The role of neutral hydrogen in radio galaxies Bjorn Emontsa,b, Raffaella Morgantib,c, Tom Oosterloob,c a Department of Astronomy, Columbia University, 550 West 120th Street, New York, N.Y. 10027, USA 7 b Kapteyn Astronomical Institute, University of Groningen, P.O. Box 800, 9700 0 0 AV Groningen, the Netherlands 2 c Netherlands Foundation for Research in Astronomy, Postbus 2, 7990 AA n Dwingeloo, the Netherlands a J 6 1 Abstract 1 v 5 We present morphological and statistical results of a study of neutral hydrogen 3 (HI) in a complete sample of nearby, non-cluster radio galaxies. We detect large- 4 1 scale HI emission in the early-type host galaxies of 25% of our sample sources. The 0 large-scale HI is mainly distributed in disk- and ring-like structures with sizes up 7 0 to 190 kpc and masses up to 2×1010M⊙. All radio galaxies with MHI & 109M⊙ / have a compact radio source. When we compare our sample of radio-loud early- h p type galaxies with samples of radio-quiet early-type galaxies there appears to be no - significant difference in HI properties (mass, morphology and detection rate). This o r suggests that that the radio-loud phase could be just a short phase that occurs at t s some point during the life-time of many, or even all, early-type galaxies. a : v i Key words: galaxies: active, galaxies: ISM, ISM: kinematics and dynamics X r a 1 Introduction Large-scale HI is detected in a growing number of early-type galaxies. In these proceedings numerous cases are presented by Oosterloo et al. and Serra et al., while many more cases are known from the literature (e.g. Morganti et al., 2006; van Gorkom & Schiminovich, 1997; Schiminovich et al., 1997). In the majority of the known cases, the HI is distributed in regular rotating disk- or ring-like structures, that can reach far beyond the optical host galaxy and have HI masses up to a few times the HI mass of the Milky Way. However, this could be an observational bias, given that irregular structures are more frequently observed if one has the sensitivity to trace low mass HI structures Preprint submitted to Elsevier 5 February 2008 of only a few million solar masses (Morganti et al., 2006). Two good expla- nations for the origin of large-scale HI structures in early-type galaxies are gas-rich galaxy mergers and the cold accretion of circum-galactic gas. In case of a major merger between gas-rich galaxies, part of the gas is trans- ported to the central region of the merging system, where a sudden burst of star formation is triggered (Mihos & Hernquist, 1996). Another part of the gas is expelled in large-scale tidal features of low surface-brightness, which can reach far beyond the optical host galaxy. If the environment is not too hostile and the gas in the tails remains gravitationally bound to the system, it can fall back onto the galaxy and settle into a disk- or ring-like struc- ture within a few galactic orbits (>1 Gyr; Barnes, 2002). In the meanwhile, the stars in the merging systems have rearranged into an early-type galaxy (Hibbard & van Gorkom, 1996). In case of cold accretion, galaxies accrete gas from the inter-galactic medium (IGM) via a cold mode, i.e. part of the gas cools along filamentary structures without being shock-heated (Kerˇes et al., 2005). This gas (with T < 105K) may cool further to form the large-scale structures of neutral hydrogen. On smaller scales, Kaufmann et al. (2006) show that through the cooling of hot halo gas, cold gas can be assembled onto a galactic disc. While both mechanisms provide a viable explanation for the formation of large-scale HI around early-type galaxies, the exact formation mechanism is in most cases not evident from the HI distribution alone. To verify the origin of these HI structures it is therefore necessary to study other tracers of the formation history of the galaxy. A good tracer in this respect is the stellar population content of the galaxy. As described above, a major merger event triggers a burst of star formation in the host galaxy, which can be traced with optical spectra. We have used both HI imaging and optical spectroscopy to study the formation history of nearby radio galaxies. 2 HI in radio galaxies Becausemajormergersareofteninvoked totriggerpowerfulradiosources(e.g. Heckman et al., 1986), it is particularly interesting to study the formation history of radio-loud early-type galaxies and compare this with that of radio- quiet early-type galaxies. For this reason we studied a complete sample of nearby radio galaxies in HI, followed-up by an optical spectroscopic study of these systems (to study their stellar populations). In this paper we will focus ontheHIresults andthecomparisonwithHIresults onradio-quiet early-type galaxies. A more detailed analysis of the HI properties of the individual radio galaxies is given in Emonts et al. (2006, 2007), while the stellar population analysis will be presented in a future paper. 2 Fig. 1. From Emonts et al. (2007). 0th-moment total intensity maps of theHI emis- sion (contours) in our HI-rich nearby radio galaxies (B2 1217+29, or NGC 4278, is presentedbyMorganti et al.,2006).RadiocontinuumisonlyshownforB21322+36 (grey contours); for the other sources, the radio continuum is unresolved (or only marginally resolved for B2 0722+30). Although HI absorption is present in all five radio galaxies, we only show the HI absorption (white contours/profile) in case it clarifies the morphology of the large-scale HI. The arrows mark the host galaxies of our sample sources, while the broken lines show the direction along which the position-velocity (PV) plots are taken. Contour levels: B2 0648+27: 0.22, 0.36, 0.52, 0.71, 0.95, 1.2, 1.5, 1.8, 2.1 ×1020 cm−2 (see also Emonts et al., 2006); B2 0258+35: from 0.34 to 3.0 in steps of 0.44 ×1020 cm−2; B2 1322+36: 1.7, 2.3, 2.8 ×1020 cm−2 (black) –continuum:from22to200instepsof44.5 mJybeam−1 (grey);NGC 3894: 0.17, 0.49, 0.87, 1.7, 3.2, 4.6 ×1020 cm−2 (black) – PV: -1.0, -5.0, -10, -14 (grey), 1.0, 2.0, 3.0, 4.5, 6.5 (black) mJy beam−1; B2 0722+30: 0.67, 1.3, 1.8, 2.3, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0 ×1020 cm−2 (part of the HI disk that is observed in absorption is not plotted for clarification) – PV plot: -0.5, -1.4, -2.4, -3.4, -4.4 (grey), 0.5, 0.7, 0.9, 1.1, 1.3 (black) mJy beam−1. Our HI sample consists of 21 radio galaxies from the B2-catalogue (F408MHz & 0.2 Jy) up to a redshift of z ≈ 0.04. This sample is complete, with the restric- tion that we left out sources in dense cluster environments (since here large- scale gaseous features are likely wiped out on relatively short time scales) and BL-Lacobjects.InadditionweobservedNGC3894,whichhasacompactradio source with radio power comparable to our B2-sample sources. We leave NGC 3894 out of the statistical analysis in Sect. 4. In total we observed 9 compact (< 15 kpc) radio sources and 13 extended (> 15 kpc) Fanaroff & Riley (1974) 3 Table 1 HI in radio galaxies. Given is the name, NGC number, total HI mass detected in emission, diameter of the HI structure (or distance to the host galaxy for B2 1322+36), peak in HI surface density, and morphology of the HI structure (D = disk, R = ring, B = “blob”). H◦ = 71 km s−1 Mpc−1 used throughout this paper. # B2 Name NGC MHI DHI ΣHI Mor. 2 (M⊙) (kpc) (M⊙/pc ) HI 1 0258+35 1167 1.8×1010 160 2.7 D 2 0648+27 a - 8.5×109 190 1.7 R 3 0722+30 - 2.3×108 15 4.1 D 4 1217+29 b 4278 6.9×108 37 - D 5 1322+36 5141 6.9×107 20 3.7 B 6 - 3894 2.2×109 105 3.8 R a). Emonts et al. (2006); b). Morganti et al. (2006). type-I radio sources. The sources have a radio power 22.0 < Log (P1.4 GHz) < 25.0 with no bias in P1.4 GHz between the compact and extended sources. The radio sources are hosted by early-type galaxies (E and S0). Observations were made during various observing runs in the period Nov. 2002 - Feb. 2005 with the Very Large Array (VLA) in C-configuration and the Westerbork Synthe- sis Radio Telescope (WSRT). A full description of the sample and observing details will be presented in a future paper. 3 Results on our radio-loud sample We detect large-scale HI emission in six of our sample galaxies. Images and properties of the large-scale HI structures are shown in Fig. 1 and Table 1. In most cases the HI is distributed in a fairly regular rotating disk or ring (with diameter up to 190 kpc and mass up to 2×1010M ), although a varying ⊙ degree of asymmetry is still visible in these structures. For one of these radio galaxies – B2 0648+27 – we already confirmed a merger origin through both the detection of a post-starburst stellar population, that dominates the light throughout the opticalhost galaxy(see Emonts et al., 2006),andthe fact that plume- or tail-like structures appear in deep optical imaging (Heisler & Vader, 1994). The merger event in B2 0648+27 must have happened more than a Gyr ago, after which the HI gas that was expelled during the merger had the time to fall back onto the host galaxy and settle in the regular rotating ring that we observe. In case a merger event is confirmed also for the other HI-rich systems (with MHI > 109M⊙), than also for these systems the regular kinematics of the HI gas suggest that the HI structures are old. It is striking 4 1 2 6 4 3 5 Fig. 2.Total HImassinemission plotted against thelinearsize oftheradiosources. In case of non-detection a firm upper limit (3σ across 400 km s−1) is given. that we find no clear cases of ongoing mergers (in the form of tidal HI-tails, -bridges or -plumes) associated with our sample sources. In fact, regardless of the formation mechanism of these structures (be it major mergers or cold accretion), the large-scale structures are much older than the current period of radio-AGN activity. Another interesting result is that we find a segregation in large-scale HI mass content with radio source size (Fig. 2). The radio galaxies in our sample with MHI & 109M⊙ all have a compact radio source, while the more extended radio sources - all of Fanaroff & Riley type-I - do not contain these amounts of large-scale HI. As explained in Emonts et al. (2007), a possible explanation for this segregation is that - due to the re-distribution of the ISM in a merger event - the central radio sources in the HI-rich radio galaxies do not grow, either because they are frustrated by ISM in the central region of the galaxy, or because the fuelling stops before the sources can expand. The lack of large amounts of HI associated with the extended FR-I sources suggests that they are likely fed through processes other than gas-rich mergers (e.g. cooling flows ortheblackhole’srotationalenergy).Ifconfirmedbystudiesoflargersamples, the neutral gas content may therefore be a specific property of the host galaxy for various types of radio sources. 4 Comparison with radio-quiet samples Recently, Morganti et al. (2006) and Oosterloo et al. (2006) have completed two studies that were aimed at studying the occurrence and the morphology of large-scale HI in early-type galaxies (not selected on radio loudness). In this Section we compare the results of these two studies with the results that we obtained on our sample of radio-loud early-type galaxies. 5 HIPASS follow-up sample: The first study by Oosterloo et al. (2006) involves the follow-up imaging of HI in early-type galaxies detected in the single-dish HI Parkes All-Sky Survey (HIPASS). This project is described in detail in these proceedings by Serra et al. (2007). The HIPASS sample of early-type galaxies is a complete sample with a typical detection limit of about 109M . ⊙ Initial results give a conservative HI detection rate in early-type galaxies of − 5 12% (Sadler, 2001). Two-third of the HI structures that are imaged in the HIPASS follow-up study are large and regular rotating disks or rings (Oosterloo et al., 2006; Serra et al., 2007). Sauron sample: The second study by Morganti et al. (2006) involves deep HI imaging of 12 early-type galaxies selected from a larger, representative sam- ple of early-type galaxies observed with the optical integral field spectrograph SAURON. With a low detection limit of a few ×106M⊙, the HI detection rate in this sample is 70%. The morphology of the HI is more diverse than in the HIPASS follow-up study, with HI morphologies ranging from regular rotating disks to irregular clouds, tails and complex distributions. The HI detection rate of our complete B2 sample of radio-loud early-type galaxies is 25%. To compare this detection rate with the detection rates in the two samples of ’normal’ early-type galaxies (i.e. not selected on radio- loudness), we plot in Fig. 3 the observed HI mass against the power of the radio source (in case of non-detection the upper limit is given). From Fig. 3 it is immediately clear that the early-type galaxies from the HIPASS and Sauron samples are radio-quiet compared with the radio galaxies in our B2 sample (one object common to both theB2 andthe Sauron sample is the nearby radio HIPASS sub−sample B2 sample Sauron sample B2 sample (radio quiet) (radio loud) (radio quiet) (radio loud) Fig.3.HImassplottedagainstradiopowerfortheearly-typegalaxies ofthevarious samples. In case of non-detection the upper limit is plotted. The values of the HIPASSfollow-upandtheSauronsamplearetakenfromOosterloo et al.(2006)and Morganti et al.(2006).FortheB2samplethecirclesrepresenttheHIdetectionsand theflatarrowsthenon-detections; fortheHIPASSfollow-up andtheSauronsample thetrianglesrepresenttheHIdetections andthepointedarrowsthenon-detections. The dividing line between the various samples is drawn for clarification and does not represent a physical division between radio-loud and radio-quiet galaxies. 6 Table 2 HI detection rates of the various samples of early-type galaxies HIPASS B2 Sauron # galaxies 818 20∗ 12 detection limit (M⊙) ∼ 109 few×108 few×106 detection rate (%) 5-12∗∗ 25 70 % with MHI > 109M⊙ 5-12 10 17 % with MHI > few×108M⊙ - 25 33-50 ∗ Complete B2 sample does not include NGC 3894 (see Sect. 2) and B2 1557+26 (which redshift of z = 0.044 is too high). ∗∗ Initial results for HIPASS, based on unconfused HI detections (Sadler, 2001). galaxy B2 1217+29/NGC4278).Table 2 summarizes the HI detection rates of the three samples. Although the detection rates are very different for the three samples, we argue that this could be the result of a difference in sensitivity, rather than a true difference in HI content. This is based on the fact that for the various samples there does not appear to be a significant difference in the percentage of galaxies with HI masses above 109M⊙ (the detection limit of the HIPASS sample) and there is only a marginal difference for HI masses above a few ×108M (the detection limit of the B2 sample) – the latter effect ⊙ beeing subject to the small number statistics of the Sauron sample. The morphology of the observed HI structures in the two radio-quiet samples is remarkably similar to that of the HI structures in our radio-loud B2 sample. In all samples, at the high-mass end (MHI & ×109M⊙) the HI is distributed in large and regular rotating disk- or ring-like structures. For lower amounts (MHI ∼ few×106−108M⊙), the samples also contain galaxies in which a more irregular HI distribution is detected (as is the case for B2 1322+36). Thus, as far as we can tell from the limited comparison between the three samples, there appears to be no major difference in both HI detection rate andHImorphologybetween theradio-quietandradio-loudearly-typegalaxies in these samples. For sure, there is no evidence that our radio-loud sample has a higher detection-rate or contains more tidally distorted HI structures than the radio-quiet samples. If confirmed by larger samples with comparable sensitivity, this indicates that the radio-loud phase could be just a short period that occurs at some point during the lifetime of many – or maybe even all? – early-type galaxies. As a final note, we would like to stress that our complete sample of radio-loud early-type galaxies did not include the more powerful radio sources of type FR-II, which are found at higher z and which are often associated with major mergers (Heckman et al., 1986). 7 5 Conclusions In a study of HI in a complete sample of nearby, non-cluster radio galaxies, we detect large-scale HI emission in 25% of the cases. The HI is mainly dis- tributed in fairly regular rotating disk- or ring-like structures. Regardless of the formation mechanism of these HI structures (be it major mergers or cold accretion), their formation must have occurred long before the onset of the current phase of radio-AGN activity. We find no signs of ongoing mergers, nor do we find a major difference in morphology or detection rate with samples of radio-quiet early-type galaxies. If confirmed by larger samples, this indicates that the radio-loud phase could be just a short period that occurs at some point during the lifetime of many (all?) early-type galaxies. Acknowledgements The author B. Emonts would like to thank Raffaella Morganti and Nanuschka Csonka for organising this nice workshop and Montse Villar-Mart´ın for giving useful comments to improve the paper. Part of this project is funded by the Netherlands Organisation for Scientific Research (NWO) under Rubicon grant 680.50.0508. References Barnes, J. E. 2006, MNRAS, 333, 481 Emonts, B. H. C., Morganti, R., Tadhunter, C. N., Holt, J., Oosterloo, T. A., van der Hulst, J. M. & Wills, K. A. 2006, A&A, 454, 125 Emonts, B. H. C., Morganti, R., Oosterloo, T. A., van der Hulst, J. M., van Moorsel, G. & Tadhunter, C. N. 2007, A&A, in press. Fanaroff B. L. & Riley, J. M. 1974, MNRAS, 167, 31 Heckman, T. M. et al. 1986, ApJ, 311, 526 Heisler, C. A. & Vader, J. P. 1994, AJ, 107, 35 Hibbard, J. E. & van Gorkom, J. H. 1996, AJ, 111, 655 Kaufmann, T. et al. 2006, MNRAS, 370, 1612 Kerˇes, D., Katz, N., Weinberg, D. H. & Dav´e, R. 2005, MNRAS, 363, 2 Mihos J.C. & Hernquist L. 1996, ApJ, 464, 641 Morganti, R. et al. 2006, MNRAS, 371, 157 Noordermeer,E.,vanderHulst,J.M.,Sansizi,R.,Swaters,R.A.&vanAlbada, T.S. 2005, A&A, 442, 137 Oosterloo, T. A., Sadler, E. M., Morganti, R., van der Hulst, J. M. & Serra, P., 2006, A&A (submitted) Sadler, E. M. 2001, ASP Conf. Ser. 240, 445 Schiminovich, D., van Gorkom, J., van der Hulst, T., Oosterloo, T. & Wilkin- son, A. 1997, ASP Conf. Ser. 116, 362 8 Serra, P., Trager, S. C., van der Hulst, J. M., Oosterloo, T. A. & Morganti, R. 2006, A&A, 453, 493 Serra,P.,et al.inproceedings of theconference “TheFateofGasinGalaxies”, New Astron. Rev., in press. van Gorkom, J. & Schiminovich, D. 1997, ASP Conf. Ser. 116, 310 9