Beryllium in turnoff stars of NGC6397: early Galaxy spallation, cosmochronology and cluster formation

Abstract

We present the first detection of beryllium in two turnoff stars of the old, metal-poor globular cluster NGC 6397. The beryllium lines are clearly detected and we determine a mean beryllium abundance of log(Be/H)=-12.35 +/- 0.2. The beryllium abundance is very similar to that of field stars of similar Fe content. We interpret the beryllium abundance observed as the result of primary spallation of cosmic rays acting on a Galactic scale, showing that beryllium can be used as a powerful cosmochronometer for the first stellar generations. With this method, we estimate that the cluster formed 0.2-0.3 Gyr after the onset of star formation in the Galaxy, in excellent agreement with the age derived from main sequence fitting. From the same spectra we also find low O (noticeably different for the two stars) and high N abundances, suggesting that the original gas was enriched in CNO processed material. Our beryllium results, together with the N, O, and Li abundances, provide insights on the formation of this globular cluster, showing that any CNO processing of the gas must have occurred in the protocluster cloud before the formation of the stars we observe now. We encounter, however, difficulties in giving a fully consistent picture of the cluster formation, able to explain the complex overall abundance pattern. Comment: to appear in A&A

Full text

Page 1

arXiv:astro-ph/0407524v1 26 Jul 2004
Astronomy & Astrophysics manuscript no.
(will be inserted by hand later)
Beryllium in turnoff stars of NGC6397: early
Galaxy spallation, cosmochronology and cluster
formation⋆
L. Pasquini1, P. Bonifacio2, S. Randich3, D. Galli3, R. G. Gratton4
1European Southern Observatory, Garching bei M¨ unchen, Germany
2INAF–Osservatorio di Trieste, Trieste, Italy
3INAF–Osservatorio di Arcetri, Firenze, Italy
4INAF–Osservatorio di Padova, Padova, Italy
the date of receipt and acceptance should be inserted later
Abstract. We present the first detection of beryllium in two turnoff stars of the old,
metal-poor globular cluster NGC 6397. The beryllium lines are clearly detected
and we determine a mean beryllium abundance of log(Be/H)= −12.35±0.2. The
beryllium abundance is very similar to that of field stars of similar Fe content. We
interpret the beryllium abundance observed as the result of primary spallation of
cosmic rays acting on a Galactic scale, showing that beryllium can be used as a
powerful cosmochronometer for the first stellar generations. With this method,
we estimate that the cluster formed 0.2–0.3 Gyr after the onset of star formation
in the Galaxy, in excellent agreement with the age derived from main sequence
fitting. From the same spectra we also find low O (noticeably different for the two
stars) and high N abundances, suggesting that the original gas was enriched in
CNO processed material. Our beryllium results, together with the N, O, and Li
abundances, provide insights on the formation of this globular cluster, showing
that any CNO processing of the gas must have occurred in the protocluster cloud
before the formation of the stars we observe now. We encounter, however, diffi-
culties in giving a fully consistent picture of the cluster formation, able to explain
the complex overall abundance pattern.
Key words. Stars: abundances – stars: globular clusters – NGC6397 – stars: for-
mation, age, late-type

Page 2

2L. Pasquini et al.: Beryllium abundances in NGC 6397
1. Introduction
Beryllium has a unique origin: it is produced in the interstellar medium (ISM) by Galactic
cosmic rays (GCRs) through the spallation of carbon, oxygen and nitrogen nuclei (Reeves
et al. 1970); unlike lithium it is not produced during the Big Bang, unlike most other
elements is not produced in stars. It is well established that6Li, beryllium and B are
produced in the Galactic halo by two spallative processes: (i) collisions of accelerated
C,N,O nuclei in GCRs with ISM protons and α-particles (a primary process leading
to a linear dependence of Be on metallicity); (ii) collisions of energetic protons and α-
particles in GCRs with ISM C,N,O nuclei (a secondary process leading to a quadratic
dependence of beryllium on metallicity)1. The primary process (i) affects the Galaxy as
a whole and is predicted to dominate the production of beryllium in the metal-poor ISM
of the early Galaxy (Suzuki, Yoshii & Kajino 1999; Suzuki & Yoshii 2001). The spallative
origin of beryllium implies a power-law relation between the abundance of beryllium and
the metal abundance of stars, a trend confirmed observationally down to [Fe/H] ≃ −3
in metal-poor field halo stars (Boesgaard et al. 1999, hereafter B99). If, however, one is
interested in the earliest phases of Galactic evolution, metallicity is no longer a reliable age
indicator. As a consequence of the dispersed character of star formation in the Galactic
halo and the lack of efficient mixing in the gas, the chemical composition of the ISM in
the first stages of Galactic evolution was strongly affected by local enrichment produced
by individual Type II supernovae. Observations of very metal metal poor stars (Ryan,
Norris, & Beers 1996; McWilliam 1997) show in fact a large, probably intrinsic, spread of
elemental abundances, interpreted as an indication of inhomogeneous chemical evolution
of the early Galaxy (e.g. Travaglio, Galli, & Burkert 2001, Suzuki & Yoshii 2001). Even
if the most recent, high quality data show that part of this spread is due to the limited
quality of the previous data (Cayrel et al. 2004), a one-to-one relation of metallicity
with age in the first stellar generations is not expected. Conversely, beryllium, B and the
isotope6Li, being produced by energetic particles generated and transported globally on
a Galactic scale, are expected to show a much smaller spread of abundances than the
products of Type II supernovae (typically oxygen), making them ideal “cosmic clocks” for
dating the first stages of halo evolution (Suzuki et al. 2001, Beers et al. 2000). Globular
clusters represent ideal test cases for this new clock because they are an extremely old
stellar population for which independent, reliable age determinations are possible (Salaris
& Weiss 2002; Gratton et al. 2003, hereafter G03).
To test beryllium as a cosmic clock it is necessary to measure beryllium in stars which
can be dated independently. Stars in old globular clusters such as NGC 6397 are very
Send offprint requests to: lpasquin@eso.org
⋆Based on observations collected at the ESO VLT, Paranal Observatory, Chile
1In addition,
6,7Li can be produced also by fusion (α + α) reactions in the ISM, a process
important for the production of Li isotopes in a metal–poor gas

Page 3

L. Pasquini et al.: Beryllium abundances in NGC 63973
good candidates, because their ages can be determined in a reliable way and they have
been shown to be formed within ∼ 1 Gyr after the Big Bang (G03). However, the search
for beryllium in metal-poor stars has been limited so far to relatively bright field stars
(V ≤ 12.5), because the only available beryllium lines are the Be II resonance doublet
at 313.1 nm. This wavelength is very close to the atmospheric cut-off at 300 nm and
the terrestrial atmosphere heavily absorbs the incoming radiation, making observations
very challenging. In addition, the equivalent width of the Be lines is very small and the
spectral region crowded, so that high spectral resolution is necessary.
The high UV efficiency of UVES at the VLT telescope Kueyen has opened a new
possibility, allowing for the first time the detection of beryllium in two turn-off (TO) stars
in NGC 6397. It is crucial to reach the cluster TO because these stars have not depleted
beryllium in their atmospheres. Conversely, brighter subgiants in the same cluster show
clear evidence of Li dilution (Castilho et al. 2000, hereafter C00). Since this more fragile
element is at its original level in TO stars, then, a fortiori, beryllium is not expected to
be depleted in their atmospheres.
2. Basic properties of NGC6397
NGC6397 is one of the closest and best studied globular clusters. In particular it has been
the subject of several recent high resolution spectroscopic studies which have shown a
very good agreement on the cluster [Fe/H] abundance and on its homogeneity along
the color-magnitude diagram. The cluster Fe abundance determined spectroscopically by
several groups (see C00 and references therein), is in the range −2.2 < [Fe/H] < −1.8. The
most recent determinations agree on the value [Fe/H] ≃ −2.0 (C00; Th´ evenin et al. 2001,
hereafter T01; Gratton et al. 2001, hereafter G01), although these studies adopt different
techniques: LTE analysis in C00 and G01, non-LTE computations in T01. A detailed
chemical analysis of NGC6397 has been carried out by T01, who found abundances of
α- and Fe-peak elements in agreement with those of field stars of similar metallicity.
The Li abundance in TO stars of NGC6397 (log(N(Li) = 2.36 ± 0.05, Molaro &
Pasquini 1994; C00; Bonifacio et al. 2002, hereafter B02) is consistent with the Spite’s Li
plateau (Spite & Spite 1982). The Li abundance decreases with the star advancing on the
red-giant branch (RGB) because of dilution, and only upper limits can be obtained for
stars brighter than the RGB bump (C00). The average cluster O abundance appears to
be lower than that of field stars of similar metallicity ([O/Fe] ≃ 0.2±0.1 according to C00
and G01), but variations of O are claimed among the subgiants, possibly anticorrelated
with Na (Carretta et al. 2004). Thus, although NGC6397 is more homogeneous than most
globular clusters, it nevertheless shows signs of contamination by gas processed through
CNO cycling.

Page 4

4L. Pasquini et al.: Beryllium abundances in NGC 6397
Table 1. NGC6397 stars, their atmospheric parameters and abundances. The atmo-
spheric parameters are from B02, while the abundances are from this work and T01
StarTeff
logg [Fe/H] [N/H][O/H]log(Be/H)
A0228 6274 4.1
−2.05a
−0.74
−2.24
−12.27
A2111 6207 4.1
−2.01a
−0.74
−1.64
−12.43
HD218502
(a) from T01
6296b
4.13
−1.85b
−1.95:
−12.36
(b) from Alonso et al. 1999
3. Sample selection and observations
We selected from the T01 sample the two brightest stars to allow the challenging beryl-
lium observations. The selected stars, together with their atmospheric parameters (B02)
and chemical abundances (T01), are listed in Table 1.
The observations were carried out in service mode at the VLT observatory in several
runs during summer 2003 with the UVES spectrograph (Dekker et al. 2000). We used
a 1′′slit providing a resolving power of 45,000. The blue CCD was binned along the
spatial direction to minimize the CCD read out noise. We used the dichroic to obtain, in
addition to the UV data, spectra with the red 860 nm setup, containing the O i 777 nm
triplet which we used to derive the O abundances. A total of 8 exposures of 90 minutes
each were obtained per star. The observations were reduced with the UVES pipeline
(Ballester et al. 2000). The reduced spectra were averaged leading to a S/N ratio in the
beryllium region of 8 and 15 per pixel for star A2111 and A228 respectively. The spectra
in the beryllium region, together with our synthetic best fits (cfr. next section) are shown
in Figure 1. In addition to the cluster stars, we obtained high S/N (∼ 130) spectra of
the bright reference star HD 218502, whose atmosperic parameters are very close to the
ones of the cluster stars (cfr. Table 1).
4. Data analysis: Stellar parameters and determination of abundances
Table 1 lists the adopted atmospheric stellar parameter (B02) and the abundance values
determined in this work.
A full synthesis of the region around the beryllium doublet has been performed using
the line list given in Table 2. In this table the ions are identified by their code, according to
Kurucz’s convention (Kurucz 1993). Almost all lines have been extracted by a version of
the Kurucz data base updated by F. Castelli (private communication). The only exception
are the OH lines, for which gf values have been computed from lifetimes of Goldman &
Gillis (1981). We used our Linux version (Sbordone et al. 2004) of the SYNTHE code
(Kurucz 1993). For each star the best abundance was found by a χ2fit to the whole
feature. The best fits, shown in Fig. 1, correspond to beryllium abundances of log(Be/H)=

Page 5

L. Pasquini et al.: Beryllium abundances in NGC 63975
Table 2. Line list used for the spectrum synthesis in the region of the Be II lines.
Wavelengthion log gf
ref. Pred.E
(nm)(cm−1)
313.0027 107.00 -2.799K 9182.250
313.005640.00 -0.700 EM4186.110
313.0167 22.00-0.468K47913.551
313.0190107.00 -3.085K* 9337.492
313.0191107.00 -2.657K*10779.396
313.0195 26.00 -3.603 K9428819.951
313.020225.01 -3.714 K8838720.020
313.021226.00-2.840 K94 58812.012
313.0254106.00 -1.180K 4205.830
313.025723.01 -0.290NBS 2808.720
313.0281108.00 -1.931GG 0.000
313.0290 106.00 -1.136K 4206.520
313.034058.01-0.152MC4266.397
313.0353 27.01-3.333 K8824074.600
313.0370106.00 -1.555K 267.550
313.0377 22.00-1.594K 11531.759
313.0419106.00-2.464K 14274.020
313.0420 4.01-0.168 BIE0.000
313.043925.00 -2.518K88 30419.609
313.0476 106.00-2.450K 14274.970
313.0519107.00 -2.701K*11114.699
313.0549 25.01 -1.146K88 52373.180
313.0550 107.00-3.586K 6932.594
313.0562 26.01-5.213 K8830388.543
313.056924.01 -2.457K88 42986.621
313.0570 108.00-1.559 GG0.000
313.0592 24.00 -1.968 K88 28682.207
313.0636 106.00-1.180K*4188.995
313.063725.00 -1.008K8834423.270
313.0648 106.00-1.447K 270.590
313.0674 106.00-1.136K*4189.812
313.0726 107.00 -2.851K*9508.673
313.078041.01 +0.410 HLB3542.500
313.0782 106.00 -1.555K* 265.368
313.080922.01 -1.140 MFW94.100
313.081364.01 -0.083 MC9328.864
313.0823 107.00-2.670K*11115.135
313.087158.01 +0.481 CC8789.380
313.0906 106.00 -0.503K*18848.234
313.0910 107.00-3.018K*9338.252
313.0928 106.00-3.136K 17.770
313.0932 108.00-3.358 GG0.000
313.101525.01 -1.217K8849291.309
313.103725.00 -1.725 K8830425.711
313.1058 106.00-1.447K* 268.362
313.10654.01 -0.468 BIE0.000
313.107090.01 -1.559MC0.000
313.1074106.00 -0.514K* 18849.781
313.110940.00 -0.400 CB4196.850
313.111026.00 -5.171 K9424574.652
313.1115 107.00-2.627K*11113.809
313.111676.00 +0.050 CB’14848.050
313.1198 107.00-3.552K6933.030