Astron. Astrophys. 347, 455-472 (1999)

5. Classification and assignment

After having determined the parameters of our cluster sample, we now discuss each cluster's possible affiliations with the galactic structure components i.e. halo, disk or bulge for each cluster. The necessary criteria are introduced in the following subsection.

5.1. The assignment criteria

5.1.1. Disk and halo: Zinn (1985)

Zinn (1985) divided the GC-system into a metal-poor ( dex) halo- and a metal-rich () disk-subsystem. This distinction also correlated with the kinematics and spatial distribution of their objects. The resulting criteria are listed in Table 11. Eq. 8 gives the orbital velocity of a cluster depending on its observed radial velocity . can be compared to the net rotation as given in Table 11.

Table 11. Kinematics and spatial distribution of the metal-rich and -poor subsystems of GCs according to Zinn (1985)

5.1.2. Bulge and (thick) disk: Minniti (1995,1996)

Minniti (1995, 1996) divided Zinn's metal-rich disk system further into GCs belonging to the (thick) disk on the one hand and to the bulge on the other. Comparing the GCs with their corresponding field population, he assigned the GCs with galactocentric distances kpc to the bulge and the ones with kpc to the thick disk.

5.1.3. Inner halo, bar and disk: Burkert & Smith

Burkert & Smith (1997) used the masses of the metal-rich GCs to distinguish between a group belonging to the inner halo and a group which can be further divided into a bar- and a ring-system using the kinematics and spatial distribution of the clusters (see Table 12).

Table 12. Criteria for subgroups of the metal-rich GCs according to Burkert & Smith (1997).

5.1.4. Radial velocities

Unfortunately, there do not exist any data on proper motions of our clusters. The only kinematic information available are radial velocities, catalogued by Harris (1996). Thus, we can only check, whether a disk orbit is compatible with a given radial velocity. This is possible by comparing the measured radial velocity with the expected one, calculated via Eq. 8 assuming that disk clusters move on circular orbits in the galactic plane.

where l is the galactic longitude, and x and y are the heliocentric coordinates. We used kpc and km/s. gives the velocities of the clusters in the plane, corresponding to the galactic rotational velocity with the values taken from Fich & Tremaine (1991).

5.1.5. Metallicity gradient

The metallicity gradient of the disk is an uncertain criteria insofar, as it is defined for the outer ranges of the galactic disk. We use a metallicity gradient referring to the population of old open clusters. The oldest of these objects have ages similar to the youngest GCs (Phelps et al. 1994). Their scale height is comparable to other thick disk objects. Assuming that they are related to a possible disk population of GCs (Friel 1995), we can use their metallicity gradient

(Friel 1995) as a criterion for whether our GCs belong to the galactic thick disk or not.

5.2. The assignment

Using the above criteria, we assigned the clusters of our sample according to Table 13. The values of the parameters necessary to decide on group membership are listed in Table 14.

Table 13. Assignment of the clusters to the systems disk, halo, bulge and bar according to the criteria in the first column. ? is used if no assignment is possible, the symbols d, h, bu and ba together with + or - relate to criteria which only can decide whether an object belongs to a certain group or not. Symbols in brackets denote uncertainties explained in the text.

Table 14. All relevant parameters for the assignment. gives the metallicity according to the -relation, the heliocentric distance, the galactocentric distance and z the distance to the disk. The M in stands for cluster mass, for the orbital velocity, derived via the rotational velocity curve of Fich & Tremaine (1991), for the observed radial velocity and for the orbital velocity of the clusters, calculated using and the assumption of circular cluster orbits (in order to compare with a net rotation). contains the expected radial velocity assuming disk orbits and the values derived via a metallicity gradient of the old open clusters (Friel 1995).

As for the metallicities of our clusters, they all belong to the disk system according to Zinn, which is obvious as the sample had been selected in this way. Not so obvious is the comparison with the net rotation of Zinn's disk group. Only NGC 5927 shows a value of which is not totally off the net rotation as given in Table 11.

The clusters belonging to the bulge according to Minniti's criterion are members of the bar following the arguments of Burkert & Smith (1997). Binney et al. (1997) quote a value of for the angle between x-axis in galactocentric coordinates and the major semiaxis of the bulge structure. Its end lying nearer to the sun is located at small galactic longitudes ( in cartesian coordinates). Fig. 34 shows the spatial distribution of our cluster sample. The coordinates of the `bar' clusters NGC 6342, 6528 and 6553 according to Burkert & Smith seem to be consistent with a structure described by Binney et al. (1997). However, as the referee pointed out, we do not know how long-lived the Milky-Way bar is, and other tracers of old populations such as RR Lyrae do not follow the bar (Alcock et al. 1998). Moreover, the distance between the `bar' clusters NGC 6528 and NGC 6553 is about 5 kpc, which is much larger than the length of the Milky-Way bar according to most authors (e.g. Binney et al. 1997). Also note in Fig. 34, that the errors in the x-coordinate are larger than those in y and z.

Fig. 34a and b. Distribution of the cluster sample in the galactocentric x-y-plane a and y-z-plane b . In the upper panel , the observer is located at . The distribution in the y-z-plane is as seen from the sun. The distances used correspond to in Table 9.

There are only two `disk' clusters remaining, assuming Burkert & Smith's definition of disk clusters: NGC 5927 and NGC 6760. However, the radial velocities corroborate this result for NGC 5927 only. For any other cluster, the radial velocities seem to exclude an assignment to the disk.

The metallicity gradient of the old open clusters leads to the conclusion that none of our clusters is to be assigned to the thick disk. Taken the whole sample of metal rich clusters (i.e. clusters with dex according to Zinn 1985, see Tables 16, 15), we find only three objects, which could be disk clusters according to the metallicity gradient criterion.

Table 15. The relevant parameters of the remaining metal rich GCs according to Harris (1996). The columns are labeled as in Table 14. Distances are given in kpc, metallicities in dex and velocities in km/s.

Table 16. Suggested assignment according to the criteria discussed above for the remaining metal rich GCs. The columns are labeled as in Table 13.

Some of the clusters do not meet any of the criteria. Interestingly, they are the most massive, but metal-poorest objects of the sample. These objects are NGC 6316, 6760, and 6441 as well as (in Table 16) NGC 104, 6356 and 6388. Although NGC 104 seems to be a disk cluster and mostly is referred to as such, the large distance to the galactic plane (3 kpc) does not support this assignment. Probably, the mentioned objects belong to the halo, being its metal-richest clusters. Zinn (1985) and Armandroff (1993) point to the fact that the division into metal-rich and poor clusters is by no means an exact one, but that there is a metal-rich sample of halo clusters as well as a metal-poorer one of disk objects. Richtler et al. (1994) discussed the existence of a subgroup of disk clusters according to Zinn (1985), based on an analysis of the metallicities and the minimum inclination angles derived from -values for these clusters. They conclude that the clusters NGC 6496, 6624 and 6637 might not be disk clusters after all, but belong to the halo. Adding their argument to the above discussion, we end up with 3 probable disk members (NGC 5927, additionally from Table 16 Liller 1 and Pal 10. Pal 11 is excluded because of its large minimum inclination angle.) and 9 clusters (NGC 104, 6316, 6356, 6388, 6441, 6496, 6624, 6637 and 6760) that more likely belong to the halo than to the (thick) disk. The rest of the clusters (NGC 6342, 6528, 6553 and the remaining ones of Table 16) fall in with the bulge/bar-group of Minniti (1995) and Burkert & Smith (1997).

© European Southern Observatory (ESO) 1999

Online publication: June 30, 1999
helpdesk.link@springer.de