tacconi, L. J., Neri R., Genzel, R., et al.: 2013 ApJ 768, 74Verhamme, A., Schaerer, D., Maselli, A.: 2006, A&A 460, 397Vlajic M., Bland-Hawthorn J., Freeman K.C.: 2011, ApJ 732, 7Wang, J., Kau ffff