A&A 436, 75-90 (2005)
DOI: 10.1051/0004-6361:20042175

New H2O masers in Seyfert and FIR bright galaxies[*]

C. Henkel1 - A. B. Peck2 - A. Tarchi3,4 - N. M. Nagar5,6,7 - J. A. Braatz8 - P. Castangia4,9 - L. Moscadelli4


1 - Max-Planck-Institut für Radioastronomie, Auf dem Hügel 69, 53121 Bonn, Germany
2 - Harvard-Smithsonian Center for Astrophysics, SAO/SMA Project, 654 N. A'ohoku Pl., Hilo, HI 96720, USA
3 - Istituto di Radioastronomia, CNR, via Gobetti 101, 40129-Bologna, Italy
4 - INAF - Osservatorio Astronomico di Cagliari, Loc. Poggio dei Pini, Strada 54, 09012 Capoterra (CA), Italy
5 - INAF, Arcetri Observatory, Largo E. Fermi 5, 50125 Florence, Italy
6 - Kapteyn Instituut, Postbus 800, 9700 AV Groningen, The Netherlands
7 - Astronomy Group, Departamento de Física, Universidad de Concepción, Casilla 160-C, Concepción, Chile
8 - National Radio Astronomy Observatory, PO Box 2, Green Bank, WV 24944, USA
9 - Universitá di Cagliari, Dipartimento di Fisica, Cittadella Universitaria, 09012 Capoterra (CA), Italy

Received 14 October 2004 / Accepted 16 February 2005

Abstract
Using the Effelsberg 100-m telescope, detections of four extragalactic water vapor masers are reported. Isotropic luminosities are $\sim$50, 1000, 1 and 230 $L_{\odot}$ for Mrk 1066 (UGC 2456), Mrk 34, NGC 3556 and Arp 299, respectively. Mrk 34 contains by far the most distant and one of the most luminous water vapor megamasers so far reported in a Seyfert galaxy. The interacting system Arp 299 appears to show two maser hotspots separated by approximately 20 $^{\prime \prime }$. With these new results and even more recent data from Braatz et al. (#!Bra04!#, ApJ, 617, L29), the detection rate in our sample of Seyferts with known jet-Narrow Line Region interactions becomes 50% (7/14), while in star forming galaxies with high ( $S_{\rm 100~\mu m}>50$ Jy) far infrared fluxes the detection rate is 22% (10/45). The jet-NLR interaction sample may not only contain "jet-masers'' but also a significant number of accretion "disk-masers'' like those seen in NGC 4258. A statistical analysis of 53 extragalactic H2O sources (excluding the Galaxy and the Magellanic Clouds) indicates (1) that the correlation between IRAS Point Source and H2O luminosities, established for individual star forming regions in the galactic disk, also holds for AGN-dominated megamaser galaxies; (2) that maser luminosities are not correlated with 60 $\mu$m/100 $\mu$m color temperatures; and (3) that only a small fraction of the luminous megamasers ( $L_{\rm H_2O} > 100$ $L_{\odot}$) detectable with 100-m sized telescopes have so far been identified. The H2O luminosity function (LF) suggests that the number of galaxies with 1  $L_{\odot} < L_{\rm H_2O} < 10$ $L_{\odot}$, the transition range between "kilomasers'' (mostly star formation) and "megamasers'' (active galactic nuclei), is small. The overall slope of the LF, $\sim$-1.5, indicates that the number of detectable masers is almost independent of their luminosity. If the LF is not steepening at very high maser luminosities and if it is possible to find suitable candidate sources, H2O megamasers at significant redshifts should be detectable even with present day state-of-the-art facilities.

Key words: masers - galaxies: active - galaxies: jets - galaxies: Seyfert - galaxies: starburst - radio lines: galaxies

1 Introduction and sample selection

Extragalactic water vapor masers, observed through the 22 GHz ( $\lambda \sim1.3$ cm) $J_{\rm K_aK_c} = 6 _{16}{-}5_{23}$ line of ortho-H2O that traces warm ( $T_{\rm kin} \ga400$ K) and dense (n(H $_2)\ga10^7$ cm-3) molecular gas (e.g. Kylafis & Norman 1987, 1991; Fiebig & Güsten 1989), are primarily seen as a means to probe nuclear accretion disks in active galaxies. The best known source, NGC 4258, shows a thin, slightly warped, nearly edge-on Keplerian disk of subparsec scale enclosing a central mass of $\sim$ $4\times10 ^{7}$ $M_{\odot}$ (e.g. Greenhill et al. 1995; Miyoshi et al. 1995; Herrnstein et al. 1999).

There is evidence, however, for additional classes of extragalactic H2O masers. There are sources in which at least a part of the H2O emission appears to be the result of an interaction between the nuclear radio jet and an encroaching molecular cloud (e.g. Mrk 348; Peck et al. 2003).

Most of the nuclear water vapor sources are characterised by (isotropic) $L_{\rm H_2O} > 10$ $L_{\odot}$ and are classified as "megamasers''. H2O masers associated with prominent star forming regions similar to those seen in the Galaxy (e.g. in M 33; Greenhill et al. 1993) are less luminous and comprise the majority of known "kilomasers'' ( $L_{\rm H_2O} \la 10$ $L_{\odot}$).

Table 1: "Jet-maser'' observations.

Providing bright, almost point-like hotspots, H2O masers are ideal probes for Very Long Baseline Interferometry (VLBI). A broad variety of astrophysical studies is possible. This includes the determination of geometric distances and 3-dimensional velocity vectors of galaxies, masses of nuclear engines, maps of accretion disks and physics of nuclear jet-molecular cloud interaction (for recent reviews, see Greenhill 2002, 2004; Maloney 2002; Henkel & Braatz 2003; Morganti et al. 2004; Henkel et al. 2005).

So far, almost 1000 active galaxies have been surveyed. In order to detect a large number of strong maser sources that could help to elucidate the nuclear environment of their parent galaxies and their geometric distance, typical detection limits were several 10 mJy or more, resulting in detection rates between zero (e.g. Henkel et al. 1998) and a few percent (e.g. Henkel et al. 1984; Braatz et al. 1996; Greenhill et al. 2002). The low detection rates are probably the result of the limited sensitivity of the surveys, rather than an intrinsic lack of extragalactic H2O masers. The technology exists to do deeper searches; what is required is a set of criteria to narrow the list of candidates from all nearby galaxies to a manageable few. Here we present the results of two quite different, but equally successful, deep searches (to estimate distances, H0 = 75 km s-1 Mpc-1 is used, whenever possible, throughout the paper).

Sample 1:
The targets of the first sample (hereafter "jet-maser'' sample) have been selected from a collection of Seyfert galaxies with declination >-30$^{\circ }$ in which both the inclination of the host galaxy and the linear extent of the radio source are known (Nagar & Wilson 1999). To maximize the detection rate, we have chosen galaxies which either show evidence at optical wavelengths of interaction between the radio jet and clouds in the narrow line region, or have a face-on (i<35$^{\circ }$) galaxy disk and extended pc to 100-pc scale radio structures, possibly suggesting that both the disk of the galaxy and the radio jet should be fairly close to the plane of the sky. This combination of geometries increases the probability that the radio jet lies close to the disk of the galaxy, thus increasing the likelihood of interaction between the radio jet and the galaxy interstellar medium (ISM). Indeed, two of the four "jet-maser'' sources known prior to this study, NGC 1068 (Gallimore et al. 2001) and Mrk 348 (Peck et al. 2003), appear in the parent sample and fit the above criteria. The complete list of 14 jet-maser target sources is given in Table 1.

Sample 2:
The second sample (hereafter "far infrared maser'' or "FIR-maser'' sample) is comprised of all galaxies with declination >-30$^{\circ }$ and IRAS point source flux density $S_{100~\mu \rm m} >50$ Jy (e.g. Fullmer & Lonsdale 1989). The same criteria were already used by Henkel et al. (1986, hereafter HWB) to perform a first FIR-flux-based survey in which they detected a new maser in the galaxy IC 10. Recently we have detected masers in IC 342 (Tarchi et al. 2002a) and NGC 2146 (Tarchi et al. 2002b), which were previously undetected although they were included in the sample of HWB. We attribute this result to an occcasional flare (IC 342) and to an improvement in the Effelsberg receiver and backend systems (NGC 2146). This motivated us to re-observe the list of previously undetected targets. The list of sources was compiled using the IRAS Point Source Catalog (e.g. Fullmer & Lonsdale 1989) and is shown in Table 2. There is a total of 45 sources, the great majority of them being spiral galaxies. Two of these already known to show maser emission, IC 342 and NGC 2146, were reobserved.

Table 2: "FIR-maser'' observations.

2 Observations

The target sources of the two samples were measured in the 616-523 line of H2O (rest frequency: 22.23508 GHz) with the 100-m telescope of the MPIfR at Effelsberg on various occasions between June 2001 and April 2004. The full width to half power beamwidth was $\sim$40 $^{\prime \prime }$ and the pointing accuracy was in most cases better than 10 $^{\prime \prime }$ (see also Sect. 4.2.3). A dual channel HEMT receiver provided system temperatures of 130-180 K on a main beam brightness temperature scale. The observations were carried out in a dual beam switching mode with a beam throw of 2$^\prime $ and a switching frequency of $\sim$1 Hz. The autocorrelator backend was split into eight bands of width 40 or 80 MHz and 512 or 256 channels each that could individually be shifted in frequency by up to $\pm$250 MHz relative to the recessional velocity of the galaxy. This yielded channel spacings of $\sim$1 or $\sim$4 km s-1. A few spectra were also taken with two bands of 20 MHz and 4096 channels each. The resulting channel spacing was then $\sim$0.07 km s-1. Flux calibration was obtained by measurements of W3(OH) (for the flux, see Mauersberger et al. 1988). Gain variations as a function of elevation were taken into account (see Eq. (1) of Gallimore et al. 2001) and the 1$\sigma$ flux calibration error is expected to not exceed $\pm$10%.

Table 3: Line parameters of newly detected masers.

3 Results

From the jet-maser sample, we have detected two new megamasers, Mrk 1066 and Mrk 34. The FIR-maser sample also yields two new detections, a megamaser (Arp 299) and a kilomaser (NGC 3556). Line profiles are shown in Figs. 1-6. Line parameters including recessional velocity and (isotropic) H2O luminosity are given in Table 3. Properties of the detected galaxies are discussed below.


  \begin{figure} \par\includegraphics[angle=-90, width=7.9cm, clip]{2175fig1.ps} \end{figure} Figure 1: 22 GHz H2O megamaser profiles toward Mrk 1066 with a channel spacing of 1.08 km s-1. $\alpha _{2000}$ = 02$^{\rm h}$59$^{\rm m}$58 $~.\!\!^{\rm s}$6, $\delta _{2000} = 36 ^{\circ }$49$^\prime $14 $^{\prime \prime }$. Velocity scales are with respect to the Local Standard of Rest (LSR) and use the optical convention that is equivalent to cz. $V_{\rm sys} = {\rm c}z_{\rm sys} = 3605$ km s-1 (NASA/IPAC Extragalactic Database (NED)). $V_{\rm LSR} - V_{\rm HEL} = -3.55$ km s-1.
Open with DEXTER

3.1 Mrk 1066 (UGC 2456)

Mrk 1066 is a FIR luminous ( $L_{\rm FIR} \sim 7 \times10 ^{10}$ $L_{\odot}$) SB0+ galaxy, containing a double nucleus (e.g. Gimeno et al. 2004). It is one of the few early-type galaxies that have been detected in CO (see Henkel & Wiklind 1997, note that their FIR luminosity (their Table 2) is too low). Its systemic velocity is 3605 km s-1 (see Table 1), corresponding to a distance of $\sim$50 Mpc. The inclination angle is 42$^{\circ }$ (Whittle 1992). Hubble Space Telescope (HST) imaging of the nuclear region (Bower et al. 1995) shows a jet-like feature in a narrow-band image which includes [O III] and H$\beta$. The distribution is bipolar, oriented at $\sim$315$^{\circ }$, and extending to an angular radius of $\sim$1 $~.\!\!^{\prime\prime}$5, with emission from the north-western side being dominant. In H$\alpha$ and [N II], the jet is equally prominent on both sides of the nucleus. The 3.6 cm radio continuum emission (Nagar et al. 1999) is extended along the same axis over $\sim$2 $~.\!\!^{\prime\prime}$5.

There is a strong narrow maser spike at 3636 km s-1 with a full width to half maximum linewidth of less than 2 km s-1 (Figs. 1 and 2) and a peak flux density of 80 mJy. The spike becomes narrower between March 7 and 10, 2002, appears to be unresolved in frequency in May and reaches only $\sim$50 mJy in September. It seems that a gradual decrease of the linewidth is finally accompanied by a decrease in peak flux density. At a distance of $\sim$50 Mpc, isotropic luminosities reach $\sim$10 $L_{\odot}$. A second much wider component, at $\sim$3550 km s-1, has a peak flux density of 10-20 mJy and an isotropic luminosity of $\sim$40 $L_{\odot}$.


  \begin{figure} \par\includegraphics[angle=-90, width=7.7cm]{2175fig2.ps} \end{figure} Figure 2: High spectral resolution profiles of the narrow maser spike in Mrk 1066 (see Fig. 1). The channel spacing is 0.067 km s-1.
Open with DEXTER

3.2 Mrk 34

Mrk 34 (IRAS 10309+6017), another luminous ($\sim$1011 $L_{\odot}$) FIR source, is a distant Seyfert 2 galaxy (z=0.0505, $D \sim200$ Mpc; Falcke et al. 1998). The optical galaxy is characterized as having an inclination angle of 57$^{\circ }$ in Whittle (1992), although a second generation Digital Sky Survey (DSS) image shows the galaxy to be compact, with poorly defined outer isophotes (Nagar & Wilson 1999). The radio emission has an extended structure ($\sim$2 $~.\!\!^{\prime\prime}$5; Ulvestad & Wilson 1984), and strong evidence for an interaction between the radio jet and NLR clouds has been found by Falcke et al. (1998).

Mrk 34 is one of the most distant and most luminous H2O megamasers ever detected. The maser shows two or three distinct spectral features (Figs. 3 and 4). One is centered at a velocity of $\sim$14 840 km s-1, another at $\sim$15 770 km s-1, and a third one is tentatively seen at $\sim$14 665 km s-1. Peak flux densities are up to 10 mJy and total isotropic luminosities are $\sim$1000 $L_{\odot}$.


  \begin{figure} \par\includegraphics[angle=-90, width=7.5cm]{2175fig3.ps} \end{figure} Figure 3: Low velocity H2O megamaser profiles toward Mrk 34 with a channel spacing of 4.65 km s-1. $\alpha _{2000}$ = 10$^{\rm h}$34$^{\rm m}$08 $~.\!\!^{\rm s}$6, $\delta _{2000}$ = 60$^{\circ }$01$^\prime $52 $^{\prime \prime }$. c $z_{\rm sys}$ = 15145 km s-1 (de Grijp et al. 1992). $V_{\rm LSR}-V_{\rm HEL} = +5.23$ km s-1.
Open with DEXTER


  \begin{figure} \par\includegraphics[angle=-90, width=7.5cm]{2175fig4.ps} \end{figure} Figure 4: High velocity megamaser feature toward Mrk 34 with a channel spacing of 18.65 km s-1.
Open with DEXTER


  \begin{figure} \par\includegraphics[angle=-90, width=7.8cm]{2175fig5.ps} \end{figure} Figure 5: H2O kilomaser spectra toward NCG 3556 ( $\alpha _{2000} = 11 ^{\rm h}$11$^{\rm m}$31 $~.\!\!^{\rm s}$2, $\delta _{2000} = 55$$^{\circ }$40$^\prime $25 $^{\prime \prime }$). Channel spacings are 4.2 km s-1 ( upper panel) and 1.06 km s-1. cz = 700 km s-1 (NED). $V_{\rm LSR}-V_{\rm HEL} = +5.90$ km s-1.
Open with DEXTER

3.3 NGC 3556 (M 108)

NGC 3556 is an edge-on spiral galaxy located at a distance of $\sim$12 Mpc. Its FIR luminosity, $L_{\rm FIR} \sim 10 ^{10}$ $L_{\odot}$, is similar to that of the Milky Way. Radio continuum, H I and X-ray data indicate a violent disk halo interaction, including a prominent radio halo (e.g. de Bruyn & Hummel 1979), large H I extensions possibly delineating expanding supershells (King & Irwin 1997), compact radio continuum sources, likely representing supernova remnants (Irwin et al. 2000), and extraplanar diffuse X-ray emission (Wang et al. 2003). 12CO and HCN observations (Gao & Solomon 2004) indicate a substantial molecular gas content. No OH maser was detected (Unger et al. 1986).

With a peak flux density of 20-40 mJy, the H2O maser has an isotropic luminosity of $\sim$1 $L_{\odot}$. One or two velocity components are seen. The profiles are shown in Fig. 5.

3.4 Arp 299 (Mrk 171)

Arp 299 is a merging system at $D \sim40$ Mpc, composed of two main sources, IC 694 and NGC 3690 (for an alternative nomenclature, see Sect. 4.2.3), that are separated by 22 $^{\prime \prime }$ in east-west direction (e.g. Sargent & Scoville 1991). A FIR luminosity of several 1011 $L_{\odot}$ (Casoli et al. 1992) places Arp 299 near the boundary between luminous (LIRGs) and ultraluminous (ULIRGs) infrared galaxies. Supporting the merging scenario, two highly extended H I tails have been identified by Hibbard & Yun (1999). Radio and infrared observations reveal three main regions of activity (e.g. Gehrz et al. 1983; Aalto et al. 1997; Casoli et al. 1999), the nuclear regions of IC 694 and NGC 3690 and an interface where IC 694 and NGC 3690 overlap. NGC 3690 contains a deeply enshrouded active galactic nucleus (AGN), while the situation with respect to the similarly obscured nuclear region of IC 694 is less clear (e.g. Della Ceca et al. 2002; Ballo et al. 2004; Gallais et al. 2004). 12CO and HCN J=1-0 emission line peaks are strongest toward these most active regions, indicating the presence of large amounts of molecular gas. The positions of strongest 13CO J =1-0 line emission are, however, displaced from these hotspots (Aalto et al. 1997).

In Arp 299, water maser profiles are extremely broad ($\sim$200 km s-1), with peak flux densities of 30 mJy (Fig. 6). Adopting a distance of 42 Mpc, the total isotropic luminosity is $\sim$250 $L_{\odot}$, placing the object among the more luminous H2O megamaser sources. The maser line is centered at a velocity of 3100 km s-1, i.e. close to the systemic velocity of the entire complex of sources constituting Arp 299.

3.5 IC 342 and NGC 2146

We also observed the previously detected H2O maser sources IC 342 and NGC 2146. IC 342 was not detected in June 2001 and March 2002, indicating that the flaring component observed at $V_{\rm LSR}$ $\sim$ 16 km s-1 (Tarchi et al. 2002a) has been quiescent since June 2001. Spectra from NGC 2146, obtained in March 2002, show no significant variations with respect to profiles observed two years earlier (see Tarchi et al. 2002b).

4 Discussion

As indicated in Sect. 1, the surveys presented here have been targeted to detect two classes of extragalactic water masers, "jet-masers'' and "FIR-masers''.

4.1 The jet-maser sample

Jet-masers provide insight into the interaction of nuclear jets with dense warm molecular gas in the central parsecs of galaxies. All jet-masers known to date arise from the innermost regions of active galaxies and yield important information about the evolution of jets and their hotspots. If continuum emission from the core of the radio source is responsible for variations in maser intensity, monitoring of continuum and line emission can provide estimates, through reverberation mapping, of the speed of the material in the jet, particularly in sources where the jet appears to lie close to the plane of the sky. If, on the other hand, the continuum flare is caused by the brightening of the hotspot or working surface in the jet as it impacts a denser molecular cloud, then the onset of the continuum and maser flares should be nearly simultaneous (Peck et al. 2003).


  \begin{figure} \par\includegraphics[angle=-90, width=7.7cm]{2175fig6.ps} \end{figure} Figure 6: H2O megamaser profile toward Arp 299 ( $\alpha _{2000} = 11$$^{\rm h}$28$^{\rm m}$31 $~.\!\!^{\rm s}$9, $\delta _{2000} = 58$$^{\circ }$33$^\prime $45 $^{\prime \prime }$). The channel spacing is 4.3 km s-1. $V_{\rm LSR}-V_{\rm HEL} = +6.97$ km s-1. For radial velocities, see Sect. 4.2.3.
Open with DEXTER

In view of these implications we need to investigate the true nature of the megamasers detected in Mrk 1066 and Mrk 34. Are these really jet-masers as suggested by the selection criteria (Sect. 1)? While only VLBI observations can provide a definite answer, a detailed look at well studied jet-masers and the maser sources recently discovered by Braatz et al. (2004) can provide relevant information.

The four jet-maser sources known prior to this survey are NGC 1068 (Gallimore et al. 1996), the Circinus galaxy (Greenhill et al. 2001, 2003a), NGC 1052 (Claussen et al. 1998) and Mrk 348 (Peck et al. 2003). The first two sources show both maser emission from a circumnuclear disk and emission arising along the edges of an ionization cone or outflow in the jet. In NGC 1068, the jet-maser velocities are blue-shifted by $\sim$160 km s-1 from systemic and the feature is broad (FWHP (full width to half power) linewidth $\sim$ 60 km s-1; e.g. Gallimore et al. 2001). In Circinus, both red- and blue-shifted features are seen at velocities up to 160 km s-1 from systemic (Greenhill et al. 2003a). In NGC 1052 and Mrk 348, all the maser emission may arise along the jet (Claussen et al. 1998; Peck et al. 2003). As in NGC 1068, this is accompanied by relatively large linewidths ($\sim$90 and $\sim$130 km s-1) and significant shifts relative to the systemic velocity ($\sim$+150-200 and +130 km s-1), respectively.

To summarize, jet-maser features tend to be broader (a few 10 km s-1) than those typically seen in disk-maser sources like NGC 4258 (a few km s-1) and are usually displaced from the systemic velocity. In Mrk 1066, it is the component at c $z \sim3550$ km s-1 that shows the properties expected in the case of a nuclear jet-type H2O maser (see Fig. 1). The intense narrow spike near the systemic velocity would have a different origin. We also note, however, that the broad blue- and the narrow red-shifted features bracket the systemic velocity (3605 km s-1). Thus a masing disk like in NGC 4258 cannot be excluded.

Toward Mrk 34, the main components at c $z\sim14~840$ and 15 770 km s-1 are wide enough for characteristic jet-maser emission. However, the intrinsic weakness of the features requires smoothing which could hide individual narrow components that might represent a significant fraction of the maser emission. Furthermore, Figs. 3 and 4 show two or three velocity components that bracket the systemic velocity ( $V_{\rm sys} = 15~145\pm90$ km s-1; de Grijp et al. 1992). The velocity displacements may not be symmetric; the red-shifted emission (Fig. 4) appears to show a larger displacement than the blue-shifted emission (Fig. 3) from systemic, which would argue against the possibility of a circumnuclear disk. However, the uncertainty in c $z_{\rm sys}$ is large so that an accretion "disk-maser'' scenario is also possible. Among the three "jet-maser'' sources detected by Braatz et al. (2004; see also Table 1), Mrk 1157 (NGC 591), Mrk 3 and NGC 4151, the first one also shows a profile reminiscent of a disk-maser source. We thus conclude that our jet-maser sample does not provide exclusively jet-maser sources. Having selected sources with jets that appear to be oriented close to the plane of the sky (Sect. 1), this is apparently also an excellent selection criterion to find disk-masers that are characterized by nuclear disks viewed edge-on. Disk-masers may constitute a significant fraction of the newly discovered `jet-maser' sources and some of these may even show both signatures (like NGC 1068) of nuclear activity.

Including all sources that fulfill the selection criteria of our jet-maser sample (Table 1), the detection rate becomes an (almost incredible) 7/14 or 50%. This is the first survey undertaken to look specifically for jet-masers. The number of sources and detections is still too small for a detailed statistical analysis. While it remains to be seen whether the masers are jet- or disk-masers, the unprecedented success rate suggests that both types of masers have been found and that a tilt of >55$^{\circ }$ between nuclear and large scale disk is a highly favorable configuration for the occurrence of H2O masers in Seyfert galaxies.

4.2 The FIR-maser sample

4.2.1 Previously detected masers

FIR emission commonly arises from dust grains heated by newly formed stars. In the Milky Way, 22 GHz H2O masers are associated with sites of (mostly massive) star formation. Therefore our sample of FIR bright galaxies (Table 2) is a suitable tool to detect extragalactic H2O masers associated with young massive stars. Such masers have the potential to pinpoint the location of prominent star forming regions and to estimate their distance through complementary measurements of proper motion and radial velocity (e.g. Greenhill et al. 1993). Monitoring such masers and determining their three dimensional velocity vectors allows us to derive the gravitational potential of galaxies or groups of galaxies and to improve our understanding of the evolution of such groups with time (for the Local Group, see Brunthaler et al. 2005).

Including the early part of our survey (Tarchi et al. 2002a,b), we detected with IC 342, NGC 2146, NGC 3556 and Arp 299 four new H2O masers in a total of 45 sources (see Table 2). The new detections are a consequence of higher sensitivity (1$\sigma$ noise levels of $\sim$10 mJy for a 1 km s-1 channel), highly improved baselines and luck (in the case of the short-lived flare observed toward IC 342). Including all previously detected sources in the complete sample shown in Table 2, we find a detection rate of 10/45 or $22\pm7$%. For sources with 100$\mu$m fluxes in excess of 100 Jy, the detection rate becomes even higher: 7/19 or $37\pm14$%. Detection rates for the jet-maser and the FIR-maser samples lie far above the corresponding rates deduced from other carefully selected samples (see e.g. Henkel et al. 1984, 1986, 1998; Haschick & Baan 1985; Braatz et al. 1996; Greenhill et al. 2002).


  \begin{figure} \par\includegraphics[angle=-90, width=16.2cm]{2175fig7.ps} \end{figure} Figure 7: H2O megamaser profiles toward Arp 299, taken on Jan. 30, 2004, during a night with excellent pointing conditions. The reference position is $\alpha _{2000} = 11 ^{\rm h}$28$^{\rm m}$31 $~.\!\!^{\rm s}$9, $\delta _{2000} = 58^{\circ }$33$^\prime $45 $^{\prime \prime }$. Offsets in arcsec ( $\Delta \alpha $, $\Delta \delta $) are given in the upper right corner of each box. Spectra with offsets $\Delta \alpha = -20$ $^{\prime \prime }$ trace emission from the "overlap'' region (north) and NGC 3690 (south). Spectra with $\Delta \alpha = +20$ $^{\prime \prime }$ are sensitive to emission from IC 694 (see Sect. 4.2.3). The channel spacing is 16.8 km s-1.
Open with DEXTER

To find out why the FIR-maser sample contains numerous 22 GHz H2O maser sources and to elucidate the nature of the sources in NGC 3556 and Arp 299, we have to classify the properties of the previously studied masers of this sample. Two of the sources, those in NGC 1068 and NGC 3079, are luminous megamasers (e.g. Gallimore et al. 2001; Trotter et al. 1998). The weaker "kilomasers'' in IC 10, IC 342, NGC 2146 and NGC 3034 (M 82) are associated with sites of massive star formation (Argon et al. 1994; Baudry & Brouillet 1996; Tarchi et al. 2002a,b). There are also two known weak nuclear kilomasers, in NGC 5194 (M 51) (Hagiwara et al. 2001b) and in NGC 253 (Henkel et al. 2004). Whether they are related to the nearby AGN or to star formation remains, however, unclear.

4.2.2 NGC 3556

The position of the H2O kilomaser in NGC 3556 is not yet accurately known. From the relative number of nuclear versus non-nuclear masers of similar luminosity, the most likely interpretation is an association with a site of massive star formation. In view of NGC 253 and NGC 5194, however, there is a small chance for a nuclear maser in NGC 3556.

4.2.3 Arp 299

As indicated in Sect. 3.4, the merging system Arp 299 is composed of two galaxies, NGC 3690 in the west, IC 694 in the east, and a star-bursting interface or overlap region 10 $^{\prime \prime }$ north of NGC 3690[*]. Since NGC 3690 and IC 694 are only half a beam size (20 $^{\prime \prime }$) apart in our observations and since the separation between NGC 3690 and the overlap region is even smaller, extremely good pointing conditions were needed to map the region. Three maps were made. Figure 7 shows the most accurate (pointing accuracy $\pm$4 $^{\prime \prime }$) and extended, albeit also the most noisy one. In spite of the rather low signal-to-noise ratios we note that (1) H2O emission may originate from more than one hotspot; (2) one of the potential sources, the one in the east, is close to IC 694, where an OH megamaser was already reported (Baan & Haschick 1990); (3) there appears to be a western peak of emission that is located near the center of the second dominant galaxy of the system, NGC 3690; (4) a broader feature near the center is likely caused by blending of the two main hotspots associated with IC 694 and NGC 3690; (5) the vigorously star forming overlap region appears to be devoid of H2O megamaser emission.

When we compare these results with the CO velocity field observed by Casoli et al. (1999), we find that the H2O velocity of the eastern peak, 2980 km s-1, is consistent with the CO velocity of the south-eastern part of IC 694 (i.e. offset w.r.t. the nucleus of IC 694 that has a velocity of cz = 3110 km s-1). Although the location of the western hotspot is close to the core of NGC 3690, the velocities of the maser emission, the CO lines and the systemic velocity of NGC 3690 do not match perfectly ($\sim$3100 km s-1 from CO, c $z_{\rm sys}$ = 3121 km s-1 (NASA/IPAC Extragalactic Database (NED)), versus $\sim$3150 km s-1 from H2O). Interestingly, velocities near 3150 km s-1 as seen in H2O are consistent with those of the overlap region.

A comparison of the profile shown in Fig. 6 with the central one in Fig. 7 suggests a slight offset in position (a few arcsec in east-west direction) and weaker peak emission in the latter case. This is within the uncertainties of pointing and calibration, but maser variability can also explain the differences.

With the H2O emission likely originating from IC 694 and NGC 3690, Arp 299 is the fourth extragalactic system beyond the Magellanic Clouds that is known to exhibit $\lambda$ = 18 cm OH and $\lambda$ = 1.3 cm H2O maser emission (in NGC 253, NGC 1068 and M 82, such masers are also observed; see Weliachew et al. 1984; Turner 1985; Baudry & Brouillet 1996; Gallimore et al. 1996; Henkel et al. 2004). In these other galaxies, however, either H2O or OH or both lines only reach kilomaser luminosities. IC 694 may thus be the first known galaxy with both an OH and an H2O megamaser (for OH, see Baan & Haschick 1990). The global OH and H2O line profiles appear to be similar, except for a weak OH feature at $\sim$3500 km s-1that is not seen in H2O.

Arp 299 is the second most luminous FIR source with a known H2O megamaser. While OH megamasers are closely associated with ultraluminous infrared galaxies (ULIRGs; see e.g. Darling & Giovanelli 2002a), the only ULIRG with a luminous H2O maser was so far NGC 6240 (Hagiwara et al. 2002, 2003a; Nakai et al. 2002; Braatz et al. 2003). In accordance with the observed anticorrelation between the occurence of OH and H2O megamasers (OH megamasers may arise from low density molecular gas, while H2O megamasers originate from gas of much higher density; e.g. Kylafis et al. 1991; Randell et al. 1995), this is one of those ULIRGs in which no OH emission is seen. The second detection of an H2O megamaser in a luminous FIR galaxy (although Arp 299 is not quite as luminous as NGC 6240) makes a dedicated survey of such luminous FIR galaxies worthwhile.


  \begin{figure} \par\includegraphics[angle=-90, width=8.3cm]{2175fig8.ps} \end{figure} Figure 8: Detection rate of the H2O FIR-maser sample (see Table 2 for the targets and Sect. 1 for selection criteria) including all galaxies exceeding a given IRAS Point Source Catalog $S_{\rm 100~\mu m}$ flux density.
Open with DEXTER

4.2.4 Detection probabilities

The high rate of maser detections in our sample of FIR luminous galaxies (Table 2) strongly suggests that a relationship between FIR flux density and maser phenomena exists, consistent with the assessment of HWB. The detection rate for masers in accretion disks is dictated by tight geometric constraints. The cumulative maser output of a star forming region may not be so narrowly confined.

Figure 8 shows the cumulative detection rate above a given 100 $\mu$m IRAS Point Source Catalog flux for the parent galaxy. The detection rate strongly declines with decreasing FIR flux. For fluxes $\sim$1000 Jy, 100-300 Jy, and 50-100 Jy, we find detection rates of 2/2 or 100%, 5/17 or 29% and 3/26 or 12% (unlike in Fig. 8, these are not cumulative detection rates but detection rates related to their specific FIR flux density interval). Maffei 2 and NGC 5236 (M 83) show no detectable maser emission near their nuclei but have $S_{\rm 100~\mu m} >200$ Jy. In view of the statistical properties of the sample, frequent monitoring of these sources would likely reveal H2O maser emission, possibly a short-lived flare like that seen in IC 342 (Tarchi et al. 2002a).

Figure 8 shows a detection probability of $\sim$50% for sources with $S_{\rm 100~\mu m} \ga120$ Jy. If the two brightest FIR sources, NGC 253 and M 82 (NGC 3034), were at $D \sim10$ Mpc (i.e. three times their estimated distance), this would imply $S_{\rm 100~\mu m} \sim100$ Jy and H2O peak fluxes of $\sim$5 mJy (broad emission feature) and $\sim$10 mJy (narrow emission feature), respectively (for the line profiles, see Ho et al. 1987; Baudry et al. 1994; Henkel et al. 2004). Thus the two sources would be just below the detection limit, consistent with the detection probability at the corresponding 100 $\mu$m flux.

While there is significant scatter (among the sources of Table 2, the most extreme source by far is NGC 3079, whose H2O maser would be detectable even at a distance corresponding to $S_{\rm 100~\mu m} \sim10$ Jy), we conclude that at present sensitivities, there is for most sources a detection threshold near $S_{\rm 100~\mu m}=100$ Jy. It appears that $S_{\rm 100~\mu m}$ and H2O peak fluxes are roughly proportional, as was already suggested by HWB on the basis of a smaller number of detected sources. Such a result is reminiscent of the $L_{\rm FIR}$- $L_{\rm H_2O}$ correlation found by Jaffe et al. (1981) for galactic star forming regions and is readily explained if most of the detected sources in our FIR sample are associated with sites of massive star formation. Four of the ten detected sources are indeed related to star formation, two to AGN, while the nature of the remaining four is uncertain. While the scatter is large, nevertheless even the AGN related megamaser galaxies roughly follow the correlation found for galactic sources, i.e. $L_{\rm FIR}$/ $L_{\rm H_2O} \sim10$9 (see Fig. 9). This is difficult to explain and might be caused by a spatially extended cascade of nuclear bars that contains warm dust and that is needed to fuel the very nuclear region (e.g. Shlosman & Heller 2002).

Given the correlation between FIR and H2O maser flux densities, an improvement in sensitivity by one order of magnitude to detect H2O masers would lower the 100 $\mu$m flux threshold from $\sim$100 Jy to $\sim$10 Jy, and would provide a $\sim$25 times richer sample of detectable targets ($\sim$250 sources at $\delta >-30^{\circ}$). This enlarged sample might then also include some of the brighter ULIRGs that are not part of this study because of too large distances and correspondingly low infrared flux densities.


  \begin{figure} \par\includegraphics[angle=-90, width=6.5cm]{2175fig9.ps} \end{figure} Figure 9: IRAS Point Source FIR luminosity versus total H2O luminosity of H2O detected galaxies (cf. Table 4). For NGC 2146, the total logarithmic H2O luminosity is 0.9 in solar units (Tarchi et al. 2002a). Stars denote the ten sources belonging to the FIR selected maser sample. The diagonal line shows the correlation found by Jaffe et al. (1981) for individual galactic star forming regions.
Open with DEXTER

Table 4: Extragalactic H2O masers beyond the magellanic clouds.

4.3 The entire extragalactic maser sample

4.3.1 The distance bias

Table 4 lists the 53 galaxies with a total of 57 groups of H2O masers detected beyond the Magellanic Clouds. The separation between megamasers (AGN environment) and kilomasers (mainly star formation) is not entirely clear. NGC 2782, an $L_{\rm H_2O} \sim12$ $L_{\odot}$ maser, might not possess an AGN (e.g. Braatz et al. 2004), while the masers in NGC 2146 (a total of $\sim$8 $L_{\odot}$; see Table 5) are known to be related to star formation (Tarchi et al. 2002b). A clear separation between jet and accretion disk masers would also be appropriate. The lack of high resolution data toward most sources, however, makes such a classification elusive. We thus group the masers according to their isotropic luminosity, assuming that all sources with $L_{\rm H_2O} \geq 10$ $L_{\odot}$ (the megamasers) are nuclear.

The H2O detections presented in Table 4 were collected from various surveys with different sensitivities and even within a survey, noise levels may differ from source to source. Furthermore, the masers are time variable and it is always more difficult to detect a broad weak feature than a stronger but narrower spectral component. In view of this highly heterogeneous data base, we adopt a characteristic linewidth of the dominant spectral feature of 20 km s-1 and a detection threshold of 50 mJy (this sensitivity is inferior to that in our surveys (Sects. 4.1 and 4.2) and reflects the higher noise levels of most other data, the exception being the Braatz et al. 2004, survey). We should then be able to detect masers with 1, 10, 100, 1000, and 10 000 $L_{\odot}$ out to maximal distances of

\begin{displaymath}D/{\rm Mpc} = \left[(L_{\rm H_2O}/L_{\odot}) /(0.023 \times\ S/{\rm Jy} \times \Delta V/{\rm km~s^{-1}})\right]^{1/2}, \end{displaymath}

i.e. 6.5, 21, 65, 210, and 650 Mpc, respectively. Note that these distance limits only depend on the square root of the adopted observational sensitivity.

We can check how consistent this is with the sample of detected masers listed in Table 4. The total H2O luminosity per galaxy is taken. Table 5 shows the results. The number of detections in the most likely distance bin is given in italics. IC 342 was not included because of the intrinsic weakness of its maser. It is apparent that for the kilomaser galaxies (here defined to show luminosities <10 $L_{\odot}$) either the number of sources in the expected bin is by far the highest or there are additional detections at both lower and higher distances (the latter a consequence of the fact that the sensitivity of the surveys is not uniform). This provides a picture that is approximately consistent with expectations.

For the more luminous megamasers the situation is different. The distance distribution of the 20 masers with 10  $L_{\odot} \leq L_{\rm H_2O} < 100~L_{\odot}$ is still consistent. Four are located at D <21 Mpc, and four at a distance higher than the estimated limiting distance of 65 Mpc. Most of the detections are obtained in their most likely distance bin. Among the 18 masers with 100  $L_{\odot} \leq L_{\rm H_2O} < 1000~L_{\odot}$, however, 13 are closer or near $D \sim65$ Mpc, the inner limit of the most likely distance range, while among the four galaxies with 1000  $L_{\odot} \leq L_{\rm H_2O} < 10~000$ $L_{\odot}$, three are closer than the corresponding $D \sim210$ Mpc limit. The fourth megamaser, 3C 403, surpasses this limit by only a small amount. None of the masers in the last two groups has a distance larger than the estimated maximum distance.

Table 5: Number of detected maser galaxies per luminosity and distance interval (see Table 4)a.

The surveys do not cover the entire sky and are therefore incomplete by an unknown amount. The fact that distances of masers with lower luminosity are consistent with expected values indicates that a bias related to the distance of these sources is negligible. Most of the luminous megamasers ( $L_{\rm H_2O} \geq100$ $L_{\odot}$), however, are observed at distances that are smaller than expected. Is this an effect of different lineshapes or a consequence of the observed sample of sources? Considering the four most luminous targets with $L_{\rm H_2O} \geq 1000$ $L_{\odot}$, only TXS2226-184 has an unusually wide profile, while the others show "normal'' lineshapes. Without going into any detail, we note that the situation is similar for sources with 100  $L_{\odot} \leq L_{\rm H_2O} < 1000$ $L_{\odot}$. We thus conclude that the bias towards "nearby'' sources in the sample of luminous water masers is caused by the selection of galaxies so far observed. The entire megamaser sample is dominated by the surveys of Braatz et al. (1994, 1996, 1997, 2003, 2004) that are mostly confined to recessional velocites $\la$7000 km s-1, i.e. out to $D \sim100$ Mpc (this also holds for the two surveys discussed in Sects. 4.1 and 4.2). Sources with significantly larger distances were rarely observed.

From the number of sources at "near'' distances we may extrapolate to the larger volumes to estimate the percentage of missing detections in this larger volume. This may provide lower limits because detections at the `nearby' distances may be incomplete as well. For the more luminous megamasers with 100  $L_{\odot} \leq L_{\rm H_2O} < 1000$ $L_{\odot}$, four sources are observed within D=21 Mpc, so that $\sim$ $4000\pm2000$ detectable targets may be expected within D=210 Mpc. This has to be compared with 16 known such objects. With 11 known sources within $\sim$65 Mpc, we still expect $\sim$$350\pm105$ objects within D=210 Mpc, a factor of $\sim$20 above the detected number. Among the most luminous four sources, those with $L_{\rm H_2O} \ga 1000$ $L_{\odot}$, three are detected inside of 210 Mpc, so we would expect $100\pm60$ detectable targets out to D=650 Mpc, of which so far only four have been identified.

While the large errors in the predicted numbers of detectable sources may raise scepticism, an analysis of the distances of the galaxies belonging to the two most luminous H2O luminosity bins (100  $L_{\odot} \leq L_{\rm H_2O} < 10~000~L_{\odot}$) yields a clear result. 14 of the 22 sources in these bins are not in the expected most distant shell (between D=65 and 210 or between D=210 and 650 Mpc, respectively) but are located more nearby. Two additional sources are located at the inner boundary of the most likely shell, while no source is detected beyond the estimated distance limit. This implies that the majority of sources, 73%, is located within a volume that encompasses only 3.2% of the volume in which the masers would be detectable. Assuming an isotropic spatial distribution and applying the Bernoulli theorem, a deviation of 8% from 73% corresponds to 1$\sigma$. The discrepancy between the expected (3.2% of the detections in the inner, 96.8% in the outer shell) and observed (73% in the inner, 27% in the outer shell) spatial distributions is therefore significant. We conclude that statistical evidence strongly indicates that only a tiny fraction of the luminous megamaser sources detectable with presently available instrumentation has been discovered to date.

So far we have not yet considered that the maser luminosities are not necessarily at the upper edge of their respective bin. This has the consequence that not all of them should be detectable out to the upper limit of the corresponding most likely distance interval. To quantify this we have to determine the H2O luminosity function.

4.3.2 The H2O maser luminosity function

The luminosity function $\Phi(L_{\rm H_2O})$ is the number density of objects with luminosity $L_{\rm H_2O}$ per logarithmic interval in $L_{\rm H_2O}$. An unbiased direct measurement of $\Phi(L_{\rm H_2O})$ would require that all objects with a given luminosity be detected within the survey volume, which is not possible in flux limited surveys like those presented in Sects. 4.1 and 4.2. Instead, each object in a survey has an effective volume in which it could have been detected and the sum of detections weighted by their available volumes $V_{\rm i}$ determines the luminosity function.

Not accounting for the incompleteness of the detected H2O megamaser sample and ignoring the possibility that in different luminosity bins the fraction of detected sources may be different, we can derive a zeroth order approximation to the luminosity function (LF) of extragalactic H2O maser sources. Such a computation is not only limited by the effects mentioned above but there are additional factors, the main three being:

(a)
non-uniform sky coverage resulting from several large and many small surveys. Most of the large surveys have used optical-magnitude-limited samples of galaxies in the northern sky. We have approximated the sky coverage to be the entire northern sky;
(b)
different sensitivity limits in the various surveys. We have considered two typical detection thresholds: 1 Jy km s-1 (from e.g. a detection limit of a 20 km s-1 line with peak flux 50 mJy) and 0.2 Jy km s-1 (i.e. five times weaker);
(c)
the use of optical-magnitude limited galaxy samples for H2O maser surveys. Given this, the H2O maser LF is best calculated via the optical LF and the bivariate H2O and optical LF (e.g. Meurs & Wilson 1984). Given the diversity of the selection criteria in different surveys, however, we are forced to directly calculate the H2O LF.
We use the standard $V/V_{\rm max}$ method (Schmidt 1968) to estimate the H2O LF. The sample objects (Table 4) were divided into bins of 0.5 dex over the range $L_{\rm H_2O} = 10 ^{-1}{-}10^{4}$ $L_{\odot}$. For each luminosity bin (Lp; p=1,10) we calculated the differential LF value as follows:

\begin{displaymath}\Phi(L_{p}) = \frac{4\pi}{\Omega} \Sigma_{i=1}^{n(L_{p)}} (1/V_{\rm max})_{i}. \end{displaymath}

Here n(Lp) is the number of galaxies with $L_{p}-0.25 < \log~L_{\rm H_2O}<L_{p}+0.25$. The term after the summation sign represents the inverse of the maximum volume ( $V_{\rm max}$) over which an individual galaxy can be detected given its maser luminosity and the detection limit of the survey. As discussed above in (b), we consider two different detection limits: 0.2 Jy km s-1 and 1 Jy km s-1. In the former case, 6 masers have to be left out of the computation since they are weaker than the assumed detection limit, and in the latter case 19 masers have to be omitted. The maser in IC 342 is outside the considered H2O luminosity range. Results are plotted in Fig. 10 for both detection limits.

Figure 10 demonstrates that the H2O luminosity function does not strongly depend on the detection limit used. From the overall slope of the luminosity function we derive $\Phi\propto (\log~L_{\rm H_2O})^{-1.6}$ which is steeper than the LF for OH megamasers (see Darling & Giovanelli 2002b). Noteworthy is the fast decay in the number of sources at the upper end of the maser luminosity function that could indicate that ultraluminous H2O "gigamasers'' are rare. Obvious is also a low number of sources in the 1-10 $L_{\odot}$ bin. This bin marks the upper end of the luminosity distribution of known star forming regions and is located slightly below the luminosity of the "weak'' megamaser sources. So there might exist a minimum of H2O emitting targets just below the megamaser luminosity threshold. Both results, the minimum at $L_{\rm H_2O} = 1{-}10~L_{\odot}$ and the fast decline at highest maser luminosities, are, however, of questionable significance. The number of sources in the $L_{\rm H_2O} = 0.1{-}10~L_{\odot}$ bins is not yet large enough to make a convincing case. And Fig. 10 does not account for the distance bias discussed for the most luminous sources in Sect. 4.3.1.


  \begin{figure} \par\includegraphics[angle=0,width=8.4cm]{2175fig10.ps} \end{figure} Figure 10: The 22 GHz H2O luminosity function (LF) for maser galaxies beyond the Magellanic Clouds (Table 4). The filled circles show the LF for a detection limit of 0.2 Jy km s-1. The dashed line shows a fit with a slope of -1.3 (for uncertainties in the slope, see Sect. 4.3.2). Error bars of individual points are derived from Poisson statistics following Condon (1989). Not included in the diagram are IC 342 (too low maser luminosity) and UGC 3255, Mrk 3, Mrk 78, NGC 4151, NGC 5256 and NGC 6240 (below the adopted detection limit). The number of masing galaxies in each bin are also shown. To illustrate the effect of changing the sensitivity, we also show the LF for a detection limit of 1 Jy km s-1 (empty circles; in this case 19 of the 53 galaxies fall below the detection limit).
Open with DEXTER

Can we hope to see H2O megamaser emission at cosmological distances? The most luminous megamaser known at present, that in TXS2226-184 with a redshift of z=0.025 (Koekemoer et al. 1995), a peak flux density of 400 mJy, and a linewidth of $\sim$80 km s-1 would be detectable by a 100-m telescope out to $z \sim0.4$. With the Square Kilometer Array (SKA) detections at significant redshifts will thus be possible.

We may also estimate the number of detectable H2O megamasers with the LF shown in Fig. 10. The lower limit to the H2O luminosity of a source at distance D varies as

\begin{displaymath}L_{\rm H_2O} = {\rm C_1} D_{\rm max}^2. \end{displaymath}

The total number of H2O masers of a given luminosity to be detected within this distance can then be expressed by

\begin{displaymath}N_{\rm tot, H_2O} = \int{N_{\rm H_2O}\ {\rm d}V_{\rm obs}} = (4/3) \pi ~N_{\rm H_2O}~D_{\rm max}^3, \end{displaymath}

where $N_{\rm H_2O}$ is the uniform space density of the masers. Assuming that the overall slope of the LF is independent of the maser luminosity, we then obtain with

\begin{displaymath}N_{\rm H_2O} = {\rm C_2} L_{\rm H_2O}^{-1.6} \end{displaymath}

(see above)

\begin{displaymath}N_{\rm tot, H_2O} = (4/3)\pi ~{\rm C_1}^{-1.5}~{\rm C_2}~~L_{\rm H_2O}^{-0.1} \propto\ L_{\rm H_2O}^{-0.1}. \end{displaymath}

This implies that the number of observable masers is almost independent of their intrinsic luminosity: the smaller source density at higher H2O luminosities is compensated by the larger volume in which they can be detected. In principle, this would permit H2O detections with 100-m sized telescopes out to large redshifts, provided that the LF is not steepening at very high maser luminosities and that it is possible to find suitable candidate sources.

In Sect. 4.3.1 evidence was found that only a small fraction of the detectable luminous H2O maser sources is known to date. This may be the main cause for the comparatively steep slope in the LF at highest maser luminosities (see Fig. 10). Ignoring therefore the two bins with highest maser luminosities in Fig. 10, the slope of the LF becomes -1.3 instead of -1.6. This provides a realistic estimate of systematic errors but does not qualitatively change our conclusion.

At the end of Sect. 4.3.1 it was mentioned that the LF is also needed to quantify the deficit of distant high luminosity masers. Adopting the result (see above) that the number of detectable masers is almost independent of the maser luminosity, we can determine with

\begin{displaymath}V/V_{i+1} = [L_{{\rm H_2O},i+1}-L_{{\rm H_2O},i}]^{-1} L_{{\r... ...},i+1}^{-1.5} \int{L_{{\rm H_2O}}^{1.5}~~{\rm d}L_{\rm H_2O}} \end{displaymath}

the fractional volume Vf = V/Vi+1 that masers can occupy in the luminosity range $L_{\rm H_2O,\it i}$ to $L_{{\rm H_2O},i+1}$ relative to the maximum volume Vi+1 defined by the upper luminosity limit $L_{{\rm H_2O},{\it i}+1}$. The indices i and i+1 indicate the lower and upper boundaries of the studied luminosity bin. The integral is calculated between the limits $L_{{\rm H_2O},i}$ and $L_{{\rm H_2O},i+1}$. For order of magnitude bins as considered in Sect. 4.3.1 an occupied average volume of 44% is reached. For the luminous megamasers with $L_{\rm H_2O} \geq100$ $L_{\odot}$ we thus expect 7% of the detections in the inner shells and 93% in the outer envelope that is still far from the observed (73% versus 27%) maser distribution. We thus conclude that the distance bias outlined in Sect. 4.3.1 is real and that the LF slope remains approximately constant within the $L_{\rm H_2O}$ range shown in Fig. 10.

5 Conclusions

This article presents a search for 22 GHz ( $\lambda \sim1.3$ cm) H2O masers towards two classes of objects, i.e. galaxies that (1) either contain nuclear jets that are oriented close to the disk of the galaxy and the plane of the sky or that (2) are bright in the far infrared. The main results are:

(1)
Two new "jet-maser'' sources were detected. One of these, Mrk 1066, shows two components that bracket the systemic velocity of its parent galaxy. Mrk 34 contains the most distant and one of the most luminous megamasers so far observed in a Seyfert galaxy. The source comprises three spectral components that cover a velocity range of $\sim$1000 km s-1.

(2)
Two new masers were also detected in the sample of FIR bright galaxies. One source is a relatively weak ( $L_{\rm H_2O}$ $\sim$ 1 $L_{\odot}$) kilomaser, while the other, Arp 299, is a luminous megamaser in a merging system with high infrared luminosity. There may be two maser components, one associated with the subsystem IC 694 and the other with NGC 3690, following the conventional nomenclature of the source.

(3)
When compared with other H2O surveys, the jet-maser and FIR maser samples show extremely high detection rates and are thus providing a strong motivation for further studies. Including previously detected sources, the jet-maser detection rate is 50% (7/14), while the FIR maser detection rate is 22% (10/45).

(4)
As far as one can judge from single-dish data (this paper and Braatz et al. 2004), a significant fraction of the "jet-maser'' sources appear to be disk-masers with a closer resemblance to NGC 4258 than to classical jet-maser sources like Mrk 348 or NGC 1052.

(5)
The detection rate in the sample of bright FIR sources is a function of the FIR flux density. This implies that more sensitive surveys will detect H2O in galaxies with smaller 100 $\mu$m fluxes. An increase in observational sensitivity by a factor of $\sim$10 should yield a 25-fold increase in the number of detections.

(6)
The correlation between IRAS Point Source and total H2O luminosity of a galaxy follows, with significant scatter, the correlation found for individual star forming regions in the Galaxy. While the agreement is expected for galaxies hosting masers associated with star formation, the agreement for galaxies with AGN dominated masers is less obvious. It may be related to spatially extended dust rich cascades of bars that fuel the central engine.

(7)
60 $\mu$m/100 $\mu$m color temperatures from the IRAS Point Source Catalog are not correlated with H2O maser luminosities (see footnote a) in Table 4).

(8)
There is an observational distance bias: Most of the detectable luminous H2O megamasers ( $L_{\rm H_2O} >100~L_{\odot}$) have not yet been found.

(9)
The extragalactic H2O maser luminosity function (LF) might show a minimum near the transition between the luminosity range of kilomasers (mostly star formation) and megamasers (AGN) in the interval 1-10 $L_{\odot}$. The overall slope is $\sim$-1.6 and implies that the number of observable masers is almost independent of their luminosity. If the LF is not steepening at very high luminosities and if there is a chance to find suitable candidate sources, masers should be detectable with existing telescopes out to cosmological distances.

Acknowledgements
We wish to thank M. Elitzur and L. J. Greenhill for useful discussions during the conception of this project and an anonymous referee for carefully reading the draft and making useful suggestions. AP and AT wish to thank the MPIfR for their hospitality during the observing run. NN was partially supported by the Italian Ministry for University and Research (MURST) under grant Cofin00-02-36 and the Italian Space Agency (ASI) under grant 1/R/27/00. This research has made use of the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet Propulsion Laboratory, Caltech, under contract with NASA. This research has also made use of NASA's Astrophysics Data System Abstract Service.

References

 
Copyright ESO 2005