L110, 1997 September 10
We present a deep color-magnitude diagram
in the VI passbands of the globular cluster M54, a member of the
Sagittarius dwarf galaxy. The data extend below the cluster's main-sequence
turnoff, allowing us to estimate the cluster's age. We find that M54
is 0.51.5
Gyr older than the Galactic globulars M68 and M5. In absolute terms, its
age is comparable to the published age estimates of the other member
clusters Arp 2 and Terzan 8 but is significantly greater than that of the
member cluster Terzan 7. An age estimate of the Sagittarius field
population relative to M54 suggests that M54 is
3 Gyr
older than the field. We discuss briefly the star formation history of
the Sagittarius dwarf galaxy.
Subject headings: galaxies:
individual (Sagittarius)
galaxies:
star clustersgalaxies:
stellar contentglobular
clusters: individual
(NGC 6715)stars:
Population II
1 Hubble Fellow.
2 Observations obtained while a staff member of the Cerro Tololo Inter-American Observatory, one of the National Optical Astronomy Observatories (NOAO). NOAO is operated by the Association of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation.
Ibata, Gilmore,
& Irwin (1994, hereafter IGI) announced the
discovery of a gas-poor galaxy lying behind the bulge of the Milky Way at
a distance of
24 kpc.
They estimated the overall stellar content of this galaxy (hereafter
referred to as the Sagittarius dwarf galaxy, or Sgr) to be similar in age,
metallicity, and total luminosity to the Fornax dwarf spheroidal galaxy.
They also noted that four previously known globular clusters (M54, Arp 2,
Ter 7, and Ter 8) had roughly the correct distances, positions on the sky,
and radial velocities to be members of this galaxy.
The presence of a galaxy so nearby provides an unparalleled opportunity for detailed study of stellar populations, chemical enrichment history, and galaxy formation and evolution. Ibata et al. (1997) provide an excellent review of the ensuing research on Sgr. The results of several of these studies have increased the likelihood that the four globular clusters are members of Sgr. In particular, Da Costa & Armandroff (1995) presented new abundances and radial velocities for the four globular cluster candidates; their velocities agree well with the systemic velocity of Sgr (IGI). The new isodensity map of Ibata et al. (1997) confirms that Sgr extends over the three outlying clusters, Arp 2, Ter 7, and Ter 8.
Table 1
summarizes our current knowledge concerning the ages of Sgr and its
globular clusters. The difficulties in comparing absolute ages of stellar
systems are well known (see, e.g., Chaboyer
1995), so we have used ages, when available, from the self-consistent
work of Chaboyer, Demarque, &
Sarajedini (1996, hereafter CDS). We use the ages
from column 2 of their Table 3, which employ
an MV(RR) -
[Fe
H]
relation nearly identical to that of Lee,
Demarque, & Zinn (1990, hereafter referred to
as LDZ). CDS do not include ages for
the Sgr field population, so the estimates for Sgr listed in
Table 1 are less homogeneous. The confusion is
compounded by the uncertainty in the metallicity of Sgr; the dependence of
derived age on assumed metallicity is well known (see, e.g.,
CDS). Still, the studies agree on an age between 10 and
14 Gyr for the dominant population of Sgr.
The only Sgr globular cluster candidate
currently without accurate main-sequence turnoff (MSTO) photometry and an
age estimate is M54 (NGC 6715). Sarajedini
& Layden (1995, hereafter SL95) presented a CCD
color-magnitude diagram (CMD) of the bright stars in
M54 (MV
+2.0).
They found M54 to be metal poor
([FeH]
= -1.79 dex) and to have a blue horizontal branch (HB) typical of old
Galactic globular clusters. The recent photometry of M54 by
Marconi et al. (1997) goes deeper, but the
photometric errors and incompleteness at the magnitude of the MSTO preclude
an accurate age analysis.
M54 is by far the most luminous of the
four Sgr clusters (SL95) and perhaps the most secure
candidate for membership in Sgr, since it lies in the highest density
region of that galaxy (IGI) and has a
distance (SL95) and radial velocity (Da
Costa & Armandroff 1995) that correspond very well with those of
Sgr. SL95 have speculated that M54 is
the 
nucleus
of Sgr, akin to the nuclei found in many dwarf elliptical galaxies.
Clearly, we cannot have a complete understanding of the star formation
history of Sgr without information about the age of M54.
In order to obtain photometry to the MSTO,
we secured deep images of M54 using the Cerro Tololo Inter-American
Observatory (CTIO) 4.0 m telescope at prime focus during the nights of 1995
June 28 and 29. We employed the Tek no. 4 20482 pixel CCD, which
provided a scale of
0
43
pixel-1. The median seeing
was 12.
We measured instrumental stellar magnitudes on each frame using the DoPHOT photometry program (Schechter, Mateo, & Saha 1993) and transformed them directly to the VI magnitude system of SL95 using the large number of stars common to both data sets. Photometry from the five VI frame pairs with the best seeing were assembled, and mean magnitudes and errors (standard errors of the mean) were computed for each star detected in three or more frame pairs. The details of the reduction procedure and the photometric data for the resulting 26,485 stars are presented in Layden & Sarajedini (1997). Comparisons show that these data are accurately tied to the SL95 photometric system.
Figure 1
presents the VI CMDs for (a) all the M54 stars
with high-quality data and (b) all the high-quality M54 stars
located between
2
5
and
43
from the cluster center. In these panels, the curves are the fiducial red
giant branches (RGBs) of M54 and Sgr derived by SL95.
These curves, together with the CMDs of Sgr and a foreground bulge control
field by Mateo et al. (1994,
hereafter MUSKKK), facilitate the interpretation of
our CMD.
Fig. 1
The curve on the left is the M54 RGB
fiducial; it guides the eye faintward to where the M54 RGB becomes well
populated. The M54 RGB turns blueward onto the subgiant branch (SGB) at
(V - I, V) = (0.9, 20.9) mag and merges with a
column of points
0.2 mag
blueward of this. As we will see, this column represents the superposed
MSTO regions of M54 and Sgr.
The curve on the right is the Sgr RGB
fiducial. The lower RGB of Sgr is not as well populated as that of M54 in
this figure, but there appears to be an excess of points roughly parallel
to the M54 RGB that terminates at (V - I, V) = (1.0,
20.9) mag and that presumably turns blueward onto the SGB at this point
(see MUSKKK). The MSTO of Sgr in
the MUSKKK field occurs at (V - I,
V) 
(0.75, 21.4) mag, the same region as the MSTO of M54 in our data. The plume
of stars at V - I = 0.75 and 20.2 < V < 20.9
mag corresponds to the young (4 Gyr) Sgr population discovered
by MUSKKK.
Other prominent sequences in
Figure 1a include the M54 blue
HB (SL95), the Sgr red HB
clump (MUSKKK; SL95), and a
population of blue stragglers or very young stars belonging to either M54
or Sgr (0.2 < V - I < 0.6, 19 < V < 21
mag). The MUSKKK bulge control field coincides
well with the column of stars at V - I
0.9 and
V < 19.5 mag that sweeps redward at fainter magnitudes across the
M54 and Sgr lower RGBs. Thus, most of the scatter with V > 20 and
V - I > 1.0 mag is attributed to the foreground bulge.
One important qualitative statement about
the relative ages of M54 and Sgr can be made at this point. The MSTOs of
these populations appear to be coincident at V - I
0.75 mag.
The reddenings are identical since the populations lie in the same field.
Yet the metallicity of M54 is at least 0.5 dex lower than that of Sgr,
so for the MSTOs to coincide, M54 must be older than Sgr.
A simple estimate of the age of M54 can be
made by directly comparing our photometry with the fiducial sequences of
other clusters. High-quality VI CCD photometry exists in
the literature for M68
([FeH]
= -2.09) and M5
([FeH]
= -1.40), which bracket M54 in
metallicity. Figure 2 shows our data plotted
with the fiducial sequence of M68 from
Walker (1994; Fig. 2,
dashed line) and the fiducial of M5 from
Sandquist et al. (1996;
Fig. 2, solid line). All the cluster data were
registered to the observational H-R diagram with the
V(HB), [FeH],
and E(V - I) values given in
Table 2 along with the relation
MV(RR)
= 0.17[FeH]
+0.82 (LDZ).
Fig. 2
The fiducial sequence comparison reveals that the age of M54 is comparable to those of M68 and M5. Since M54 is almost exactly between M68 and M5 in metallicity, one expects the data for M54 to lie midway between the M68 and M5 fiducials. However, the M54 data, particularly for the SGB, appears to be skewed slightly toward the M5 fiducial. This suggests that M54 may be slightly older than M68 or M5. In the next section, we will quantify this age difference.
Another method for measuring globular
cluster ages is isochrone fitting. Figure 3
shows the revised Yale isochrones (RYI; Green,
Demarque, & King 1987) for Y = 0.23,
[FeH]
= -1.50, and ages of
1018
Gyr, superposed on the data from Figure 1b.
The isochrones were shifted to the observed plane with the V(HB) and
E(V - I) values listed in Table 2
and the LDZ relation
between MV(RR) and
[FeH]
(assumed for consistency with the RYI, see King,
Demarque, & King 1988). The ridge line of M54 SGB stars suggests an
age
of 1314
Gyr. The 12 Gyr isochrone forms an envelope about the MSTO points, setting
a hard lower limit for the age of M54 under the stated assumptions.
Fig. 3
We used isochrones with
[FeH]
= -1.50 because the RYI employ scaled solar abundance ratios, whereas
observations suggest that Galactic globular clusters have an enhancement of
elements of
[Fe]
+0.4 dex
(see, e.g., Pagel &
Tautvai
ien
1995). Salaris, Chieffi, & Straniero
(1993) showed that for a given iron abundance, scaled solar isochrones
that are 0.29 dex more metal rich in
[FeH]
closely
mimic -enhanced
isochrones with
[Fe]
= +0.4 dex. RYI
with [FeH]
= -1.79 indicate ages
12
Gyr greater than those shown here.
Analogous RYI fits to the data of M68
(Walker 1994) and M5 (Sandquist et
al. 1996), again using the parameters from Table 2
and
the -enhanced
metallicities, produced ages of
13 Gyr
for M68 and
12 Gyr
for M5. As in § 3, M54 appears to be comparable in
age to these clusters, or perhaps slightly older.
We are concerned by the tendency for the
isochrones to be bluer than the data at V
22.
This could be due to (1) differential incompleteness in our data as a
function of color, (2) a tendency for Sgr stars to dominate the red side of
the lower main sequence and thus to bias the data redward, or (3)
inadequacies in the isochrones or adopted reddening and distance modulus.
Though adopting a larger reddening (e.g., by 0.05 mag
2
)
corrects the main-sequence color problem and makes the derived age younger
(2 Gyr),
it degrades the fit to the lower RGB. Adjusting the reddening and distance
modulus in concert enables us to obtain a better overall fit; the age
obtained is 15 Gyr for E(V - I) = 0.18 and a
distance modulus 0.15 mag smaller than that employed
in Figure 3.
Our isochrone age estimates are supported
by estimates based on the luminosity of the SGB. The difference between the
magnitude of the subgiants at a well-defined color and that of the HB is
similar for M54, M68, and M5. When calibrated with the RYI, we find that
M54
is 12
Gyr older than M68 and
01
Gyr older than M5. Details of this procedure are presented in
Layden & Sarajedini (1997).
Given the uncertainties in determining
absolute ages, we would like to compare our age for M54 to that of the Sgr
field population in a relative sense. In Figure 1, the
lower RGB of the Sgr field population appears to terminate abruptly at
V
20.9 mag. RYI with metallicities and ages with lower RGBs terminating at
this magnitude can be used to place an upper limit on the age of the
Sgr field. For
[FeH]
= -0.50 (SL95), we find a maximum age of 6 Gyr.
For [FeH]
= -1.2 and
[Fe]
= +0.4, we find an age of 9 Gyr. For
[FeH]
= -1.2 and
[Fe]
= +0.0, we find an age of 11 Gyr. The latter is the oldest age obtainable
for Sgr that is consistent with currently quoted abundance estimates. This
age is also in good agreement with the Sgr ages listed
in Table 1. Clearly, M54 must be older than the Sgr
field stars by
3
Gyr.
All three of the methods discussed above
suggest that M54
is 0.51.5
Gyr older than the comparison clusters M68 and M5. CDS
find the age of M5 to be typical of Galactic globular clusters of its
metallicity, while M68 may be somewhat younger than average. Thus, M54 has
an age typical of Galactic globulars of its metallicity (see Fig. 1
of CDS).
The absolute age estimates discussed in
§ 4, with the use of the LDZ
relation between MV(RR) and
[FeH]
and [Fe]
= +0.4, suggest that M54 has an age of
14
Gyr. Comparing this with the ages of the other Sgr globulars shown
in Table 1 indicates that M54, Ter 8, and Arp 2 are all
comparably old (for more details, see Layden &
Sarajedini 1997), while Ter 7 is significantly younger. Given the
uncertainties in the existing photometry, we cannot rule out the
possibility that the three old clusters in Sgr are coeval.
Comparing our absolute age for M54 with
the age estimates for the dominant Sgr field population shown in
Table 1 suggests that M54 is older than the metal-rich
field population in which it is embedded. This result is supported by
our analysis in § 4, where we estimated the maximum
age of the Sgr field as a function of assumed
[FeH]
and found that M54 is at least 3 Gyr older than Sgr.
Taken at face value, these ages suggest
that the metal-poor clusters represent the earliest epoch of significant
star formation in Sgr. Vigorous star formation in the field appears to have
begun several gigayears later. Given this age difference, it seems
reasonable to expect that gas expelled from evolving metal-poor cluster
stars enriched the interstellar medium and thus the first generation of Sgr
field stars. This explains, at least in part, why the Sgr field is so much
more metal rich than the old clusters. As was the case for many of the
Galactic satellite dwarf spheroidals (see,
e.g., Smecker-Hane et al. 1994), Sgr
managed to retain a significant portion of its gas for many gigayears,
enabling the formation of Ter 7 and the
4 Gyr
field population discussed by MUSKKK and represented by
the blue plume of stars above the MSTO in Figure
1. Given the age and abundance of Ter 7, it is perhaps more appropriate
to compare this cluster with the
populous
clusters
of the SMC or ESO 121-SC03 in the LMC (Da Costa
1991) than with traditional globular clusters. Finally, we note that
the HB morphologies of the three old Sgr globulars are quite blue for
their metallicity
(SL95; Buonanno et al.
1995; Ortolani & Gratton 1990), in
better agreement with the Galactic globular clusters than those of the
Fornax dwarf galaxy. In this respect, Sgr may be a better example of a
surviving building block of the Galactic halo than Fornax
(see Zinn 1993).
We thank Mario Mateo for his thoughtful comments. A. C. L. was supported by NASA grant HF-01082.01-96A, and A. S. was supported by NASA grant HF-01077.01-94A from the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5-26555.
ien,
G. 1995, MNRAS, 276, 505 First citation in article | NASA ADSGalaxy
Connection, ed. G. H. Smith & J. P. Brodie (San Francisco: ASP),
302 First citation in article
Full image (77kb) | Discussion in text
FIG.
1.(a)
Color-magnitude diagram of 18,796 stars in the direction of M54. Only stars
with at least three
detections, V
< 0.050 mag,
and V
- I < 0.071 mag are shown. (b) The same as in
(a), but only the 7551 stars located between
25
and
43
from the cluster center are shown. The curves are the RGB fiducials of M54
(left) and Sgr (right) from SL95.
Full image (37kb) | Discussion in text
FIG.
2.As
in Figure 1b, but the observed data have been
shifted to the theoretical plane as discussed in §
3. The solid line is the fiducial for M5
([FeH]
= -1.40, Walker 1994), and the dashed line is the
fiducial for M68
([FeH]
= -2.09, Sandquist et al. 1996).
Full image (45kb) | Discussion in text
FIG.
3.Data
from Figure 1b plotted with revised Yale
isochrones shifted to the observed plane (see § 4).
The isochrones are for 10, 12, 14, 16, and 18 Gyr (left to right)
and
[FeH]
= -1.50. The latter is equivalent to an isochrone with
[FeH]
= -1.79 and
[Fe]
= +0.4.
| Object | [FeH] | Age | References |
| Sgr... | -1.2 | 10 | 1 |
| Sgr... | -0.5 | 12 | 2 |
| Sgr... | -1.0 | 1014 | 3 |
| Sgr... | -1.1 | 10 | 4 |
| Arp 2... | -1.70 | 12.3 ± 0.8 | 5 |
| Ter 7... | -0.36 | 7.2 ± 0.5 | 5 |
| Ter 8... | -1.99 | 16.9 ± 1.5 | 5 |
(1) MUSKKK; (2) Mateo et al.
1996; (3) Fahlman et al. 1996;
(4) Marconi et al. 1997; (5) CDS.
| Cluster | V(HB) | [FeH] | E(V - I) a |
| M68 | 15.64 ± 0.01 | -2.09 ± 0.11 | 0.09 |
| M54 | 18.17 ± 0.05 | -1.79 ± 0.08 | 0.17 |
| M5 | 15.09 ± 0.02 | -1.40 ± 0.06 | 0.04 |