1: %\documentclass{aastex}
2: \documentclass[12pt,preprint]{aastex}
3: %\documentclass[preprint2]{aastex}
4: %\usepackage{natbib}
5: \shorttitle{Abundances on the Main Sequence of Omega Centauri}
6: \shortauthors{Stanford et al.}
7: \newcommand{\wcen}{$\omega$ Cen}
8: \newcommand{\feh}{[Fe/H]}
9: \newcommand{\cah}{[Ca/H]}
10: \newcommand{\eq}{$\approx$}
11: \newcommand{\hs}{\hspace*{12pt} }
12: \newcommand{\tuc}{47~Tuc}
13:
14: \begin{document}
15:
16: \title{Abundances on the Main Sequence of Omega Centauri}
17:
18: \author{Laura M. Stanford, G. S. Da
19: Costa and John E. Norris}
20: \affil{Research School of Astronomy and Astrophysics, Australian
21: National University, Weston, ACT, 2611, Australia}
22: \email{stanford, gdc, jen@mso.anu.edu.au}
23: \and
24:
25: \author{Russell D. Cannon} \affil{Anglo-Australian
26: Observatory, P.O. Box 296, Epping, NSW, 2121, Australia}
27: \email{rdc@aaoepp.gov.au}
28:
29:
30: \begin{abstract}
31: Abundance ratios of carbon, nitrogen and strontium relative to iron,
32: calculated using spectrum synthesis techniques, are given for a sample
33: of main sequence and turnoff stars that belong to the globular cluster
34: $\omega$ Centauri. The variations of carbon, nitrogen and/or
35: strontium show several different abundance patterns as a function of
36: [Fe/H]. The source of the enhancements/depletions in carbon, nitrogen
37: and/or strontium may be enrichment from asymptotic giant branch stars
38: of low (1--3 M$_{\odot}$) and intermediate (3--8 M$_\odot$) mass.
39: Massive rotating stars which produce excess nitrogen without carbon
40: and oxygen overabundances may also play a role. These abundances
41: enable different contributors to be considered and incorporated into
42: the evolutionary picture of $\omega$ Cen.
43:
44: \end{abstract}
45:
46: \keywords{globular clusters: general ---
47: globular clusters: individual ($\omega$ Centauri)}
48:
49: \section{Introduction} \label{intro}
50: \defcitealias{nd95}{ND95}
51:
52: The globular cluster $\omega$ Centauri is known to have
53: characteristics not shared by other clusters. Its unusual nature was
54: first noted by \citet{woo66} and \citet{cs73}, who demonstrated that
55: its red giant branch (RGB) has an intrinsic color spread. Early
56: abundance studies of member stars, obtained spectroscopically, have
57: shown a range in metallicity \citep{fr75, le77, bde78, cb86, fss88,
58: pn89}. In the last decade detailed spectroscopic and photometric work
59: on {\wcen} has revealed the large extent of the abundance range within
60: the system.
61:
62: The cluster metallicity distribution has been studied in more detail
63: recently, using spectroscopic analysis of the Ca II triplet lines
64: \citep{nfm96, sk96}, photometry \citep{lee99, pan00, hr00, sol05a} and
65: spectroscopic analysis of the Ca II K line \citep{sta06a}. These
66: studies found few, if any, stars with {\feh}$<$--2.0. The
67: distribution then rises to a peak at {\feh}=--1.7 with a long tail to
68: higher metallicities, up to {\feh}$\approx$--0.4. The distribution
69: also has a second peak at {\feh}=--1.2 \citep{nfm96}. Three main
70: populations within the cluster have been identified \citep{nfm96,
71: sk96,pan02}. The bulk of the cluster stars ($\sim$70\%) has
72: metallicities at [Fe/H]$\approx$--1.7, the metal-intermediate
73: population has [Fe/H]$\approx$--1.2 and represents roughly 20\% of the
74: stars, and the remaining 5\% of the stars are metal-rich at
75: [Fe/H]$\approx$--0.6. \citet{sol05a} found the RGB to be discrete in
76: nature and identified up to five distinct populations, three of which
77: are within the metal-intermediate population described above.
78: Elemental studies have shown a range in abundances within the cluster
79: members. Carbon, nitrogen and oxygen exhibit large scatters for a
80: given {\feh} (\citealt{per80, bw93}; \citealt[hereafter ND95]{nd95}).
81: The $\alpha$-elements (Mg, Si, Ca and Ti) have constant values at a
82: given [Fe/H] for most metallicities (\citealt{bw93, scl95};
83: \citetalias{nd95}; \citealt{smi00}) but [$\alpha$/Fe] decreases at
84: higher metallicities ({\feh} $>$ --1.0) \citep{pan02}. As the
85: $\alpha$ element abundances are constant with iron below
86: {\feh}\eq--1.0, they are produced by the same source, most likely
87: supernovae Type II. At higher abundances, decreasing [$\alpha$/Fe] is
88: indicative of supernovae Type Ia contributions.
89:
90:
91: The heavy neutron-capture elements (e.g. Y, Zr, Ba, La, Pr, Nd) give
92: information on enrichment sources, shedding light on whether a slow or
93: rapid neutron-capture process occurred (or both). \citetalias{nd95}
94: and \citet{scl95} found the s-process element abundance ratios with
95: respect to iron increase with increasing {\feh} and then flatten above
96: {\feh}=--1.2. These results are in contrast to normal globular
97: clusters, and indicate that the stellar winds from asymptotic giant
98: branch (AGB) stars (sources of s-process elements) were involved with
99: the enrichment process of {\wcen}. \citetalias{nd95} concluded also
100: that there were no correlations between the abundance ratios with
101: respect to iron for the s-process elements and those for C, N, O, Na
102: and Al. Eu is an r-process element, and abundances of this element
103: are found to be lower in {\wcen} stars than in the field
104: \citep{scl95}, and ratios of Ba/La show that the enrichment process is
105: dominated by the s-process \citepalias{nd95}.
106:
107:
108:
109: Another piece in the {\wcen} puzzle is the double main sequence seen
110: in deep photometry of the cluster \citep{and02, bed04}. This is
111: thought to be the result of helium variations of the order of
112: $\Delta$Y$\sim$0.12~dex between two populations \citep{nor04, pio05}.
113: The source of the large amount of additional helium required may be
114: AGB stars with masses larger than 6M$_{\odot}$, stellar winds
115: associated with early evolutionary phases of massive stars, or perhaps
116: Type II supernovae \citep{nor04,pio05, mm06}. \citet{bn06} suggest
117: that the helium enhanced population formed from gas ejected by the
118: stellar populations that surrounded {\wcen} when it was the nucleus of
119: a dwarf galaxy. \citet{mm06} have argued that in models with low
120: metallicity, massive stars with high rotation can produce the large
121: amounts of helium required to explain the blue main sequence (bMS),
122: depending on the slope of the initial mass function. These models
123: also produce a large excess of nitrogen but little carbon, consistent
124: with the phenomena observed in spectra of the stars in the bMS
125: \citep{pio05}.
126:
127:
128: Variations of the CNO group, Na, Mg and Al abundances have been found
129: in most {\it normal} globular clusters on the RGB to different
130: degrees. In the last few decades, studies of main sequence turn off
131: (MSTO) stars in a number of globular clusters have shown that
132: variations exist in C and N in this region of the CMD as well. These
133: clusters include 47 Tuc \citep{can98, bri04}, NGC~6752 \citep{ss91,
134: gra01a}, M71 \citep{coh99, bc01, rc02}, M5 \citep{cbs02} and M13
135: \citep{bcs04}. Anticorrelations between Na and O have been shown to
136: exist on the main sequence (MS) in 47 Tuc \citep{car04}, NGC~6397 and
137: NGC~6752 \citep{gra01b}. These latter two clusters also have an
138: anticorrelation between Mg and Al \citep{gra01b}.
139:
140: There are three possibilities for the origin of the abundance
141: variations in {\wcen} and other globular clusters: internal
142: processing, accretion of matter onto surface layers, and primordial
143: formation. Firstly, internal processing within the individual stars
144: themselves, with the products then being mixed to the surface layers,
145: may be the source of the abundance variations. This may account, in
146: part, for the variations in the RGB stars in some clusters, as the
147: level of C depletion on the RGB increases with decreasing magnitude
148: (e.g. \citealt{lan86}, M92). However, mixing of the internal layers
149: does not occur in the current MS stars (e.g. \citealt{dd82}).
150: Therefore, internal processing cannot explain all the variations seen
151: in globular clusters. The second option is that the enriched objects
152: accreted matter onto their surface layers from the stellar winds of
153: AGB stars. The stars may have been a part of a binary system or in
154: the vicinity of AGB stars. The accreted material will be on the
155: surface layers only, and therefore, in a comparison of MS with RGB
156: abundance ratios, one should generally find higher abundances for the
157: unevolved stars compared with the evolved ones as increased convective
158: mixing as the stars evolve up the RGB will dilute the accreted
159: material. Primordial variations account for the third and final
160: possibility. Here, the stars formed from material that had been
161: previously enriched from sources such as winds and ejecta of AGB
162: stars, massive rotating stars and/or supernovae.
163:
164:
165: Most spectroscopic studies have concentrated on the red giant branch
166: (RGB) stars in {\wcen} because of their greater brightness. We have
167: obtained spectra of a sample of turnoff stars to examine the
168: relationship between age and metallicity in the cluster
169: \citep{sta06a}, and to determine abundances of several elements and
170: compare them with those found on the RGB. This will give further
171: insights into the evolutionary history of {\wcen} and enable us to
172: distinguish between surface enhancement and primordial enrichment
173: scenarios within the cluster.
174:
175: The present paper describes the analysis of these spectra to determine
176: abundances of C, N and strontium, relative to iron, for the MS and
177: turnoff (TO) sample and an attempt to untangle the complex
178: evolutionary history of {\wcen}. Early results of this work are given
179: in \citet{dac05}. In \S~2 we describe the observations and reduction
180: techniques. \S~3 outlines the derivation of metallicities and ages for
181: the sample and discusses the metallicity populations. We describe the
182: techniques used to determine C, N and Sr abundances in \S~4. Finally,
183: \S~5 summarizes the results and discusses a chemical evolutionary
184: history for the cluster.
185:
186:
187:
188: \section{Observations and Reduction} \label{OR_sect}
189:
190: The observations and reduction of our data are described in detail in
191: \citet{sta06a} to which we refer the reader. Suffice it here to say
192: the photometry of the cluster was obtained with the 1m
193: telescope/Tektronix CCD combination at Siding Spring Observatory in
194: the V and B bands. From these data a CMD, shown in Figure \ref{cmd1},
195: was constructed for objects within an annulus 15--25 arcminutes from
196: the cluster center. Two regions were defined on the upper MS with a
197: view to determining the metallicity and age range for the cluster.
198: Stars in each region were observed using the Two degree Field (2dF)
199: Multiobject spectrograph \citep{lew02} on the Anglo-Australian
200: Telescope . This spectrograph has the capability of simultaneously
201: observing up to 400 objects using a fibre fed system. The first
202: sample was observed in May 1998 and April 1999 (hereafter 98/99
203: sample), and the second in March 2002 (hereafter 2002 sample). Figure
204: \ref{cmd1} shows the two boxes from which candidates were selected.
205: Spectra were obtained with 1200 line/mm gratings employed in the
206: spectrographs and covered the wavelength range
207: $\lambda\lambda$3800--4600\AA. They have a resolution of
208: $\sim$2.5\AA\ FWHM and signal-to-noise (S/N) of 30-50. A final sample
209: of 424 radial velocity selected members was found. The members are
210: shown as large dots in Figure \ref{cmd1}. The small dots represent the
211: photometry of all the remaining objects that do not have membership
212: information.
213:
214: Data for stars on the MS were also obtained for three other globular
215: clusters --- NGC 6397, NGC 6752 and 47 Tuc. These clusters were
216: observed in a similar manner to $\omega$ Cen, and membership was
217: determined based on radial velocities and metallicities. The data
218: obtained for the clusters were used to test the reliability of the
219: metallicity calibration, and for comparison of abundances.
220:
221: \section{Metallicities and Populations} \label{met_sect}
222:
223: The process for calculating metallicities has been described in detail
224: in \citet{sta06a}. To summarize, the metallicities were calculated
225: using a combination of two methods (see \citet{bee99} for details of
226: these methods briefly described below). The first uses the Ca
227: \textsc{ii} K line strength and color of an object to assign a
228: metallicity. Due to the saturation of the Ca \textsc{ii} K line these
229: metallicities become uncertain above [Fe/H]$\approx$--1.0. The second
230: method utilizes the weak metal lines in the spectrum. This method is
231: reliant on spectra having good signal-to-noise ($>$30) as the noise
232: can become confused with the weak metal lines. As metal-poor stars
233: have few and very weak metal lines, this method is unreliable at lower
234: metallicities ([Fe/H]$<$--2.0) and the abundance from the Ca
235: \textsc{ii} K line strength was adopted. Between [Fe/H]=--2.0 and
236: [Fe/H]=--1.0 a weighted mean of the two values was used and above
237: \feh=--1.0 metallicities from the second method were adopted.
238: Metallicities were calculated for the {\wcen} sample, and for the
239: calibrating clusters NGC~6397, NGC~6752 and {\tuc}. An age was then
240: calculated for each {\wcen} star based on its position on the CMD and
241: metallicity using theoretical isochrones (cf. \citet{sta06a}).
242:
243: The metallicities are used here to divide the samples into three
244: groups, approximate the three main populations in {\wcen}. The
245: metal-poor population (1st Pop.) contains stars with [Fe/H]$<$--1.5.
246: The metal-intermediate population (2nd Pop.) contains objects that
247: have --1.5$\leq$[Fe/H]$<$--1.1. The third population (3rd Pop.)
248: contains the stars that have the highest metallicities in our sample
249: with [Fe/H]$\geq$--1.1. The histograms of metallicities for our two
250: samples (98/99 and 2002) are shown in Figure \ref{hist}. The figure
251: also shows the cuts in metallicity that make up the three populations,
252: with 255/424 in the 1st Population, 144/424 in the second and 25/424
253: in the metal-rich population. The calibrating clusters were used to
254: compare abundances at similar metallicities. The values for these
255: clusters come from \citet{har96}. NGC~6397, with [Fe/H]=--1.96, was
256: compared with the first population. NGC~6752 is borderline between
257: the 1st Pop. and 2nd Pop. with [Fe/H]=--1.56, and can be used for
258: comparison purposes with both of these populations. {\tuc}, with
259: [Fe/H]=--0.76, was used to compare with the 3rd Pop.
260:
261: The photometry for the three {\wcen} populations is plotted in Figure
262: \ref{cmd2}. An isochrone (Y$^{2}$, \citealt{yi01}) was fitted to each
263: population on the CMD with parameters depending on the mean
264: metallicity and age for the population. These isochrones are
265: overplotted in Figure \ref{cmd2}. Ten points were chosen along the
266: isochrone that encompassed our sample in luminosity and color and the
267: corresponding temperature and gravity were used in the abundance
268: analysis. For the first and second populations (at {\feh}=--1.7
269: and~--1.2 respectively) the [$\alpha$/Fe] ratio was taken to be 0.3
270: (\citetalias{nd95}; \citealt{scl95, smi00}). For the most metal-rich
271: population with a mean [Fe/H]=--0.8, the alpha enhancement was taken
272: to be [$\alpha$/Fe]=0.18.
273:
274:
275:
276: \subsection{Indices}
277:
278: Indices were measured for the G band (CH) at $\sim$4300{\AA}, violet
279: CN band at 3883{\AA}, the Sr \textsc{ii} 4077{\AA} and 4215{\AA}
280: lines, and Ba \textsc{ii} 4554{\AA} for the {\wcen} members and
281: calibrating cluster stars. The weaker Sr \textsc{ii} 4215{\AA} line
282: was used as confirmation of the Sr abundance, which was determined
283: primarily from the Sr\textsc{ii} 4077{\AA} feature. The bandpasses
284: for CH, Sr and Ba are given in Table \ref{tbl-1} and the indices are
285: defined following \citet{bee99}. The CN index is S3839 as defined by
286: \citet{nor81}. The {\wcen} data are plotted as a function of
287: metallicity in Figure \ref{indfeh}, and are split into main-sequence
288: (V$>$18) and SGB (V$\leq$18) samples. The corresponding indices
289: determined for NGC~6397, NGC~6752 and {\tuc} are represented by
290: boxplots showing the area in [Fe/H]--index space they occupy for
291: comparison with the MS {\wcen} stars, as we do not have data for the
292: SGB stars in these clusters. The center line within the rectangle for
293: each boxplot represents the median for each dataset. The lower and
294: upper sides of the box represent the first and third quartiles while
295: the vertical lines extending from the upper and lower edges of the box
296: represent the upper and lower limits of the data. The error bars in
297: the figure represent the standard error of the mean of the indices for
298: individual observations, and the derived error in metallicity.
299:
300: Studying the two CH plots, one can see that the SGB group has, in
301: general, higher indices than on the MS one. This is due to the cooler
302: temperatures on the SGB than the MS. At lower metallicities on the
303: MS, there are two members that have unusually high C indices. These
304: fall well outside the ranges in [Fe/H] and the CH index that is
305: occupied by the calibrating clusters. Outliers can also be identified
306: on the SGB. The S3839 index shows several objects which appear to
307: have high abundances of carbon and/or nitrogen in both the MS and SGB
308: groups. The strontium index plot shows a large scatter, with several
309: outliers in each. For the SGB group, a least-squares fit to the data
310: was performed, shown by the solid line. This line indicates there are
311: several objects with considerably higher Sr indices than the main
312: population. There is little spread in the Ba index, most likely due
313: to the low sensitivity of our data to the Ba abundance.
314:
315: The CN/CH bimodality known to exist on the RGB of {\tuc} is also found
316: on the MS \citep{can98, dac04}. CN strong and CN weak stars exist in
317: {\wcen} on the RGB (\citealt{nb75, per80, cb86}; \citetalias{nd95}),
318: but have not yet been seen on the MS in {\wcen}. Figure \ref{cnch}
319: plots the CN index versus the CH one for the {\wcen} data, and is
320: split into the three metallicity bins as well as the MS and SGB groups
321: (left and right columns respectively). Also plotted, as open circles,
322: are the {\tuc} data showing the bimodality within this cluster for
323: comparison with the metal-rich, MS population in {\wcen}. The solid
324: line is a fit to the {\tuc} anticorrelation and plotted in each panel
325: as a reference. The dotted line is the one used to define the CN
326: weak/CH strong, and vice versa, groups.
327:
328:
329:
330:
331: \section{First Pass Abundances} \label{abund_sect}
332:
333: To obtain an initial estimate of the abundance ratios of stars in our
334: sample, in particular for the stars that have enhancements in carbon,
335: nitrogen and/or strontium, we compared indices obtained from synthetic
336: spectra with those from the observed data. As described in
337: \S\ref{met_sect}, the {\wcen} sample was divided by metallicity into
338: three populations, as shown in Figure \ref{cmd2}. Isochrones with the
339: mean metallicity and ages of each population were fitted to the data.
340: Discrete points along these isochrones gave a series of stellar
341: parameters (temperature and gravity) that are representative of the
342: observed data for each population.
343:
344:
345: Kurucz models (1993) were used with atomic line lists from R. A. Bell
346: (2000, private communication) and Kurucz molecular lines lists to
347: generate synthetic spectra to compare with our observations. The
348: spectrum synthesis code was originally developed by \citet{cn78}. A
349: synthetic spectrum with appropriate stellar parameters was compared
350: with an atlas of the Sun \citep{beckers}. Using an adopted solar C
351: abundance of log(N/N$_{tot}$)=--3.36, the CH feature showed a small
352: difference between the spectrum of the Sun, and the synthetic
353: spectrum, and as a result the log(gf) values were reduced by
354: 0.5~dex. Similarly, for an adopted N abundance of
355: log(N/N$_{tot}$)=--4.04, the synthetic violet CN feature was too
356: strong with respect to the solar spectrum and the log(gf) values were
357: reduced by 0.25~dex. The log(gf) values for the blue CN feature
358: ($\lambda$4215{\AA}) were reduced by 0.10~dex in a similar manner. The
359: Sr {\sc ii} 4077 and 4215 lines and Ba {\sc ii} 4554 line were also
360: compared and found to be within acceptable agreement. An observed
361: spectrum of Arcturus ($\alpha$ Boo) was then compared with a synthetic
362: spectrum with appropriate stellar parameters \citepalias{nd95} and was
363: used as an independent check of the adjusted values. It was found to
364: provide acceptable agreement with abundance studies of this star.
365: Table \ref{tbl-2} lists the gf values and solar abundances used here
366: for Sr and Ba.
367:
368:
369: The derived abundance of C is influenced by the amount of oxygen
370: within the star as the CO molecule forms preferentially over the CH
371: one. However, for our range of stellar parameters, changing the O
372: abundance was demonstrated to have no effect on our calculations.
373: This was tested by generating a series of synthetic spectra at
374: metallicities that spanned the {\wcen} metallicity range with several
375: temperatures and gravities typical of the observed sample. The oxygen
376: abundance was varied for each one (between [O/Fe]=0.0 and 0.5 dex),
377: and the strength of the G band was compared between synthetic spectra
378: with the same stellar parameters. It was found that oxygen did not
379: play a significant role in the abundance determined from the CH
380: feature for any of the stellar parameters considered. This test was
381: also conducted for the CN feature and again no noticeable difference
382: was found with varying oxygen abundance. We have assumed that oxygen
383: follows the general $\alpha$-enhancement trend. That is, [O/Fe]=0.3
384: for objects with metallicities [Fe/H]$\leq$--1.0, and [O/Fe]=+0.18 for
385: the higher metallicity stars. \citetalias{nd95} show that there is a
386: number of stars with oxygen deficiencies (values of [O/Fe]$<$0.0) for
387: RGB stars. This indicates that the ON cycle may have occurred in
388: these objects. Although the assumed [O/Fe] abundance does not play a
389: significant role in the abundances determined, it would be beneficial
390: to determine the oxygen abundances for individual stars in
391: future studies.
392:
393: The stellar parameters derived from the isochrones were used to
394: generate synthetic spectra for ten carbon abundances at each point
395: along an isochrone, for each population. G band indices were
396: determined for each of these synthetic spectra and plotted along with
397: the indices from the observed spectra for each population. This
398: enabled a mean carbon abundance to be obtained for each population,
399: which was then used when another set of synthetic spectra were
400: generated with the nitrogen abundance being the varying parameter.
401: The S3839 index was measured in these synthetic spectra and the
402: results for the synthetic spectra were again plotted with the observed
403: data to obtain a mean nitrogen abundance for each population. This
404: procedure was followed again, but with varying strontium abundance and
405: the strontium indices plotted for the synthetic and observed data.
406:
407: Interpolation between the synthetic isoindex lines for each element
408: and population enabled a first estimate at the abundances of carbon,
409: nitrogen and strontium for each object. From these abundances the
410: objects with possible enhancements (or in some cases depletions) in
411: carbon, nitrogen and strontium could be identified for further in
412: depth analysis. The synthetic spectra give an initial estimate of the
413: abundances, but should be viewed with caution since there is a
414: 0.4--0.6 dex range in {\feh} for the member stars of each population
415: but the isoindex lines are for a fixed abundance. Also, some stars
416: have temperatures and gravities that deviate from those used for the
417: isoindex lines. The larger the difference in {\feh}, temperature
418: and/or gravity from the isoindex value, the larger the uncertainty in
419: the initial estimate.
420:
421:
422: The C abundance for the synthetic spectra was varied between
423: [C/Fe]=--0.9 and +1.8. The G band index was measured for each
424: spectrum and plotted with the observed data, shown in the left column
425: of Figure \ref{c1st}. The diamonds represent those stars for which
426: individual abundances were subsequently determined (see
427: \S\ref{iaa_sect}). The synthetic data for each abundance are plotted
428: as solid lines. A large spread in C abundance can be seen within both
429: the MS stars and the SGB stars and there are two MS stars that have
430: possible enhanced features. The C abundance for the $\omega$ Cen
431: stars were calculated by interpolation in luminosity and G band
432: strength to give an initial approximation to the abundance of that
433: element.
434:
435:
436:
437:
438: The right column of Figure \ref{c1st} shows the three calibrating
439: globular clusters (NGC~6397, NGC~6752 and {\tuc}). NGC~6397 and
440: NGC~6752 bracket the metallicity range of the 1st Pop., and their
441: results as normal globular clusters can be compared to those found for
442: {\wcen} in order to determine which stars are {\it abnormal} on the MS
443: of the latter. The error bars in each panel represent errors in
444: measured index and magnitude. They do not include the uncertainty in
445: the inferred abundance induced by an object's distance in color,
446: magnitude and metallicity from the isochrone used to generate the
447: isoindex lines.
448:
449: The normal globular clusters show no significant spread (1$\sigma$) in
450: the CH index, while the first population of {\wcen}, on the other
451: hand, shows several objects that are outliers from the bulk of the
452: population and a 2$\sigma$ spread on the MS and a 4$\sigma$ spread on
453: the SGB. NGC~6752 can also be compared with the 2nd pop. Again the
454: small spread in the normal GC is in contrast to that seen in {\wcen}.
455: The 3rd Pop. can be compared with the results from {\tuc}, for which
456: the indices have been split according to the CN--CH bimodality for
457: clarity. On close inspection, one sees that the {\wcen} results have
458: a similar range in indices to {\tuc}.
459:
460: The mean C abundance for the bulk of each population was determined
461: and then used when calculating synthetic spectra with varying
462: nitrogen. Again, the CN index was determined for each synthetic
463: spectrum and plotted against the observed data, shown in the left
464: column of Figure \ref{n1st}. Once more, the error bars in each panel
465: represent errors in measured index and magnitude. The diamonds in this
466: plot indicate those stars that had individual N abundances determined
467: (see \S\ref{iaa_sect}). As with the plot of the CH indices, the top
468: panel shows the most metal-poor stars, the middle one the intermediate
469: metallicity population, and the bottom panel the most metal-rich
470: stars. Inspection of the top panel shows the synthetic lines for CN
471: are closely spaced on the MS, showing the low sensitivity of the CN
472: feature at these abundances and temperatures. Consequently, finding
473: any objects with high N abundances is difficult, and any spread is
474: comparable to the measurement errors (1$\sigma$). Determining N
475: abundances becomes a slightly easier task on the SGB and the presence
476: of several outliers is clear. The outliers are more pronounced in the
477: 2nd Pop. and there is a large range in apparent N abundance in the 3rd
478: population with 4$\sigma$ and 9$\sigma$ spreads on the MS and SGB,
479: respectively. Initial estimates of the N abundance for each {\wcen}
480: star were calculated, and all objects with a first pass N abundance,
481: [N/Fe], greater than 1.2~dex were included for further study, although
482: this includes a fair number of objects in the metal-poor, MS category
483: that were subsequently shown to have normal N abundances due to the
484: low sensitivity of the feature.
485:
486:
487:
488: The right column of Figure \ref{n1st} shows the results for the
489: calibrating globular clusters. Both NGC~6397 and NGC~6752 show little
490: scatter in their indices, as would be expected when the synthetic
491: indices are taken into account. There is very little sensitivity in
492: this feature at these metallicities, temperatures and gravities.
493: {\wcen}, as with the G band, shows several stars that have CN indices
494: that are higher than the mean, even with the low sensitivity in this
495: area. The indices for {\tuc} are again split into the CN--CH
496: bimodality. Two sets of synthetic lines were plotted for these data,
497: with different mean C abundance. We see a larger range within the
498: {\wcen} data than for the {\tuc} indices.
499:
500: Sr and Ba were analyzed simultaneously, although the Ba index was too
501: weak to obtain meaningful results. The synthetic Sr indices are
502: plotted in the left column of Figure \ref{s1st} along with the {\wcen}
503: data, where the error bars in each panel represent errors in measured
504: index and magnitude. A large spread of the Sr index is apparent in
505: Figure \ref{s1st}, and is mainly due to the low S/N and intermediate
506: resolution of our spectra. The spread in the measured index is
507: comparable to the measurement error in the first two populations. The
508: third population shows a 3$\sigma$ spread in comparison with the
509: measured errors. Despite this, we are able to identify objects that
510: may have large Sr enhancements, and use these for individual spectrum
511: synthesis analysis. The diamonds represent those objects that had
512: individual Sr abundances calculated, which are discussed in
513: \S\ref{iaa_sect}.
514:
515:
516:
517:
518: The normal globular clusters all show no Sr enhancement and any
519: scatter is comparable (of order 1$\sigma$) to the measurement errors
520: of the index. {\tuc} shows several objects that may be considered to
521: have higher Sr abundance ratios, but closer inspection reveals that
522: these objects have higher indices due to low S/N. {\wcen}, in all
523: cases, shows larger scatter than the normal globular clusters, with
524: several stars having higher indices being objects of particular
525: interest. Inspection of spectra shows these are not the result of
526: lower S/N.
527:
528:
529: The choice of isochrones that were used for each of the three
530: populations in the first pass abundance analysis was determined from
531: the mean metallicity and ages of the respective populations. The
532: impact of a different choice of parameters for the three sets of
533: isochrones was evaluated. The first population was assigned an
534: isochrone with [Fe/H]=--1.9 and age=15 Gyr (as against --1.7 and 13.5
535: Gyr), the second population an isochrone with [Fe/H]=--1.2, age=11 Gyr
536: *cf. --1.4 and 13.0 Gyr) and the third an isochrone with [Fe/H]=--0.6
537: and age=9 Gyr (cf. --0.9 and 12 Gyr). Most stars showed differences
538: in first pass carbon abundances of less than 0.5~dex. Strontium
539: abundances were calculated with differences less than 0.8~dex.
540: Differences in nitrogen values were affected the greatest of the three
541: elements, with differences ranging up to $\pm$2.0~dex in the most
542: extreme case, and $\pm$1.0 on average. In practice, the result for
543: the nitrogen abundances was not significant as objects with enhanced N
544: selected for further analysis were determined using the S3839 index
545: rather than the derived N abundance, as described in the following
546: section. The use of a different distance modulus and reddening were
547: also investigated and the differences in derived abundance was found
548: to be less than 0.2~dex for carbon and nitrogen, and less than 0.5~dex
549: for strontium. These differences are all within the error margins
550: placed on the selection of stars with possible enhancements in C, N or
551: Sr.
552:
553:
554:
555:
556: \section{Individual Abundance Analysis} \label{iaa_sect}
557:
558: \subsection{Abundance Determination of [C/Fe], [N/Fe] and [Sr/Fe]}
559:
560: Following the analysis in \S\ref{abund_sect}, the subsample of stars
561: with possibly enhanced C, N or Sr abundances were individually
562: analyzed using spectrum synthesis techniques. These objects were
563: chosen according to their first pass abundances. For carbon, if a
564: star's interpolated abundance was [C/Fe]$>$0.3 it was included. The
565: cutoff for Sr was [Sr/Fe]$>$0.8. The CN index was more complicated
566: due to the weakness of the feature and hence its low sensitivity to C
567: and N. Any interpolated abundance had very large error bars due to
568: the small separation between isoindex lines. Therefore, these objects
569: were chosen based on the S3839 index rather than interpolated
570: abundance and the cutoff imposed was S3839$>$--0.05. All objects in
571: the 3rd Pop. were included for individual analysis.
572:
573: Stars with [C/Fe]$<$--0.75 from the first pass abundance estimation
574: were also included from the first two populations. However it was
575: found from the comparisons of observed and synthetic spectra for
576: individual stars that in most instances there was not sufficient
577: sensitivity in the data to reliably measure depletions. The low
578: sensitivity of the CN and Sr features on the MS also meant that it was
579: not possible to measure such depletions in N and Sr, respectively, and
580: as such we can not say anything about whether depletions exist amongst
581: these populations.
582:
583: An age was assigned to each star based on its position on the CMD and
584: metallicity using theoretical isochrones \citep{yi01} as described in
585: \citet{sta06a}. It was relatively straightforward to also obtain
586: temperatures and gravities from the isochrones simultaneously with the
587: ages. A grid of isochrones was used which span the metallicity range
588: --2.6$<$[Fe/H]$<$0.3 in 0.05 dex increments. For each metallicity
589: there were 34 isochrones with ages 2--19~Gyrs in 0.5~Gyr steps. Alpha
590: enhancement was taken to be constant ([$\alpha$/Fe]=0.3) for
591: [Fe/H]$\leq$--1.0, and declining linearly for higher [Fe/H] until it
592: reached the solar value at [Fe/H]=0. To assign an age to each star,
593: its metallicity was used to select the nearest isochrone in our
594: grid. The isochrones with this metallicity but with differing ages
595: were then compared to the star's $(B-V)_0$ and M$_{V}$ on the CMD to
596: find the one closest. Usually a star's position did not fall directly
597: on one isochrone and linear interpolation in color or magnitude was
598: performed between the two closest ones to determine its age, along
599: with the temperature and gravity. In almost every case the derived
600: temperature and gravity are not the same as those from the average
601: isochrones used in Figure \ref{cmd2}. Comparisons were made with
602: temperatures obtained from color temperature relations in
603: \citet{alo96}. These were, on average, smaller by 100K. This value is
604: within the 150K uncertainty in temperature used in the error analysis.
605: Microturbulence was assumed to be 1.0 kms$^{-1}$.
606:
607:
608:
609: Each candidate had the abundances of C, N and Sr determined. The C
610: feature (CH) at $\sim$4300{\AA} was analyzed first. Figure
611: \ref{cspec} gives examples of a MS star (upper panel) and SGB star
612: (lower panel) with enhancements in carbon. The heavy, solid lines
613: depict the observed spectrum, while the light solid lines show the
614: synthetic spectra. The carbon abundance ratio was modified in the
615: synthetic spectra while all other parameters were held constant. The
616: determined carbon abundance ratio is plotted, along with two others
617: bracketed the value by $\pm$0.3~dex. A synthetic spectra having
618: [C/Fe]=0.0 is also shown for comparison.
619:
620:
621:
622: Given [C/Fe], the CN feature at $\sim$3883{\AA} was used to obtain the
623: N abundances. It proved to be quite difficult to obtain accurate
624: abundances at the resolution of our spectra. There was little
625: sensitivity of the feature at abundances lower than [N/Fe]=1.0~dex,
626: especially for objects with low {\feh}. Therefore abundances were only
627: recorded when [N/Fe]$\geq$1.0~dex. Figure \ref{nspec} gives examples
628: of a MS star (top panel) and SGB star (lower panel) with enhancements
629: in nitrogen. The heavy, solid lines indicate the observed spectrum,
630: while the light solid lines show the synthetic spectra. The
631: determined nitrogen abundance ratio is plotted, along with two others
632: bracketing the value by $\pm$0.3~dex. A synthetic spectrum with
633: [N/Fe]=0.0 is also shown for comparison.
634:
635:
636:
637:
638: The Sr abundance was obtained from the Sr {\sc ii} line at 4077{\AA}
639: line. However, at the S/N of the data it was indistinguishable from
640: the noise at low abundances and results were only recorded when
641: [Sr/Fe]$\geq$1.0~dex. Further, the abundance obtained from this line
642: was only accepted if the (weaker) Sr {\sc ii} line at 4215{\AA}
643: confirmed the abundance of the first line. The Ba {\sc ii} at
644: 4554{\AA} line was also examined. However its sensitivity was too low
645: to obtain reliable abundances in most cases. It was able to be used
646: as an extra confirmation of the Sr abundance, as the Ba abundance is
647: expected to be strongly correlated with that of Sr at these
648: metallicities. Figure \ref{sspec} gives examples of a MS star (left
649: column) and SGB star (right column) with enhancements in strontium.
650: The heavy, solid lines indicate the observed spectrum, while the light
651: solid lines show the synthetic spectra. The top panels show the Sr
652: \textsc{ii} 4077{\AA} line, the middle panels the Sr \textsc{ii} 4215
653: {\AA} one. The strontium abundance ratio determined for these stars
654: is plotted, along with two others bracketing the value by
655: $\pm$0.3~dex. A synthetic spectrum with [Sr/Fe]=0.0 is also shown for
656: comparison. The lower panels show the Ba \textsc{ii} 4554{\AA} line
657: with the same enhancements in [Ba/Fe] as used for [Sr/Fe].
658:
659:
660:
661:
662:
663: \subsubsection{Errors}
664:
665: Typical errors were calculated for several objects with different
666: metallicities. The temperature, gravity and metallicity were varied
667: independently. The measured index was also varied by its errors and
668: included in the error calculations. The temperature was varied by
669: 150K, determined by reassigning an isochrone to an individual star
670: based on the error in photometry. The gravity varied by 0.2~dex
671: (determined in a similar way to the uncertainty in temperature), and
672: the metallicity by the individual abundance error for the star being
673: analyzed. The error contributions were added in quadrature. Typical
674: errors were 0.27~dex for the C abundance, 0.37~dex for the N abundance
675: and 0.38~dex for Sr. Note that these errors refer only to the objects
676: that had abundances determined for them.
677:
678: Abundances were also calculated for a small number of {\it normal}
679: stars across the three metallicity populations to verify that we would
680: obtain unenhanced abundances for these objects. This was found to be
681: the case, but again the limited sensitivity of the CN and Sr features
682: may not reveal possible enhancements of N and Sr, respectively.
683:
684:
685: \subsubsection{Results}
686:
687: Our abundances are presented in Table \ref{tbl-3}, which gives the
688: identification number for each star in column 1, and the V magnitude
689: and B--V in columns 2 and 3. The temperature, gravity and metallicity
690: are listed in columns 4--6. In columns 7--10 indices for the G band
691: (CH), S3839 (CN), Sr 4077{\AA} and Ba 4554{\AA} are listed. Abundance
692: ratios of C, N and Sr, relative to iron, are given in columns 11, 12
693: and 13, respectively, if calculated. To untangle the abundance
694: patterns, each metallicity population was treated separately and
695: abundance information then compared between populations. The
696: positions, photometry, metallicity and ages for these stars are
697: available in the electronic version of Table 2 in \citet{sta06a}.
698:
699:
700: When performing the spectrum synthesis analysis, comparisons by eye
701: between the Ca \textsc{ii} K line in the observed and synthetic
702: spectra were used as a check on the metallicities and temperatures.
703: In a small number of cases there was a large discrepancy between the
704: observed and synthetic, and therefore those stars were not considered
705: further. In total, 32 stars did not have an abundance analysis
706: performed due to unreliable metallicities and/or ages. It is unclear
707: whether the source of the error is driven by the metallicities,
708: related to a 3$\sigma$ error in the photometry or both. After
709: excluding these stars, this left a total of 392 stars in our sample.
710:
711:
712: Table \ref{tbl-4} gives the number and percentage of normal, C, N and
713: Sr enhancements for each metallicity population. Due to the varying
714: sensitivity of the CH, CN and Sr features, the emphasis of this paper
715: is on objects with enhancements: [C/Fe]$\geq$0.5~dex,
716: [N/Fe]$\geq$1.0~dex, and [Sr/Fe]$\geq$1.0~dex, even though abundance
717: ratios were calculated $\sim$1$\sigma$ below these values. These
718: limits were put in place to ensure we had complete numbers of objects
719: within our sample above these values. There are objects with
720: enhancements in carbon and/or nitrogen, and strontium which therefore
721: are counted in both the carbon/nitrogen enhanced and strontium
722: enhanced groups. This means the percentages often sum to more than
723: 100\%.
724:
725: Using a Gaussian distribution with a kernel equal to the error in
726: each abundance, the effect of errors on the number of objects found to
727: be enhanced was investigated. The first distribution was centered at
728: [C/Fe]=0.0, with $\sigma_C$=0.27~dex, and produced 5\% of the sample
729: as enhanced objects under our criteria of [C/Fe]$\geq$0.5. This is
730: comparable to the percentage of enhanced objects found on the
731: MS. However, the two metal-poor objects on the MS (9005309 and
732: 7007334, with [C/Fe]=1.30 and 1.35, respectively) have $\sim5\sigma$ C
733: enhancements and are not likely to result from the uncertainties. The
734: number of objects on the SGB that were found to be enhanced is
735: significantly higher than 5\%, but may be due to the bias against the
736: bulk of the metal-poor objects in the SGB sample introduced by the
737: blue cutoff at (B--V)=0.6 for the 2002 sample.
738:
739: A similar procedure was employed for the nitrogen and strontium
740: abundances, using $\sigma_{N}$=0.37~dex and
741: $\sigma_{Sr}$=0.38~dex. Both the N and Sr simulations indicated that
742: it would be unlikely that any objects would be included in the
743: enhanced sample as spurious detections, since less than 1\% of the
744: simulated sample had abundances greater than the criteria of
745: [N/Fe]$\geq$1.0~dex and [Sr/Fe]$\geq$1.0~dex.
746:
747: Table \ref{tbl-5} lists the abundance patterns found on the main
748: sequence and subgiant branches. One should note that often when we do
749: not find any stars of a particular abundance pattern, it is due to low
750: sensitivity of the relevant observed features. The table only lists
751: numbers of stars with detectable enhancements of [C/Fe]$\geq$0.5~dex,
752: [N/Fe]$\geq$1.0~dex and/or [Sr/Fe]$\geq$1.0~dex. Due to the selection
753: of the regions on the CMD, meaningful comparisons between the MS and
754: SGB are not possible. The blue color limit for the SGB box is redder
755: than the bulk of the metal-poor population in the cluster. Due to
756: these selection effects there are probably more detections of unusual
757: stars on the SGB than on the MS.
758:
759:
760:
761: \subsection{Abundance Patterns}
762:
763: Figure \ref{abpat} plots the abundance ratios (relative to iron) of
764: carbon, nitrogen and strontium for each of the three populations, as a
765: function of luminosity. Care must be taken with interpretation of
766: this diagram. The shaded regions in the figure show the areas of
767: limited sensitivity and are plotted below the cutoff values by an
768: amount of $\sigma$/2. Consequently, we do not detect all the
769: enhancements/depletions that are potentially present in our sample.
770: Therefore, only a few conclusions can be drawn from the recorded
771: abundance ratios. Stars have been plotted with different symbols in
772: order to better facilitate the comparisons between different elements
773: and abundances. For the 1st and 2nd populations the open triangles
774: represent those stars that are enriched in carbon ([C/Fe]$\geq$0.5)
775: and strontium ([Sr/Fe]$\geq$1.0). Objects that are enhanced in both
776: nitrogen and strontium are shown by solid stars. The one object with
777: enhancements in carbon, nitrogen and strontium, 8001811, is
778: represented by a plus sign inside a circle. All other stars are
779: represented by solid dots with no cross-referencing between panels.
780: Since all stars in the 3rd population have abundance determinations,
781: the symbols are used to show the corresponding object between carbon
782: of any value with enhanced Sr and unenhanced N (as open triangles), or
783: enhanced N and enhanced Sr (as solid stars). Solid triangles
784: represent stars with carbon abundances and nitrogen enhancements that
785: do not have high Sr abundances.
786:
787:
788:
789:
790: \subsubsection{Carbon}
791:
792: Panels a, b and c of Figure \ref{abpat} show the [C/Fe] abundances for
793: each population. The objects in Fig. \ref{abpat} with individual
794: abundance determinations are shown as diamonds in the carbon, nitrogen
795: and strontium indices plots (Figures \ref{c1st}, \ref{n1st} and
796: \ref{s1st}). The reader is reminded that the spectrum synthesis was
797: performed for the entire 3rd Pop. sample, but only for a select group
798: of {\it enhanced} objects for the 1st and 2nd Pops. Inspection of the
799: C enhanced objects in the 1st Pop. (Fig. \ref{abpat}, panel a) shows
800: there are relatively more objects with elevated abundances on the SGB
801: than on the MS. This same effect is seen to a lesser degree in the
802: 2nd Pop. data in Fig. \ref{abpat}, panel b. If the process that led
803: to the formation of the C-rich stars was primordial (i.e. they formed
804: from matter already enhanced in C and therefore have uniform
805: abundances throughout) one would expect equal numbers of these objects
806: on the MS and SGB. On the other hand, if these objects had elevated C
807: abundances due to the accretion of material onto their surface layers,
808: as they move off the MS and onto the SGB the convective layer deepens
809: and such enhancements become diminished. Consequently, one would
810: expect not to find as many SGB objects with C enhancements,
811: contradicting the results found here. Instead, the result may be due
812: to the diminished sensitivity of the CH feature on the MS compared
813: with the SGB preventing detection of enhanced C in main sequence
814: objects.
815:
816: The metal-poor objects on the main sequence with large carbon
817: enhancements are possible counterparts of the CH stars found on the
818: RGB in {\wcen}. Two objects on the main sequence, 7007334
819: ([C/Fe]=1.35) and 9005309 ([C/Fe]=1.30), seen in Fig. \ref{abpat}
820: (panel a), show the largest carbon enhancement in our sample, with no
821: other metal-poor main sequence objects showing any enhancement. These
822: objects also show no detectable enhancements of nitrogen or strontium.
823: The study by \citet{bd74} found that for two of the CH giants in
824: {\wcen} (ROA 55 and 70), [C/H]$\sim$--0.8 and [N/H]$\sim$0.0. Adopting
825: [Fe/H]=--1.9 for ROA 70 \citep{gra82}, this gives [C/Fe]$\sim$1.1 and
826: [N/Fe]$\sim$1.9. The carbon enhancements are similar to the values
827: found here, although the N enhancement is apparently not, though are N
828: sensitivity at this [Fe/H] is low. As for the s-process elements,
829: \citet{gra82} found [s/Fe]=1.1 for ROA 70 (where s represents the
830: averaged abundances of the s-process elements), but for our MS stars
831: we do not detect any enhancement of Sr above [Sr/Fe]=1.0~dex. A
832: metal-poor SGB star, 8001811, shows enhancements in carbon, nitrogen
833: and strontium, similar to those found for RGB CH stars in {\wcen}.
834: This star may also be similar to the ``subgiant CH'' stars reported by
835: \citet{lb91} in the field, that exhibit carbon and s-process
836: enhancement.
837:
838:
839: Figure \ref{c1st} indicates that there are some stars in each of the
840: three populations with possible C depletions. Upon further
841: investigation it was found that in the first population with the
842: lowest metallicities, C abundances were not able to be reliably
843: determined below [C/Fe]=0.0 for the majority of candidates. For the
844: second metallicity population, with \mbox{--1.5$\leq${\feh}$<$--1.1},
845: [C/Fe] started to lose its sensitivity around --0.2. C abundances for
846: the most metal-rich population, with {\feh}$\geq$--1.1, were able to
847: be determined down to [C/Fe]$\sim$--0.7.
848:
849: It is significant that we only find one object in the 3rd Pop. with C
850: enhancements greater than 0.0~dex. All objects within the 3rd
851: Pop. went through the individual abundance determination process,
852: although again the CN and Sr features of some objects were not useful
853: for abundance determination. Most of the objects in the 3rd Pop. were
854: C depleted, as can be seen in the panel c of Fig. \ref{abpat}. Some
855: of these C depleted stars have various large N and/or Sr enhancements.
856: However, it is interesting to note that we find several objects (4/9)
857: with C depletions that do not have corresponding N and/or Sr
858: enhancements. This may be due to the low sensitivity of CN and Sr
859: features.
860:
861: \subsubsection{Nitrogen}
862:
863:
864:
865: Inspection of the middle column of Figure \ref{abpat} shows the [N/Fe]
866: abundance ratios as functions of luminosity for each of the three
867: populations. The limited sensitivity of the CN feature is indicated
868: by the shaded regions, and for this reason only the extremely N
869: enhanced objects are identified. On the SGB in the metal-poor and
870: metal-intermediate populations it can be seen there is a fraction
871: ($\sim$20\%) of N enhanced objects, with [N/Fe] up to $\sim$2.0~dex.
872: Similarly, the 3rd Pop. contains a significant fraction of objects
873: with N enhancements, of order 50\%. These N abundances are
874: $\sim$0.5--1.0~dex higher than those found for RGB stars in
875: \citetalias{nd95}, and up to $\sim$0.75~dex higher than in the
876: analysis of RGB stars of \citet{bw93}. This large discrepancy
877: between the present work and other studies may be due to differing
878: analysis techniques. Analysis of RGB stars using the method employed
879: here would be useful in resolving this discrepancy.
880:
881: Nitrogen enhancements are not seen in the 1st Pop. on the MS, although
882: such stars are seen on the SGB at this metallicity and on the MS for
883: the higher abundance groups. Whether this is due to the low
884: sensitivity of the CN feature at the resolution and S/N used here and
885: our subsequent inability to detect the objects with overabundances in
886: N, or is a real effect remains unclear. That said, canonical stellar
887: evolution results do not predict the processing of C to N between the
888: MS and SGB phases of such low mass stars. Only one star on the SGB was
889: found with [N/Fe]$\geq$2.0. Therefore the fact that no enhancements of
890: [N/Fe]$>$2.0 were seen for MS stars may be a statistical effect.
891: Higher resolution data are needed to address this question.
892: \citet{pio05}, in a study of the double main sequence, found nitrogen
893: enhancements of [N/Fe]=1.0 for the metal-intermediate bMS and
894: [N/Fe]$<$1.0 for the metal-poor rMS. Therefore it seems likely that
895: our technique, which could only measure enhancements above [N/Fe]=1.0,
896: did not have the required sensitivity to measure any enhancements in N
897: on the main sequence.
898:
899:
900: \subsubsection{Strontium}
901:
902:
903: For Sr there is a abundance trend similar to that found for N across
904: the three populations, as may be seen in Figure \ref{abpat} (right
905: column). Here one finds a significant number of objects that show Sr
906: enhancement in all three metallicity populations. In particular, we
907: draw attention to the most Sr-rich object, the main sequence star
908: 2015448, which has [Sr/Fe]=1.6. This object has the added anomalous
909: behavior that while the abundance of Ba usually tracks that of Sr in
910: heavy neutron-capture enhanced objects, we do not detect Ba, and
911: estimate [Ba/Fe]$<$0.6. This object is discussed in more detail in
912: \citet{sta06b}.
913:
914:
915: \subsubsection{Metal rich Population}
916:
917: The 3rd Pop. differs from the other two populations in that there were
918: essentially no enhancements in C across the MS or SGB. One 3rd
919: Pop. object, 2016543, was identified with mild C enhancement. This
920: object had no N or Sr enhancements.
921:
922: Figure \ref{pspec} shows an example of a more typical member of the
923: 3rd Pop, which has depleted carbon ([C/Fe]=--0.4). This was used to
924: obtain a N abundance from the CN features at 3883{\AA} and 4215{\AA},
925: yielding [N/Fe]=1.7 (middle panels). This object also has enhanced Sr
926: and Ba, shown in the lower panels. Another star in our sample,
927: 3001426, has similar stellar parameters and abundances.
928:
929:
930:
931:
932:
933: \section{Discussion} \label{disc_sect}
934:
935: The abundance patterns found here, while somewhat limited, when taken
936: together with the data for the red giants, have the potential to
937: constrain putative sources of enrichment. We begin with a short
938: discussion of possible sources of enrichment and follow with
939: constraints that the present results place on the enrichment
940: processes.
941:
942: The origin of the abundance enhancements found here are most likely
943: primordial or the result of accretion events. These abundance
944: patterns are unlikely to have been generated within the MS stars
945: themselves as the small convective region does not mix processed
946: materials from the stellar interior to its surface. Comparing these
947: results with those found on the RGB (\citealt{scl95},
948: \citetalias{nd95}) seems to suggest that we are finding similar
949: abundance enhancements in N and the light s-process elements, at least
950: on the SGB. However, most stars studied on the RGB are C depleted, or
951: near solar, and N is enhanced by $\sim$1.0~dex
952: (\citealt{cb86,pn89,bw93}; \citetalias{nd95}). In RGB stars where the
953: C is depleted and N enhanced, the CN and ON cycles have been at work.
954: Whether this was in the stars themselves, in nearby objects whose
955: material was accreted, or occurred in objects from which these stars
956: formed is unclear. However, primordial or accretion of C and/or O is
957: needed to explain the CO-strong stars that exist on the RGB
958: \citep{per80}. While the details of the cluster chemical evolution
959: remain in question, we discuss the various type of objects that might
960: have been responsible for the nucleosynthesis of the various elements
961: that are varying within the system.
962:
963: \subsection{Possible Sources of Enrichment}
964:
965: \subsubsection{Low Mass AGB stars}
966:
967: Low mass (1--3 M$_{\odot}$) AGB stars undergo dredge-up episodes which
968: bring C and s-process elements into the convective region
969: \citep{gal98}. Because of their low mass, these objects do not have
970: hot bottom burning (HBB) episodes which would produce N and reduce C
971: via the CNO cycle. The amount of s-process elements that are produced
972: is dependent on neutron density and the availability of the seed Fe
973: nuclei \citep{smi05}. The neutron source is the
974: \mbox{$^{13}$C($\alpha$,n)$^{16}$O} reaction. The s-process is
975: believed to occur in a $^{13}$C pocket which is produced at the third
976: dredge-up episode \citep{gal98,ll05}. Specifically, hydrogen downflow
977: from the envelope penetrates the $^{12}$C intershell region and at H
978: reignition a $^{13}$C enhanced zone is formed. This $^{13}$C source
979: releases neutrons giving rise to efficient s-processing that is
980: dependent on the mass fractions of H, $^{4}$He, $^{13}$C, $^{14}$N and
981: metallicity. For the C enhanced stars where we do not detect any Sr
982: enhancement (s-process), it may be that we simply do not have the
983: sensitivity to detect any enrichment. Alternatively, if there was
984: little or no $^{13}$C pocket there was no source of neutrons for the
985: s-process to occur. Thus, depending on the size and efficiency of the
986: $^{13}$C pocket, these AGB stars may account for the enhancements of C
987: with and without Sr found in our sample.
988:
989:
990: \subsubsection{Intermediate Mass AGB stars}
991:
992: Intermediate mass (3--8 M$_{\odot}$) AGB stars are more likely to have
993: HBB, thereby producing a site for the operation of the CN cycle and
994: processing of C into N \citep{vdm02}. Production of s-process
995: elements is not expected to be very high in these objects due to the
996: shorter duration of the $^{13}$C pocket \citep{lat04}. Added to this
997: is lower neutron density from the $^{22}$Ne($\alpha$,n)$^{25}$Mg
998: reaction as the convective region becomes smaller in more massive
999: stars. These stars may account for the objects with N enhancements
1000: with little or no s-process element enhancements.
1001:
1002:
1003: \subsubsection{Massive Stars}
1004:
1005: \citet{mm06} have recently presented models of rotating massive
1006: ($\sim$60 M$_{\odot}$) stars. These objects produce stellar winds
1007: with large excess of He and N, but no C or O excess. Their products
1008: are consistent with the abundances seen on the blue main sequence
1009: (bMS) (see \citealt{pio05}), and may be responsible for some of the
1010: enhanced objects found here. Recently, abundance patterns of C, N, O,
1011: Na and Li obtained from models of fast rotating, massive stars were
1012: shown by \citet{dec06} to be similar the chemical anomalies observed
1013: in normal globular cluster stars. Whether these two sets of models
1014: also synthesize s-process elements is unclear but they may account for
1015: objects with enhancements in N.
1016:
1017:
1018: \citet{smi06} discussed the prospect of Wolf-Rayet and OB stars as
1019: enrichment sources for globular cluster anomalies. These objects (WN)
1020: may produce large amounts of nitrogen, but a top-heavy stellar mass
1021: function is required to generate the large amounts of N needed to
1022: explain the observations in globular clusters. A similar scenario is
1023: needed in the \citet{mm06} models.
1024:
1025:
1026: \subsection{Observational Constraints on the Enrichment of {\wcen}}
1027:
1028: Similarities exist between the carbon-rich objects found on the MS and
1029: those found on the RGB. Here, two stars out of 195 ($\sim$1\%) were
1030: found to have [C/Fe]$>$1.0 (these are the metal-poor stars 7007334
1031: and 9005309). There is one possible subgiant CH star, 8001811, which
1032: translates to $\sim$0.5\% of the total number of SGB stars with CH
1033: star properties. These percentages are similar to that found on the
1034: RGB for stars brighter than V$\sim$13 in the Woolley et al. (1966)
1035: catalogue, where there are four CH stars (55, 70, 279, 577, see
1036: e.g. \citetalias{nd95}) out of a possible $\sim$600, equating to
1037: 0.7\%. This suggests that one might expect to find an approximately
1038: one percent incidence of CH stars in any sample at most stages of
1039: evolution in the cluster (i.e. excluding blue horizontal branch
1040: objects), and indicates that the CH stars are not limited to a
1041: particular evolutionary phase.
1042:
1043:
1044: It is interesting that the C-rich ([C/Fe]$\geq$0.5) stars occur
1045: predominately at low metallicities and we find virtually none in the
1046: 3rd Pop. This may result from small number statistics. In the first
1047: and second populations we find that 10\% of objects have enhanced C.
1048: If the same percentage applies to the third population this would
1049: equate to only two stars and it is not outside the realm of
1050: possibility that they were simply not found in our search. On the
1051: other hand, it may be a result of the material from which these stars
1052: formed being low in C to begin with. The source of the carbon
1053: enrichment could be low-mass AGB stars or perhaps SNe II. In the
1054: latter case, the outer layers escaped while the material close to the
1055: iron core did not. Low mass AGB stars also produce s-process
1056: elements, depending on initial mass and metallicity. Therefore, they
1057: are able to account for the range in s-process enhancement seen in
1058: many of the C-rich stars.
1059:
1060:
1061: Inspection of Figure \ref{abpat} and Table \ref{tbl-4} show there are
1062: $\sim$10\% of the total sample (excluding the two metal-poor CH stars)
1063: with [C/Fe]$\geq$0.5. This is unusual when compared with the RGB
1064: abundances in {\wcen}, as previous studies have not found any stars
1065: (excluding CH stars) with carbon enhancements at these levels. Carbon
1066: abundances of main sequence and subgiant branch stars in other
1067: globular clusters also do not show any such enhancements.
1068: \citet{car05} studied carbon, nitrogen and oxygen in dwarfs and
1069: subgiant branch stars in NGC~6397, NGC~6752 and {\tuc}. For dwarfs,
1070: carbon abundances were less than [C/Fe]=0.50 for NGC~6397, [C/Fe]=0.2
1071: for NGC~6752 and [C/Fe]=--0.1 for {\tuc}. The subgiants showed
1072: abundance levels of [C/Fe]$\leq$0.15 for NGC~6397, and
1073: [C/Fe]$\leq$--0.10 for NGC~6752 and {\tuc}.
1074:
1075: These C enhanced objects on the MSTO of {\wcen} may be the
1076: evolutionary precursors to the CO-strong objects on the RGB
1077: \citep{per80}. Of the stars that are investigated in \citet{per80},
1078: 28 come from the unbiased sample of \citet{cs73}. The CO-strong stars
1079: represent 8 out of these 28 stars, or $\sim$30\% of RGB objects. If
1080: the 10\% of C-rich ([C/Fe]$\geq$0.5) stars found on the MSTO are the
1081: evolutionary precursors, one would expect that more of these objects
1082: should have been found. However, our choice of the C-rich cutoff (at
1083: [C/Fe]=0.5) is arbitrary and it is not known what the link is between
1084: CO-strong status and carbon abundance on the MSTO. If the C-rich
1085: cutoff was lower we would have found a higher percentage of objects
1086: that are C-rich on the MSTO. It would be useful in future
1087: investigations to determine carbon abundances for the whole sample to
1088: confirm the connection between the C-enhanced objects at the MSTO and
1089: the CO-strong objects on the RGB.
1090:
1091: The carbon abundances found here in the 3rd Pop. are low when
1092: compared with those determined for dwarfs by \citet{car05} in
1093: NGC~6397, NGC~6752 and {\tuc}. The lowest abundances found for
1094: NGC~6752 and {\tuc} were [C/Fe]=--0.2 and [C/Fe]=--0.13, respectively.
1095: The origin of these depletions for the most metal-rich group is not
1096: well understood. \citet{pan03} did not investigate carbon abundances
1097: for the most metal-rich RGB sample in {\wcen}, and of the few
1098: metal-rich RGB stars studied by \citetalias{nd95} carbon abundances
1099: were found to be [C/Fe]$\sim$--0.3. If evolutionary mixing on ascent
1100: of the RGB occurred, one would expect even lower carbon abundances
1101: than those found on the MS, but for the two metal-rich RGB stars
1102: studied by \citetalias{nd95} this was not found to be the case.
1103: Estimates of the level of carbon depletions for the 1st and 2nd
1104: Pops. would be of great interest to be able to compare them not only
1105: with those found in the 3rd Pop., but also with those on the RGB.
1106: \citet{pio05} found [C/Fe]=0.0 for both the metal-poor and
1107: metal-intermediate main sequences (rMS and bMS, respectively) using
1108: composite spectra of stars belonging to both populations.
1109:
1110: The nitrogen enhancements seen at the MSTO across all three
1111: populations are quite substantial with [N/Fe] as large as 2.2. These
1112: abundances are slightly larger than those seen in other clusters
1113: \citep{car05}. The nitrogen enhancements in NGC~6752 reach [N/Fe]=1.7
1114: for dwarfs and [N/Fe]=1.3 for subgiants. The maximum N enhancements
1115: for {\tuc} are lower and only reach [N/Fe]=1.1 for the subgiants,
1116: while our third population shows enhancements greater than [N/Fe]=1.1
1117: for 37.5\% of the group. \citet{pio05} investigated the two distinct
1118: main sequences in {\wcen} and found abundances of C, N and Ba using
1119: composite spectra of a number of stars in each sequence. They found
1120: [N/Fe]$\sim$1.0 for the blue main sequence (bMS), which is equivalent
1121: in metallicity to the 2nd Pop. discussed here. This is lower by a
1122: factor of 3--10 for the abundances of N found here. This may be due
1123: to a systematic offset in the abundances found here compared with
1124: those found by \citet{pio05}. On the {\wcen} RGB, \citetalias{nd95}
1125: also found N abundances to be less than those found here by
1126: $\sim$0.7~dex.
1127:
1128: While Sr abundances have not been extensively investigated on the RGB,
1129: lines of other light s-process elements such as Y and Zr have been
1130: analyzed (\citetalias{nd95} \citealt{scl95, smi00}). In general, it
1131: was found that the abundance of the s-process elements relative to
1132: iron increased as a function of metallicity. There are no
1133: correlations of s-process elemental abundances with other light
1134: elements where variations are seen, such as C, N, O, Na and Al
1135: \citepalias{nd95}. Qualitatively, this is similar to what is found
1136: here. The incidence of Sr enhancement relative to iron increases as a
1137: function of metallicity (see Table \ref{tbl-4}). This fits in with
1138: the idea that low to intermediate mass AGB stars contributed to the
1139: gas from which the later generations of stars formed. A comparison
1140: between the Sr abundances found for RGB and MS stars will be discussed
1141: further in a forthcoming paper (Stanford et al. 2007, in preparation).
1142:
1143:
1144:
1145: It is not clear if intermediate mass AGB stars are responsible since
1146: s-processing is not very efficient in these objects. To account for
1147: the N and Sr enhanced stars favorable conditions of N and s-process
1148: element production in intermediate mass AGB stars are needed.
1149: Alternatively, the enriched objects may be accounted for by several
1150: different sources. For instance, massive rotating stars may account
1151: for the N overabundances while low-mass AGB may account for the
1152: s-process enrichment. This latter scenario is supported by the fact
1153: that no strong correlation is seen between nitrogen and strontium
1154: (s-process) abundances.
1155:
1156:
1157: Other globular clusters, that do not show metallicity (i.e. iron)
1158: spreads, show variations in the abundances of light elements to
1159: varying degrees, as described in \S \ref{intro}. Several clusters
1160: show CN--CH bimodality not only on the RGB but also on the MS
1161: \citep{can98, bri04, ss91, gra01a, coh99, bc01, rc02, cbs02, bcs04,
1162: dac04}. In this respect, the spread in C and N on the MSTO of {\wcen}
1163: is not unique and these results demonstrate the variations are a
1164: product of environment since field stars do not show similar patterns.
1165: Variations in other light elements such as Na, O, Mg and Al are also
1166: found on the MS in some clusters \citep{gra01b}. These elements have
1167: yet to be studied on the MS level in {\wcen} to determine if the same
1168: effect is seen in this cluster. The theoretical yields of AGB stars
1169: at present fail to explain the abundance patterns seen in the light
1170: elements in globular clusters (see e.g. \citealt{fen04}). That is,
1171: there is more to the enrichment scenario in other clusters that is yet
1172: to be explained and this complication may also be the case for
1173: {\wcen}.
1174:
1175: \subsubsection{Constraints on the Role of Binarity}
1176:
1177: If the enhancements are due to past accretion events as part of binary
1178: systems, the fraction of objects with overabundances indicates {\wcen}
1179: should have a large binary fraction, of order $\sim$20\%. In a radial
1180: velocity study of red giants in {\wcen} spanning over a decade,
1181: \citet{may96} estimated the global binary frequency is 3--4\%.
1182: Further, they reported that most of the giants with known chemical
1183: peculiarities, such as Ba, CH and S stars, have constant velocities,
1184: suggesting that they are not members of binary systems. However,
1185: there is the possibility that these objects may have been former
1186: binaries that were disrupted by close encounters with nearby stars.
1187: The same argument can be applied to the MS stars, where the fraction
1188: of C, N and Sr enhanced stars exceeds the observed binary frequency
1189: for giants in the cluster.
1190:
1191: \section{Conclusion}
1192:
1193: A low resolution spectroscopic abundance analysis of a sample of 392
1194: {\wcen} members revealed objects enhanced in carbon, nitrogen and/or
1195: strontium. These enhanced objects were analyzed in detail and
1196: abundances ratios of carbon, nitrogen and strontium relative to iron
1197: determined. The abundances revealed patterns that can possibly be
1198: explained by low and intermediate mass AGB and/or rotating massive
1199: star nucleosynthesis. The enhancements are either primordial or from
1200: accretion events, rather than evolutionary. Higher resolution, and
1201: higher S/N data are needed to further constrain the possible sources
1202: of these abundances anomalies. From such data, abundances of a large
1203: number of elements such as the light elements (O, Na, Mg, Al), other
1204: light and heavy s-process elements, and iron peak elements, will be
1205: essential in obtaining another piece of the $\omega$ Cen puzzle.
1206:
1207:
1208: \acknowledgements
1209: We thank the Director and staff of the Anglo-Australian Observatory
1210: for the use of their facilities.
1211:
1212: \begin{thebibliography}{}
1213: \bibitem[Alonso et al.(1996)]{alo96}
1214: Alonso, A., Arribas, S., Martinez-Roger, C. 1996, A\&A, 313, 873
1215: \bibitem[Anderson et al.(2002)]{and02}
1216: Anderson, J., 2002, in van Leeuwen, F., Hughes, J. D., \& Piotto, G., eds., ASP
1217: Conf. Ser. Vol. 265, A Unique Window into
1218: Astrophysics. (Astron. Soc. Pac.: San Francisco), 87
1219: \bibitem[Beckers etal.(1976)]{beckers}
1220: Beckers, J. M., Bridges, C. A., \& Gilliam, L. B. 1976, A high resolution spectral atlas of the solar irradiance from 380 to 700 nanometers. (Hanscom AFB, MS: Sacramento Peak Obs.)
1221: \bibitem[Bedin et al.(2004)]{bed04}
1222: Bedin, L. R., Piotto, G., Anderson, J., Cassisi, S., King, I. R., Momany, Y., \& Carraro, G. 2004, \apj, 605, L125
1223: \bibitem[Beers \& Christlieb(2005)]{bc05}
1224: Beers, T. C. \& Christlieb, N. 2005, ARA\&A, 43, 531
1225: \bibitem[Beers et al.(1999)]{bee99}
1226: Beers, T. C., Rossi, S., Norris, J. E., Ryan, S. G., \& Shefler, T. 1999, \aj, 117, 981
1227: \bibitem[Bekki \& Norris(2006)]{bn06}
1228: Bekki, K., \& Norris, J. E. 2006, \apj, 637, L109
1229: \bibitem[Bell \& Dickens(1974)]{bd74}
1230: Bell, R. A., \& Dickens, R. J. 1974, \mnras, 166, 89
1231: \bibitem[Bessell \& Norris(1976)]{bn76}
1232: Bessell, M. S., \& Norris, J. 1976, \apj, 208, 369
1233: \bibitem[Briley \& Cohen(2001)]{bc01}
1234: Briley, M. M., \& Cohen, J. G. 2001, \aj, 122, 242
1235: \bibitem[Briley, Cohen \& Stetson(2004)]{bcs04}
1236: Briley, M. M., Cohen, J. G., \& Stetson, P. B. 2004, \aj, 127, 1579
1237: \bibitem[Briley et al(2004)]{bri04}
1238: Briley, M. M., Harbeck, D., Smith, G. H., \& Grebel, E. K. 2004, \aj, 127, 1588
1239: \bibitem[Brown \& Wallerstein(1993)]{bw93}
1240: Brown, J. A., \& Wallerstein, G. 1993, \aj, 106, 133
1241: \bibitem[Butler, Dickens \& Epps(1978)]{bde78}
1242: Butler, D., Dickens, R. J., \& Epps, E. 1978, \apj, 225, 148
1243: \bibitem[Cameron(1982)]{cam82}
1244: Cameron, A. G. W. 1982, \apss, 82 123
1245: \bibitem[Cannon et al(1998)]{can98}
1246: Cannon, R. D., Croke, B. F. W., Bell, R. A., Hesser, J. E., \& Stathakis, R. A. 1998, \mnras, 298, 601
1247: \bibitem[Cannon \& Stobie(1973)]{cs73}
1248: Cannon, R. D., \& Stobie, R. S. 1973, \mnras, 162, 207
1249: \bibitem[Carretta et al.(2004)]{car04}
1250: Carretta, E., Gratton, R. G., Bragaglia, A., Bonifacio, P., \& Pasquini, L. 2004, A\&A, 416, 925
1251: \bibitem[Carretta et al.(2005)]{car05}
1252: Carretta, E., Gratton, R. G., Lucatello, S., Bragaglia, A., \& Bonifacio, P. 2005, A\&A, 433, 597
1253: \bibitem[Cohen (1999)]{coh99}
1254: Cohen, J. G. 1999, \aj, 117, 2434
1255: \bibitem[Cohen \& Bell(1986)]{cb86}
1256: Cohen, J. G., \& Bell, R. A. 1986, \apj, 305, 698
1257: \bibitem[Cohen, Briley \& Stetson(2002)]{cbs02}
1258: Cohen, J. G., Briley, M. M., \& Stetson, P. B. 2002, \aj, 123, 2525
1259: \bibitem[Cottrell \& Norris(1978)]{cn78}
1260: Cottrell, P. \& Norris, J. 1978, \apj, 221, 893
1261: \bibitem[Cunha et al.(2002)]{cun02}
1262: Cunha, K., Smith, V. V., Suntzeff, N. B., Norris, J. E., Da Costa, G. S., \& Plez, B. 2002, \aj, 124, 379
1263: \bibitem[Da Costa et al.(2004)]{dac04}
1264: Da Costa, G. S., Cannon, R. D., Croke, B., \& Norris, J. E. 2004, Mem. Soc. Astron. Italiana, 75, 370
1265: \bibitem[Da Costa \& Demarque(1982)]{dd82}
1266: Da Costa, G. S. \& Demarque, P. 1982, ApJ, 259, 193
1267: \bibitem[Da Costa et al.(2005)]{dac05}
1268: Da Costa, G. S., Stanford, L. M., Norris, J. E., Cannon, R. D. 2005, in IAU Symposium No. 228, From Lithium to Uranium: Elemental Tracers of Early Cosmic Evolution., Hill, V., Francois, P., \& Primas, F., eds., 369, Cambridge: Cambridge University Press
1269: \bibitem[Decressin et al.(2006)]{dec06}
1270: Decressin, T., Meynet, G., Charbonnel, C., Prantzos, N., \& Ekstr$\ddot{\mbox{o}}$m, S. 2006, astro-ph/0611379
1271: \bibitem[Dickens \& Bell(1976)]{db76}
1272: Dickens, R. J., \& Bell, R. A. 1976, \apj, 207, 506
1273: \bibitem[Dinescu et al.(1999a)]{din99a}
1274: Dinescu, D. I., van Altena, W. F., Girard, T. M., \& L\'{o}pez, C. E. 1999, \aj, 117, 277
1275: \bibitem[Fenner et al.(2004)]{fen04}
1276: Fenner, Y., Campbell, S. W., Karakas, A. I., Lattanzio, J. C., \& Gibson, B. C. 2004, \mnras, 353, 789
1277: \bibitem[Francois, Spite \& Spite(1988)]{fss88}
1278: Francois, P., Spite, M., \& Spite, F. 1988, A\&A, 191, 267
1279: \bibitem[Freeman \& Rodgers(1975)]{fr75}
1280: Freeman, K. C., \& Rodgers, A. W. 1975, ApJ, 201, L71
1281: \bibitem[Gallino etal.(1998)]{gal98}
1282: Gallino, R., Arlandini, C., Busson, M., Lugaro, M., Travaglio, C., Straniero, O., Chieffi, A., \& Limongi, M. 1998, \apj, 497, 388
1283: \bibitem[Gratton(1982)]{gra82}
1284: Gratton, R. G. 1982, A\&A, 115, 336
1285: \bibitem[Gratton et al(2001b)]{gra01b}
1286: Gratton, R. G., Bonifacio, P., Bragaglia, A., Carretta, E., Castellani, V., Centurion, M., Chieffi, A., Claudi, R., Clementini, G., D'Antona, F., Desidera, S., Francois, P., Grunahl, F., Lucatello, S., Molaro, P., Pasquini, L., Sneden, C., Spite, F., \& Straniero, O. 2001, A\&A, 369, 87
1287: \bibitem[Gratton et al(2001a)]{gra01a}
1288: Gratton, R. G., Sneden, C., Carretta, E., \& Bragaglia, A. 2001, A\&A, 354, 169
1289: \bibitem[Harris(1996)]{har96}
1290: Harris, W. E., 1996, \aj, 112, 1478
1291: \bibitem[Hilker \& Richtler(2000)]{hr00}
1292: Hilker, M., \& Richtler, T. 2000, A\&A, 362, 895
1293: \bibitem[Langer et al.(1986)]{lan86}
1294: Langer, G. E., Kraft, R. P., Carbon, D. F., Friel. E., \& Oke, J. B. \pasp, 1986, 98
1295: \bibitem[Lattanzio et al. (2004)]{lat04}
1296: Lattanzio, J., Karakas, A., Campbell, S., Elliot, L., \& Chieffi, A. 2004, Mem. Soc. Astron. Italiana, 75, 322
1297: \bibitem[Lattanzio \& Lugaro(2005)]{ll05}
1298: Lattanzio, J. C., \& Lugaro, M. A. 2005, NuPhA, 758, L477
1299: \bibitem[Lee et al.(1999)]{lee99}
1300: Lee, Y.-W., Joo, J.-M., Sohn, Y.-J., Rey, S.-C., Lee, H.-C., Walker, A. R. 1999, Nature, 402, 55
1301: \bibitem[Lewis et al.(2002)]{lew02}
1302: Lewis et al., 2002, \mnras, 333, 279
1303: \bibitem[Lloyd Evans(1977)]{le77}
1304: Lloyd Evans, T. 1977, \mnras, 181, 591
1305: \bibitem[Lub(2001)]{lub02}
1306: Lub, J., 2002, in ASP Conf. Ser. Vol. 265, A Unique Window into Astrophysics. van Leeuwen, F., Hughes, J. D., \& Piotto, G., eds., (Astron. Soc. Pac.: San Francisco), 95
1307: \bibitem[Luck \& Bond(1991)]{lb91}
1308: Luck, R. E., \& Bond, H. E. 1991, \apjs, 77, 515
1309: \bibitem[Maeder \& Meynet(2006)]{mm06}
1310: Maeder, A., \& Meynet, G. 2006, A\&A, 448, L37
1311: \bibitem[Mayor et al.(1996)]{may96}
1312: Mayor, M., Duguennoy, A., Udry, S., Anderson, J., \& Nordstrom, B. 1996, ASP Conf. Ser. Vol. 90, The Origins, Evolution, and Destinies of Binary Stars in Clusters, in Milone, G., \& Mermilliod, J.-C., eds., (Astron. Soc. Pac.: San Francisco), 190
1313: \bibitem[Norris(2004)]{nor04}
1314: Norris, J. E. 2004, \apj, 612, L25
1315: \bibitem[Norris \& Bessell(1975)]{nb75}
1316: Norris, J., \& Bessell, M. S. 1975, \apj, 201, L75
1317: \bibitem[Norris et al.(1981)]{nor81}
1318: Norris, J. E., Cottrell, P. L., Freeman, K. C., Da Costa, G. S. 1981, \apj, 244, 205
1319: \bibitem[Norris \& Da Costa(1995a)]{nd95}
1320: Norris, J. E., \& Da Costa, G. S. 1995, ApJ, 447, 680
1321: \bibitem[Norris \& Da Costa(1995b)]{nd95b}
1322: Norris, J. E., \& Da Costa, G. S. 1995, \apj, 441, L81, ND95
1323: \bibitem[Norris, Freeman \& Mighell(1996)]{nfm96}
1324: Norris, J. E., Freeman, K. C., \& Mighell, K. J. 1996, ApJ, 462, 241
1325: \bibitem[Norris, Ryan \& Beers(1997)]{nrb97}
1326: Norris, J. E., Ryan, S. G., \& Beers, T. C. 1997, \apj, 488, 350
1327: \bibitem[Paltoglou \& Norris(1989)]{pn89}
1328: Paltoglou, G., \& Norris, J. E. 1989, \apj, 336, 185
1329: \bibitem[Pancino (2003)]{pan03}
1330: Pancino, E. 2003, Ph.D. Thesis, Univ. Bologna
1331: \bibitem[Pancino et al.(2000)]{pan00}
1332: Pancino, E., Ferraro, F. R., Bellazzini, M., Piotto, G., \& Zoccali, M. 2000, ApJ, 534, 83
1333: \bibitem[Pancino et al.(2002)]{pan02}
1334: Pancino, E., Pasquini, L., Hill, V., Ferraro, F. R., Bellazzini, M. 2002, \apj, 568, L101
1335: \bibitem[Persson et al.(1980)]{per80}
1336: Persson, S. E., Frogel, J. A., Cohen, J. G., Aaronson, M., \& Matthews, K. 1980, \apj, 235, 452
1337: \bibitem[Piotto~et~al.(2005)]{pio05}
1338: Piotto, G., Villanova, S., Bedin, L. R., Gratton, R., Cassisi, S., Momany, Y., Recio-Blanco, A., Lucatello, S., Anderson, J., King, I. R., Peitrinferni, A. \& Carraro, G. 2005, \apj, 621, 777
1339: \bibitem[Ram\'{\i}rez \& Cohen(2002)]{rc02}
1340: Ram\'{\i}rez, S. V. \& Cohen J. G. 20002, \aj, 123, 3277
1341: \bibitem[Rey et al.(2004)]{rey04}
1342: Rey, S.-C., Lee, Y.-W., Ree, C. H., Joo, M.-J., \& Sohn, Y.-J. 2004, \aj, 127, 958
1343: \bibitem[Smith(2006)]{smi06}
1344: Smith, G. H. 2006, \pasp, 118, 1225
1345: \bibitem[Smith(2005)]{smi05}
1346: Smith, V. V. 2005, in ASP Conf. Ser. Vol. 336, Cosmic Abundances as Records of Stellar Evolution and Nucleosynthesis in honor of David L. Lambert. Barnes, T. G. \& Bash, F. N., eds., (Astron. Soc. Pac.: San Francisco), 165
1347: \bibitem[Smith et al.(1995)]{scl95}
1348: Smith, V. V., Cunha, K., \& Lambert, D. 1995, AJ, 110, 2827
1349: \bibitem[Smith et al.(2000)]{smi00}
1350: Smith, V. V., Suntzeff, N., Cunha, K., Gallino, R., Busso, M., Lambert, D., \& Straniero, O. 2000, AJ, 119, 1239
1351: \bibitem[Sollima et al.(2005a)]{sol05a}
1352: Sollima, A., Ferraro, F. R., Pancino, E. \& Bellazzini, M. 2005, \mnras, 357, 265
1353: \bibitem[Stanford et al.(2006a)]{sta06a}
1354: Stanford, L. M., Da Costa, G. S., Norris, J. E., \& Cannon, R. D. 2006a, \apj, 647, 1075
1355: \bibitem[Stanford et al.(2006b)]{sta06b}
1356: Stanford, L. M., Da Costa, G. S., Norris, J. E., \& Cannon, R. D. 2006b, \apj, 653, L117
1357: \bibitem[Suntzeff \& Kraft(1996)]{sk96}
1358: Suntzeff, N. B., \& Kraft, R., P. 1996, \aj, 111, 1913
1359: \bibitem[Suntzeff \& Smith(1991)]{ss91}
1360: Suntzeff, N. B., \& Smith, V. V. 1991, \apj, 381, 160
1361: \bibitem[Ventura et al.(2002)]{vdm02}
1362: Ventura, P., D'Antona, F., \& Mazzitelli, I. 2002, A\&A, 393, 215
1363: \bibitem[Woolley et al.(1966)]{woo66}
1364: Woolley, R. v. d. R., et al.\ 1996, R. Obs. Ann., No 2
1365: \bibitem[Yi et al.(2001)]{yi01}
1366: Yi, S., Demarque, P., Kim, Y.-C., Lee, Y.-W., Ree, C.-H., Lejeune, T., \& Barnes, S. 2001, \apj, 136, 417
1367: \end{thebibliography}
1368:
1369:
1370:
1371: % Tables
1372:
1373: \clearpage
1374: \input{tab1.tex}
1375:
1376: %\clearpage
1377: \input{tab2.tex}
1378:
1379: %\clearpage
1380: \input{stub.tab3.tex}
1381:
1382: \clearpage
1383: \input{tab4.tex}
1384:
1385: %\clearpage
1386: \input{tab5.tex}
1387:
1388:
1389: % Figures
1390:
1391: \clearpage
1392: \begin{figure}
1393: \begin{center}
1394: \includegraphics[scale=.60,angle=0]{f1.ps}
1395: \figcaption[f1.ps]{Color-magnitude diagram for stars near the MSTO and
1396: subgiant branch of {\wcen}. Large circles denote stars which are
1397: radial velocity members from two samples selected within the regions
1398: outlined. \label{cmd1} }
1399: \end{center}
1400: \end{figure}
1401:
1402: \clearpage
1403: \begin{figure}
1404: \begin{center}
1405: \includegraphics[scale=.60,angle=0]{f2.ps}
1406: \figcaption[f2.ps]{Generalized histograms of the metallicity
1407: distribution for the 98/99 (panel~a) and 2002 (panel~b) samples
1408: (gaussian kernel of $\sigma$=0.15--0.20, based on the individual
1409: errors for each metallicity). For comparison, the \citet{nfm96} red
1410: giant branch distribution is also plotted ($\sigma$=0.14). Indicated
1411: on each histogram is the division into three populations with the
1412: first having [Fe/H]$<$--1.5, the second --1.5$\leq$[Fe/H]$<$--1.1, and
1413: the third [Fe/H]$\geq$--1.1.
1414: \label{hist} }
1415: \end{center}
1416: \end{figure}
1417:
1418:
1419: \clearpage
1420: \begin{figure}
1421: \begin{center}
1422: \includegraphics[scale=.60,angle=0]{f3.ps}
1423: \figcaption[f3.ps]{Color-magnitude diagrams of the three populations
1424: of {\wcen} defined by their metallicities. Isochrones with
1425: metallicities and ages representing the means of each population are
1426: also plotted.
1427: \label{cmd2} }
1428: \end{center}
1429: \end{figure}
1430:
1431:
1432:
1433: \clearpage
1434: \begin{figure}
1435: \begin{center}
1436: \includegraphics[scale=.80,angle=0]{f4.ps} \figcaption[f4.ps]{ Indices
1437: for the G band (CH), CN at $\sim$3883{\AA}, Sr {\sc ii} 4077{\AA} and
1438: Ba {\sc ii} 4554{\AA}, plotted against metallicity for the main
1439: sequence (V$>$18) and subgiant branch stars (V$\leq$18). The boxed
1440: regions on the MS plot represent indices calculated for three other
1441: globular clusters --- NGC~6397 ([Fe/H]=--1.96), NGC~6752
1442: ([Fe/H]=--1.56) and {\tuc} ([Fe/H]=--0.76). \label{indfeh} }
1443: \end{center}
1444: \end{figure}
1445:
1446: \clearpage
1447: \begin{figure}
1448: \begin{center}
1449: \includegraphics[scale=.60,angle=0]{f5.ps} \figcaption[f5.ps]{CN vs CH
1450: indices for {\wcen} (solid dots) and {\tuc} (open circles). The
1451: {\tuc} data, plotted only in the metal-rich, MS panel, are split
1452: according to the CN/CH bimodality with the division shown by the
1453: dotted line defined in the bottom left panel, and reproduced in the
1454: other panels. The solid line is a fit to the CN, CH anticorrelation
1455: and plotted in each panel as a reference. The {\wcen} data are
1456: divided according to metallicity and position on the CMD (MS or SGB).
1457: \label{cnch} }
1458: \end{center}
1459: \end{figure}
1460:
1461: \clearpage
1462: \begin{figure}
1463: \begin{center}
1464: \includegraphics[scale=.60,angle=0]{f6.ps} \figcaption[f6.ps]{G band
1465: indices plotted against absolute magnitude (M$_{V}$) for each of the
1466: three populations of {\wcen} in the left panels, and calibrating
1467: globular clusters in the right panels. The diamonds represent those
1468: stars for which individual abundances were determined, while filled
1469: circles represent all other objects. Lines of common abundance are
1470: also plotted for the synthetic spectra. The vertical dashed line
1471: represents the cutoff between main sequence and turnoff stars. See
1472: text for details.
1473: \label{c1st} }
1474: \end{center}
1475: \end{figure}
1476:
1477: \clearpage
1478: \begin{figure}
1479: \begin{center}
1480: \includegraphics[scale=.60,angle=0]{f7.ps} \figcaption[f7.ps]{CN
1481: indices plotted against absolute magnitude (M$_{V}$) for each of the
1482: three populations in the left hand panels and the calibrating globular
1483: clusters in the right. The diamonds represent those stars for which
1484: individual abundances were determined, while filled
1485: circles represent all other objects. Lines of common abundance are
1486: also plotted for the synthetic spectra. The vertical dashed line
1487: represents the cutoff between main sequence and turnoff stars. See
1488: text for details. \label{n1st} }
1489: \end{center}
1490: \end{figure}
1491:
1492: \clearpage
1493: \begin{figure}
1494: \begin{center}
1495: \includegraphics[scale=.60,angle=0]{f8.ps} \figcaption[f8.ps]{ Sr {\sc
1496: ii} 4077{\AA}, indices plotted against absolute magnitude (M$_{V}$)
1497: for each of the three populations in the left hand panels and the
1498: calibrating globular clusters in the right. The diamonds represent
1499: those stars for which individual abundances were determined, while
1500: filled circles represent all other objects. Lines of common abundance
1501: are also plotted for the synthetic spectra. The vertical dashed line
1502: represents the cutoff between main sequence and turnoff stars. See
1503: text for details.\label{s1st}}
1504: \end{center}
1505: \end{figure}
1506:
1507:
1508:
1509:
1510:
1511:
1512: \clearpage
1513: \begin{figure}
1514: \begin{center}
1515: \includegraphics[scale=.60,angle=0]{f9.ps}
1516: \figcaption[f9.ps]{Examples of observed and synthetic spectra for two
1517: objects with enhanced carbon abundances. The bold line represents the
1518: observed spectrum in each case. The top panel shows an example of a
1519: MS star, 7007334, with [C/Fe]=1.35, while the lower panel shows a SGB
1520: star, 2010136, with [C/Fe]=0.9. Synthetic spectra with four carbon
1521: abundances are shown for each object, including [C/Fe]=0.0 for
1522: completeness.
1523: \label{cspec} }
1524: \end{center}
1525: \end{figure}
1526:
1527: \clearpage
1528: \begin{figure}
1529: \begin{center}
1530: \includegraphics[scale=.60,angle=0]{f10.ps}
1531: \figcaption[f10.ps]{Example observed and synthetic spectra of two
1532: objects with enhanced nitrogen abundances. The top panel shows the
1533: spectrum of a MS star in bold, 5012643, and synthetic spectra as
1534: light lines for a range of nitrogen abundances including [N/Fe]=0.0.
1535: The bottom panel shows a SGB star, 5018984, with [N/Fe]=2.0~dex. For
1536: both stars, [C/Fe]=0.0.
1537: \label{nspec} }
1538: \end{center}
1539: \end{figure}
1540:
1541: \clearpage
1542: \begin{figure}
1543: \begin{center}
1544: \includegraphics[scale=.70,angle=0]{f11.ps}
1545: \figcaption[f11.ps]{Observed and synthetic spectra showing a MS
1546: object, 7007295, on the left. The right panel shows a SGB star,
1547: 8007880, with enhanced s-process elements. The top panel in each
1548: shows the Sr {\sc ii} 4077{\AA} line, the middle shows Sr {\sc ii}
1549: 4215{\AA} and the bottom shows \mbox{Ba {\sc ii} 4554{\AA}}.
1550: \label{sspec} }
1551: \end{center}
1552: \end{figure}
1553:
1554: \clearpage
1555: \begin{figure}
1556: \begin{center}
1557: \includegraphics[scale=.70,angle=0]{f12.ps} \figcaption[f12.ps]{
1558: Abundance ratios (relative to Fe) of C, N and Sr for the three
1559: metallicity populations as a function of magnitude. The shaded
1560: regions indicate areas of low sensitivity where the abundances are not
1561: able to be determined reliably, and as such care should be taken in
1562: interpretation of the abundance patterns. The dotted horizontal line
1563: represents solar abundance ratios. For main sequence stars V$>$18
1564: (M$_{V}>$4.04) and turnoff stars V$\leq$18 (M$_{V}\leq$4.04). Also,
1565: one should note the limited sensitivity of the CN and Sr features,
1566: particularly at low metallicities, which makes it difficult to
1567: recognize all objects of interest. Finally, the biased color
1568: selection of the SGB sample that excludes many metal-poor stars should
1569: not be overlooked. In particular, comparisons between the MS and SGB
1570: groups should not be made. See text for more details for definitions
1571: of the symbols.
1572: \label{abpat} }
1573: \end{center}
1574: \end{figure}
1575:
1576:
1577: \clearpage
1578: \begin{figure}
1579: \begin{center}
1580: \includegraphics[scale=.70,angle=0]{f13.ps}
1581: \figcaption[f13.ps]{Observed and synthetic spectra of star 2014590
1582: from the most metal-rich population, with [Fe/H]=--1.0~dex. The top
1583: left panel shows the spectrum of this object in the range
1584: 3800--4600{\AA}. The top right panel shows the observed spectrum
1585: (bold line) with synthetic spectra (light lines) in the region of the
1586: G band. The middle two panels show the CN features at $\sim$3883{\AA}
1587: (left panel) and $\sim$4216{\AA} (right panel). The bottom left panel
1588: shows the Sr {\sc ii} 4077{\AA} line and the bottom right panel shows
1589: Ba {\sc ii} 4554{\AA}. \label{pspec} }
1590: \end{center}
1591: \end{figure}
1592:
1593:
1594:
1595: \end{document}
1596:
1597: