1: %% The command below calls the preprint style
2: %% which will produce a one-column, single-spaced document.
3: %% Examples of commands for other substyles follow. Use
4: %% whichever is most appropriate for your purposes.
5: %%
6: %\usepackage{/usr/share/texmf/ptex/platex209/article}
7:
8: %\documentclass[12pt,preprint]{../bibtex/emulateapj}
9: \documentclass[12pt,preprint]{aastex}
10:
11: %% manuscript produces a one-column, double-spaced document:
12:
13: %\documentclass[manuscript]{aastex}
14:
15: %% preprint2 produces a double-column, single-spaced document:
16:
17: % \documentclass[preprint2]{aastex}
18:
19: %% Sometimes a paper's abstract is too long to fit on the
20: %% title page in preprint2 mode. When that is the case,
21: %% use the longabstract style option.
22:
23: % \documentclass[preprint2,longabstract]{aastex}
24: %\usepackage{lscape,/usr/share/texmf/tex/latex/psnfss/times}
25: \usepackage[bf,small]{caption}
26: \newcommand{\vdag}{(v)^\dagger}
27: \newcommand{\myemail}{b.davies@leeds.ac.uk}
28: \newcommand{\microns}{$\mu$m}
29: \def\EW{$W_{\lambda}$}
30: \def\deg{$^{\circ}$}
31: \def\ga{\mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$}}}\hbox{$>$}}}}
32: \def\la{\mathrel{\hbox{\rlap{\hbox{\lower4pt\hbox{$\sim$}}}\hbox{$<$}}}}
33: \def\msunyr{$M$ \mbox{$_{\normalsize\odot}$}\rm{yr}$^{-1}$}
34: \def\msun{$M$\mbox{$_{\normalsize\odot}$}}
35: \def\minit{$M_{\rm init}$}
36: \def\lsun{$L$\mbox{$_{\normalsize\odot}$}}
37: \def\kms{\,km~s$^{-1}$}
38: \def\EW{$W_{\lambda}$}
39: \def\arcsec{$^{\prime \prime}$}
40: \def\arcmin{$^{\prime}$}
41: %\newcommand{\figref}{#1}{Fig.\,\ref{#1}}
42: %\newcommand{\tabref}{#1}{Tab.\,\ref{#1}}
43:
44: %% You can insert a short comment on the title page using the command below.
45:
46: \slugcomment{}
47:
48: \shorttitle{Chemical abundances of Galactic Centre RSGs}
49: \shortauthors{Davies et al.}
50:
51: %% This is the end of the preamble. Indicate the beginning of the
52: %% paper itself with \begin{document}.
53:
54: \begin{document}
55:
56: %% LaTeX will automatically break titles if they run longer than
57: %% one line. However, you may use \\ to force a line break if
58: %% you desire.
59:
60: \title{The chemical abundances in the Galactic Centre from the
61: atmospheres of Red Supergiants}
62:
63: %% Use \author, \affil, and the \and command to format
64: %% author and affiliation information.
65: %% Note that \email has replaced the old \authoremail command
66: %% from AASTeX v4.0. You can use \email to mark an email address
67: %% anywhere in the paper, not just in the front matter.
68: %% As in the title, use \\ to force line breaks.
69:
70: \author{Ben Davies\altaffilmark{1,2}, Livia Origlia\altaffilmark{3},
71: Rolf-Peter Kudritzki\altaffilmark{4}, Don F.\ Figer\altaffilmark{2},
72: R.\ Michael Rich\altaffilmark{5}, Francisco Najarro\altaffilmark{6} }
73:
74: \affil{$^{1}$School of Physics \& Astronomy, University of Leeds,
75: Woodhouse Lane, Leeds LS2 9JT, U.K.}
76: \affil{$^{2}$Chester F.\ Carlson Center for Imaging Science, Rochester
77: Institute of Technology, 54 Lomb Memorial Drive, Rochester NY, 14623,
78: USA}
79: \affil{$^{3}$INAF-Osservatorio Astronomico di Bologna, Via Ranzani 1, I-40127 Bologna, Italy}
80: \affil{$^{4}$Institute for Astronomy, University of Hawaii, 2680
81: Woodlawn Drive, Honolulu, HI, 96822, USA}
82: \affil{$^{5}$Department of Physics and Astronomy, University of California at Los Angeles, Los Angeles, CA 90095-1547, USA}
83: \affil{$^{6}$Instituto de Estructura de la Materia, Consejo Superior
84: de Investigaciones Cientificas, Calle Serrano 121, 28006 Madrid,
85: Spain.}
86:
87:
88: \begin{abstract}
89: The Galactic Centre (GC) has experienced a high degree of recent
90: star-forming activity, as evidenced by the large number of massive
91: stars currently residing there. The relative abundances of chemical
92: elements in the GC may provide insights into the origins of this
93: activity. Here, we present high-resolution $H$-band spectra of two Red
94: Supergiants in the GC (IRS~7 and VR~5-7), and in combination with
95: spectral synthesis we derive abundances for Fe and C, as well as other
96: $\alpha$-elements Ca, Si, Mg Ti and O. We find that the C-depletion in
97: VR~5-7 is consistent with the predictions of evolutionary models of
98: RSGs, while the heavy depletion of C and O in IRS~7's atmosphere is
99: indicative of deep mixing, possibly due to fast initial rotation
100: and/or enhanced mass-loss. Our results indicate that the {\it current}
101: surface Fe/H content of each star is slightly above Solar. However,
102: comparisons to evolutionary models indicate that the {\it initial}
103: Fe/H ratio was likely closer to Solar, and has been driven higher by
104: H-depletion at the stars' surface. Overall, we find $\alpha$/Fe
105: ratios for both stars which are consistent with the thin Galactic
106: disk. These results are consistent with other chemical studies of the
107: GC, given the precision to which abundances can currently be
108: determined. We argue that the GC abundances are consistent with a
109: scenario in which the recent star-forming activity in the GC was
110: fuelled by either material travelling down the Bar from the inner
111: disk, or from the winds of stars in the inner Bulge -- with no need to
112: invoke top-heavy stellar Initial Mass Functions to explain anomalous
113: abundance ratios.
114:
115:
116: %Thus, the recent star-forming
117: %activity in the GC could have been fuelled by material travelling down
118: %the Bar from the inner disk or from the winds of stars in the inner
119: %Bulge, with no need to invoke, for example, top-heavy stellar Initial
120: %Mass Functions to explain an abnormal $\alpha$/Fe ratio.
121: \end{abstract}
122:
123: %% Keywords should appear after the \end{abstract} command. The uncommented
124: %% example has been keyed in ApJ style. See the instructions to authors
125: %% for the journal to which you are submitting your paper to determine
126: %% what keyword punctuation is appropriate.
127:
128: \keywords{Galaxy: center, Galaxy: evolution, supergiants, stars:evolution,
129: stars:late-type}
130:
131: \section{Introduction} \label{sec:intro}
132: Understanding the star-formation history at the Galactic Centre (GC)
133: is key to understanding the Galaxy's secular evolution. The inner
134: 40\,pc has clearly experienced a high degree of recent star formation,
135: apparent from the young stellar clusters residing there. These
136: clusters host some of the most massive stars known to exist in the
137: Galaxy -- the Arches cluster \citep{Figer02}, the Quintuplet cluster
138: \citep{Figer99} and the Central cluster \citep{Paumard06}. It remains
139: an open question how these clusters were formed, given the harsh
140: environment of the GC, or whether they were formed {\it in situ} in
141: the GC at all.
142:
143: The key to answering this question may lie in chemical abundance
144: measurements. By probing the relative abundances of certain elements
145: we may infer the star formation history of the Galactic
146: Centre. Broadly, $\alpha$-elements (such as Mg, Ti, Ca, Si) are
147: produced predominantly in the core-collapse supernovae (CCSNe) of
148: massive stars. Meanwhile, Fe-peak elements are produced in Type-Ia
149: SNe, that is the thermonuclear explosion of a low-mass white dwarf
150: which has accreted a sufficient amount of mass from a binary companion
151: to reach the Chandrasekhar limit. Hence, $\alpha$ enrichment happens
152: on much shorter timescales ($\sim$10 Myrs), due to the comparitively
153: brief lifetimes of massive stars; while Fe-enrichment occurs on much
154: longer timescales ($\sim$Gyrs). In the Galaxy, the ratio of
155: $\alpha$/Fe is found to be lower in the thin disk than in the other
156: Galactic environments of the bulge and thick disk
157: \citep[e.g.][]{Bensby04,R-O05,Luck06}. This is commonly explained as
158: being due to the rapid formation of the bulge and halo, where star
159: formation in the early Galaxy was intense but brief, and the elevated
160: $\alpha$ abundances from the early CCSNe were `frozen-in'. In the thin
161: disk however, star formation has continued throughout the lifetime of
162: the Galaxy, allowing Type-Ia SNe to contribute to the chemical
163: evolution, and driving the $\alpha$/Fe ratio to below that of the halo
164: and bulge. Meanwhile, a gradient of increasing metal content is found
165: at lower Galacto-centric distances ($R_{\rm GC}$) within the disk
166: \citep[e.g.][]{Luck06}, suggesting that the rate of star formation in
167: the thin disk has been higher toward the GC. However, the absence of
168: suitable probes of metallicity inwards of $R_{\rm GC} \approx 4$kpc,
169: combined with the large extiction towards the GC, have meant that it
170: has been unclear whether the trend of increasing metal content flows
171: all the way to the center of the Galaxy.
172:
173: In recent years, advancements in infra-red astronomy have meant that
174: it is now possible to make abundance measurements of stars in the GC,
175: in order to place this region within the evolutionary framework of the
176: whole Galaxy. Infra-red studies are necessary to work around the 30
177: magnitudes of visual extinction toward the GC, with many such studies
178: concentrating on the large number of high-mass stars present
179: there. Massive stars provide excellent probes of chemical abundances;
180: their spectra are rich in emission/absorption lines, while their short
181: lifetimes mean that they provide an up-to-date picture of the local
182: metallicity.
183:
184: In studies of hot stars, \citet{Najarro04} and \citet{Martins08} both
185: attempted to infer the initial O abundance of the Arches cluster from
186: the asymptotic nitrogen abundances of the cluster's most evolved
187: stars. When comparing the results of both studies to the latest
188: stellar evolutionary models, in which the relative abundance ratios of
189: the heavy elements are fixed, each study suggests a slightly
190: super-Solar global metal content $Z/Z_{\odot}$, though neither study
191: was able to give abundances of specific $\alpha$ or Fe-peak
192: elements. In a study of two Luminous Blue Variables (LBVs) in the
193: Quintuplet cluster \citet{Najarro08} was able to make direct
194: measurements of Fe content, plus the $\alpha$-elements Si and Mg. They
195: found Fe to be approximately Solar, while Si and Mg were enhanced by
196: $\approx$ 0.3$\pm$0.2dex with respect to Solar
197: values. \citet{Geballe06} studied the star IRS~8 in the Central
198: cluster using low-resolution $K$-band spectroscopy. They argued that
199: the 2.116\microns\ feature, which is thought to be a blend of C~{\sc
200: iii}, N~{\sc iii} and O~{\sc iii} transitions, was sensitive to O
201: content, and used it to derive an O abundance which was super-Solar at
202: the 2$\sigma$ level when compared to the \citet{Asplund05} Solar
203: abundances\footnote{When the abundances from the 1-D Solar model of
204: \citet{G-S98} are used, the derived O abundance is consistent with
205: Solar.}.
206:
207: There have also been several studies of cool massive stars,
208: specifically Red Supergiants (RSGs). \citet{Carr00} performed the
209: first abundance study of a star in the GC, modelling the high
210: resolution $H$ and $K$ spectrum of the Red Supergiant (RSG) in the
211: Central cluster IRS~7. They found approximately Solar values for Fe to
212: within their quoted uncertainties, with a depletion of O with respect
213: to solar likely due to internal CNO processing. \citet{Ramirez00}
214: analysed a further five RSGs in the central cluster, using $K$-band
215: high resolution data, again finding Fe content consistent with
216: Solar. Ramirez et al.\ also studied the Quintuplet RSG VR5-7, finding
217: an Fe abundance consistent with the rest of their
218: sample. \citet{Cunha07} analysed this same set of stars as the Ramirez
219: et al.\ sample, adding $H$-band spectra to the dataset. Again, Solar
220: Fe content was found, but they also found marginal evidence for
221: $\alpha$ enhancement: the ratios of both O/Fe and Ca/Fe were found to
222: be above Solar at the 2$\sigma$ level, again in comparison to the
223: Asplund et al.\ Solar values.
224:
225: Studies of GC H{\sc ii} regions have also yielded marginal evidence
226: for super-Solar $\alpha$ abundances
227: \citep{Simpson95,Rudolph06}. Analysis of X-ray spectroscopy of the
228: Sg~A~East supernova (SN) remnant by \citet{Sakano04} shows strong
229: evidence for $\alpha$-enrichment, however if the object is a remnant
230: of a core-collapse SN one expects to see significant self-enrichment
231: of the $\alpha$-elements.
232:
233: The results described above are all somewhat consistent with a picture
234: of chemical abundances in the GC which are Solar to within the
235: uncertainties, with marginal evidence for
236: $\alpha$-enhancement. Several authors have suggested that the
237: super-Solar $\alpha$/Fe ratio is real, and is indicative of the star
238: forming history in the GC. Several explanations have been suggested to
239: explain this result:
240:
241: \begin{itemize}
242: \item A recent burst of star-formation produced a great number of
243: massive stars, which evolved to core-collapse SNe. This has enriched
244: the local environment in $\alpha$-elements, an effect which can now
245: be seen in the surface abundances of subsequent generations of
246: stars. However, this is generally inconsistent with the mix of old,
247: intermediate and young stellar populations found in the GC. Analyses
248: of the GC's H-R diagram and luminosity function have led various
249: authors to argue for a constant rate of star-formation over the last
250: $\sim$Gyr, rather than an isolated burst
251: \citep[e.g.][]{Blum03,Figer04,Maness07}.
252:
253: \item The harsh environment of the GC results in a Jeans mass larger
254: than in the rest of the Galaxy, and produces a top-heavy Initial
255: Mass Function (IMF) \citep{Morris93}. The increased rate of CCSNe
256: due to the higher fraction of massive stars results in a $\alpha$/Fe
257: ratio above that of the disk. This is consistent with the results of
258: \citet{Maness07}, who argued for a top-heavy IMF when comparing the
259: observed GC H-R diagram with quantitative model predictions.
260:
261: \item The GC's source of star-forming material is provided by the
262: winds of Red Giants in the Bulge, which tend to have high
263: $\alpha$/Fe ratios. Thus the natal material was
264: already $\alpha$-enriched before it arrived at the GC \citep{M-S96}.
265: \end{itemize}
266:
267: However, as stated above, the evidence for a high $\alpha$/Fe ratio is
268: marginal. Works which study both the Fe and $\alpha$ content directly
269: are limited to \citet{Cunha07} and \citet{Najarro08}; the former
270: studied O and Ca, while the latter studied Si and Mg. Oxygen can be a
271: poor tracer of $\alpha$ abundances in massive stars, as to some extent
272: it is processed by nuclear burning and can give results which are not
273: representative of the other $\alpha$ elements.
274:
275: Here we present an abundance study of two of the GC RSGs, IRS~7 in the
276: Central cluster and VR~5-7 in the Quintuplet cluster. For the first
277: time we study a range of $\alpha$-elements -- Ca, Si, Mg, O and Ti --
278: while also studying Fe and C.
279:
280: Our method is both independent and complementary to the previous
281: studies of these stars by \citet{Carr00}, \citet{Ramirez00} and
282: \citet{Cunha07}, providing full spectral synthesis of the observed
283: spectra and equivalent width measurements of selected lines. This
284: method has been proven to be effective in determining chemical
285: abundances of low mass giants in the Galactic bulge
286: \citep[e.g.][]{Origlia02,Origlia05a,Rich07}
287: as well as of young stellar clusters dominated by red supergiants
288: \citep{Larsen06,Larsen08}.
289:
290: We begin in Sect.\ \ref{sec:obs} with a description of our
291: observations and data-reduction steps, followed by a description of
292: our abundance analysis methods in Sect.\ \ref{sec:an}. The results are
293: presented and discussed in Sect.\ \ref{sec:results}.
294:
295:
296: \section{Observations \& data reduction} \label{sec:obs}
297: Observations were taken with NIRSPEC, the cross-dispersed echelle
298: spectrograph mounted on Keck-II, during the night of 3rd June 1999. We
299: used the instrument in high-resolution mode, with the NIRSPEC-5
300: filter, in conjunction with the 0.576\arcsec\ $\times$ 12\arcsec\
301: slit. The dispersion angle was set to 62.53\deg, with cross-dispersion
302: angle set to 35.53\deg. This gave us a spectral resolution of
303: $\sim$17,000 of select regions in the wavelength range of
304: 1.5-1.7\microns.
305:
306: We integrated on each star for 20s in each of two nod-positions along
307: the slit. In addition to the cluster stars, we also observed
308: HD~104337, a B2~IV star, as a telluric standard on each
309: night. Flat-fields were taken with a continuum lamp. For wavelength
310: calibration purposes, arc frames were taken with Ar, Ne, Xe and Kr
311: lamps to provide as many template lines as possible in the narrow
312: wavelength range of each spectral order. As high-precision radial
313: velocity measurements were not required for this study, no etalon
314: frames were taken.
315:
316: We subtracted nod-pairs of frames to remove the sky background, dark
317: current and detector bias level. Each frame was then flat-fielded
318: using the continuum-lamp exposures. We corrected for the warping of
319: each order in a process known as {\it rectification}. The warping in
320: the spatial and dispersion directions were characterized by fitting
321: 3rd-degree polynomials to the two star-traces in a nod-pair and the
322: arc lines respectively. These fits were then used to resample each
323: order onto a linear grid. As the arc-line wavelengths are known, this
324: process also wavelength-calibrates the data.
325:
326: Cosmic ray hits and bad pixels were identified by taking the ratio of
327: two spectra of the same object, and identifying values lying outside
328: 5$\sigma$ of the residual spectrum. These values were replaced with
329: the median value of the three neighbouring pixels either side. We
330: removed the H and He~{\sc i} absorption features of the telluric
331: standard via linear interpolation. The atmospheric absorption features
332: in the science frames were then removed by dividing through by the
333: telluric standard. Finally, the spectra were normalised by dividing
334: through by the median continuum value. The dense absorption-line
335: spectra make signal-to-noise ratio (SNR) estimates difficult, but we
336: estimate from weak features we know are real that the SNR is better
337: than 100 for all stars observed.
338:
339: \subsection{Data analysis} \label{sec:an}
340: Abundance analysis is performed by using full spectral synthesis
341: techniques and equivalent width measurements of representative lines.
342: Indeed, at the NIRSPEC resolution of ~17,000 a few single
343: roto-vibrational OH lines and CO bandheads can be measured and used to
344: derive accurate oxygen and carbon abundances. Abundances of other
345: metals can be derived from the atomic lines of Fe~I, Mg~I, Si~I, Ti~I,
346: and Ca~I.
347:
348: Photometric estimates of the stellar parameters are initially used as
349: input to compute a grid of synthetic spectra of red supergiant stars
350: with varying atmospheric parameters and abundances, by using an
351: updated \citep{Origlia02,Origlia03} version of the code described in
352: \citet{Origlia93}. Briefly, the code uses the LTE approximation and is
353: based on molecular blanketed model atmospheres of \citet{Johnson80}. The
354: code includes several thousands of near IR atomic lines and molecular
355: roto-vibrational transitions due to CO, OH and CN. Three main
356: compilations of atomic oscillator strengths are used, namely the
357: Kurucz database\footnote{\tt
358: http://cfa-www.harward.edu/amdata/ampdata/kurucz23/sekur.html}, and
359: those published by \citet{B-G73} and \citet{M-B99}.
360:
361: Abundance estimates are mainly obtained by best-fitting the full
362: observed spectrum and by measuring the equivalent widths of a few
363: selected features, dominated by a specific chemical element -- see
364: Table \ref{tab:ew} for a list of some of the the diagnostic lines,
365: along with their oscillator stregths and excitation potentials. The
366: H-band spectra of RSGs contain thousands of absorption lines, and we
367: concentrate on fitting those spectral features which are relatively
368: unblended and which have reliable atomic/molecular data. We do not
369: attempt to fit features which are blends of poorly-understood
370: transitions. Typical equivalent width values of fitted features range
371: between 300 and 800 m\AA\ with a conservative error of
372: $\pm$20~m\AA\ to also account for a $\pm$2\% uncertainty in the
373: continuum positioning.
374:
375: For initial estimates of $T_{\rm eff}$ we use the values quoted by
376: \citet{Cunha07}. We also use the intrinsic luminosities determined in
377: this paper in combination with the Geneva evolutionary models to
378: estimate the surface gravity $\log g$. The values of $\log g$ and the
379: microturbulent velocity $\xi$ are fine-tuned using the CO and OH
380: lines, which are sensitive to these parameters. We found that $\log g
381: = 0.0$ and $\xi = 3$\kms\ provided good fits for both stars.
382:
383: We note that the parameter of macroturbulence is not specifically
384: included in our current analysis. In previous spectral synthesis
385: studies of the objects in this paper authors have required the
386: inclusion of macroturbulent velocities in the region of
387: 15-25\kms. However, as these values are comparable to the spectral
388: resolution of our observations (18\kms), we do not attempt to
389: constrain this parameter. We do find some evidence for an additional
390: broadening component in the spectrum of IRS~7, this is discussed
391: further in Sect.\ \ref{sec:results}.
392:
393: The model which better reproduces the overall observed spectrum and
394: the equivalent widths of selected lines is chosen as the best fit
395: model for that particular spectrum. In order to check the statistical
396: significance of our best-fit solution, we compute a set of six {\it
397: test} models with varying stellar parameters of T$_{\rm
398: eff}$=$\pm$200\,K, $\log g = \pm$0.5dex, and $\xi = \pm$1\kms. For
399: each test model, the abundances $A(X)$ are fine-tuned to produce the
400: best fit in that model, typically by $\pm$0.2dex. We then compute the
401: residuals between these models and the observed spectrum. We find that
402: our best-fit model always produces significantly lower residuals than
403: the test models (see Fig.\ \ref{fig:err}). This process was repeated
404: with smaller variations in the stellar parameters of the test models
405: until we found a set of test models which produced residuals
406: comparable to those of the best-fit model. We found that models with
407: $\delta$T$_{\rm eff}$=$\pm$100\,K, $\delta (\log g) = \pm$0.3dex) and
408: $\delta \xi = \pm$0.5\kms\ had the same statistical significance as
409: our best-fit model . We adopt these values as the experimental
410: uncertainties in each parameter. The uncertainties in abundance levels
411: varies from element to element, and depends on other factors such as
412: blending and continuum placing. The abundance uncertainties are
413: typically in the range 0.1-0.2dex.
414:
415: \begin{figure}[p]
416: \centering
417: \includegraphics[width=15cm]{fig1.eps}
418: \caption{Select regions of the two stars studied (dotted-line),
419: overplotted with the best-fitting model for each (solid-line). The
420: left panel shows IRS~7 in the Central Cluster; the right panel
421: shows VR~5-7 in the Quintuplet Cluster.}
422: \label{fig:spec}
423: \end{figure}
424:
425:
426: %stellar parameter uncertainty of
427: %$\pm$200~K in temperature (T$_{eff}$), $\pm$0.5 ~dex in log-gravity
428: %(log~g) and $\pm$0.5~km~s$^{-1}$ in microturbulence velocity ($\xi$),
429: %can introduce a further systematic $\le$0.2~dex uncertainty in the
430: %absolute abundances. However, {\it test models} with stellar
431: %parameters varying by $\Delta $T$_{\rm eff}$$\pm$200~K, $\Delta
432: %$log~g=$\pm$0.5 and $\Delta \xi$$\mp$1.0~~km~s$^{-1}$ with respect to
433: %the best-fit, and abundances varying accordingly by
434: %$\approx\pm0.2$~dex, in order to still reproduce the depth of the
435: %observed features, are always less statistical significant than the
436: %best fit solution. Hence, as a conservative estimate of the
437: %systematic error in the derived best-fit abundances, due to the
438: %residual uncertainty in the adopted stellar parameters, one can assume
439: %a value of $\approx \pm 0.1$~dex.
440:
441: \begin{table*}
442: \scriptsize
443: \begin{center}
444: \caption[]{Oscillator strengths (Log {\it gf}), excitation potentials
445: ($\chi$ in eV) and equivalent widths (EW) of some unblended
446: representative lines for the observed stars in the Galactic Centre.}
447: \label{tab:ew}
448: \begin{tabular}{lcccccccc}
449: \hline\hline
450: & Ca~{\sc i} & Fe~{\sc i} & Fe~{\sc i} & Mg~{\sc i} & Si~{\sc i} & OH & OH & Ti~{\sc i} \\
451: & $\lambda $1.61508& $\lambda $1.61532&$\lambda $1.55317&$\lambda $1.57658&$\lambda $1.58884&$\lambda $1.55580&$\lambda $1.55036&$\lambda $1.55437\\
452: \hline \\
453: %& \multicolumn{9}{c}{Log {\it gf}}\\
454: %\hline
455: Log {\it gf} &0.362&-0.821&-0.357&0.380&-0.030&-5.492&-7.687&-1.480\\
456: $\chi$ (eV) &5.302&5.35 &5.64&5.93&5.08&0.30&0.84&1.88\\ \\
457: %\hline
458: %\multicolumn{9}{c}{EW (m\AA)} \vspace{2mm} \\
459: {\it EW (m\AA)} \vspace{2mm} \\
460: %\hline
461: ~~IRS7 & 412 & 420 & 360 & 520 & 745 & 529 & 505 & 352 \\
462: ~~VR5-7& 468 & 410 & 303 & 532 & 704 & 597 & 628 & 345 \vspace{2mm} \\
463: \hline
464: \end{tabular}
465: \end{center}
466: \end{table*}
467:
468:
469:
470:
471: %\end{itemize}
472:
473: %{\bf ...LIVIA... some values have been corrected, especially the Solar values of Grevesse & Sauval 1998....}
474:
475:
476: \begin{deluxetable}{lcccccccccccccc}
477: \tabletypesize{\scriptsize}
478: \tablecaption{Derived temperatures and abundances for the stars studied, and
479: comparisons to previous works. \label{tab:results}}
480: \tablewidth{0pt}
481: \tablehead{
482: \colhead{Reference} &
483: \colhead{$\rm T_{eff} (K)$}&
484: \colhead{$\log g$}&
485: \colhead{$\xi$/\kms}&
486: \colhead{A(Fe)}&
487: \colhead{A(O)}&
488: \colhead{A(Si)}&
489: \colhead{A(Mg)}&
490: \colhead{A(Ca)}&
491: \colhead{A(Ti)}&
492: \colhead{A(C)}
493: }
494: \startdata
495: \multicolumn{9}{l}{\small IRS 7} \vspace{1mm} \\
496: This work & 3600 & 0.0 & 3.0 & 7.59 & 8.11 & 7.75 & 7.53 & 6.46 & 5.22 & 7.77 \\
497: \vspace{1mm} & $\pm$ 200 & $\pm$ 0.3 & $\pm$ 0.5 & $\pm$ 0.10 & $\pm$ 0.13 & $\pm$ 0.18 & $\pm$ 0.15 & $\pm$ 0.16 & $\pm$ 0.11 & $\pm$ 0.07 \\
498:
499: Carr et al.\ 2000 & 3600 & -0.6 & 3.0 & 7.50 & 8.13 & -- & -- & -- & -- & 7.78 \\
500: \vspace{1mm} & $\pm$ 230 & $\pm$ 0.2 & $\pm$ 0.3 & $\pm$ 0.13 & $\pm$ 0.32 & -- & -- & -- & -- & $\pm$ 0.13 \\
501:
502: Cunha et al.\ 2007 & 3650 & -0.5 & 3.2 & 7.66 & -- & -- & -- & 6.86 & -- & -- \\
503: \vspace{1mm} & $\pm$ 150 & $\pm$ 0.3 & $\pm$0.3 & $\pm$ 0.15 & -- & -- & -- & $\pm$ 0.15 & -- & -- \\
504:
505: Ramirez et al.\ 2000 & 3470 & -0.6 & 3.3 & 7.61 & -- & -- & -- & -- & -- & -- \\
506: \vspace{3mm} & $\pm$ 250 & $\pm$ 0.2 & $\pm$0.4 & $\pm$ 0.27 & -- & -- & -- & -- & -- & -- \\
507:
508: \multicolumn{9}{l}{\it Average Central Cluster abundances:} \vspace{1mm} \\
509: Cunha et al.\ 2007 & -- & -- & -- & 7.59 & 9.04 & -- & -- & 6.57 & -- & -- \\
510: \vspace{1mm} & -- & -- & -- & $\pm$ 0.15 & $\pm$ 0.19 & -- & -- & $\pm$ 0.14 & -- & -- \\
511: Ramirez et al.\ 2000 & -- & -- & -- & 7.61 & -- & -- & -- & -- & -- & -- \\
512: \vspace{1mm} & -- & -- & -- & $\pm$ 0.22 & -- & -- & -- & -- & -- & -- \\
513: \\
514: \hline
515: \\
516: \multicolumn{9}{l}{\small VR\,5-7} \vspace{1mm} \\
517: This work & 3400 & 0.0 & 3.0 & 7.55 & 9.09 & 7.65 & 7.53 & 6.50 & 5.15 & 8.22 \\
518: \vspace{1mm} & $\pm$ 200 & $\pm$ 0.3 & $\pm$ 0.5 & $\pm$ 0.10 & $\pm$ 0.11 & $\pm$ 0.14 & $\pm$ 0.14 & $\pm$ 0.13 & $\pm$ 0.09 & $\pm$ 0.07 \\\
519:
520: Cunha et al.\ 2007 & 3600 & -0.15 & 2.6 & 7.60 & -- & -- & -- & 6.51 & -- & -- \\
521: \vspace{1mm} & $\pm$ 150 & $\pm$ 0.3 & $\pm$ 0.3 & $\pm$ 0.15 & -- & -- & -- & $\pm$ 0.15 & -- & -- \\
522:
523: Ramirez et al.\ 2000 & 3500 & -0.2 & 2.9 & 7.61 & -- & -- & -- & -- & -- & -- \\
524: \vspace{3mm} & $\pm$ 300 & $\pm$0.3 & $\pm$0.5 & $\pm$ 0.23 & -- & -- & -- & -- & -- & -- \\
525:
526: \multicolumn{9}{l}{\it Average Quintuplet Cluster abundances:} \vspace{1mm} \\
527: Najarro et al.\ 2008 & -- & -- & -- & 7.54 & -- & 7.85 & 7.84 & -- & -- & -- \\
528: \vspace{1mm} & -- & -- & -- & $\pm$ 0.15 & -- & $\pm$ 0.25 & $\pm$ 0.20 & -- & -- & -- \\
529: \\
530: \hline
531: \\
532: \multicolumn{9}{l}{\small Solar values} \vspace{1mm} \\
533: Grevesse \& Sauval 1998 & -- & -- & -- & 7.50 & 8.83 & 7.55 & 7.52 &
534: 6.36 & 5.02 & 8.57 \\
535: \vspace{1mm} & -- & -- & -- & $\pm$ 0.05 & $\pm$ 0.06 & $\pm$ 0.05 &
536: $\pm$ 0.05 & $\pm$ 0.02 & $\pm$ 0.06 & $\pm$ 0.06 \\
537:
538: Asplund et al.\ 2005 & -- & -- & -- & 7.45 & 8.66 & 7.51 & 7.53 & 6.31
539: & 4.90 & 8.39 \\
540: \vspace{1mm} & -- & -- & -- & $\pm$ 0.05 & $\pm$ 0.05 & $\pm$ 0.04 &
541: $\pm$ 0.09 & $\pm$ 0.04 & $\pm$ 0.06 & $\pm$ 0.05 \\
542: \enddata
543: \end{deluxetable}
544:
545:
546:
547:
548: \begin{figure}[t]
549: \centering
550: \includegraphics[width=12cm]{fig2.eps}
551: \caption{The observed spectrum of VR~5-7 ({\it histogram})
552: overplotted with the best-fit model ({\it solid-line}) and test
553: models, which vary the stellar parameters and abundances,
554: illustrating the robustness of the best-fit solution. Bottom-left
555: panel: varying abundances only; bottom right panel: effective
556: temperature varied and abundances adjusted accordingly; top-left
557: panel: microturbulence varied with abundances adjusted; top-right
558: panel: gravity varied and abundances adjusted.}
559: \label{fig:err}
560: \end{figure}
561:
562:
563: \section{Results \& Discussion} \label{sec:results}
564: In Fig.\ \ref{fig:spec} we show select regions of each stars' spectrum
565: containing diagnostic lines, with the best-fit model overplotted. In
566: the case of VR~5-7 we found we were able to fit the diagnostic lines
567: without the inclusion of any additional broadening. Therefore, any
568: macroturbulence present must be below or comparable to the
569: instrumental broadening -- the spectral resolution of our
570: observations, 18\kms. This is consistent with previous studies of this
571: object at higher spectral resolution, which have required
572: macroturbulent velocities of 12-15\kms\ \citep{Ramirez00,Cunha07}.
573:
574: For IRS~7, we found that the fit could be improved by convolving the
575: spectrum with a broadening profile -- either using a Doppler profile
576: of 20\kms, or by increasing the gaussian (instrumental) broadening to
577: 27\kms. These values are consistent with macroturbulent velocities of
578: 20-25\kms\ used by Ramirez et al.\ and Cunha et al.\ in their fits to
579: higher resolution spectra. The fit to IRS~7 shown in
580: Fig.\ \ref{fig:spec} includes this extra Doppler broadening. In
581: general, we found that we were unable to achieve the same quality of
582: fit for IRS~7 as for VR~5-7, and this is reflected in the slightly
583: increased uncertainties in the derived elemental abundances. One
584: possible explanation for this could be the interaction of the star's
585: outer atmosphere with nearby hot stars, augmenting the ionization
586: structure and/or inducing departures from spherical symmetry in
587: IRS~7's envelope \citep{Serabyn91,Y-Z91,Y-Z92}.
588:
589: In Table \ref{tab:results}, the derived abundances are shown for each
590: element studied, as well as the derived stellar temperature. We find
591: temperatures gravities and microturbulent velocities for each star
592: which are in good agreement with previous studies of the same
593: objects. We quantify the abundance of each element $X$ in the form
594: $A({X}) = \log(X/{\rm H}) + 12$. We also tabulate the abundances
595: derived for the same stars by other authors using independent
596: methods. Finally, we also show the average abundances derived for the
597: stars' host clusters from other independent studies in the
598: literature. Below, we describe the elemental abundances individually.
599:
600: \subsection{Iron}
601: From Table \ref{tab:results} we see that the Fe abundances we derive
602: are in excellent agreement with previous studies of the same
603: objects. They are also consistent with the average abundances of their
604: host clusters: our measurement of $A({\rm Fe}) = 7.59 \pm 0.10$ for
605: IRS~7 agrees perfectly with the average $A$(Fe) of 10 RSGs in the
606: Central cluster studied by \citet[][$\bar{A}({\rm
607: Fe})=7.59\pm0.15$]{Cunha07}, and the 9 RSGs observed by
608: \citet[][$\bar{A}({\rm Fe})=7.61\pm0.22$]{Ramirez00}. Our measurement
609: for VR~5-7 ($A({\rm Fe}) = 7.55 \pm 0.10$) is also consistent with the
610: average of the two LBVs studied by \citet[][$\bar{A}({\rm Fe}) = 7.54
611: \pm 0.15$]{Najarro08}.
612:
613: When comparing the measured Fe composition for the two GC RSGs to
614: Solar abundances, we use the latest Solar Fe composition from the 3-D
615: Solar model of \citet[][ see Table~{\ref{tab:results}}]{Asplund05}. We
616: find Fe abundances with respect to Solar [Fe/H] of $+0.14 \pm 0.11$
617: for IRS~7, and $+0.10 \pm 0.11$ for VR~5-7, where the uncertainties
618: are the quadrature sum of those in our measurements and those in the
619: Solar values. Neither of these values represent a
620: statistically-significant detection of a departure from Solar
621: abundances. In order to minimise uncertainties, we consider our
622: analysis in combination with previous studies whose analysis
623: techniques were complementary to our own. As the studies by
624: \citet{Carr00}, \citet{Ramirez00} and \citet{Cunha07} were essentially
625: all made using similar methodology to one another, we do not consider
626: these three studies as being mutually independent. Instead, we take
627: the measurements of \citet{Cunha07} to be the state-of-the-art among
628: these studies. The average of Cunha et al.'s result for IRS~7 and ours
629: is (${\bar{A}({\rm Fe})} = 7.63 \pm 0.05$), which agrees well with the
630: average Central cluster value they found. Similarly, for VR~5-7 the
631: average is (${\bar{A}({\rm Fe})} = 7.58 \pm 0.04$). With respect to
632: the Solar values of \citet{Asplund05}, this gives [Fe/H]$_{IRS~7} =
633: 0.18 \pm 0.07$ and [Fe/H]$_{VR~5-7} = 0.13 \pm 0.06$.
634:
635: From these measurements, it appears that there is evidence for slight
636: Fe enhancement within the massive evolved stars at the GC. The
637: uncertainties we state above are the simple root-mean-square of the
638: abundances derived by \citet{Cunha07} and ourselves. It is likely
639: that, at the uncertainty level of a few percent, we are dominated by
640: systematic uncertainties which in some way afflict both studies, such
641: as departures from LTE in the stars' atmospheres. Thus, we conclude
642: that the two stars studied have surface Fe abundances which are
643: 30-50\% above that of Solar, though at this level we are hitting the
644: precision limits of stellar abundance analyses.
645:
646: However, when interpreting these results one must consider the evolved
647: nature of the objects studied. As it is likely that the atmospheres of
648: the stars suffer from H depletion due to core H burning and deep
649: convective mixing, the {\it initial} Fe abundances of the stars may
650: well have been entirely consistent with the Solar level. This is
651: illustrated in Fig.\ \ref{fig:feh}, in a plot of Fe/H against stellar
652: luminosity for the rotating Geneva stellar evolutionary models
653: presented in \citet{Hirschi04}. The plot shows how the Fe/H abundance
654: ratio increases as stars evolve off the main-sequence and the products
655: of nuclear burning begin to become apparent at the surface. On
656: average, the surface Fe/H ratio is $\sim$0.1dex lower during the RSG
657: phase than on the main-sequence.
658:
659: From comparison with our Fe measurements for the two GC RSGs, it
660: appears that while both stars currently show marginal evidence for Fe
661: enhancement, their initial Fe/H composition was likely much closer to
662: Solar. Similarly, the previous studies of stars in the GC by e.g.\
663: \citet{Najarro08} and \citet{Cunha07} are consistent with Solar
664: initial Fe abundances if one considers the evolved nature of the
665: objects studied.
666:
667: Finally, we note that a depletion in H due to stellar nucleosynthesis
668: implies a corresponding increase in He in the stellar envelope. This
669: has no impact on our analysis method, as the opacity in RSG
670: atmospheres is fully dominated by H$^{-}$, even if H is marginally
671: depleted.
672:
673:
674: \begin{figure}[t]
675: \centering
676: \includegraphics[width=8cm,bb=10 10 546 453]{fig3.eps}
677: \caption{Surface hydrogen depletion of massive stars as a function
678: of their evolution, illustrating that the measured Fe/H abundance
679: for evolved stars can be larger than that of their initial
680: metallicity. The overplotted mass-tracks are from \citet{Hirschi04}. }
681: \label{fig:feh}
682: \end{figure}
683:
684: \subsection{$\alpha$-elements}
685: In this study we have determined the abundances of several
686: $\alpha$-elements: Ca, Si, Mg, O, Ti and C. The interpretation of the
687: abundances of C and O is not as straight-forward as with the other
688: $\alpha$ elements, as they are altered by stellar nucleosynthesis. For
689: this reason we will discuss O separately at the end of this section,
690: and C in Sect.\ \ref{sec:carb}. For the rest of this study we define
691: $\alpha \equiv $(Ca, Si, Mg, Ti).
692:
693: Literature $\alpha$-element studies of these two stars are lacking;
694: the only previous study being the derivation of Ca abundances by
695: \citet{Cunha07}. From Table~\ref{tab:results}, we find excellent
696: agreement with their Ca abundance for VR~5-7. For IRS~7, we find
697: discrepancies between our two measurements of $\Delta A({\rm Ca}) =
698: 0.4 \pm 0.2$. However, we note that \citet{Cunha07} find an average
699: $A$(Ca) for the Central cluster of $\bar{A}({\rm Ca}) = 6.57 \pm
700: 0.14$, consistent with our measurement for IRS~7 (${A}({\rm Ca})_{\rm
701: IRS~7} = 6.46 \pm 0.16$). If the Ca abundances in the Central cluster
702: are homogeneous, it may be that, in Cunha et al.'s sample, IRS~7
703: represents a random statistical outlier at the 2$\sigma$ level --
704: entirely plausible in a sample size of 10. We consider our $A$(Ca)
705: measurements to be in good agreement with those of Cunha et al.
706:
707: \citet{Najarro08} studied the $\alpha$-elements Si and Mg in the
708: spectra of the two LBVs in the Quintuplet cluster, finding marginal
709: evidence for enhancement with respect to Solar, but with large
710: uncertainties of 0.2-0.25\,dex. Given these uncertainties, our
711: measurements of Si and Mg in the spectrum of VR~5-7 are consistent
712: with those determined by Najarro et al.\ for stars in the same
713: cluster.
714:
715: Overall, on average we find that the $\alpha$-elements studied here
716: are overabundant with respect to Solar by $\sim0.15$dex. This excess
717: is clearly within the uncertainties associated with each
718: measurement. If, for the moment, we assume that each element
719: represents an independent measurement, we can combine the abundances
720: for all four $\alpha$-elements studied here to determine the average
721: $\alpha$-abundance for each star. We find average abundances for each
722: star of [$\alpha$/H]$_{\rm IRS~7} = +0.18 \pm 0.07$; and
723: [$\alpha$/H]$_{\rm VR~5-7} = +0.15 \pm 0.05$, where the quoted errors
724: are the rms scatter of the individual $\alpha$-element
725: abundances. However, as was described in the previous section, these
726: abundance ratios may be tainted by $\sim$0.1dex of H self-depletion
727: due to the stars' evolution. Thus, we consider the {\it initial}
728: $\alpha$/H abundances of the two stars to be in excellent agreement
729: with the Solar value.
730:
731: The parameter of relevence to the study of the GC's star-formation
732: history is the ratio $\alpha$/Fe: as stated in the introduction,
733: $\alpha$-elements are enriched by CCSNe, while Fe is enriched by
734: Type-Ia SN. Hence, a short burst of massive star formation should
735: produce an over-abundance of $\alpha$-elements with respect to Fe;
736: while a long star-forming epoch over $\ga$Gyr eventually pulls the
737: $\alpha$/Fe ratio back down. From the average Fe/H and $\alpha$/H for
738: each star in this study, we find [$\alpha$/Fe]$_{\rm IRS~7} = -0.04
739: \pm 0.13$; and [$\alpha$/Fe]$_{\rm VR~5-7} = +0.01 \pm 0.12$. We find
740: no evidence for $\alpha$-enrichment with respect to iron in these
741: stars.
742:
743: For oxygen, comparisons of relative abundances between objects are
744: hindered by confusion over the Solar O abundance used as a reference
745: point. In the 1-D Solar model, the O fraction was found to be
746: $A$(O)$_{\odot} = 8.83 \pm 0.05$, a value which stayed constant to
747: within the errors over several revisions
748: \citep{A-G89,G-N93,G-S98}. However, in the 3-D Solar model, the O
749: fraction has been revised downwards to $A$(O)$_{\odot} = 8.66 \pm
750: 0.05$ \citep{Asplund05}. For the rest of this discussion, we use the
751: Asplund et al.\ value.
752:
753: For IRS~7, we find O/H to be depleted by 0.55$\pm$0.14\,dex with
754: respect to Solar, in good agreement with the study by \citet{Carr00}
755: of the same star. However, this is much lower ($\sim$0.9dex) than the
756: average O abundance of the star's host cluster: \citet{Cunha07} found
757: the Central cluster to have an average super-Solar O/H ratio of
758: 0.38$\pm$0.20\,dex. Interestingly, we find the O abundance of VR~5-7
759: in the neighbouring Quintuplet cluster to be similar to the average of
760: Central cluster ($A({\rm O})_{\rm VR~5-7} = +0.43 \pm 0.12$). The
761: anomolous O abundance of IRS~7 with respect to the rest of the GC RSGs
762: could be explained as global super-Solar O/H of RSGs in the GC, with
763: severe O depletion at the surface of IRS~7 due its advanced
764: evolutionary state -- this stars' evolution is discussed in more
765: detail in the next section. We find O abundances relative to Fe for
766: the two objects of [O/Fe]$_{\rm IRS~7} = -0.69 \pm 0.18$; and
767: [O/Fe]$_{\rm VR~5-7} = +0.33 \pm 0.16$.
768:
769: In summary, we find that the $\alpha$ abundances in the GC are
770: consistent with Solar, both as a fraction of H and Fe. We suggest that
771: there may be a slight enhancement of O with respect to H, though we
772: treat this result with caution due to the recent controversy over the
773: Solar O abundance.
774:
775:
776:
777:
778: \begin{figure}[t]
779: \centering
780: \includegraphics[width=8cm,bb=10 10 546 453]{fig4a.eps}
781: \includegraphics[width=8cm,bb=10 10 546 453]{fig4b.eps}
782: \caption{ {\it Left}: The surface carbon abundances of the two stars
783: in this study, in comparison to the evolutionary models of
784: \citet{Mey-Mae00}. We plot the mass tracks for the three initial
785: masses shown, using the Solar metallicity models with initial
786: rotational velocities of 300\kms. The arrow denotes the magnitude of
787: the change in the updated Solar carbon abundance with respect to
788: those used by Meynet \& Maeder in constructing the mass-tracks.
789: {\it Right}: Surface C abundances versus surface O abundances. The
790: overplotted mass-tracks are from the same evolutionary models as the
791: left-hand panel. The arrow again denotes the change in the measured
792: Solar composition with the updated Solar model.}
793: \label{fig:cl}
794: \end{figure}
795:
796: \subsection{Carbon} \label{sec:carb}
797: As with oxygen, the Solar carbon abundance has also been revised
798: downwards in the 3-D Solar model, from $A$(C)$_{\odot} = 8.57 \pm
799: 0.05$ \citep{A-G89,G-N93,G-S98} to $A$(C)$_{\odot} = 8.39 \pm 0.05$
800: \citep{Asplund05}. We use the Asplund et al.\ value in this
801: discussion.
802:
803: Regardless of the adopted value of $A$(C)$_{\odot}$, we find strong
804: evidence for C depletion in IRS~7. For this object we derive $A$(C)$ =
805: 7.77 \pm 0.07$, in good agreement with the previous study of by
806: \citet{Carr00}. This represents a carbon abundance relative to Solar
807: of {[C/H] = -0.62 $\pm$ 0.09}. This is in contrast to VR~5-7, which
808: is only marginally depleted in C compared to the Solar value.
809:
810: In Fig.\ \ref{fig:cl} we plot $A$(C) for each star studied against
811: their luminosities, which we take from \citet[][]{Cunha07}. To
812: illustrate the level of C-depletion as a function of the stars'
813: evolution, we overplot the theoretical mass-tracks of
814: \citet{Hirschi04}, calculated at Solar metallicity and which
815: incorporate initial rotational velocities of 300\kms. We note that
816: these tracks were created using the relative abundances of
817: \citet{A-G89} and \citet{G-N93}, and therefore use the higher initial
818: C fractions described above. Thus, all other things being
819: equal\footnote{We make the assumption that changes of a few $\times$
820: 0.1dex in the relative abundances of C and O will not affect, for
821: example, reaction rates or envelope opacties such that the
822: evolutionary path of the star would be significantly altered from
823: those plotted in Fig.\ \ref{fig:cl}, other than the linear
824: displacement in abundance-space.}, the tracks should be displaced by
825: -0.18dex in $A$(C) if the latest relative C abundance is used. This is
826: indicated by the arrow in the plot panel.
827:
828: From the figure, we see that VR~5-7 is in good agreement with the
829: 15\msun\ track, even if the track is moved downwards to reflect the
830: updated C fraction. However, the heavy depletion of C in IRS~7 is not
831: reproduced by the mass-tracks, even when the lower value for
832: $A$(C)$_{\odot}$ is taken into account. The sharp downturn in $A$(C)
833: of the 25\msun\ track corresponds to the removal of the H-rich
834: envelope and the evolution of the star {\it away} from the RSG stage
835: towards a Wolf-Rayet phase. On the 25\msun\ mass-track, the minimum
836: $A$(C) reached {\it during the RSG phase} is similar to the minimum
837: $A$(C) of the 15\msun\ and 20\msun\ tracks, at $A$(C)$\approx$8.45.
838:
839: We explore the subject of the stars' evolutionary state further in the
840: right-hand panel of Fig.\ \ref{fig:cl}, where we plot the carbon
841: fraction against the oxygen fraction for the two stars. The location
842: in the panel synonymous with the RSG phase corresponds to the
843: lower-left end of the 15\msun\ and 20\msun\ tracks at [$A$(O), $A$(C)]
844: $\approx (8.9, 8.4)$. If, as above, we again make the assumption that
845: the mass-tracks can simply be displaced to represent the updated Solar
846: abundances, the star VR~5-7 appears to be enhanced in O with respect
847: to the evolutionary tracks. IRS~7 on the other hand shows severe
848: depletion in both C and O, much more than is predicted by the
849: evolutionary models for an RSG with Solar initial abundances. Indeed,
850: the position of IRS~7 in $A$(C)-$A$(O) space is more consistent with
851: the $A$(C) minimum of the 25\msun\ track. In the model, this
852: corresponds to a nitrogen-rich Wolf-Rayet (WN) phase, where the
853: products of CNO burning are more readily seen due to the removal of
854: the outer envelope by the stellar wind. Our results suggest that the
855: envelope of IRS~7 has experienced a greater level of pollution by
856: CNO-processed material than is predicted by the standard set of
857: rotating evolutionary models. This may have been caused by
858: particularly deep convection zones in the star's atmosphere, or by
859: enhanced rotational mixing due to a high initial rotation
860: velocity. Alternatively, a high mass-loss phase could have peeled away
861: the outer layers, revealing more chemically-processed material, but
862: without removing enough H to cause the drop the envelope opacity
863: required for the star to evolve back to the blue.
864:
865:
866: \subsection{Summary and discussion: chemical abundances in the
867: Galactic Centre} \label{sec:ZGC}
868: The results of our abundance analyses of two RSGs in the GC can be
869: summarized as follows:
870:
871: \begin{itemize}
872: \item The observed Fe/H ratios of the two stars studied are marginally
873: super-Solar, in agreement with previous studies.
874:
875: \item For the $\alpha$-elements Ca, Si, Mg and Ti together, we find
876: enhancement with respect to H by 40-60\%, albeit at the 2$\sigma$
877: confidence level. The $\alpha$/Fe ratios are Solar to within the
878: errors ($\pm$25\%), a result which is again consistent with previous
879: studies.
880:
881: \item Weak evidence for O-enrichment is found in VR~5-7, while IRS~7
882: is heavily O-depleted - most likely a result of its advanced
883: evolutionary state.
884:
885: \item We find significant C depletion in the atmospheres of both
886: stars. In the case of VR~5-7, this is consistent with the
887: predictions of a 15-20\msun\ RSG in the rotating Geneva evolutionary
888: models. IRS~7 however has much greater C depletion. In combination
889: with this star's low O abundance, this suggests that the products of
890: severe CNO processing are been seen at the star's surface.
891: \end{itemize}
892:
893: As summarized in Sect.\ \ref{sec:intro}, several previous studies of
894: objects in the GC have argued for super-Solar metal abundances in the
895: GC \citep{Carr00,Ramirez00,Cunha07,Najarro08}. We emphasize here that,
896: where available, we measure elemental abundances $X/H$ for Fe and
897: several $\alpha$-elements that are consistent with these
898: studies. However, in the interpretation of their results these
899: previous studies did not take into account the self-depletion of H at
900: the surfaces of the stars due to nuclear burning and internal
901: mixing. We have shown, with the aid of evolutionary models, that this
902: level of self-depletion RSGs may be as large as 0.1dex, and is
903: comparable to the observed level metal enhancement with respect to
904: H. Hence, the super-Solar $X/H$ ratios may simply be an artifact of
905: H-depletion rather than being due to enriched metal abundances, and
906: the {\it initial} $X/H$ abundance ratios of the stars were consistent
907: with Solar levels. When taking this effect into account, we interpret
908: the results differently to previous works and conclude that the
909: metallic elemental abundances in the GC are Solar, at least to within
910: the precision to which we are currently able to measure them. In
911: addition, we find that the ratio $\alpha$/Fe is also Solar to within
912: the errors.
913:
914: Thus, with a Solar $\alpha$/Fe ratio in the GC it is not necessary to
915: invoke a peculiar star-forming history, such as a significantly
916: flattened IMF. Nor do we need to reconcile a super-Solar $\alpha$/Fe
917: ratio with star-formation which has been continuous for the last
918: $\sim$Gyr \citep[see Sect.\ \ref{sec:intro}, and ][]{Figer04}. We
919: again emphasize that our results of Solar $\alpha$/Fe and Solar Fe/H
920: to within the uncertainties {\it are in agreement with previous
921: studies} when H-depletion due to nuclear processing is taken into
922: account.
923:
924: So, how do these abundance patterns fit in with the rest of the
925: Galaxy? A Solar Fe abundance in the GC is certainly inconsistent with
926: a simple extrapolation of the disk's Fe gradient to lower
927: Galacto-centric distances. \citet{Luck06} summarize the recent results
928: of their group and show a clear trend of increasing Fe/H towards the
929: GC from a large sample of Cepheids. They derive a gradient of
930: -0.068dex\,kpc$^{-1}$, which when extrapolated all the way to the GC
931: implies $[{\rm Fe/H}] \approx +0.6$. Our results, as well as those of
932: other recent studies, show that the Fe/H ratio in the GC is more
933: consistent with that of the inner disk, at $R_{\rm GC} \sim 5$kpc (see
934: below).
935:
936: As noted by \citet{Cunha07}, reconciliation between the disk Fe/H and
937: that of the GC require that the Fe/H gradient become significantly
938: flattened inside $R_{\rm GC} \sim 5$kpc. However, there is no reason
939: to assume that this abundance gradient can simply be extrapolated all
940: the way to the GC. Shortward of $R_{\rm GC} \sim 5$kpc there is a
941: distinct lack of recent star-forming activity: there is a dearth of
942: giant HII regions and massive stars between 0.1kpc$\la R_{\rm GC}
943: \la5$kpc. Presumably this is related to the presence of the Galactic
944: Bar, which extends to $R_{\rm GC} \sim 4$kpc \citep{Benjamin05}. This
945: obvious and sharp transition in star-formation history makes it
946: unlikely that the abundance gradients in the outer Galaxy are valid
947: within $R_{\rm GC} \sim 4$kpc.
948:
949: We are left with the problem of explaining the abundance patterns in
950: the GC, with respect to those in the rest of the Galaxy. In external
951: barred-spiral galaxies, it is commonly accepted that the effect of the
952: bar is to transport gas from the inner disk to the galaxy's centre --
953: see for example the observational studies of M100 \citep{Allard06} and
954: NGC1365 \citep{ZS08}, as well as numerical work
955: \citep[e.g.][]{Athanassoula92}. Material infalling onto the galactic
956: centre becomes shocked, and a nuclear starburst is triggered. This
957: starburst can continue for as long as the centre is supplied with
958: material. For an extensive review of the secular evolution of barred
959: spiral galaxies, the reader is directed to \citet{K-K04}.
960:
961: If the star-formation in the centre of our own Galaxy is being fuelled
962: by matter transported down the Bar, one would expect to find similar
963: abundance patterns at the GC as at the edges of the Bar. That is, the
964: abundances we find in the thin disk at Galacto-centric distances of
965: $R_{\rm GC} \sim 4$kpc. Abundance studies at these radii are few. The
966: cepheid data from \citet{Luck06} becomes scarce at $R_{\rm GC} \la
967: 6$kpc with only one data-point at $R_{\rm GC} < 5$kpc. From their
968: inner-most data they find [Fe/H]$\sim$0.2-0.3dex and
969: [Ca/Fe]$\sim$0.0. \citet{Smartt01}, \citet{D-C04} and \citet{Munn04}
970: have observed a total of 6 OB stars within $R_{\rm GC} < 5$kpc,
971: finding weak evidence for a super-Solar O/H ratio, but with a large
972: spread. Analysis of the Scutum Red Supergiant clusters, located at
973: $R_{\rm GC} \approx 3.5-4$\, kpc, has revealed abundances which are
974: $\approx$0.2dex {\it below} Solar (Davies et al., in prep). In their
975: discussion of this result in the context of other abundance
976: measurements in the GC, Davies et al.\ suggest that the abundances at
977: $R_{\rm GC}$ from 3 to 5\,kpc are approximately Solar, with Solar
978: $\alpha$/Fe, but with a spread of $\pm$0.25dex. The abundance gradient
979: between the inner disk and our location in the Galaxy must then be
980: very shallow, as is typical for galaxies with strong bars
981: \citep{Zaritsky94}. \citet{Andrievsky02} suggested that the gradient
982: is extremely flat in the Solar neighbourhood
983: ($\approx$-0.02dex\,kpc$^{-1}$), and steepens at $R_{\rm GC} \la
984: 6$kpc. However, we will show in a forthcoming paper (Davies et al., in
985: prep) that this steepening may be an artifact of their spatial
986: sampling, that large azimuthal variations may be present within
987: $R_{\rm GC} \la 6$kpc, and that this behaviour is seen in the inner
988: disks of other barred spirals.
989:
990: The abundances found at $R_{\rm GC} \sim 4$kpc are then similar with
991: those we find at the GC; namely Fe/H, $\alpha$/Fe and O/Fe which are
992: all approximately Solar to within the uncertainties of
993: $\pm$20-30\%. Hence, from a chemistry point-of-view it is entirely
994: plausible that the star-formation in the GC is fuelled by material
995: channelled from the inner-disk by the Bar, in a manner similar to that
996: seen in external barred spirals. Further, the rate at which material
997: flows down the Bar seems to be consistent with the global
998: star-formation rate in the GC. \citet{M-S96} use order-of-magnitude
999: dynamical arguments to estimate the rate of material flowing from the
1000: Inner Lindblad Resonance (ILR) to be $\dot{M}_{\rm ILR} \sim
1001: 0.1-1$\msunyr; while \citet{Gusten89} estimate the GC star-formation
1002: rate to be 0.3-0.6\msunyr from the inferred number of Lyman
1003: photons. Hence, the rate of inflowing material appear to be sufficient
1004: to power the GC star-formation, even when one takes into account the
1005: outward-flowing Galactic wind \citep[$\dot{M}_{\rm GW} \la
1006: 0.1$\msunyr][]{M-S96}.
1007:
1008: Clearly, there are large uncertainties on these numbers, and we cannot
1009: discard the contribution of the winds from Bulge stars. Abundance
1010: studies of these objects have found similar average values of [Fe/H]
1011: and [O/Fe], albeit with a large spread in [Fe/H] of $\pm$0.4dex
1012: \citep{R-O05,C-S06,Lecureur07,Fulbright07}. The total mass-loss rate
1013: from these objects with $R_{\rm GC} \sim 2$kpc has been estimated to
1014: be $\sim$0.1\msunyr, though it is unknown how much of this material
1015: finds its way to the GC. If one wishes to add a further layer of
1016: complexity to the problem, it is possible that metal-rich gas from
1017: recent generations of massive stars mixes with metal-poor gas from the
1018: bulge stars' winds to contrive current abundance levels which appear
1019: approximately Solar \citep[see also][]{vL03}.
1020:
1021: In summary, from our abundance analysis of two RSGs in the GC,
1022: combined with the results of other studies, we find that both the Fe
1023: abundance and the $\alpha$/Fe ratio are consistent with Solar values,
1024: and those of the inner Galactic disk at $R_{\rm GC} \sim 4-5$kpc. When
1025: combined with mass-budget arguments, these results suggest that the
1026: recent star-formation in the GC was fuelled by material channelled
1027: along the Galactic Bar from the inner disk, though we cannot rule out
1028: a contribution from the winds of Bulge giants. Thus, there is no need
1029: to invoke of a top-heavy IMF to produce the relative chemical
1030: abundances. However, if a top-heavy IMF {\it does} exist in the GC,
1031: abundance levels in the GC may be held at Solar by a contribution of
1032: metal-poor gas from the winds of bulge giants.
1033:
1034: %\section{Conclusions}
1035:
1036:
1037: \acknowledgments Acknowledgements: we would like to thank Katia Cunha
1038: and Rob Kennicutt for useful discussions, and Raphael Hirschi for
1039: providing the Geneva mass-tracks at advanced evolutionary stages. We
1040: would also like to thank the referee for a careful reading of the
1041: manuscript and suggestions which helped improve the paper. The
1042: material in this work is supported by NASA under award NNG~05-GC37G,
1043: through the Long-Term Space Astrophysics program. This research was
1044: performed in the Rochester Imaging Detector Laboratory with support
1045: from a NYSTAR Faculty Development Program grant. The data presented
1046: here were obtained at the W.\ M.\ Keck Observatory, which is operated
1047: as a scientific partnership among the California Institute of
1048: Technology, the University of California, and the National Aeronautics
1049: and Space Administration. The Observatory was made possible by the
1050: generous financial support of the W.\ M.\ Keck Foundation. This
1051: research has made use of the IDL software package, and the GSFC IDL
1052: library.
1053:
1054:
1055:
1056:
1057:
1058: %\bibliographystyle{/fat/Data/bibtex/apj}
1059: %\bibliography{/fat/Data/bibtex/biblio}
1060: \bibliographystyle{/fat/Data/bibtex/apj}
1061: \bibliography{/fat/Data/bibtex/biblio}
1062:
1063:
1064: \end{document}
1065:
1066:
1067: