Astron. Astrophys. 321, 189-201 (1997)

4. Discussion

4.1. Forbidden line emission

Kwan & Tademaru (1988) proposed that the double-peaked profile observed in the [OI] 6300 lines of many TTSs is in fact due to two separate outflow processes, each with distinct velocities. Their model invokes for the fast component an initially radially expanding wind which is collimated by the magnetic field of the circumstellar disk. For the slow component, they propose a wind from the circumstellar disk and/or a warm disk corona. Such a model naturally predicts the double-peaked structure observed, for example, by Edwards et al. (1987). Kwan & Tademaru (1995) have recently developed a model of a rotating disk wind to explain the origin of the LVC emission in cTTSs, caused by a magnetic torque and centrifugal flinging. The mass loss rate for such a disk wind is considerably smaller ( 10%) than that of the jet responsible for the HVC emission.

To what degree is this model applicable to HAEBESs? Before examining this idea we should consider what our observations tell us about the possibility of disks around HAEBESs. As described in the Results section, the [OI] 6300 forbidden line emission falls into one of four categories, examples of which are shown in Fig. 1. Of those stars in emission category I, all are known to drive either stellar jets and/or molecular outflows (see, for example, Mundt & Ray 1994, Fukui et al. 1993). The relative scarcity of HVC forbidden emission lines in HAEBESs may be taken, in terms of the Kwan and Tademaru (1988) model, as indication that the number of HH jets amongst this class is small. This is born out by imaging studies (e.g. Mundt & Ray 1994). Moreover Bastien & Ménard (1990) have observed the polarization patterns, in both the optical and infrared, of a number of HAEBESs and we note that three of the category I stars (V645 Cyg, LkH 233 and PV Cep) and one of the category II stars (R Mon) all show polarization patterns indicative of multiple scattering from a flattened disk-like structure.

In category II we find the largest single group of the observed stars, with 14/28 stars (50%) showing low (-10 kms - -55 kms ) velocity blueshifted lines. There is essentially no corresponding redshifted group; the majority of the remaining [OI] 6300 profiles clustering closely around zero velocity. Only BD 164 shows a redshifted velocity over +10 kms . All the category II stars that have detected [SII] 6716/6731 emission (the 4 category II stars in Table 2 plus ST 202) have other indications of mass loss and extended outflows, apart from ST 202 which has not been included in any published molecular or optical outflow study. In contrast none of the category II stars without [SII] 6716/6731 emission show any indications of jets or outflows, except MWC 1080 which has weak [SII] emission detected by Poetzel et al. (1992). It is therefore interesting to note that in the case of cTTSs, those stars with low veiling continua, and consequently low inferred accretion rates, show unshifted and symmetric profiles of [OI] 6300 and weak or absent [SII] 6716/6731 emission (Edwards et al. 1987; Hartigan et al. 1995) .

Knowing that essentially all the category I stars and many of the category IIa stars are outflow sources and also show [SII] 6716/6731 emission, it is tempting to suggest that the presence of [SII] 6716/6731 indicates a certain degree of outflow activity that the stars without [SII] 6716/6731 emission lack. Whether this is simply a matter of the relatively higher rates of accretion producing relatively higher mass loss rates and consequently brighter forbidden lines (Corcoran & Ray 1997a) or because the origin of the observed [SII] emission (and more generally, all forbidden line emission) differs between the stars with outflows and the other categories (IIb, III and IV) is not clear.

The mechanism that gives rise to the forbidden LVC emission appears much the same in both category I and II, with respect to the line widths and profiles, although as noted the LVC blueshifted velocities are uniformly higher in the category I sources. It may be that the mechanism for accelerating the HVC material entrains LVC material and accelerates it to the higher velocities seen, as described in the models of Königl (1989, 1991) and Safier (1993a, b).

The centroid velocity of the [OI] 6300 emission in our few category III stars is only marginally redshifted (+10 kms 15 kms ). The simplest explanation is that these small redshifts are due to deviations from the zero velocity caused by the method used; e.g. velocity differences between the stellar velocity and the bulk gas velocity responsible for the Na D absorption lines. No molecular or optical outflow is associated with any of the category III stars. The 7 stars in category IV have symmetric and unshifted profiles, centred on the stellar rest velocity. In this regard it is interesting to note that LkH 234 (Ray et al. 1990) and BD 3471 (Goodrich 1992) are both in category IV and have been regarded as possible sources of a jet and a HH object respectively. However, a close-by embedded infrared companion to LkH 234, recently discovered by Weintraub et al. (1994) now appears to be the true source of the jet. Whether an infrared companion can also explain the case of BD 3471 is worth investigating. None of the other category IV stars is associated with a molecular or optical outflow. Moreover, none of the category III or IV stars show [SII] 6716/6731 emission.

It was mentioned in Sect. 3.1 that our results conflict with those of Böhm & Catala (1994), and in particular the clear preference we find for the centroid velocity of the LVC to be blueshifted argues strongly against the spherical wind model of Böhm & Catala (1994). At best this model can only apply to a small subset of the HAEBESs. It must be noted that our sample contains many more Hillenbrand Group II stars (Hillenbrand et al. 1992), which are those HAEBESs that show flat or rising spectra in their spectral energy distributions, than the sample of Böhm & Catala (1994). This may in part influence the distribution of [OI] 6300 emission, as the Group II stars are believed to be less evolved and potentially more active in terms of outflows (Hillenbrand et al. 1992). Certainly all the stars in our sample that show high velocity blueshifted emission (category I) are Hillenbrand Group II stars. In the sample presented here 10 out of 28 stars are the sources of molecular outflows, jets and/or HH objects. Böhm & Catala have only 3 such sources out of 17 stars. Moreover none of the stars observed by Böhm & Catala with the MUSICOS echelle spectrograph show any [SII] 6716/6731 emission. As already mentioned, emission from [SII] 6716/6731 appears to be connected to the conditions that give rise to jets.

Although the centroid velocities of the LVC are blueshifted in most cases, it is nevertheless very interesting that the LVC is very broad with significant blue and red wings, at least for the category I and II stars. If, as we do here, one attributes the LVC to a disk wind along the lines of the model of Kwan & Tademaru (1988, 1995) then the broadness of the component cannot be understood in terms of a poorly collimated wind but instead may be due to the rotation of the disk (see below).

A number of important results stem from the observation of low velocity blueshifted components in the forbidden lines of HAEBESs. Firstly, the systematic blueshift of the [OI] 6300 line in a majority of cases parallels what is seen amongst the cTTSs. Moreover the analogy between the two groups is strengthened further with the discovery that the equivalent width of the [OI] 6300 line scales with the near-infrared colour in the same way as the cTTSs (Corcoran & Ray 1997a). In cTTSs the consistent blueshifts of the forbidden lines are taken as evidence of the presence of optically thick circumstellar disks, which occlude the receding outflow, leaving only the blueshifted material visible to the observer. While in the case of the HAEBESs only 10% of the observed stars show strongly blueshifted lines similar to those observed in classical TTSs, a majority of the observed Herbig stars do show a low velocity blueshift. In Fig. 2 we displayed a histogram of the distribution of centroid velocities for the [OI] 6300 LVCs in our sample. As the plot shows there is a distinct asymmetry to the distribution of the LVC centroid velocities, with no group of HAEBESs with moderately redshifted forbidden emission lines. Only 5 stars show centroid velocities blueshifted by -40 kms or greater, 3 of which are the LVC components of double component (i.e. HVC plus LVC) profiles (V645 Cyg, Z CMa & PV Cep).

Thus, although the presence of disks around HAEBESs in all cases may still be in question, the observations presented here of a blueshifted asymmetry in both the HVCs and LVCs show that disks almost certainly exist around many HAEBESs. These disks could be purely reprocessing or actively accreting or a combination of both processes. The outflow sources have jets, forbidden lines, molecular outflows and other indicators of activity very similar to the cTTSs that also show the best evidence for circumstellar disks.

Turning now to the possibility of an evolutionary sequence amongst the surveyed stars, if, as might be suspected from their degree of embeddedness (Hillenbrand et al. 1992), the Hillenbrand Group II stars represent the relatively less evolved HAEBESs, it may be possible to construct an evolutionary sequence for the various HAEBESs using their degree of outflow activity. We place the category I stars, as the least evolved stars, at the beginning of the sequence. Consisting mainly of Hillenbrand Group II stars they are believed to be comprised of a central star surrounded by a disk plus a dusty envelope (Hillenbrand et al. 1992; Natta et al. 1993a, b). Of the 4 stars in category II with [SII] 6716/6731 emission, 2 stars (R Mon & V376 Cas) also have indications of outflows and/or jets (Brugel et al. 1984; Corcoran et al. 1995). These stars we also place at the beginning of the evolutionary sequence.

Those category II stars that have no detectable levels of [SII] 6716/6731 emission but do have blueshifted [OI] 6300 emission at low velocities are probably representative of the more evolved state where the conditions for high velocity extended outflows have died away, be it from a weakening of a disk magnetic field that might be responsible for accelerating a jet or a reduction in the accretion rate that would have powered the high velocity outflow. The inner regions of the disk may still produce weak [OI] 6300 emission in a disk wind or warm disk corona (see § 4.2) as proposed by Kwan & Tademaru (1988) but any associated [SII] 6716/6731 emission is too weak to be observed. Alternatively the category II stars may represent quiescent jet/outflow sources. The latter idea, however, seems less likely given the lack of known molecular outflows associated with these stars. Specifically, according to Padman & Richer (1994) molecular outflows have characteristic timescales that are much longer than the inferred dynamical timescales. The lifetimes of molecular outflows ( yrs) are such that intermittancy in a stellar jet driving the outflow would not allow the molecular outflow to dissipate in the short time when the jet is in a quiescent state ( yrs, the dynamical time scale for a typically observed jet, Mundt et al. 1990).

As the system evolves the disk may at some point have lost sufficient mass through the action of a stellar wind or other mass loss process to render it optically thin (see Strom et al. 1993 and Wolk & Walter 1996 for possible examples of TTSs with such disks). At such a stage both the redshifted and blueshifted parts of a stellar or disk wind should be visible and perhaps no longer collimated by the action of the disk or its magnetic field. The resulting profile produced by the combination of the two portions of the wind will be unshifted and symmetric, broadened to roughly twice the outflow velocity, if viewed edge-on. The category IV stars show just such emission and may be considered to be the most evolved of the [OI] 6300 stars. Alternatively the category IV stars may have completely dispersed their circumstellar disks, and produce the [OI] 6300 emission in the manner proposed by Böhm & Catala (1994) as a spherically symmetric wind or a sum of spherically symmetric distributed streams forming at the surface of the star which produce shocks at great distances from the star. Given however the apparent power law proportionality between infrared excess and the strength of the forbidden line emission (Corcoran & Ray 1997a) for all the [OI] 6300 emitting HAEBE stars, we favour the idea that one mechanism is responsible for the outflow properties of all categories. Fig. 3 shows a possible evolutionary sequence from the category I to category II & IV stars.

Fig. 3. Three evolutionary phases of the outflow envisaged from the observations. Firstly, top left, the category I stars associated with jets and molecular outflows have a powerful jet responsible of the HVC emission, with characteristic velocities of 300 kms . A rotating disk wind or disk corona gives rise to the LVC emission with typical outflow velocities of 50 kms and line broadening due to the rotation. As the system evolves the high velocity outflow dies off, leaving only the low velocity wind responsible for the LVC emission. The average radial LVC velocity of category IIa and IIb stars is lower than that of the category I stars. Eventually the circumstellar disk may become optically thin at radii where forbidden emission lines form. Narrow and centred forbidden emission lines result.

The proposed evolutionary sequence involves the evolution of the stellar wind. Initially the wind, presumably driven by accretion, has a high velocity and may be collimated and drive an outflow, as is found in the partially embedded (Hillenbrand Group II) stars. Then as the star and circumstellar material evolve the wind velocity decreases, perhaps as the accretion rate drops, and the wind is characterized by the lower blueshifted velocities seen in the category II stars ([OI] 6300 centroid velocities of -50 kms ). In the later stages either a very low wind velocity or an optically thin circumstellar disk could produce the centred forbidden lines seen in the category IV stars presented here.

Finally we mention the suggestion (see Königl 1996) that the absence of a HVC in the [OI] emission of many HAEBESs may not be an evolutionary effect. Instead it is proposed that the absence of an [OI] HVC is caused by the lack of neutral oxygen, due to the photoionization near the outflow axis of the star. Königl (1996) proposes that in contrast the disk wind, responsible for the LVC emission, may not have this problem as it may be shielded, at least close to the disk, from the central radiation field. Although a plausible explanation, at least for the most luminous sources like MWC 297, if it were to be the case in general then one would expect most of our category II sources to be associated with extended outflows (seen, for example, in CO) and this is not the case. That is to say, there genuinely appears to be a lack of a HVC in many of these stars.

4.2. The nature and origin of the HVC and LVC

It is generally accepted that the observation of double-peaked forbidden line profiles in many cTTSs and certain HAEBESs provide evidence of two distinct velocity gas flows from such stars. The high velocity component is most readily explained as a jet forming deep in the gravitational well of the star, either at the stellar surface or within a few stellar radii and emerging from the system along the star's rotational axis. Certainly associating the HVC emission with the presence of a stellar jet appears warranted. In many observations where both components (HVC & LVC) are observed the radial velocities of the HVC match those observed for associated stellar jets and HH objects (e.g. HL Tau, Mundt et al. 1990; Z CMa, Poetzel et al. 1989; DG Tau, Mundt et al. 1987 and Solf & Böhm 1993). Where the HVC is extended (CW Tau, DO Tau, Hirth et al. 1994b; PV Cep, LkH 233, Corcoran & Ray 1997b) there is clear evidence for the association of the HVC with the larger scale optical jet. A number of competing models have been proposed to explain the formation of high velocity jets and these have been addressed comprehensively in reviews by Königl & Ruden (1993) and Ray & Mundt (1993). In synopsis current theories favour either a stellar wind or disk wind as the source of the jet material and collimation via a magnetic field, either stellar or disk-generated in origin. Optical jets are observed to be well collimated on scales smaller than 150-200 AU (Ray et al. 1996). Fendt et al. (1995) have examined the relativistic force-free equilibrium equation for the collimation of YSO winds by a relativistically rotating magnetosphere, originating in the star and reproduce the observed jet diameters well.

While the various models proposed have had success in modeling the jet component's emission, all the models that assume a single velocity law and single outflow from the young star naturally have difficulty in modeling the LVC emission. The LVC, however, is not readily explainable in term of a simple jet model. The low outflow velocities ( 10 kms in cTTSs, 80 kms in HAEBESs) contrast with the broad width of the lines. Line widths of 100 kms for the LVCs of cTTSs and 150-200 kms for the LVCs of HAEBESs make a simple poorly collimated wind unlikely as edge-on viewing angles are always necessary to explain the large line widths but small centroid velocities. The simplest assumption to explain the broad line widths is that of rotational broadening. This in turn suggests a disk origin for the LVC, at least in the case of the cTTSs, as these low mass stars do not rotate at such high velocities. The arguments supporting such an origin are threefold. Firstly there is the combination of the low outflow velocity and the broad line widths as already mentioned. Secondly the observed luminosity of the forbidden lines, for example [SII] 6716/6731, indicate an emitting region of several tens of AU in diameter (Edwards et al. 1987). Thirdly Hirth et al. (1994a) & Hartigan et al. (1995) have shown that there is a progression in the blueshifted velocities of LVC forbidden lines of young stars showing HVC+LVC emission. The species with the highest critical densities (e.g. [OI] 5577) show the smallest blueshifted velocities, those with intermediate critical densities (such as [OI] 6300) show higher blueshifted velocities and the species with the lowest critical density (e.g. [SII] 6716/6731) show the highest blueshifted velocities. Similar behaviour is observed for all the stars listed in Table 2 except Z CMa and KK Oph (compare the [OI] 6300 radial velocities in Table 1 and radial velocities of the [SII] lines in Table 2). The progression of increasing velocity with decreasing density is consistent with an accelerating wind from a disk (Solf & Böhm 1993; Böhm & Solf 1994; Hirth et al. 1994a; Hartigan et al. 1995). Kwan & Tademaru (1995) have recently extended their qualitative model to a simulation of a rotating disk wind to explain the origin of the LVC emission in cTTSs. The disk wind is driven by a magnetic torque and centrifugal flinging. Essentially the LVC is driven in a manner similar to the jet but, due to the lower initial z-velocity of the wind and its location further from the star, the overall acceleration is much less than that experienced by the HVC.

Having adopted an hypothesis of a disk wind as the origin of the LVC emission we can directly relate the width of the LVC line to the inner radius of the disk at which forbidden line emission is observed. Assuming the disk is in Keplerian rotation around the star then kms (Kwan & Tademaru 1995) gives the Keplerian velocity at radius r, assuming an eccentricity of 0. Taking the case of LkH 233, where the line width of the LVC is about 150 kms , the inner radius of the emission would be 0.1 AU or about 30 stellar radii (M = 2.6 and R = 2.6 , Hillenbrand et al. 1992). Assuming a velocity of 10 kms for the slowest material the outer edge of the forbidden line emission is at about 24 AU.

© European Southern Observatory (ESO) 1997

Online publication: June 30, 1998
helpdesk.link@springer.de