A&A 397, 987-995 (2003)
DOI: 10.1051/0004-6361:20021562
G. Cutispoto1 - G. Tagliaferri2 - J. R. de Medeiros3 - L. Pastori2 - L. Pasquini4 - J. Andersen5
1 - INAF - Catania Astrophysical Observatory, v. S. Sofia 78,
95125 Catania, Italy
2 - INAF - Brera Astronomical Observatory, via Bianchi 46,
22055 Merate (LC), Italy
3 - University Federal of Rio Grande do Norte, Department of Physics,
59072-970 Natal, R.N., Brazil
4 - European Southern Observatory, Karl-Schwarzschild-Strasse 2,
85748 Garching bei München, Germany
5 - Astronomical Observatory, NBIfAFG, Juliane Maries Vej 30,
2100 Copenhagen, Denmark
Received 8 July 2002 / Accepted 28 October 2002
Abstract
We present an analysis of our high-resolution spectroscopic and
high-precision
photometric observations of a sample of
110 nearby late-F and G-type stars selected for their large rotational
velocity. The relationships between Li abundance, X-ray luminosity, and
are investigated. We find that, as expected, the stars in our sample show
statistically higher Li abundance and activity level than field star samples
with similar characteristics, but slower rotation. Surprisingly, however, we
also find four rapidly-rotating single main-sequence stars with
very low Li abundance. For both single and binary stars we find a large spread
of Li abundance for stars with rotation lower than about 18
.
The well-established correlation between X-ray luminosity and rotation rate
is clearly observed. All single unevolved solar type stars with
larger than 18
are strong X-ray emitters and have high Li abundance.
Finally, we find also five evolved stars with very low Li abundance that are
still rather fast rotators.
The results from our sample confirm the presence of
young very active stars close to the Sun, in agreement with recent findings
from EUV and X-ray surveys, although our sample does not show such extreme
characteristics as those selected from EUV and X-ray surveys at the current
flux limits.
Key words: stars: abundances - stars: activity - stars: fundamental parameters - stars: variables: general - X-rays: stars
Hence, we believe that it would be extremely interesting to study a sample of stars selected directly on the basis of high rotational velocity rather than from such parameters as high (X-ray) activity or young age. We would expect such a sample to be composed, essentially without exception, by stars with high coronal activity, including both young single (or binary) stars as well as short-period older binaries. The former should then have a very high Li abundance (i.e. higher than the Hyades), the latter a large spread of Li abundance. In any case, the average Li abundance should be higher than that expected in field stars of the same spectral type, selected only on the basis of their distance from the Sun. We stress that in such a sample we would not expect to find neither low-activity stars nor single stars with low Li abundance.
There are several reasons to expect that our sample will show higher Li abundances than randomly selected star samples with similar spectral type. For single stars the reason is that, although a strict one-to-one correlation does not hold, it has been shown that young stars tend to have higher Li abundances than old stars (see the discussion in Pasquini et al. 1997). Hence, because our sample is selected on the basis of fast rotation (i.e. a rotational velocity similar to or higher than the value observed for stars of comparable spectral type in the Hyades), the single stars are expected to be rather young, typically younger than 1 Gyr. Therefore, they are expected to show Li abundance levels similar to or higher than those observed in the Hyades. For binary stars the reason is that the rapid rotators which are not young are likely to be members of tidally locked systems. Although for some of them we expect to find low Li abundances, it has been recognized that the Li content in such systems is, at least on a statistical basis, larger than in randomly selected stars (Randich et al. 1993, 1994; Fernandez-Figueroa et al. 1993; Barrado y Navascues et al. 1998).
However, we note that it is
not clear which is the physical cause for this difference;
several theories predict that rotation itself (or the rotational history of
the star) may influence the Li depletion through either meridional circulation
or rotationally induced mixing (Charbonnel et al. 2000; Pinsonneault et al. 2000). Evidence for some dependence of Li abundance on rotation has
been observed among K-type stars in young clusters
(Soderblom et al. 1994; Garcia-Lopez et al. 1994) and even among
slow rotators in the Per cluster (Randich et al. 1998). We
also need to consider that the
clusters in which this rotational dependence of Li has been observed are all
rather young, and stars in such clusters may have rotations up to several tens
of
.
Finally, it is not clear if such dependence of Li on rotation is an
artifact due to the use of inappropriate atmospheric models or if phenomena
like stellar activity play an important role in this context (cf. Pasquini
2000; Randich 2001; Cutispoto 2002).
A sample of nearby fast-rotating solar-type stars was selected and studied by
Cutispoto et al. (2002), hereafter referred to as Paper I. Accurate
spectral classifications, effective temperatures (
), rotational
velocities
(
), Li abundances (
), radial velocities (RV), and X-ray luminosities
(
)
were determined. In this paper we discuss
the log
vs.
diagrams, compare the
of the
stars of our sample with those of the Pleiades and Hyades
clusters and with a sample of stars not selected on the basis of
;
we also
discuss the relationships between
,
and
and the
global properties of our sample.
In order to identify young solar-type stars in the solar neighborhood we
started from the CORAVEL survey of 3200 F to G-type stars brighter than
(Nordström et al. 1999). We defined the sample by
selecting all stars with significant
rotational velocity as measured from the width of the CORAVEL
cross-correlation profile. Specifically, a lower limit of 8
was chosen
for the
of the Gaussian fit to the profile (see Sect. 2.1 in Paper I
for further details). This choice ensures that
all the selected stars are likely to fulfil one of the signatures for youth,
significant rotational velocity.
Among the stars thus selected, 129 can be observed in the southern hemisphere,
and we observed them spectroscopically with the 1.4-m ESO CAT telescope,
and photometrically with the 50-cm ESO telescope, in various observing runs.
For each star we derived an accurate spectral classification and
determined RV, ,
and
(see Paper I for details).
All data used in the analysis and figures presented in this paper are taken
from Table 1 of Paper I.
Out of the 129 stars observed, we found that 19 are not true solar-type stars (see Sect. 3.2 in Paper I), and they are not further studied in this paper. The remaining 110 stars here presented comprise 42 single stars, 33 visual binaries (VB), and 35 spectroscopic binaries (SB). Among the VB components, 35 are not SBs (see Table 1 in Paper I).
In the following analysis we treated the 35 single components of VBs, whether primaries or secondaries, as single stars, because their rotation rate is not influenced by the presence of the companion. Moreover, we were able to obtain separate spectral classifications for 61 of the components of SBs (see Table 1 and Appendix 2 in Paper I for details). In summary, therefore, data for 138 objects, including 77 single stars (42 + 35 as explained above) and 61 close binary components are available for the analysis performed in this paper.
Figure 1a shows the B-V vs. MV colour-magnitude diagram (CMD) for these 138 stars. In this figure we also plot the MS region and the giant sequences determined from Hipparcos data (Houk et al. 1997). Our sample contains a total of 99 MS stars (59 single and 40 close binaries) and 39 non-main sequence stars (18 single and 21 close binaries). It also contains a higher percentage of close binary stars compared to what is expected from a sample of randomly selected field stars. This is expected, not only because the binaries are on average brighter than single stars, but also because our selection criterion (i.e. high rotation rate, see Sect. 4.1 in Paper I) favours the inclusion of tidally synchronised binary stars. In this context, our working sample is one of the most complete samples of solar-type stars studied yet. In Fig. 1b we highlight the CMD diagram for the stars we believe to be very young (see Sect. 3.1).
![]() |
Figure 1: The B-V vs. MV diagram for the stars in our sample. Single stars and primary and secondary componentes of VBs and of SBs are indentified by different symbols. The continuous line and the long-dashed line outline the main sequence and giant regions, respectively, from Hipparcos data (Houk et al. 1997); the short-dashed lines indicate the limits of the dispersion of main sequence stars from Hipparcos data. Panel a) shows the complete sample; panel b) shows the stars we believe to be very young. |
Open with DEXTER |
![]() |
Figure 2:
![]() ![]() ![]() ![]() |
Open with DEXTER |
Most of our MS single stars are above the Hyades track.
We know that stars with low
exist among old, inactive, slow-rotating
field stars (Pasquini et al. 1994). Intermediate-age stars have
comparable, albeit slightly lower
,
than the Hyades, as confirmed
by the observations of Li in intermediate-age open clusters (Randich et al.
2000), which, however, do not show objects with low
in the colour
range we consider. In contrast, a large scatter in
is clearly present in
the solar-age open cluster M 67 (Pasquini et al. 1997). In
summary, while
is not a good age tracer for stars older than about 1 Gyr, it has been
shown that stars with
comparable to or above that of the
Hyades can be confidently classified as being as young as the
Hyades or even younger.
Accordingly, it is immediately clear that many single stars in our sample
are indeed quite young, with ages lower than 1 Gyr. In particular,
we find that among the single MS stars studied, 6 (10%) have an
higher than that of the Pleiades, 38 (
64%) have an
between that of the Pleiades and Hyades, and 10 (
17%) have an
lower than that of the Hyades,
while for only 5 stars (
9%) we were just able to measure an upper
limit for
.
This result can be compared with the observations by
Pasquini et al. (1994)
of a sample of field stars of similar spectral type, but not selected on
the basis of high
.
Of the 42 MS single stars they studied, none has an
higher than that of the Pleiades, 10 (
24%) have an
in between that of the Pleiades and Hyades, and 20
(
48%) have an
below that of the Hyades, while for 12
stars (
28%) only an upper limit for
could be measured.
The six stars we find above the Pleiades top envelope are likely
bona-fide ZAMS stars (see also Fig. 1b, where we plot all stars
which we consider to be very young). This is reflected by both their fast
rotation and high
.
Listed in order of decreasing
these stars are:
HD 171488 (
,
), HD 36869
(
,
), HD 116402 (
,
), HD 217343 (
,
),
HD 222259 A (
,
)
and HD 202917
(
,
).
It is quite surprising that our sample includes also four apparently single
MS stars with significant rotation and an
much lower than that of the
Hyades, and comparable to or lower than that of the Sun. Their low
is not easily understood, and they deserve a separate discussion.
Listed in order of decreasing
these four
stars are: HD 199672
(
), HD 108361 A (
),
HD 184525 (
)
and HD 207377 A
(
). The last two
are particularly intriguing because they are both chromospherically
active (showing Ca II K emission line), and HD 184525 is
also a bright X-ray source. Hence, their very low
is quite puzzling.
HD 108361 A has no emission in the Ca II K line, while for
HD 199672 we do not have a spectrum of the Ca II H&K region.
These four stars are very interesting. They could be
objects which suffered peculiar angular momentum histories, and they
may represent the elusive population of young Li-poor stars predicted
by rotational mixing induced models (e.g. Pinsonneault et al. 2000;
Delyiannis et al. 2000).
Considering that we do not expect to have any selection bias in
our sample, apart from rotation, and making the assumption
that all our single stars are due to a recent star formation burst,
we can conclude that these young Li-poor stars represent much less
than 10% of the entire population. These objects
would have to be born as extremely fast rotators (if rotational
mixing was the mechanism responsible for Li depletion), considering
that, even after a strong spin-down, they still maintain rotational
velocities of the order of 10 km s-1.
To conclude, it is not obvious how to explain the low
found in these four single rapidly rotating MS stars.
They clearly deserve further investigation, in particular HD 184525.
As it can be seen from Fig. 2a, there is a fifth star (HD 127352 B,
log
,
B-V=0.79) with a very low
;
but
for such cool stars most of the Li
is already depleted at the age of the Hyades, therefore the upper
limit we found is not unsual even for a fast rotator.
Figure 2b shows the
vs. log
distribution for the
single stars above
the MS. A large dispersion is present, but in this case the low
which
occurs for five stars, in order of decreasing
,
HD 20837 B
(
), HD 68676 (
),
HD 84353 (
), HD 101117
(
)
and HD 153926 (
)
is not
surprising. If we compare our results with those of Mallik (1999),
we see that in the restricted
range of our sample the two
distributions are quite similar.
Many surveys of evolved stars have shown a rather sharp decrease
in
in the subgiant phase just after the objects leave
the MS (Randich et al. 2000; Lèbre et al. 1999).
It is interesting to note that the drastic drop in
after the
MS has been associated with effects related to the stellar rotational history.
In other words, it happens because the stars are slowed down from an initial
high rotation (Talon & Charbonnel 1998). However, the five stars
mentioned above are still pretty fast rotators despite their low
.
Single evolved stars with enhanced rotation are rare and pose, until now,
a difficult challenge for astrophysics. A dredge-up of angular momentum
from a rapidly rotating deep interior to the stellar surface is hypothesized by
different authors (e.g., Simon & Drake 1989), but other hypotheses
like the accretion of brown dwarfs or planets by giant stars can lead to
significant spin-up of the stellar surface (Siess & Livio 1999). Both
scenarios are still open for observational test. For the five fast rotators
discussed above these two scenarios may be hypothesized to explain their
enhanced rotation. Nevertheless, because these stars show low
,
an
additional hypothesis seems necessary: if dredge-up of angular momentum did
occur, one should not expect lithium production in the stellar interior; on the
other hand, if accretion of brown dwarfs or planets is the origin of the
observed high rotation, these objects have had no significant
.
A crucial test for such a hypothesis would be, in the present case, an
observation of CNO ratios combined with beryllium measurements.
Finally, Fig. 2b shows two stars,
HD 106506 (
)
and HD 141943
(
), with
.
Such a high value of
has never been found among evolved stars in the
spectral range we are considering. As they also show extremely fast
rotation, we consider these objects to be PMS stars. Their position in
the B-V vs. MV CMD is shown in Fig. 1b.
Figure 3 shows the
vs. log
distribution for
the close binaries (i.e. for
components of SBs, including SBs that are members of VBs). Panels a and b
show the distribution of close binary components on (40 objects) and above
(21 objects) the MS, respectively.
![]() |
Figure 3:
![]() ![]() ![]() ![]() |
Open with DEXTER |
As expected, the close binaries contain a higher proportion of stars with
low
than the single fast-rotating stars.
In fact, high rotation can be sustained by tidal interaction also in old
systems, which would otherwise have had enough time to destroy their Li.
Thus, if one selects stars based on their high rotation rate, one may expect
to select both young and old objects in the case of binaries. Moreover, we
know that binarity can affect both
and its links with rotation
and activity (e.g. Randich et al. 1993, 1994;
Fernandez-Figueroa et al. 1993; Barrado y Navascues et al. 1998),
and that tidal effects are expected to influence
in close binaries, which
should retain more Li than single stars of the same age (Zahn 1994).
For all these reasons it is not surprising that among the MS single
components of SBs in our sample, 3 (
8%) have an
higher than
that of the Pleiades, 16 (
40%) have an
between the
Pleiades and Hyades, and 5 (
12%) have an
below that of the Hyades, while for 16 stars (
40%)
only an upper limit to
could be determined.
The distribution of
in binaries is different from that in single stars:
more Li-poor objects are found in the binaries. The three stars above the
Pleiades upper envelope are the binary components
HD 202908 Ab (
,
), HD 84323 a
(
,
)
and HD 13183 a (
,
). While HD 84323 a and HD 13183 a are
likely to be ZAMS binaries (see their position in Fig. 1b), the
classification of HD 202908 is more difficult. In fact,
HD 202908 Ab is
part of a multiple system and is itself a VB whose primary
component is an SB2 system (see Table 1 and Appendix 2 in Paper I). The
values we computed for HD 202908 are 3.0 (
),
3.2 (
)
and 2.4 (
)
for the Aa, Ab and B
component, respectively. This system deserves further investigation
and its evolutionary status is not clear at this time.
Finally, between the non-MS binaries (Fig. 3b) there is one,
HD 104467, with
and
,
which
we believe to be a PMS binary star (see also Fig. 1b).
These results seem to indicate that another stellar parameter plays a
significant role in the connection between rotation and
.
This parameter
seems to be important only for single solar-type stars with
smaller
that about 18
,
while for higher rotation rates
is always
very high. These results are also found for evolved solar-type stars
(de Medeiros et al. 2000; do Nascimento et al. 2000). We note that
HD 106506
(
,
)
is not included in Fig. 4a
not to compress the dynamic range of the figure too much.
For both MS and evolved binary stars, as shown in Fig. 4b, the decrease of
the spread in the
distribution at high rotation rates is less marked
than for single stars. Again this can be explained by the effect of tidally
locked rotation: very old binaries can be strongly depleted in lithium while
still maintaining a high rotation rate.
![]() |
Figure 4:
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Open with DEXTER |
Pasquini et al. (1994) found a clear tendency for chromospherically
active solar-type stars to have high
and vice versa, despite the
large scattering observed in the diagram of
vs. Ca II surface flux.
Similar results are found in samples of X-ray or EUV selected active stars
(Favata et al. 1993, 1995; Tagliaferri et al. 1994,
2000; Jeffries 1995).
The dependence of
on stellar activity is consistent with the
predictions of standard evolutionary models, according to which
in MS
stars should depend on stellar temperature, metallicity and age. Here,
stellar activity is considered a good indicator of age for solar-type
stars (e.g. Soderblom et al. 1991; Pizzolato et al. 2000). What
is not clear is the physical cause of the large spread of the
values for
single stars with
.
Of course, one should expect that for some of these stars the low rotation
rate reflects the
effect, i.e. stars seen pole-on. If the stars
with
and high
are all seen pole-on, this could
reduce the observed spread somewhat. But it is difficult to believe that
all the Li-rich slow-rotating stars analysed in this work should have
such a preferred orientation of their axes.
One possible explanation is that
the observed spread was produced during early MS lifetime. We recall
that our sample is not homogeneous in age, and that substantial depletion
occurs in G-type stars between the ZAMS and the Hyades age (Pasquini
2000). The observations of young open clusters show that cool
(
)
MS stars present a clear spread of
(Soderblom et al. 1994;
Garcia Lopez et al. 1994; Randich et al. 1998) and that the fast
rotators (i.e. stars rotating faster that
15
)
are all close to the
upper envelope of the
distribution, while among the slower rotators a
large spread is observed. Such a spread among the slower rotators has been
discussed in the framework of the time needed to dissipate circumstellar disks
during the PMS phase and its consequences on the stellar momentum evolution
and the associated mixing in the early MS phase (Randich et al. 1998).
However,
the observations of young open clusters show no evidence for scattering
among the hottest G-type stars, and these are the objects forming
the bulk of our sample (in fact only a few of our stars are as cool as
5300 K). Finally, we recall that Jones et al. (1999) have
speculated that extra mixing on the MS could be driven by spin-down and
angular momentum loss. If this is indeed true, the observed scattering in the
versus
diagram could be due to different initial rotation rates.
![]() |
Figure 5:
![]() ![]() ![]() ![]() |
Open with DEXTER |
In Figs. 5a, b we plot
as a function of
.
We note that
among the single stars with
(Fig. 5a), all the 10
stars detected, except one, have
close to or larger than
,
a very large value for single stars.
The exception is HD 136160 (
;
), which,
among single
stars, has the earliest (F6V) spectral type in our sample and moderately
slow rotation for that spectral type. The well-established correlation
between
and rotation rate (Pallavicini et al. 1981;
Schmitt et al. 1985; Hempelmann et al. 1995;
Stauffer et al. 1997) is clearly observed.
There also seems to be a sort of saturation limit, which would appear even
more clearly if we had included HD 106506, which has
and
(we have omitted this star in order to obtain a reasonable
dynamic range of the figures and preserve clarity).
As expected, in particular at saturation level, single evolved stars have
larger
compared to single MS stars for similar values
of
.
Among the binary stars (Fig. 5b) with
all the 13
stars detected, except two, have an
near or larger than
8
1029
.
The exceptions are HD 218602, an SB1 (F9V + ?) system with
= 3
1029
and
HD 81997, an F5V + K5V binary system with
.
The
vs.
correlation is clearly observed also for binary stars.
The
distributions for MS and evolved binaries are similar,
although the largest
are observed for evolved binaries.
The spread seen for
could be partially due,
for both single and binary stars, to the presence of a solar-like activity
cycle, which in case of the Sun causes
to vary by more than
a dex (Peres et al. 2000).
![]() |
Figure 6:
![]() ![]() ![]() ![]() |
Open with DEXTER |
![]() |
Figure 7:
![]() ![]() |
Open with DEXTER |
![]() |
Figure 8:
![]() ![]() ![]() ![]() |
Open with DEXTER |
It is clear from Figs. 7a,b that no correlation between
and
exists for binary stars. This is an interesting, but not
surprising result, because old binaries, which have suffered strong Li
depletion, can still sustain high activity levels through high, tidally
induced rotation rates. In contrast, the single fast-to-moderately rotating
nearby solar-type stars show a definitive tendency for a correlation
between
and rotation, as well as between
and
.
We conclude that all single unevolved solar type stars with
are strong X-ray emitters and have high
.
Most likely,
they are all very young objects, either ZAMS or PMS.
Figure 8a shows the stars we believe to be PMS objects,
and Fig. 8b the stars we found to lie beyond the upper limit of the
Hipparcos MS (see Fig. 1). The PMS stars have ages in the 10-18 Myr interval
and masses in the
interval. HD 106506, which is
the fastest rotator in our sample, is also found to be the
youngest and the most massive star among the PMS objects. Among the evolved
stars HD 141710 and HD 74534, with masses of about
and
,
respectively, are the most massive stars in
our sample. Their estimated ages are <0.9 Gyr and <2 Gyr,
respectively. Among the most evolved single stars we find HD 150108
which has
and an
Gyr. With an age >10 Gyr, HD 20837 B (
)
is the oldest
in our sample of evolved stars.
We have tried to detect any correlation between the stellar mass and the residuals of the individual stars from the correlations discussed above. None was found, showing that the observed surface activity is much more strongly controlled by the rotation and magnetic field in the stellar envelope than by the interior structure, which is primarily dependent on the mass of the star.
The single stars of our sample show statistically higher
and activity
level than unbiased samples of nearby field stars with similar spectral types.
In particular, among the MS single stars of
our sample we find that about 10% of them have
an
higher than that of the
Pleiades, while about 64% have
between that of the
Pleiades and Hyades. Quite surprisingly, we also find
four rapidly-rotating single MS stars with very low
.
These four stars
could represent examples of stars which suffered peculiar angular momentum
histories, but it is not clear how to explain the observed characteristics.
These stars deserve further investigation.
The distribution of
for
single evolved stars is characterized by a large spread, as already
observed by other authors for evolved field stars. We find five evolved
stars with very low
that are still rather fast
rotators. Their low
cannot be explained by effects related to the
stellar rotational history, and these stars also deserve further detailed
investigation. We finally identify two new single PMS stars.
The MS binaries of our sample show a larger fraction of stars with
low
.
Only 8% of the binaries
have
higher than that of the Pleiades, while about 40%
have
between that of the Pleiades and Hyades.
This is not surprising, as high stellar rotation can be maintained by tidal
interaction also in old binaries which have had enough time to destroy
their Li. Thus, in the case of binaries our selection criteria allow both
young and old systems to be selected. We also find one new PMS binary star.
The behaviour of
as a function of
for single stars is characterized
by a large scattering in
for rotations lower than about 18
.
The
scattering
decreases with increasing rotation, and the single very fast rotators
are all stars with high
.
These results seem to indicate that another
stellar parameter plays a relevant role for the rotation-
connection in
single solar-type stars with
.
Although the physical cause of
the large scattering is not clear, it could reflect different initial rotation
rates. For both MS and evolved
binaries the spread of
remains large also at high rotation rates.
This is because even very old close binaries can maintain a high
through tidal synchronization.
We have also searched the ROSAT All Sky Survey catalog and derived the
PSRC X-ray luminosity for the stars detected. We find that 81 of the 110
possible stars were detected in the RASS (for X-ray data we cannot separate
the contribution of VBs and SBs components). Of the 25 stars in our sample
with
only three are not detected. Out of
the 85 stars with
,
59 are detected.
The well-established
correlation between
and
is thus clearly observed, in
particular for single stars. There also seems to be a kind
of saturation limit.
Our results, in agreement with recent findings from EUV and X-ray surveys, confirm the presence of young very active stars close to the Sun, although our sample has less extreme characteristics than the samples so far selected in the EUV and X-ray bands at the current flux limits (essentially the ROSAT all-sky survey limits).
Acknowledgements
Stellar activity research at INAF-Catania Astrophysical Observatory is supported by the Italian MIUR ("Ministry for Education, Universities and Research") and by the Regional Government of Sicily ("Regione Sicilia"). Stellar activity research at INAF-Brera Astronomical Observatory is supported by the Italian MIUR and by ASI ("Italian Space Agency"). Stellar activity research at the University Federal of Rio Grande do Norte is partially supported by the CNPq brazilian agency. The programmes on Galactic structure at Copenhagen University are financially supported by the Danish Natural Science Research Council and by Carlsberg Foundation. This research has made use of the SIMBAD database, operated at CDS, Strasbourg, France. Special thanks are due to the ESO staff for the collaboration and technical support during the observations, to Dr. Francesca D'Antona (INAF - Rome Astronomical Observatory) and Dr. Bjarne R. Jørgensen (Lund Observatory) for useful discussions. We also thank the referee, Dr. R. Garcia-Lopez, for his useful comments and Ms. Luigia Santagati for revision of the manuscript.