In this section we elaborate on three specific aspects of these results: the nature of the LIC and G clouds; the nature and distribution of other nearby clouds; and the complicated interstellar and circumstellar spectrum of 51 Oph.
4.1. The LIC and G clouds
As discussed in Sect. 1, one of the unsolved questions concerning the structure of the local interstellar medium is the relationship between the Local Interstellar Cloud itself, and the G cloud identified by Lallement & Bertin (1992). As discussed by Lallement et al. (1995; their Sect. 1.3), the main difficulty is the small projected velocity difference between the LIC and G cloud velocity vectors. As this is always less than 3 km s-1 (Sect. 1), lower-resolution studies have found it difficult to to determine whether both clouds are present towards any given star. Detection of both components would argue strongly for the LIC and G clouds being separate entities, while detection of only one or the other would argue for a velocity gradient within a single cloud.
With this in mind, Table 1 lists the velocities of the LIC and G clouds, projected towards the stars observed here, and Table 2 identifies those components which fall within the uncertainties on the projected cloud velocities. It will be seen that components at the LIC and/or G cloud velocities are detected towards seven of the eight stars (the exception being Oph, which is discussed further in Sect. 4.2), but only in one case (51 Oph) are absorption components found which lie within the velocity ranges expected for both clouds. Since a single exception could be explained away as a coincidence, this lack of dual detections would seem to argue against the LIC and G clouds being separate entities. However, the situation is complicated by the fact that: (1) two stars ( Cen and Cen) only fail to exhibit both components because in each case one lies outside the predicted range by km s-1 (these components are identified by question marks in Table 2); and (2) Crawford & Dunkin (1995) plausibly identified both components in the line of sight to Oph (, , pc).
In spite of these uncertainties, some trends are suggested by the present data. In particular, there is a tendency for the LIC to be observed either on its own or not at all at high negative latitudes (: i.e. towards Aqr, Gru, Gru; cf. Crawford & Dunkin 1995 for the latter star), whereas the stars which may plausibly be interpreted as exhibiting components due to both clouds ( Cen, Cen, 51 Oph, Oph) lie at low positive latitudes ). However, the fact that a component at the G cloud velocity is detected without the LIC towards stars as well-separated in latitude as Cen () and Hyi (), and that neither cloud was detected towards Oph or Eri (Crawford & Dunkin 1995), means that a simple picture in which lines of sight towards stars at low positive latitudes pass through both the LIC and the G clouds is probably untenable (unless both clouds are very inhomogeneous). Only by obtaining many more observations of nearby stars, performed with sufficient spectral resolution to discriminate between the LIC and G cloud velocities, will it be possible to shed further light on the spatial extent of these two clouds, and we hope to obtain such observations in the near future.
Regardless of the spatial distribution of the LIC and G Cloud velocity components, the present observations are able to provide new information on the physical conditions prevailing within this material. For Ca ions, the velocity dispersion is related to the kinetic temperature, , and the line-of-sight rms turbulent velocity, , by
The five components identified with the LIC in Table 2 have well-defined b -values in the range 1.2 to 2.4 km s-1. Allowing for the errors, all but one of these values is consistent with other indications of the physical conditions prevailing in the LIC (cf. Sect. 1). Specifically, Equation 1 gives km s-1 for K and km s-1, and this falls within the errors on the observed b -values for every LIC component except that towards Aqr. We may also note that these values of temperature and turbulence are consistent with the b -values obtained for velocity components identified with the G cloud, underscoring the fact that, if the two clouds are physically separate, they nevertheless have similar internal physical conditions.
The single absorption component observed towards Aqr is exceptional because, while the velocity is essentially identical to that expected for the LIC, its b -value (1.2+-00:6:4km s-1) implies an upper limit to the temperature (3500+-14400900K) which is lower than that obtained for other LIC sightlines. It is true that temperatures at the upper end of this range are consistent with the canonical LIC value of 7000 K, but Equation 1 shows that such a temperature is only possible if km s-1. Thus it seems that the region of the LIC present towards Aqr is cooler and/or less turbulent than elsewhere. Observations of additional stars, close to Aqr on the sky, will be required to determine the physical extent of this cooler/less turbulent region.
4.2. Other nearby clouds
The present observations reveal the presence of 14 velocity components (towards six of the eight stars) in addition to the ten already tentatively identified with the LIC and G clouds (Table 2). Excluding the three towards 51 Oph, which are probably circumstellar in origin (and which are discussed in Sect. 4.3), and one possibly circumstellar component towards Gru (Bertin et al. 1993), this leaves ten components which arise in additional low density clouds within the Local Bubble. As discussed in Sect. 3.2, many of these components have here been resolved for the first time, and, as might be expected, the two most distant stars ( Cen and Cen, both at about 80 pc) exhibit the largest number of components (cf. Table 2).
Most of these components have velocity dispersions which suggest physical conditions similar to those deduced for the LIC and G clouds. Specifically, all but three have b -values that are consistent with K and km s-1 (i.e. km s-1, allowing for the errors). The +3.4 km s-1 component towards Cen ( +-10:9:1 km s-1) is broader than might be expected, but, as noted in Sect. 3.2, this probably results from the presence of an additional unresolved velocity component. Of greater interest are the two components (i.e. those at -7.7 km s-1 towards Cen, and +0.3 km s-1 towards Cen) which have b -values (1.4+-00:2:3and 1.2+-00:5:4km s-1, respectively) which are lower than expected, and similar to that of the anomalously narrow LIC component identified towards Aqr. Indeed, these two clouds must be rather cooler and/or less turbulent than that towards Aqr as the b -value upper limits are lower - implying rigorous temperature upper limits of K for Cen, and K for Cen. Conversely, by assuming , we obtain rigorous upper limits to of 1.1 and 1.2 km s-1 for these components. Since, in reality, there will be some trade-off between and , the actual values of both must lie below these limits.
The line of sight towards Oph deserves special comment, because of what it tells us about the complicated state of the LISM in the direction of Ophiuchus. As noted in Sect. 4.1, Oph is the only star in the present sample which does not exhibit clear evidence for an absorption component at either of the LIC or G cloud velocities. This is all the more remarkable when one considers that Oph is bracketed on the sky by Oph and 51 Oph (with angular separations of and , respectively), and that both of these neighbouring stars do have components at the G and LIC velocities (for Oph, see Crawford & Dunkin 1995). It is true that the spectrum of Oph shown in Fig. 1 does reveal some evidence for very weak ( -2 % deep) absorption in the velocity range occupied by the LIC and G clouds (i.e. -26.7 to -24.2 km s-1 ; cf. Table 1), but, given the signal-to-noise ratio of the data, this is barely significant. Formally, the column density upper limits for components assumed to be present at the LIC and G cloud velocities are (assuming km s-1 ; the upper limits are lower for narrower lines). This is comparable to the LIC column density deduced for Aqr, but is approximately an order of magnitude lower than those found for Oph and 51 Oph. Thus, while the present non-detection is still consistent with the LIC being present towards Oph (as it would have to be if it truly surrounds the Sun), the upper limit implies order-of-magnitude spatial and/or density inhomogeneities on a scale of pc (based on the angular separation of Oph and Oph, and an assumed LIC extent of pc in this direction; cf. Lallement et al. 1994).
The two stronger, more blue-shifted, components which are observed towards Oph must be due to other, probably more distant, clouds. Indeed, components with similar velocities (-32.0 and -28.4 km s-1), and similar column densities and b -values, were observed towards Oph by Crawford & Dunkin (1995), so it appears likely that the same two clouds have also been sampled by this line of sight. At the distance of the nearer of the two stars ( Oph, pc) the lines of sight are separated by 2.7 pc, which is consistent with other estimates of the sizes of clouds within the local bubble. However, we note that Barlow et al. (1995) detected a weak ( cm-2) Ca II component at -31.8 km s-1 towards Oph (, , pc), which may imply that this line of sight has also passed through the material responsible for the most negative velocity component present towards Oph. As the separation of these two lines of sight at the distance of Oph is 6 pc, and as this is rather larger than previous estimates for the sizes of the local clouds, this cloud may actually be quite a lot closer than Oph.
The strongest absorption component towards Oph (i.e. that at -26.2 km s-1) is not observed towards Oph, and this is consistent with the view of Frisch, York & Fowler (1987) that Oph lies behind a 'wisp' of H I emission which they identified in the 21-cm maps of Colomb, Pöppel & Heiles (1980). However, it will be necessary to observe the interstellar spectra of many more nearby stars in Ophiuchus, and these at a range of distances, before we can be certain that the 21-cm structures observed in this direction by Colomb et al. are really as close (i.e. within 15 pc of the Sun) as this argument would imply.
4.3. The spectrum of 51 Oph
The spectrum of 51 Oph is interesting for a variety of reasons. In addition to the two components tentatively identified with the LIC and G clouds (Sect. 4.1), three other absorption components are present towards this star. Perhaps of greatest interest is the fact that the narrow central (-21.3 km s-1) component identified by Lagrange-Henri et al. (1990) is asymmetric, which implies the presence of additional velocity structure. Fig. 2a shows this component plotted on an expanded scale, which emphasises the asymmetry of the profile, and also shows our two-component fit (Table 2). Both of these components are extremely narrow, and the velocity dispersion of the -20.3 km s-1 component ( +-00:4:1km s-1) implies an upper limit to the kinetic temperature (obtained by assuming in Equation 1) of 870+-2155070K, which is much lower than that expected for clouds within the Local Bubble (Sect. 1). It is true that several comparably narrow Ca II components were detected in UHRF observations of Oph by Barlow et al. (1995), but these were found to be associated with the complex of components centered at about -14 km s-1, and which are known to be associated with more distant cold molecular clouds. Thus, either the line-of-sight to 51 Oph has passed through an unusually cool cloud within 25pc of the Sun (which is not observed towards other nearby stars in Ophiuchus), or these narrow components arise in the circumstellar environment.
As 51 Oph is a well known Vega-excess star (i.e. a main-sequence star exhibiting an infrared excess due to circumstellar dust; Coté & Waters 1987, Waters, Coté & Geballe 1988), it is quite likely that some of the absorption components have a circumstellar origin (Lagrange-Henri et al. 1990). Indeed, quite convincing evidence that at least the -15.8 km s-1 component, and possibly the narrow central components also, are circumstellar comes from the detection of the 4583.837 Å line of Fe II by Dunkin, Barlow & Ryan (1996). This line is not observed in the general interstellar medium because it arises from a metastable level () which lies 2.8 eV above the ground state. The fact that this level is populated strongly suggests excitation in the circumstellar environment, a point already made by Grady & Silvis (1993) in connection with their detection of UV lines from excited levels of Fe II towards this star. Fig. 2b shows the Fe II line observed at (S.K. Dunkin, personal communication) superimposed on the UHRF spectrum of the Ca K line. Even allowing for the lower resolution of the Fe II spectrum, it is clear that this line is centered at the velocity of the -15.8 km s-1 Ca II component (this is especially obvious if the Fe II line is scaled to the same depth as the Ca II component; dotted line in Fig. 2b). Inspection of Fig. 2b also shows that the Fe II line is asymmetric, indicative of additional Fe II absorption blueward of -15.8 km s-1. This is the velocity range occupied by the narrow central components, and may suggest that these are also circumstellar. Only UHRF observations of the Fe II line itself will be able to determine unambiguously whether or not this is the case.
Given these arguments for a circumstellar origin, it is of particular interest to determine whether or not these lines exhibit line-profile variations similar to those found for the well-studied Pictoris system (e.g. Ferlet, Hobbs & Vidal-Madjar 1987, Lagrange-Henri et al. 1992, Crawford et al. 1994). This is especially important as Grady & Silvis (1993) have found clear evidence for in-falling gas (with velocities up to 100 km s-1) from time-variable, redshifted UV absorption lines in 51 Oph which are very similar to those found for Pic. In order to check for such variations in the Ca K line, we degraded our UHRF spectrum to an effective resolution of , and compared it with that obtained by Lagrange-Henri et al. (1990). This comparison revealed no evidence for spectral changes, either in the velocities or in the relative intensities of the absorption components. Thus, while further ultra-high-resolution monitoring of the line profile will be needed before a definite conclusion can be reached, the present observations show no evidence for Pic-type variability in the circumstellar Ca II line of 51 Oph.
As a final point, we note that the constancy in velocity of the (presumed) circumstellar line profiles (Table 2) is doubly surprising given that Buscombe (1963) found 51 Oph to have a variable velocity. If this star really is a spectroscopic binary, and if, as seems almost certain, the -15.8 km s-1 component arises in the circumstellar environment, it would appear that the circumstellar material must surround both stars. On the other hand, it may be that the assignation of a variable stellar velocity is spurious. Clearly it would be desirable to obtain more accurate measurements of the stellar radial velocity in order to settle this question.
© European Southern Observatory (ESO) 1997
Online publication: July 8, 1998