VLBA OBSERVATIONS OF H i IN THE ARCHETYPE COMPACT SYMMETRIC OBJECT B2352+495

According to Curran et al. (2008), H i absorption associated with sources at redshifts greater than ∼0.1 has been detected toward 37 sources, 13 of them classified as CSOs, GPS, or high-frequency-peaked (HFP) sources. An overview of H i absorption in compact radio sources was published by Vermeulen et al. (2003b). They reported an H i detection rate of 33% based on Westerbork Synthesis Radio Telescope (WSRT) observations of 59 sources (i.e., 19 detections). Of the 59 sources, 11 were classified as CSO by Augusto et al. (2006) and Polatidis & Conway (2003). From this subsample of 11 CSOs, 6 (54%) show H i absorption. Other studies show similar results, e.g., Peck & Taylor (2001) reported a CSO H i detection rate of ∼50% and Gupta et al. (2006) reported a detection rate of ∼45% in GPS sources (see also Vermeulen 2002).

The optical depth range of the 6 CSOs with H i absorption from Vermeulen et al. (2003b) is 0.2%–1.7%, with line widths between 13 km s⁻¹ and 297 km s⁻¹ and column densities between 2.8 × 10¹⁹ cm^-2 and 2.6 × 10²⁰ cm⁻². To estimate H i column densities, Vermeulen et al. (2003b) assumed that the atomic gas was uniformly covering the radio continuum (covering a factor of 1) and a spin temperature of 100 K that may underestimate the spin temperature by more than an order of magnitude if the gas were located at parsec scales from the active galactic nucleus (AGN). Thus, the column densities reported by Vermeulen et al. (2003b) are lower limits.

Including the results reported in this work, H i imaging at high angular resolution has been reported toward five CSOs (Table 1). We have also listed in Table 1 all non-CSO AGNs and starbursts that have been imaged at high angular resolution in H i, i.e., in interferometric observations using baselines greater than 100 km. In the rest of this section, we review the most salient characteristics of the CSOs that have been studied in H i at high angular resolution:

• 1.  
  B1946+708. High angular resolution H i observations were conducted by Peck et al. (1999; see also Peck & Taylor 2001). They found extended and multi-peaked absorption throughout the radio continuum source. The peak optical depths range between 3% and 7%. The maximum column density (N_H i ∼ 3 × 10²³ cm⁻²; Peck & Taylor 2001) was found toward the AGN core, which also shows the greatest velocity dispersion (FWHM ∼ 350 km s⁻¹). The extent of the H i absorption in B1946+708 implies that the atomic gas is tracing a thick torus (thickness >80 pc).
• 2.  
  4C 31.04. The host galaxy is a bright elliptical at z = 0.0592. It was classified as a CSO based on very long baseline interferometer (VLBI) continuum observations that revealed a double radio lobe with a flat-spectrum radio core (Cotton et al. 1995; Giovannini et al. 2001; Giroletti et al. 2003; see also Augusto et al. 1998). Detection of H i in 4C 31.04 was reported by Mirabel (1990) based on Arecibo observations. Mirabel (1990) found two absorption features, a broad (FWHM = 133 km s⁻¹) component and a double-peaked narrow component (FWHM = 6 and 16 km s⁻¹) redshifted approximately 400 km s⁻¹ from the optical velocity. Subsequent VLBI observations by Conway (1999) showed that the broad line covers one of the radio lobes and just partially the second, indicating a sharp edge of the atomic gas torus. The narrow component is detected toward both radio lobes. The maximum optical depth of the lines is ∼5%–7%. Mirabel (1990) interpreted the narrow absorption as being caused by clouds similar to high-velocity H i clouds in our Galaxy.
• 3.  
  4C 37.11. In contrast to the two previous CSOs, H i in 4C 37.11 has been detected only toward one of the radio lobes. The absorption is characterized by two broad (FWHM > 100  km s⁻¹) H i features from two separated (in line of sight and velocity) clouds. The peak optical depth of the H i lines is ∼2%. Rodríguez et al. (2009) concluded that the H i absorption likely originates in a thick circumnuclear torus.
• 4.  
  PKS 1413+135. As in 4C 37.11, H i absorption is detected only toward one radio lobe (Perlman et al. 2002). The H i optical depth is τ ≈ 1 and the line is quite narrow, between 16 km s⁻¹ and 18 km s⁻¹. Perlman et al. (2002) concluded that the H i absorption is likely from a giant molecular cloud (GMC) in the outer disk of the host galaxy. Although PKS 1413 + 135 was classified by Perlman et al. (1996) and Gugliucci et al. (2005) as a CSO, it has several characteristics that differ form a typical CSO. In particular, it was originally classified as a BL Lac object based on its optical spectrum (Bregman et al. 1981), the nucleus dominates by more than 50% of the total cm flux density, the radio emission is variable, and the host appears to be a spiral galaxy (see Perlman et al. 1996).

Finally, two additional sources (B1934−638 and 4C 12.50) are worth discussing in detail. B1934−638 is a compact-double GPS source that has been classified as a CSO (Augusto et al. 2006); however, no flat-spectrum radio core has been detected (Tzioumis et al. 2002). Even though no H i imaging has been reported at high angular resolution, the H i absorption is against the compact radio continuum as shown by detection of absorption with three baselines of the LBA (Véron-Cetty et al. 2000). The absorption is narrow (FWHM = 18 km s⁻¹) and redshifted 260 km s⁻¹ from the optical systemic velocity; thus, the H i is probably not located in the nuclear region of the galaxy. We include B1934−638 in Table 1 although no VLBI H i imaging is available and the radio core has not been detected.

In the case of 4C 12.50, high angular resolution H i absorption observations were reported by Morganti et al. (2004). They found H i absorption against a weak counterjet and concluded that the H i likely traces a jet/cloud interaction and not a circumnuclear torus or disk. This source was classified as a CSO by Lister et al. (2003) despite several atypical characteristics. For instance, while the 1.3  GHz continuum image reveals an S-shaped distribution with a total extent of ∼300 pc (Morganti et al. 2004), the 15 GHz Very Long Baseline Array (VLBA) observations are consistent with a core–jet morphology plus a weak counterjet (i.e., inconsistent with the typical symmetrical morphology of CSOs). 4C 12.50 has no clear hot spots, it is classified as a compact steep spectrum (CSS), the radio jet exhibits superluminal motions, some isolated features have unusually high (∼60%) linear polarization in comparison to other CSOs, and more importantly, there appears to be continuity between the ∼300 pc compact radio jets and large-scale (>10 kpc) radio lobes (Stanghellini et al. 2005). Therefore, we do not consider 4C 12.50 a bona fide member of the CSO class (see Table 1).

B2352+495 has been studied in detail by a number of authors (e.g., Readhead et al. 1996; Taylor et al. 1996). In this section, we summarize the characteristics of this object.

Radio continuum. The most salient property of B2352+495 is its highly symmetric morphology at radio wavelengths. It exhibits an S-shaped morphology with a total extent of ∼120 pc (Readhead et al. 1996). High angular resolution radio continuum observations from 610 MHz to 15 GHz reveal symmetric hot spots and lobes; remarkably similar to the morphology of FR II objects (but at much smaller scale). The dynamical age of the system is ∼1200 years based on proper motion studies of the hot spots in the radio lobes (Taylor et al. 2000). Taylor et al. (1996) detected the radio core of the system, i.e., the putative location of the central engine. The core is approximately located symmetrically between the radio lobes, and drives a collimated jet that is brighter toward the northern side of the core. The radio jet is detected up to the brightest radio emission in the system (B1 and B2; Taylor et al. 1996; see also Section 3) which is also located between the two radio lobes. The hot spots in the radio lobes show no superluminal motion (Conway et al. 1992). Multi-frequency Very Large Array (VLA) observations give fractional polarization limits of <1% for this source (Rudnick & Jones 1983). The radio spectral energy distribution of B2352+495 peaks at ∼1 GHz, thus by definition, it is a GPS object (Conway et al. 1992). The total flux density of B2352+495 shows only modest (≲20%) variability on time scales of months; no systematic long term variability was found over a period of ∼15 years (Waltman et al. 1991; Conway et al. 1992). The radio luminosity is ∼10²⁶ h⁻² W Hz⁻¹ (ν_o=5 GHz at the emitted frame), i.e., some 20 times brighter than the luminosity boundary between FR Is and FR IIs (Readhead et al. 1996).

Optical/infrared counterpart. B2352+495 was not detected by IRAS to sensitivity limits well below the corresponding luminosity of Arp 220 (at the redshifted frequency), thus B2352+495 is significantly less luminous than ultraluminous infrared galaxies (Readhead et al. 1996). Snellen et al. (2003) conducted imaging and spectroscopic optical observations of B2352+495. They found that the radio source resides in an optical host in the fundamental plane of elliptical galaxies, i.e., the host is a normal elliptical.  Snellen et al. (2003) reported a redshift of 0.23790 ± 0.00016, a stellar velocity dispersion of 201 ± 17 km s⁻¹, and an effective radius of 16 ± 03 (5.8 ± 1.1 kpc).⁷ Based on the stellar velocity dispersion, the mass of the supermassive black hole at the center of B2352+495 is of the order of 10⁸ M_☉. The absolute magnitude of the galaxy is M_V = −19.8 with an apparent magnitude of m_V = 20.1 (Readhead et al. 1996). The optical luminosity of B2352+495 is ∼0.35 L*, which is smaller than the typical optical luminosity of classical doubles (FR IIs, L ∼ L*; Owen & White 1991) and cD galaxies (L ∼ 5–10 L*; Schechter 1976). The absence of Balmer lines and strong blue continuum in the spectrum demonstrates no starburst activity. In contrast, a number of emission lines are detected typical of active elliptical galaxies (Readhead et al. 1996). The AGN contribution to the total optical continuum is ≲10% (Snellen et al. 2003).

X-ray properties. High-sensitivity XMM-Newton observations of B2352+495 were conducted by Vink et al. (2006). They detected an X-ray source with a luminosity of 4.6 × 10⁴² erg s⁻¹ (2–10 keV band, ignoring absorption). Based on the X-ray detection, they estimated an intrinsic absorption column density of (0.66 ± 0.27)×10²² cm⁻², which is similar to the column densities found toward extended radio galaxies.

Prior H i observations. Absorption of the H i transition in B2352+495 was first detected by Vermeulen et al. (2003b) based on WSRT observations. The H i absorption in B2352+495 consists of a broad line (FWHM = 82 km s⁻¹) and a redshifted narrow component (FWHM = 13 km s⁻¹). The peak flux density of the two absorption components is similar (∼40 mJy). Preliminary results of the data reported in this paper were discussed by Vermeulen (2002).

It is difficult to reliably investigate the velocity field of the broad H i line given that the absorption is only detected toward the central continuum source and, as shown in Figure 2, the velocity field is complex. Nevertheless, a velocity gradient may be present: the two channels with stronger absorption at velocities above 10  km s⁻¹ from systemic appear to have stronger absorption toward the west, whereas most of the blueshifted absorption is stronger toward the east (velocity channels <−10  km s⁻¹ from systemic, see Figure 2). Figure 4 shows the position–velocity diagram of the broad H i absorption line in the east–west direction. As also seen in the individual channel maps, there is a hint of a velocity gradient between blueshifted and redshifted material. To illustrate the possible gradient, we show in Figure 3 (top panel) the opacity channels that correspond to the rest-frame velocities of 17.6  km s⁻¹ and −31.4  km s⁻¹ (Figure 2).

Zoom In Zoom Out Reset image size

The existence of a velocity gradient is tantalizing given that it is approximately centered at the systemic velocity of the galaxy, and the position angle of the gradient is almost perpendicular to the north–south jet, i.e., the H i gas could be tracing some kind of rotating structure around the supermassive black hole. We explore this hypothesis in the rest of this section; however, we stress that more sensitive global-VLBI observations are required to confirm the velocity gradient.

To investigate whether rotation around the supermassive back hole is physically plausible, we followed the formulation presented by Rodríguez et al. (2009; see their Appendix). To model the H i rotating structure, we used the position and velocity of the peak optical depth of the 17.6  km s⁻¹ and −31.4  km s⁻¹ channels (Figure 2). The position of the radio core (Figure 3 (upper panel); see also Taylor et al. 1996) was assumed to be coincident with the center of mass of the system. For a black hole mass of 10⁸ M_☉ (Snellen et al. 2003) and assuming circular trajectories coplanar with the supermassive black hole, we obtain that the orbit of the H i clouds from the 17.6  km s⁻¹ and −31.4 km s⁻¹ channels would have a radius of ∼20 pc with an orbital inclination of ∼65° with respect to the plane of the sky. Such an orbit is reasonable given the expected distribution of atomic clouds around AGNs (e.g., Peck 2005). Similar results are obtained in the case that the orbital plane of the H i clouds is not coplanar with the supermassive black hole but located above by distances of the order of the projected separation between the AGN core and the H i absorption. This estimate is clearly a zero-order approximation because non-radial inhomogeneities in the gravitational field due to stars and gas are neglected. In addition, the radius of influence of the supermassive black hole (r_BH = GM_BH/σ²_*, where G is the gravitational constant, M_BH is the mass of the black hole, and σ_* is the stellar velocity dispersion; e.g., Neumayer et al. 2007) is ∼10 pc, thus, the stellar mass enclosed by the atomic gas is likely comparable to the black hole mass.

In Figure 3 (bottom panel), we show two possibilities for the nature of the H i distribution, i.e., a thick torus (characterized by an H i scale height similar or greater than the length of the radio source) or a thin (most likely warped, e.g., Pringle 1996) disk. The obvious method to discriminate between these possibilities is to check whether there is H i toward the southern radio lobe. However, as mentioned above, our optical depth sensitivity limit does not set significant constraints on the H i distribution, i.e., clouds like those detected toward the brightest radio continuum emission could be present throughout the region.

An indirect approach to discriminate between the models is to compare our observations with lower angular resolution data. As mentioned in Section 2, the peak flux density measured with the WSRT (Vermeulen et al. 2003b) is approximately the same peak intensity we measured with the VLBA (Figure 1) and the line widths are also consistent. Thus, in contrast to other systems where VLBI observations resolve/filter out significant H i signal (e.g., IC 5063; Oosterloo et al. 2000), the H i in B2352+495 is compact and mostly distributed along the central continuum source, i.e., consistent with the thin-disk interpretation.

The thin-disk interpretation is also circumstantially supported by X-ray observations. Vink et al. (2006) reported an X-ray absorption column density of (6.6 ± 2.7) × 10²¹ cm⁻², which is similar to the column density of the narrow H i component, but several times smaller than the value of the broad line. If the H i absorption originates from an inclined (and warped) thin disk that does not substantially obscure the AGN (Figure 3, bottom panel; see also Figure 5 of Orienti et al. 2006) then as observed, the column density toward the AGN derived from X-ray observations is expected to be smaller than that of the neutral disk.

We note that the thin-disk interpretation does not necessarily imply a misalignment between the jet and the disk, i.e., the position angle of the jet changes as a function of distance from the core (Figure 3) and the disk could be warped; thus, the jet and disk could be perpendicular at (sub)parsec scales. In addition, given that the detection rate of H i absorption in CSOs is ∼50% (see Section 1.1), it is certainly possible that in some cases the atomic gas is distributed in a circumnuclear structure that is seen in projection only toward one radio lobe/jet.

Assuming the orbital parameters from the thin disk model, we now explore the stability of the H i disk with respect to gravitational fragmentation. We used the Toomre stability formulation as presented by Gallimore et al. (1999), i.e.,

$$\begin{equation} \frac{\Sigma _{\mathsc{H\,i}}}{\Sigma _c} \simeq \frac{0.028 N_{\mathsc{H\,{\sc i}}} v_{\rm rot}}{r_{{\rm pc}}}, \end{equation} \tag{ 1 }$$

where Σ_H i is the atomic hydrogen surface density, Σ_c is the critical surface density above which the disk is unstable to fragmentation, N_H i is the H i column density in units of 10²¹ cm⁻², v_rot is the rotation velocity in  km s⁻¹, and r_pc is the orbital radius in parsecs. In the case of B2352+495, we obtain Σ_H i/Σ_c ∼ 10; thus, the disk is expected to be unstable to fragmentation. We note that reducing the assumed spin temperature by an order of magnitude would not significantly modify this conclusion, i.e., Σ_H i/Σ_c would still be ∼1. This simple stability test indicates that the H i cannot be distributed in a homogeneous disk. Indeed, the opacity variations from channel to channel (Figure 2) suggest that the H i gas is not homogeneously distributed. We therefore concur with the standard schematic of clumpy atomic gas around AGNs (e.g., Peck et al. 1999).

In this section, we discuss the properties of H i absorption in B2352+495 with respect to other H i extragalactic absorbers that have been studied at very high angular resolution (i.e., interferometric observations with baselines >100 km; Table 1). As is clear from Table 1 and the discussion in Section 1.1, high angular resolution observations of H i in CSOs reveal that the absorption does not trace a specific kind of cloud; instead, the absorbing clouds appear to be in quite different locations throughout the host galaxy, from infalling HVCs and GMCs at kiloparsec scales, to torus and disks closer (≲100 pc) to the nucleus.

Only two of the five CSOs imaged at high angular resolution exhibit H i absorption toward both radio lobes (Table 1; B1934−638 has not be imaged at high angular resolution, see Section 1.1). In general (four out of six, Table 1), CSOs studied at high angular resolution are characterized by a broad (ΔV≳ 100  km s⁻¹) H i absorption component, and tend to be multi-peaked, with narrow (ΔV< 100  km s⁻¹) lines detected in five of six cases.

High angular resolution observations of H i toward non-CSO radio galaxies (Table 1) show a similar trend as the CSO counterparts, i.e., the H i absorption appears to trace a variety of environments, from disks directly associated with the AGN to clouds at kiloparsec-scale distances from the active nucleus. An apparent difference is the occurrence of H i clouds interacting with radio jets. In the case of several non-CSO with double jet/lobe radio emission, the H i appears to trace regions of interaction between atomic gas and radio jets, whereas such an interpretation has not been preferred in any of the CSOs imaged in H i at high angular resolution. This trend does not appear to be due to bias of the researchers but, in most cases, seems due to real differences in the samples. For instance, the H i absorption associated with tori/disks in CSOs is found very close (≲100 pc) to the radio core, close to the systemic velocity, and/or has evidence of rotation. Conversely, the H i in non-CSO double jet/lobe sources with possible jet/cloud iteration is found at larger distances from the core (≳500 pc) shows no evidence of rotation, and/or has significantly different velocity from systemic. Although a more robust statistical sample is needed to achieve strong conclusions, the apparent absence of H i interaction with the radio jets supports the current view that CSOs are young radio galaxies instead of being confined by dense circumnuclear material as proposed in the frustration scenario.

Arguments for the youth scenario are mostly based on the overall radio continuum properties of the sources, power budget estimates of the jets, expansion velocities, and properties of the nuclear molecular/atomic gas. CSOs lack strong extended radio emission related to energy deposited by the jets over long periods, and VLBI studies reveal proper motions that imply short dynamic ages (of the order of 10³ yr, e.g., Gugliucci et al. 2005). In addition, if the advance of the radio lobes would be truly frustrated, then for them to be in ram pressure equilibrium the density of the surrounding medium should be above 10² cm⁻³. However, if the high-density material is confined to a disk/torus then the frustration scenario is not viable. B2352+495 has evidence of being a young source based on all these arguments: it has no detected extended radio emission, the expansion velocity implies a dynamic age of ∼1200 years (Taylor et al. 2000), and, as discussed above, the high-density atomic gas appears to be distributed along a disk or torus that does not interact with the radio jet. Hence, the evidence favors the youth interpretation in this case.