4.1. Structure of the Chamaeleon clouds
The IRAS 100µm map of the Chamaeleon region shows several filamentary clouds which extend over an area of more than 100 square degrees. It is an open question whether the individual structures termed Cha I , Cha II and Cha III are really physically related to each other. There is another cloud, DC 300.2-16.9 (Hartley et al. 1986), located between Cha I and Cha II roughly at the position of T Cha.
The Hipparcos parallax of T Cha implies a relatively small distance of 66 pc. Note however that the large parallax error for this star puts an upper limit of 85 pc while the lower limit is 54 pc. The latter would place T Cha practically on the main sequence, which is absolutely inconsistent with the pronounced PMS characteristics of this star (Alcalá et al. 1993). Even the mean distance of 66 pc would give an extremely old age of some 40 Myr for this star. Since T Cha is definitely a T Tauri star, we think that the upper limit of 85 pc should be closer to the true distance of T Cha.
On the other hand, an upper limit of 180 pc has been established for the distance of the cloud DC 300.2-16.9 (Boulanger et al. 1998), to which T Cha seems to be associated. Thus, it may well be that this cloud is also located closer than the Cha I cloud, maybe also at about 90 pc from the Sun. The Hipparcos parallaxes of the other stars in subgroup 2 as well as our analysis of the proper motions in Sect. 3.1.1 support this scenario of stars and even some cloud material at distances of about 90 pc.
The question now is how can a SFR be such large in volume? We discuss the models for the formation scenario of the Chamaeleon cloud complex which possibly could explain the existence of PMS stars far off the observed molecular clouds in the next section.
Alternatively, we may note that it is also possible that the observed cloud material belongs to distinct structures as considered by Whittet et al. (1997). In this case the stars in subgroups 1 & 2 would have the same space velocities although they are not associated with the same cloud material. However, it is not unusual that young stars exhibit rather low velocities relative to the field stars in the same region (cf. the Taurus SFR or the Scorpius-Centaurus OB association).
Moreover, placing the stars of subgroup 2 at a mean distance of 90 pc rises their mean ages by about a factor of 6 to 18 Myr as compared to a mean age of 3 Myr for a mean distance of 170 pc. This could easily be explained if they belong to another structure than the stars of subgroup 1.
There are too few stars in the Chamaeleon region with distance information available to decide whether a population of PMS stars with distances intermediate between the two subgroups at around 130 pc exists. In principle, the majority of dispersed T Tauri stars detected with the flux-limited RASS are expected to be located between 90 pc and 150 pc, and the optical, IR and deep X-ray pointed observations have been sufficiently sensitive to detect PMS stars at more than 150 pc. However, most of these stars were too faint to be included in the Hipparcos Input Catalogue.
4.2. Implications for the formation scenario of the Chamaeleon cloud complex
The discovery of large populations of WTTS distributed over regions of 10-20 degrees or even more in extent centered around active cloud cores has raised questions about the scenario of their formation. Several scenarios have been suggested to account for the existence of very young stars far away from the known sites of star formation.
Sterzik & Durisen (1995) proposed that the WTTS halo observed around star formation regions might be due to high velocity ( 3 km s-1 ) escapers (run-away TTS or RATTS) produced by dynamical interactions in small stellar systems. This would imply that the velocity vectors of the stars point away from the dense molecular cloud cores from where they were ejected.
From our proper motion study there is no indication for such an overall correlation between positions and proper motions. Only for some 2 or 3 stars in our sample the ejection scenario may be invoked, namely Sz 41, CHXR 56, and perhaps RXJ 0837.0-7856, if their proper motions are not spurious due to binarity. On the contrary, if we assume that subgroup 2 is at the same distance as subgroup 1 (ignoring the Hipparcos parallaxes for the moment), the stars of subgroup 2 would move with higher space velocities in the direction of lower right ascension than the stars of subgroup 1 while being located at higher right ascension (cf. Fig. 1). This means that they would approach the Cha I molecular cloud. Given the direction of motion, we cannot exclude that some stars may have been ejected from the Cha II cloud. However, the fact that these stars seem to form a co-moving group is inconsistent with the prediction of any ejection model, in which the motion would be completely random, so that we exclude the ejection scenario as the dominant process for producing the dispersed population of WTTS in Chamaeleon.
Lépine & Duvert (1994) tried to explain the displacement with respect to the galactic plane of several near-by star forming regions including Chamaeleon by infall of high velocity clouds (HVC) on the galactic plane. There is no detailed prediction for the kinematics of the stars in the HVC impact scenario, except for the fact that, subsequent to the impact, clouds and stars will oscillate around the galactic plane and tend to separate from each other (combing-out). Given the fact that at least the stars in subgroups 1 & 2 display practically no net motion perpendicular to the galactic plane after correcting their proper motions for the solar reflex motion, one might speculate that they are just reversing their direction of motion. The large distance of the Chamaeleon association from the galactic plane may support this point of view. Nevertheless, these indications are far from being conclusive and depend strongly on the adopted distances.
Feigelson (1996) proposed that low mass stars may form in dispersed cloudlets in a turbulent environment. Also, it has long been suspected that Bok globules are the sites of isolated star formation. Recent studies (Launhardt & Henning 1997, Yun et al. 1997) have shown that such globules can be associated with embedded IR and IRAS point sources in which very young low mass stars are found. It is however not clear if these globules are related to Feigelson's cloudlets.
In order to explain the observed distribution of WTTS Feigelson (1996) considers models with a velocity dispersion of the order of 1 km s-1 (due to internal thermal motions in the gas of the parent cloud), with thermal velocity dispersal in combination with dynamical ejection, and with star formation in small cloudlets distributed over a larger region. Comparing the predictions of his models with the properties of the observed WTTS population, he found that the first two dispersal models encounter serious problems. In particular, the thermal dispersal model can explain the number of WTTS found far from the active clouds, but not their low ages. In order to overcome this difficulty within the framework of the current model one would have to assume an unplausibly high velocity dispersion even at the time of their formation. An improved dispersal model, where a certain fraction of the dispersing stars is made up of high velocity escapers, cannot account for the observations either, unless the ejection rate significantly increased recently. The model can indeed explain the existence of some very young stars far from their sites of origin, but simultaneously it produces a population of older ejected stars, which would lead to an unplausibly high star formation efficiency.
As already pointed out above, the proper motion data of the stars discussed in the previous sections are inconsistent with any dispersal model either, as the velocity vectors are not oriented away from any single point.
In the most promising model investigated by Feigelson star formation takes place in long-lived active cloud cores as well as in a number of small short-lived cloudlets distributed over a rather large region. These cloudlets are believed to possess high velocities relative to their parental giant molecular clouds because of its turbulent structure. After producing some stars with very low internal velocity dispersion the cloudlets disappear, leaving behind streams of T Tauri stars with high relative velocities between each other.
As proper motions are available only for a very small fraction of all the young stars in the Chamaeleon region (probably for less than 10% according to the estimate by Feigelson of several hundred stars yet to be discovered), it is difficult to verify the predictions of the model quantitavely. If we ignore the stars in subgroups 1 & 2 for the moment, which we assume to have formed in the dense cloud cores, we are left with not more than 33 stars which possibly originated in small cloudlets. Feigelson estimates the number of cloudlets to be of the order of 50, so that we do not expect to have more than one or two stars of the same cloudlet in our sample. Although we cannot confirm the model decisively, from the proper motion diagram (Fig. 3) one could select good candidate stars which possibly were formed in cloudlets.
Another limiting factor in our kinematical study is the lack of precise distances for the majority of our stars. For a more detailed comparision with the model of Feigelson one needs to correct the proper motions of the wider distributed TTS and not only of the stars in the two subgroups for galactic rotation and the reflex of the solar motion, which requires knowledge of the individual distances. Similarly, comparisions with the ejection scenario are also hampered by the lack of precise distances, as the relative velocities can change sign when putting the stars at higher or lower distances.
© European Southern Observatory (ESO) 1998
Online publication: September 14, 1998