0803.3145/ms.tex
1: %\documentclass[12pt,preprint]{aastex}
2: 
3: %% manuscript produces a one-column, double-spaced document:
4: 
5: %\documentclass[manuscript]{aastex}
6: 
7: %% preprint2 produces a double-column, single-spaced document:
8: 
9: \documentclass[preprint2]{aastex}
10: 
11: %%\documentclass[preprint2,longabstract]{aastex}
12: 
13: %% If you want to create your own macros, you can do so
14: %% using \newcommand. Your macros should appear before
15: %% the \begin{document} command.
16: %%
17: %% If you are submitting to a journal that translates manuscripts
18: %% into SGML, you need to follow certain guidelines when preparing
19: %% your macros. See the AASTeX v5.x Author Guide
20: %% for information.
21: 
22: 
23: 
24: \shorttitle{Vega}
25: \shortauthors{Yoon et al.}
26: 
27: \begin{document}
28: 
29: \newcommand{\degree}{$^{\circ}$}
30: \newcommand{\ie}{i.\ e.,}
31: \newcommand{\eg}{e.\ g.,}
32: \newcommand{\etal}{et al.}
33: \newcommand{\I}{{\scriptsize I}}
34: \newcommand{\II}{{\scriptsize II}}
35: \newcommand{\III}{{\scriptsize III}}
36: \newcommand{\W}{$ \lambda\, $}
37: \newcommand{\LL}{$ \lambda\lambda\, $}
38: \newcommand{\vel}{\,km\,${\rm s^{-1}}\, $}
39: 
40: \title{The effect of rotation on the spectrum of Vega\altaffilmark{8}}
41: 
42: \author{Jinmi Yoon\altaffilmark{1,2}, Deane M. Peterson\altaffilmark{1,3},
43: Robert J. Zagarello\altaffilmark{4}, J. Thomas
44: Armstrong\altaffilmark{5,6}, and Thomas Pauls\altaffilmark{5,7}}
45: \altaffiltext{1}{Department of Physics and Astronomy, Stony Brook University,
46:     Stony Brook, NY 11794-3800}
47: \altaffiltext{2}{jyoon@grad.physics.sunysb.edu }
48: \altaffiltext{3}{dpeterson@astro.sunysb.edu}
49: \altaffiltext{4}{PCPION, South Setauket, NY 11720-1325; rzagarello@mail.astro.sunysb.edu}
50: \altaffiltext{5}{Remote Sensing Division, Code 7215, Naval Research Laboratory,
51:         4555 Overlook Avenue SW, Washington, DC 20375}
52: \altaffiltext{6}{tom.armstrong@nrl.navy.mil}
53: \altaffiltext{7}{pauls@nrl.navy.mil}
54: \altaffiltext{8}{Based 
55: on spectral data retrieved from the ELODIE archive at Observatoire de
56: Haute-Provence (OHP)}
57: 
58: \begin{abstract}
59: The discovery that Vega is a rapidly rotating pole-on star has raised
60: a number of questions about this fundamental standard, including such
61: issues as its composition, and in turn its mass and age. We report
62: here a reanalysis of Vega's composition. A full spectral synthesis
63: based on the Roche model derived earlier from NPOI interferometry is
64: used. We find the line shapes in Vega's spectrum to be more complex
65: than just flat-bottomed, which have been previously reported; profiles
66: range from slightly self-reversed to simple ``V'' shapes. A high SNR
67: spectrum, obtained by stacking spectra from the ELODIE archive, shows
68: excellent agreement with the calculations, provided we add about
69: 10\vel of macroturbulence to the predicted spectra. From the abundance
70: analysis, we find that Vega shows the peculiar abundance pattern of a
71: \W Bootis star as previously suggested. We investigate the effects of
72: rotation on the deduced abundances and show that the dominant
73: ionization states are only slightly affected compared to analyses
74: using non-rotating models. We argue that the rapid rotation requires
75: the star be fully mixed. The composition leads to masses and
76: particularly ages that are quite different compared to what are
77: usually assumed.
78: \end{abstract}
79: 
80: 
81: \keywords{line: profiles --- stars: abundances ---
82: stars: chemically peculiar --- stars: early-type --- 
83: stars: individual (Vega) --- stars: rotation}
84: 
85: \section{Introduction}
86: With the announcement of the detection of the interferometric
87: signature of rapid rotation in Vega
88: \citep{Peterson2004, Peterson2006b,Aufdenberg2006}, a number of
89: questions were raised about the fundamental standard. Earlier
90: suggestions of rapid rotation were based on the high luminosity
91: \citep{Petrie1964, Gray1988} of the object and the unusual shapes of the weak 
92: lines in the spectrum \citep{Gulliver1994, Hill2004}. The high
93: luminosity is immediately explained using Roche models for the figure
94: of the rotating star, von Zeipel's theorem \citep{Zeipel1924} to
95: characterize the temperature distribution, and adopting a nearly
96: pole-on geometry, required by the small line widths \citep{Gray1988}.
97: 
98: But this model, a star rotating near breakup, raises the question of
99: whether such fundamental issues as Vega's composition, mass, and age,
100: are accurately known. It has been recognized for some time that Vega
101: appears metal poor\,\citep{Sadakane1981, Adelman1990}. And although it
102: has been known since the early 20th century that masses deduced from
103: luminosity and radius measurements are strongly affected by
104: composition, recent mass and age determinations have largely assumed
105: solar composition, the assumption being that sharp lined A stars often
106: show abundance peculiarities that are assumed due to diffusion and
107: generally confined to surface layers. The recognition of rotation
108: velocities approaching breakup renders that assumption unlikely, since
109: rotation driven circulation is likely to mix the envelope completely
110: and deeply over times short compared to operable diffusion timescales.
111: 
112: Furthermore, the large surface temperature gradients that would be
113: associated with high rotation raise a new question: how seriously are
114: simple, single model atmosphere analyses of the spectrum affected by
115: the composite nature of the atmosphere? The peculiar line shapes add
116: to this concern.  A full analysis of the spectrum, or at least
117: representative spectral features, seems necessary to demonstrate that
118: we understand the peculiar line profiles and are able to derive
119: reliable abundances.
120: 
121: This in turn requires high resolution, low noise; spectra comparable to
122: those used by \citet{Hill2004}. The spectra of Vega available on the
123: ELODIE archive provide us the necessary resolution and low noise, we
124: describe those data in \S\,2.  The computation of the synthetic
125: spectra based on a rotational model are described in \S\,3 and the
126: deduced abundances and other characteristics we reported in \S\,4,
127: including the discovery that significant macroturbulence must be
128: adopted. In \S\,5 we discuss the implications of the abundance profile
129: and argue that the suggestion that Vega belongs to the \W Bootis class
130: of objects is probably correct. We note that the effect of rotation on
131: the line strengths depends strongly on the line considered and propose
132: a simple resolution to the prediction of large departures from LTE in
133: the Fe \I\, spectrum that have not been seen in practice. We examine more
134: closely the issue of rotational mixing and conclude that the abundances
135: we find here likely represent the material out of which Vega was
136: formed.  Lastly, we estimate the mass and age of Vega based on this
137: composition. 
138: 
139: After submitting this manuscript we became aware of a paper 
140: \citep{Takeda2008} that had been recently accepted for publication 
141: in this {\it Journal} which undertakes an analysis of their previously
142: published \citep{Takeda2007} spectra toward understanding Vega's
143: rotation, much along the lines taken earlier \citep{Gulliver1994,
144: Hill2004}.  These authors draw a number of conclusions in agreement
145: with what we find here.  But they also arrive at quite a different
146: physical model of Vega, concluding in the process that errors were
147: made in the reductions of the published interferometry.  We will
148: comment on these results at the appropriate points.
149: 
150: \section{The Observational Data}
151: The Vega spectra we used are from the ELODIE archive
152: \citep{Moultaka2004}, which contains high-resolution ($R \sim 42,000$)
153: echelle spectra from the ELODIE spectrograph obtained at the
154: Observatoire de Haute-Provence 1.93\,m telescope. The ELODIE data
155: pipeline automatically extracts the spectra, establishes the
156: dispersion, and corrects for scattered light. The spectra used here
157: were obtained between 1996 and 2004.  The wavelength rectified spectra
158: covering \LL 4000-6800 are provided with 0.05\,\AA \,sampling.
159:  
160: Barycenter corrections were required before co-adding the spectra. To
161: improve the signal-to-noise ratio (SNR) which is 250 for a typical
162: spectrum, we co-added 49 out of 71 available spectra. In the
163: process we rejected spectra whose SNR was less than 100 and those
164: showing noticeable fringing. We also replaced bad pixels whose
165: residuals in individual spectra were 5 times larger than the typical
166: noise by interpolating adjacent pixels. The co-added spectra were
167: converted to a residual intensity scale by normalizing them to the
168: scale of the synthetic spectra described in \S\,3. The resultant
169: spectra, segments of which are shown in Figures \ref{fig1} and
170: \ref{fig2}, were then compared to the synthesized spectra for the
171: abundance analysis. The SNRs of the co-added spectra were estimated to
172: range from 750 to 2,200 depending on the spectral regions.
173: 
174: \begin{figure*}
175: \epsscale{1.50}
176: \plotone{f1.eps}
177: %\includegraphics[scale =.8]{f1.eps}
178: \caption{Shown are line profiles representative of the range of shapes
179: encountered for weak lines in the ELODIE spectra of Vega.  The shapes
180: run from weakly ``self-reversed'' (e.g., Fe~\I\ \W4528 and Ba~\II\ 
181: \W4554) through flat-bottomed (Cr~\II\ \W4565 and S~\I\ \W6052) to             
182: ``V''-shaped (O~\I\ \W6046).  Where known, blends are indicated in              
183: parenthesis.  Wavelengths are in the star's rest frame.\label{fig1}}
184: \end{figure*}
185: 
186: 
187: \begin{figure*}
188: \epsscale{1.50}
189: \plotone{f2.eps}
190: %\includegraphics[scale =.7]{f2.eps}
191: \caption{ Plotted here are additional segments of spectra (dotted
192: lines) showing the range of shapes of weak lines, as in
193: Figure~\ref{fig1}, only now overplotted with the synthetic spectra
194: (continuous lines). Note particularly He I \W\,4713 which, with an
195: excitation potential of 21\,eV, is formed in a small region around the
196: rotational pole and displays the corresponding ``V'' shape.  At the
197: other extreme Ca~\I\ \W6162 shows the weak double-horned
198: (``self-reversed'') shape reflecting its very low excitation
199: potential, 1.9\,eV; it is contributed exclusively by the cooler
200: equatorial regions.  Other lines showing this behavior are Ti~\II\
201: \W4708, Fe~\I\ \W5586, and Ca~\I\ \W5588, although all three are
202: (weakly) blended.  Two iron lines, Fe~\II\ $\lambda\lambda$\,6147 and
203: 6149, at intermediate excitations of 3.9\,eV above the 7.8\,eV
204: ionization potential of Fe~\I, show the expected flat-bottomed shapes,
205: although seen against a slight variability in the background
206: continuum.  The weak Ca~\I\ lines indicated with ``:'' were not
207: included in the abundance determination. \label{fig2}}
208: \end{figure*}
209: 
210: Besides the ELODIE spectra, spectra of comparably high SNR and
211: resolution of Vega have been obtained at the Dominion Astrophysical
212: Observatory\,\citep[DAO;][]{Gulliver1994, Hill2004} and the Okayama
213: Astrophysical Observatory\,\citep[OAO;][]{Takeda2007}. The DAO spectra
214: (SNR $\sim 3,300$) have not yet been released publicly. However they
215: are available as part of a graphic
216: toolkit\footnote{\url{http://www.brandonu.ca/physics/gulliver/ccd\_atlases.html}}
217: which allows one to examine sections of spectra at high resolution and
218: identify lines and probable blends. We made extensive use of this tool
219: during this investigation.\footnote{At the same time R.L. Kurucz
220: (2007, private communication) provided a high resolution synthesized
221: spectrum for Vega in the 450--500\,nm region based on a line list and
222: gf values calibrated to a solar spectrum which proved extremely
223: useful.}
224: 
225: Also recently published are OAO spectra (SNR from 1,000 to 2,000 on
226: average) covering \LL 3900-8800. However, these spectra display
227: emission (\eg \,$\sim$\, \W 4560) and absorption (\eg
228: \,$\sim$\,\W6060) features and show the head of the Paschen continuum to be
229: strongly in emission, features not reported elsewhere. So we have
230: chosen to focus exclusively on the ELODIE data set.
231: 
232: \section{Computations}
233: We assume Vega can be described by a gravity-darkened Roche spheroid
234: in solid-body rotation, with a point mass gravitational potential,
235: showing a temperature distribution varying according to von Zeipel's
236: theorem \citep{Zeipel1924}, and seen nearly pole-on
237: \citep[\eg][]{Peterson2006b, Aufdenberg2006}. Because the recent
238: interferometric measurements taken at the Navy Prototype Optical
239: Interferometer \citep[NPOI;][]{Armstrong1998} and the Center for High
240: Angular Resolution Astronomy \citep{tenBrummelaar2005} array yield
241: closely similar model parameters, we adopt the parameters obtained
242: from the NPOI data \citep{Peterson2006b} for synthesizing spectra; the
243: model has a fractional rotation velocity, $\omega =0.926$, a polar
244: surface gravity of $\log g_{\rm{p}} = 4.074$, a polar effective
245: temperature of $T_{{\rm p}} = 9988$\,K, an inclination of the
246: rotational axis to the line of sight $i = 4.54$\,\degree, and a
247: projected rotational velocity of $v\sin i = 21.7$\vel. For details of
248: fitting Roche models to the NPOI data see \citet{Peterson2006a}, and
249: for issues specific to Vega see
250: \citet{Peterson2006b}, respectively.
251: 
252: To calculate the emergent spectrum we constructed a square $256 \times
253: 256$ grid which contains the apparent disk of the star, calculated the
254: stellar parameters at the center of each cell that actually fell on
255: the flattened disk, and computed an emergent flux as a function of
256: $\lambda$, $\mu$ (cosine of the angle between the local normal and the
257: line of sight), $T_{{\rm eff}}$, $g_{{\rm eff}}$ (local gravity
258: reduced by centrifugal force), and projected velocity using the ATLAS9
259: model atmosphere grid \citep{Castelli2003} and the atomic line data
260: given in the extensive compilation of \citet{Kurucz1995}. The fluxes
261: were integrated over the disk to yield the synthetic model
262: spectrum. In these calculations LTE, hydrostatic equilibrium, and
263: plane-parallel atmospheres were assumed to represent the star's
264: surface locally.
265: 
266: A concerns have been raised recently \citep{Aufdenberg2006, Monnier2007}
267: about the rigorous applicability of the von Zeipel theorem in the parts
268: of the disk of a rotating star that are rendered cool enough to
269: generate convection.  We believe the issue is not relevant to Vega.  In
270: our model the temperature drops to about 7600\,K at the equator, and the
271: effective gravity in turn decreases to about $\log g \sim 3.5$.  From a
272: model atmosphere with $T_{\rm eff}=7500$\,K, $\log g=3.5$ we find the
273: reduced density and in turn increased fraction of hydrogen ionized
274: compared to the main sequence, substantially decreases the extent of
275: the convective region and the efficiency of the resulting convection.
276: Convection carries significant flux only in the range of $1 \leq
277: \tau_{Rosseland} \leq 30$, well out from the interior where the flux
278: requirement is established.
279: 
280: \section{Results}
281: \subsection{Line Shapes}
282: The abundance analysis was done by adjusting each element abundance
283: until the model spectra fit the co-added spectra. Since
284: Vega's lines are sharp and blending is minimal, the process of
285: adjusting the abundances was straightforward.  Several representative
286: regions of the co-added ELODIE spectrum are shown in Figures
287: \ref{fig1} and \ref{fig2}.  Weak lines throughout the spectrum show not
288: only the flat-bottomed shapes (Cr \II\ \W4565, S \I\ \W6052, and Fe
289: \II\ \W6147) as noted in recent studies \citep{Gulliver1994,Hill2004} 
290: but also weakly ``self-reversed'' shapes such as Mg \I\ \W4702 and Ca
291: \I\ \W6162 and ``V'' shapes such as He\,\I\ \W4713 and O \I\ \W6046.
292:         
293: The unusual shapes of the weak lines are strongly correlated with
294: excitation and ionization potential and can be understood in terms of
295: how the Boltzmann factors amplify the temperature gradient across the
296: disk. Since Vega is seen nearly pole on, the center of the apparent
297: disk is almost exactly at one pole, the hottest point on the star. On
298: the other hand, the limb is nearly the equator which is not only
299: 2,400\,K cooler than the pole, but the visible gas is actually cooler
300: still owing to the simple projection effects associated with
301: limb-darkening. Therefore the bound states responsible for the lines
302: seen from the light elements such as He \I, O \I, Mg \II, Al \II, and
303: Si \II\ whose ionization and excitation potentials are quite high are
304: excited mostly at the axis with zero projected velocity. There is
305: almost no contribution to the line profiles from the rotationally
306: shifted equatorial region, resulting in ``V'' shapes. The lower the
307: excitation potentials the lines have, the more enhanced the
308: contribution from the more rapidly rotating equatorial regions becomes
309: and the wider and more square shaped the line profiles get. For the
310: elements such as Ca \I, Fe \I, and Ba \II\ with the lowest excitation
311: potentials one sees a mild double-horned shape (``self-reversed'') as
312: the contribution from the equatorial region completely dominates the
313: profile. In this sequence the flat-bottomed shape is formed at
314: intermediate excitation potentials such as those of the lines 
315: of Cr \II\ and Fe \II. Our synthetic spectra predict well this sequence of
316: line shapes as shown in Figure \ref{fig2}, where three regions of the
317: ELODIE spectrum, overplotted with our synthetic spectrum, are
318: shown. We see that weak Fe \II\ lines tend to have a flat-bottomed
319: shape while weak Fe \I\ lines show a self-reversed shape. Examination
320: of data presented by \citet{Gulliver1994} and particularly in
321: \citet{Hill2004} suggests these shapes are present in their data as
322: well.
323: 
324: \subsection{Macroturbulence}
325: In the process of the spectral synthesis, in order to fit the shapes
326: of the weak line we found we had to reduce the resolution of the
327: spectra well below the nominal resolution of 42,000 of the ELODIE
328: spectra, ultimately adopting a resolution of about 25,000 as shown in
329: Figure \ref{fig3}. We interpret this additional broadening, which was
330: accomplished by convolving the synthetic spectrum with a Gaussian, as
331: adding 10\,km\,${\rm s^{-1}}$ of macroturbulence to the nominal ELODIE
332: resolution (also assumed to be a Gaussian). The effect of this
333: additional broadening is most noticeable in steep-sided line profiles
334: (\eg\, Mg \I\ \W4703 and Ni \I\ \W4713), as shown in Figure
335: \ref{fig3}. As a result the value determined for the macroturbulence
336: comes from low excitation lines and hence refers more to the
337: equatorial regions than the polar regions. As might be expected,``V''
338: shaped lines such as He \I\ \W4713 are insensitive to the
339: macroturbulence, as also shown in Figure \ref{fig3}. This is a very
340: interesting result which we discuss at more length below. One caution
341: is immediately apparent though: line widths might not be reliable
342: indicators of actual projected velocity, at least for stars seen at
343: low inclination.
344: 
345: \begin{figure*}
346: %\plotone{f3.eps}
347: \includegraphics[angle=270,scale=.65]{f3.eps}
348: \caption{These figures plotted as in Figure~\ref{fig2} 
349: illustrate the need for line broadening in addition to 
350: rotation and microturbulence (in both panels the lower spectra are
351: offset by 0.1). Mg~\I\ \W4702 shows the problem most 
352: clearly although it is also evident in Ni~\I\ \W4714.  The nominal 
353: ELODIE resolution of 42,000 (assumed to be Gaussian) allows too much 
354: structure in the steep-sided line profiles.  Reducing the resolution to 
355: 25,000 appears to be required, which we interpret as a contribution of 
356: about 10 km\,s$^{-1}$ of macroturbulence.  The effect of adding this 
357: macroturbulence is to improve the fit dramatically in the bottoms of 
358: the weak, low-excitation lines while causing the line widths to be a 
359: bit wide.  This suggests the actual projected rotation rate is below 
360: the adopted 21.7 km\,s$^{-1}$, as was suggested in the initial 
361: interferometric data reductions reported by \citet{Peterson2006b}. 
362: Note that high excitation lines like He~\I\ \W4713 are not affected by 
363: the added macroturbulence. \label{fig3}}
364: \end{figure*}
365: %\clearpage
366: At this point our analysis deviates sharply from the recent
367: contribution from \citet{Takeda2008}, who seem not to have considered
368: the possibility of large scale non-thermal line broadening.  That
369: there could, and even should, be turbulence on large scales in the
370: atmosphere of Vega seems easy to justify. Even very slow, cm\,s$^{-1}$,
371: subsurface circulation currents will be magnified by the many order of
372: magnitude drop in density found in the outer envelope, as required by
373: the equation of continuity. Add to this a very strong Coriolis force
374: owing to the rapid rotation and a surface covered with large eddies -
375: cyclones - is to be expected.  Ignoring this possibility,
376: \citet{Takeda2008} were forced to adopt a relatively slowly rotating
377: model, creating a clear conflict with the interferometric measurements
378: \citep{Peterson2006b,Aufdenberg2006}.
379: 
380: \subsection{Abundance Analysis and Microturbulence}
381: As is often the case, we found that it was generally not possible to
382: find abundances for elements (or even the same ion of an element)
383: which gave good fits to both strong and weak lines
384: simultaneously. This is usually taken as a signal that some
385: microturbulence needs to be introduced. To this end, we determined the
386: abundances from the Fe \II\ lines for two choices of the
387: microturbulence, as shown in Figure \ref{fig4}. Here the abundances
388: are given as the logarithm of the ratio of the number of an element to
389: that of total elements, $\log\frac{N_{{\rm el}}}{N_{{\rm
390: tot}}}$. \citet{Castelli1979} and
391: \citet{Sadakane1981} have previously noted that the influence of
392: the microturbulence is less important in the visual region for
393: lines of intermediate strengths about $40 \sim 70 $\,m\AA\,, which
394: we also found. For Fe \II, which has the widest range of equivalent
395: widths, we find both the scatter and any trend with equivalent
396: width are significantly reduced for a microturbulence of about
397: 2\vel which we subsequently adopt.  The O \I\ triplets also support
398: 2\vel (\eg\ Figure \ref{fig2} which shows only the case for
399: 2\vel).
400: 
401: 
402: 
403: \begin{figure}
404: \epsscale{.80}
405: \plotone{f4.eps}
406: \caption{ These plots show the derived abundances (data points) versus
407: equivalent widths of Fe \II\ lines for two different assumed values of
408: the microturbulence. Panels a) and b) show the derived abundances for
409: microturbulence values of 2\vel and 4\vel, respectively. The dashed
410: line shows the unweighted average abundances and the solid line shows
411: the trend with equivalent width.  We adopt a microturbulence of 2\vel
412: in our abundance determinations. \label{fig4}}
413: \end{figure}
414: 
415: Table \ref{tbl-1} shows the deduced abundances for Vega with
416: a microturbulence of $\xi_T =$ 2\,km\,s$^{-1}$.  In selecting lines we
417: eliminated severe blends but included weak blends where we felt
418: reliable abundances could be obtained. The columns are the laboratory
419: wavelength, lower excitation potential, equivalent width, $\log gf$,
420: and the deduced abundance ($\log\frac{N_{{\rm el}}}{N_{{\rm tot}}}$).
421: Blends we have decided to retain are noted in the last column. The
422: abundances for elements with only single lines such as Al \II, S \I,
423: Mn \I, and Ni \I\ must be considered uncertain. Even where there was
424: no obvious blending, abundances were determined exclusively by
425: spectral synthesis. Nevertheless, we give equivalent widths for
426: comparison with recent work; agreement is within 1--2\,m\AA\
427: typically.  Equivalent widths are missing where lines were not able to
428: be measured due to ``one-sided'' blends or difficulty in defining the
429: local continuum level.
430: 
431: Notable in Table \ref{tbl-1} is the discrepancy between Fe \I\ and Fe
432: \II\ abundances. The abundances of Fe reported by \citet{Adelman1990}
433: do not show this dramatic lack of balance, and this might be viewed as
434: supporting the smaller temperature gradient derived by
435: \citet{Gulliver1994} and \citet{Hill2004}. In contrast with Fe, the
436: abundances of Mg \I\ and Mg \II\ shown in Table \ref{tbl-1} do not show
437: similar behavior. We discuss this result further below.
438: 
439: \section{Discussion}
440: \subsection{How Does Rotation Affect Abundances?}
441: The main difference between a pole-on rapidly rotating star, as
442: modeled here, and a classical plane-parallel stellar atmosphere model
443: is that for the same integrated colors, the rotating model has some
444: fraction of its surface at both higher and lower local effective
445: temperatures than the non-rotating model. The expected effect is that
446: there will be spectral lines in a range of excitation and ionization
447: energies where the two models give similar results. For Vega this is
448: the case when the sum of the excitation energy and ionization energy
449: (for lines of ions such as Fe \II) is about 10\,eV. But for both
450: higher and lower energy features the expanded range of temperatures
451: will enhance line strengths, resulting in a decrease in the deduced
452: abundances from those lines
453: \citep[described as ``intensification'' by][]{Takeda2008}.  
454: 
455: We see that trend here.  Table \ref{tbl-2} summarizes the results
456: of two recent analyses of Vega with non-rotating models
457: \citep{Sadakane1981,Adelman1990} along with the element-by-element
458: results determined here.  Our results from Mg~\I,
459: Mg~\II, Al~\II, Si~\II, Ti~\II, Cr~\II, Mn~\I, and Fe~\II\ are about
460: -0.6 dex below solar, roughly that found by earlier authors.  These
461: are mostly the dominant ionization stages and thus yield fairly stable
462: abundances.  Other ions, including C~\I, Ca~\I, Sc~\II, Fe~\I, Ni~\I,
463: and Ba~\II\ are deficient by about -1.0 dex, or even more, than the
464: solar abundances (C~\I\ by -0.6), and are depressed by typically
465: several tenths dex compared to the earlier studies, in this case
466: because of their relatively small excitation and ionization energies.
467: 
468: On the other side, the He abundance we deduce, $N_{{\rm He}}/N_{{\rm
469: tot}} =0.072 \pm 0.004$, which is essentially solar (0.078), is
470: substantially higher than that found by \citet{Adelman1990}, running
471: counter to expectations.  The result determined here is based on five of
472: the six lines in the ELODIE spectral window that are expected to be
473: measurable (\W5875 is heavily involved with atmospheric water vapor
474: lines), while the helium abundance of \citet{Adelman1990} is based on
475: \W4471 only. Otherwise, we have no explanation for why we obtain a larger
476: helium abundance.
477: 
478: In broadest terms, we find that if one can determine abundances from
479: lines of the dominant ionization stage of an element, the errors
480: induced by not accounting for rotation are small.  Where lines from
481: the dominant ionization stage are not accessible (\eg Ba~\III),
482: one can expect large corrections to be required when standard,
483: model-atmosphere analyses are applied to objects rotating near
484: breakup.
485: 
486: One interesting example of the problems that can arise because of
487: the corrections required between different ionization stages of
488: the same element, involves the ionization balance between Fe~\I\ and
489: Fe~\II.  Problems with the Fe~\I/Fe~\II\ ionization balance have
490: been reported for a wide range of stars
491: \citep[\eg][]{Gigas1986,Allende1999,Thevenin1999,Johnson2002}. For
492: Vega, departures from LTE are predicted to produce about 0.3 dex
493: errors in abundances deduced from Fe~\I\ lines while Fe~\II\ lines are
494: barely affected \citep{Gigas1986}.  However, these calculations are
495: difficult owing to the complexity of the atom and the lack of accurate
496: collision and photoionization cross sections.  For example,
497: \citet{Pradhan1995} have found that many of the photoionization cross
498: sections of Fe \I\ are significantly higher than those previously
499: adopted \citep[\eg][]{Gigas1986} with the possibility that the actual
500: corrections from departures from LTE are larger still.
501: 
502: The problem with the Fe balance in Vega is confusing since at first
503: glance, straightforward LTE analyses
504: \citep{Sadakane1981,Adelman1990} provide apparent agreement between
505: the abundances deduced from the two ions. This is in contrast to the
506: sizable departures from LTE required in other similar objects.
507: 
508: However, even though we assume LTE in our analysis here, we also find
509: a serious iron ionization imbalance amounting to $\sim 0.4$\,dex, but
510: in the opposite sense of that induced by non-LTE. To understand the
511: origin of this imbalance we reanalyzed representative lines from the
512: two iron ionization states and, as a check, from the two magnesium
513: ions present, using a standard plane-parallel model. We find that
514: rotation induces an apparent 0.35\,dex error in the Fe~\I/Fe~\II\
515: ionization balance, while the corresponding effect in the
516: Mg~\I/Mg~\II\ balance is only about 0.1\,dex.
517: 
518: Thus we reach the amusing conclusion that a simple LTE analysis of
519: Vega using models which do not account for rotation give a good
520: ionization balance because of a nearly complete cancellation of the
521: effects of photoionization-driven departures from LTE in the Fe~\I\
522: ion, on the one hand, and an enhanced Fe~\I\ line spectrum
523: contributed by the extensive cool equatorial regions of the model
524: owing to the favorable viewing geometry, on the other. Note however,
525: the near balance between these two effects may disappear when one
526: analyses lines in either the ultraviolet or infrared, owing to the
527: changing relative contribution of the equatorial regions to the
528: overall light.
529: 
530: We note that \citet{Takeda2008} have independently commented on the
531: near cancellation of departures from LTE versus the effects of
532: rotation in the iron ionization balance.  However, in their
533: calculation the rotation induced errors are predicted to be about
534: half those calculated here, owing to the much lower rotation
535: velocity and the corresponding dramatically reduced temperature
536: gradient ($\sim900$\,K) in their model.
537: 
538: 
539: 
540: \subsection{Is Vega a \W Bootis Star?}
541: 
542: Since \citet{Baschek1988} remarked that Vega showed an
543: abundance pattern similar to the \W Bootis stars, several studies
544: \citep[\eg][]{Venn1990,Ilijic1998} reported that Vega may be a mild
545: \W Bootis star.  We confirm that result here.  The abundance
546: pattern we deduce matches well the main characteristics of the
547: abundance patterns of \W Bootis stars as summarized, for example by
548: \citet{Heiter2002}. Elements such as Si, S, Ca, and Sc fall in the
549: middle of their respective typical ranges while O, Mg, Ti, Cr, Mn, and
550: Fe are on the high side of normal and Ni and Ba are on the low side.
551: While most elements fit the \W Bootis abundance pattern well, C and Al
552: are somewhat out of the reported range. The Al abundance is based on
553: one line and is not certain, while carbon is off the lower end of the
554: pattern reported by \citet{Heiter2002}.  However \citet{Paunzen1999},
555: in an extensive discussion of carbon and oxygen in this group of
556: objects, find several objects with carbon abundances as low as -0.7 dex
557:  with respect to the Sun. We conclude that Vega would not be rejected as a \W
558: Bootis star on the basis of its carbon abundance and the rest of the
559: abundances determined here are very much in keeping with membership in
560: this group.
561: 
562: \subsection{Is Vega Well Mixed?}
563: 
564: A presumption, often unstated, about the nature of $\lambda$\ Boo stars
565: is that the deviations from solar composition are limited to surface
566: layers \citep[\eg][]{Baschek1992, Holweger1993}, much the same as
567: has been concluded for the Ap and Am stars which also occupy this part
568: of the H-R diagram.  But there has always been some concern about that
569: assumption since unlike the latter groups the $\lambda$\ Boo stars
570: appear to have a distribution of rotation velocities similar to normal
571: stars \citep[\eg][]{Holweger1993}.
572: 
573: We argue here that since Vega is rotating at a significant fraction of
574: breakup and yet displays fairly typical \W Boo characteristics, it is
575: unlikely that these composition anomalies are limited to the surface;
576: more likely, Vega is well mixed. The literature on rotationally induced
577: mixing has generally focused on the surface layers and the question of
578: whether the Ap and Am phenomena could be understood as due to diffusive
579: separation \citep[\eg][]{Charbonneau1993} and not on how fast an 
580: inhomogeneity introduced on the surface would be mixed throughout the 
581: envelope.  
582: 
583: However, recent efforts to include the effects of rotation in
584: evolutionary calculations of massive stars \citep{Meynet1997} have led
585: to an examination of how inhomogeneities will be redistributed through
586: a star \citep{Talon1997,Ekstrom2008}, suggesting that extensive mixing
587: is to be expected. In fact, at the highest velocities in models down to
588: 3 $M_{\odot}$, the lowest mass examined, the mixing is predicted to be
589: so deep there is the possibility that some of the nuclear products from
590: the CNO burning region might be mixed to the surface.
591: 
592: This is an interesting possibility, given the low carbon abundance we
593: have found.  From this point of view, missing is an estimate of the
594: nitrogen abundance, the lines of which are out of the ELODIE spectral
595: range.  However measurements of nitrogen line equivalent widths have
596: been reported elsewhere.  To fill in the abundance of this important
597: nuclide taking full account of the effects of rotation, we have
598: calculated the abundances for the nitrogen equivalent widths reported
599: in \citet{Venn1990} for \W7442.28 ($\log {N/N_{{\rm tot}}}=-4.05$,
600: $[N/N_{{\rm tot}}]= +0.07$) and \W7468.29 ($\log N/N_{{\rm tot}}=-4.02$, 
601: $[N/N_{{\rm tot}}]=+0.1$) (\W7423 appears to be blended and we exclude
602: it here), finding values quite close to those deduced by
603: \citet{Venn1990} at about 0.085 dex above solar.  This is an
604: intriguing result.  Although it is difficult to know what ``normal''
605: is in this star, normalizing to oxygen gives [N/O] $\sim$ +0.2 and
606: [C/O] $\sim$ -0.5, which may very well indicate that some CN cycle
607: processed material has been mixed into the envelope of the star.  In
608: this regard we note that Vega represents a rather unique object; a few
609: other \W\ Boo objects have projected velocities in the vicinity of
610: 200\,km\,s$^{-1}$, but Vega is the one object known to rotate as fast
611: as 275\,km\,s$^{-1}$, less than 10\% from breakup in terms of angular
612: velocity.  However, without a better understanding of the composition
613: of the material Vega started with, or other supporting information, we
614: must leave this as just an intriguing possibility.
615: 
616: In summary, we believe a fairly strong case can be made for the outer
617: layers of Vega being well mixed, possibly even down to the edge of its
618: nuclear burning core.  If this is so then we are looking at about 2
619: $M_{\odot}$ of material of highly unusual composition in an object that
620: is much too young to display such extreme depletion in heavy elements.
621: In this case the various mechanisms put forth to explain the $\lambda$\
622: Boo phenomena that rely on its being limited to the superficial layers
623: \citep[\eg][]{Kamp2002} seem excluded for Vega.  Some form of dust -
624: gas separation, such as suggested by \citet{Venn1990} or
625: \citet{Holweger1992}, may be involved but if so the mechanism likely
626: must work at the time of Vega's formation since so much mass is
627: involved.  
628: 
629: \subsection{Determination of the Age and Mass of Vega}
630: We estimate the mass and age of Vega by locating its measured
631: luminosity and polar radius in an appropriate evolution grid, 
632: as described in \citet{Peterson2006b}. The interior models we adopt
633: are from the BASTI database\footnote{www.te.astro.it/BASTI/index.php} 
634: \citep[][and references therein]{Pietrinferni2006} which include 
635: evolutionary calculations using scaled solar and alpha-enhanced
636: compositions for stellar masses up to 2.4\,$M_\odot$. The composition
637: found here is not a perfect fit to either of the those mixtures, but
638: the large enhancement of oxygen is about the same compared to the
639: heavy metals as the alpha-enhanced mixture adopted there. Missing are
640: the other alpha-rich elements at the enhanced levels, but given the
641: dominance of oxygen even among these nuclei, that grid should give
642: results more than adequate. To quantify how much the mismatch in the
643: details of the distribution of abundances might affect the estimate we
644: also calculate the mass and age using the scaled solar grid. In both
645: cases, the heavy element fraction used is ${\rm
646: Z}=0.0093^{+0.0006}_{-0.0005}$ as calculated from Table
647: \ref{tbl-2} and assuming $[N_{{\rm el}}/N_{{\rm tot}}] = -0.7$ 
648: for abundances not obtained here.
649: 
650: For the alpha-enhanced composition we obtain $2.09\pm 0.03 M_{\odot}$
651: and $536\pm29$\,Myr for the mass and age. The simple scaled
652: solar abundances in turn yield $2.14 M_\odot$ and 541\,Myr and
653: since the alpha-enhanced models are a much closer match to Vega's
654: composition it is clear the errors introduced by the slight
655: mismatch are small compared to the other uncertainties.
656: 
657: As described in \citet{Peterson2006b}, whether Vega is solar
658: composition throughout or the derived abundances represent the actual
659: overall composition, results in quite different estimates for the
660: star's mass and age.  Most previous authors have assumed an underlying
661: solar composition yielding estimates of $2.3 M_{\odot}$ for the mass
662: and an age in the neighborhood of 360\,Myr.  Since there is a distinct
663: possibility that the composition we have derived applies to the star as
664: a whole, Vega's estimated mass may be reduced and its implied age
665: increased substantially.  One immediate consequence of this is a
666: growing clash with the properties of the so-called ``Castor moving
667: group'' \citep{BarradoyNavascues1998}, which includes Castor ($\alpha$
668: Gem), Fomalhaut ($\alpha$ PsA) and Alderamin ($\alpha$ Cep), in
669: addition to Vega, and whose members are estimated to have an age of
670: $200\pm100$\,Myr.  Even with an assumed solar composition Vega's age
671: was not a comfortable fit for inclusion in this group.  The increased
672: age we propose would make it an unlikely member.
673: 
674: 
675: \acknowledgments
676: This research was supported in part by a grant from the Naval Research
677: Laboratory to D. M. P. and in part by NSF grant 06-07612 to Dr. Michal
678: Simon. We also thank Dr. Robert L. Kurucz for extensive discussions.
679: 
680: \begin{thebibliography}{}
681: \bibitem[Adelman \& Gulliver(1990)]{Adelman1990} Adelman, S. J., \&
682: Gulliver, A. F. 1990, \apj, 348, 712
683: \bibitem[Allende Prieto et al.(1999)]{Allende1999} Allende Prieto, C., et
684: al. 1999, \apj, 527, 879
685: \bibitem[Armstrong et al.(1998)]{Armstrong1998} Armstrong, J. T., et
686:   al. 1998, \apj, 496, 550
687: \bibitem[Aufdenberg et al.(2006)]{Aufdenberg2006} Aufdenberg, J. P., et
688: al. 2006, \apj, 645, 664
689: \bibitem[Barrado y Navascu\'es(1998)]{BarradoyNavascues1998}Barrado y
690: Navascu\'es, D.\ 1998, \aap, 339, 831
691: \bibitem[Baschek \& Slettebak(1988)]{Baschek1988} Baschek, B., \&
692: Slettebak, A. 1988, \aap, 207, 112
693: \bibitem[Baschek(1992)]{Baschek1992} Baschek, B. 1992, The Atmospheres of
694: Early-Type Stars, in  Lecture Notes in Physics, ed. U. Heber and
695: C. S. Jeffery. (Springer-Verlag, Berlin Heidelberg New York) 401, 224
696: \bibitem[Castelli \& Faraggiana(1979)]{Castelli1979} Castelli, F., \&
697: Faraggiana, R. 1979, \aap, 79,174
698: \bibitem[Charbonneau(1993)]{Charbonneau1993} Charbonneau, P. 1993, 
699: IAU Colloq. 138, Peculiar versus Normal  Phenomena in A-type 
700: and Related Stars, ed. M. M. Dworetsky, 
701: F. Castelli, \& R. Faraggiana (San Francisco: ASP), 474
702: \bibitem[Castelli \& Kurucz(2003)]{Castelli2003} Castelli, F., \&
703: Kurucz, R. L. 2003, in IAU Symp. 210, Modeling of Stellar
704: Atmospheres, ed. N. E. Piskunov., W. W. Weiss, \& D. F. Gray (San
705: Francisco: ASP), poster A20 on the enclosed
706: CD-ROM (astro-ph/040587)
707: \bibitem[Ekstr\"om et al.(2008)]{Ekstrom2008} Ekstr\"om, S., et al. 2008, 
708: \aap, 478, 467
709: \bibitem[Gigas(1986)]{Gigas1986} Gigas, D. 1986, \aap, 165, 170 
710: \bibitem[Gray(1988)]{Gray1988} Gray, R. O. 1988, JRASC, 82, 336
711: \bibitem[Grevesse \& Sauval(1998)]{Grevesse1998} Grevesse, N., \&
712: Sauval, A. J. 1998, Space Sci. Rev., 85, 161
713: \bibitem[Gulliver, Hill, \& Adelman(1994)]{Gulliver1994} Gulliver,
714: A. F., Hill, G., \& Adelman, S. J. 1994, \apj, 429, L81
715: \bibitem[Heiter(2002)]{Heiter2002} Heiter, U. 2002, \aap, 381, 959
716: \bibitem[Hill, Gulliver, \& Adelman(2004)]{Hill2004} Hill, G., Gulliver, A. F.,
717: \& Adelman, S. J. 2004, in IAU Symp. 224, The A-Star Puzzle,
718: ed. J. Zverko, J. Ziznovsky, S. J. Adelman, \& W. W. Weiss (Cambridge:
719: Cambridge Univ. Press), 35
720: \bibitem[Holweger(1992)]{Holweger1992} Holweger, H. 1992, The Atmospheres of
721: Early-Type Stars, in  Lecture Notes in Physics, ed. U. Heber and
722: C. S. Jeffery. (Springer-Verlag, Berlin Heidelberg New York) 401, 48
723: \bibitem[Holweger \& St\"urenburg(1993)]{Holweger1993} Holweger, H., \& 
724: St\"urenburg, S. 1993, IAU Colloq. 138, Peculiar versus Normal 
725: Phenomena in A-type and Related Stars, ed. M. M. Dworetsky, 
726: F. Castelli, \& R. Faraggiana (San Francisco: ASP), 356
727: \bibitem[Iliji\'{c} et al.(1998)]{Ilijic1998} Iliji\'{c}, S., et
728: al. 1998, CoSka, 27, 461
729: \bibitem[Johnson(2002)]{Johnson2002} Johnson, J. A. 2002, \apjs, 139,
730: 219
731: \bibitem[Kamp \& Paunzen(2002)]{Kamp2002} Kamp, I., \& Paunzen,
732: E. 2002, \mnras, 335, L45
733: \bibitem[Kurucz \& Bell(1995)]{Kurucz1995}Kurucz, R. L., \& Bell, B. 1995,
734: CD-Rom No. 23, Smithsonian Astrophysical Observatory
735: \bibitem[Meynet \& Maeder(1997)]{Meynet1997} Meynet, G., \& Maeder, A. 
736: 1997, \aap, 321, 465
737: \bibitem[Monnier et al.(2007)]{Monnier2007} Monnier, J. D., et
738: al. 2007, Science, 317, 342
739: \bibitem[Moultaka et al.(2004)]{Moultaka2004} Moultaka, J., et al. 2004, \pasp,
740:     116, 693
741: \bibitem[Pauzen et al.(1999)]{Paunzen1999} Paunzen, E., et al. 1999,
742: \aap, 345, 597
743: \bibitem[Peterson et al.(2004)]{Peterson2004} Peterson, D. M., et al
744: 2004, SPIE, 5491, 65
745: \bibitem[Peterson et al.(2006a)]{Peterson2006a} Peterson, D. M., et al.
746: 2006a, \apj, 636, 1087
747: \bibitem[Peterson et al.(2006b)]{Peterson2006b} Peterson, D. M., et al.
748: 2006b, Nature, 440, 896
749: \bibitem[Petrie(1964)]{Petrie1964} Petrie, R. M. 1964,
750: Publ. Dom. Astrophys. Obs. Victoria, 12, 317
751: \bibitem[Pietrinferni(2006)]{Pietrinferni2006} Pietrinferni, A., et
752: al. 2006, \aj, 642, 797
753: \bibitem[Pradhan et al.(1995)]{Pradhan1995} Pradhan, A., et al. 1995,
754: BAAS, 27, 841
755: \bibitem[Sadakane \& Nishimura(1981)]{Sadakane1981} Sadakane, K., \&
756: Nishimura, M. 1981, \pasj, 33, 189
757: \bibitem[Takeda, Kawanomoto, \& Ohishi(2007)]{Takeda2007} Takeda, Y.,
758: Kawanomoto, S., \& Ohishi, N. 2007, \pasj, 59, 245
759: \bibitem[Takeda, Kawanomoto, \& Ohishi(2008)]{Takeda2008} Takeda, Y.,
760: Kawanomoto, S., \& Ohishi, N. 2008, \apj, preprint doi:10.1086/'528949'
761: \bibitem[Talon et al.(1997)]{Talon1997} Talon, S., et al. 1997, \aap, 322, 209
762: \bibitem[ten Brummelaar et al.(2005)]{tenBrummelaar2005} ten
763: Brummelaar, T. A., et al. 2005, \apj, 628, 453
764: \bibitem[Th\'{e}venin \& Idiart(1999)]{Thevenin1999} Th\'{e}venin,
765: F., \& Idiart, T. P. 1999, \apj, 521, 753
766: \bibitem[Venn \& Lambert(1990)]{Venn1990} Venn, K. A., \& Lambert,
767: D. L. 1993, \apj, 363, 234
768: \bibitem[von Zeipel(1924)]{Zeipel1924} von Zeipel, H. 1924, \mnras, 84,
769: 684
770: \bibitem[Zahn(1992)]{Zahn1992} Zahn, J. -P. 1992, \aap, 256, 115
771: \end{thebibliography}
772: 
773: 
774: \clearpage
775: \begin{deluxetable}{lrrrrl}
776: \tabletypesize{\scriptsize}
777: \tablecaption{The Abundance Analysis of Vega\label{tbl-1}}
778: \tablewidth{0pt}
779: 
780: \tablehead{
781: \colhead{$\lambda$} & \colhead{EP} &
782: \colhead{$w_{\lambda}$} &
783: \colhead{$\log gf$} &\colhead{log $\frac{N_{{\rm el}}}{N_{{\rm
784: tot}}}$\tablenotemark{a}} & \colhead{Blends} \\
785: \colhead{(\AA)} & \colhead{(${\rm cm^{-1}}$)}&
786: \colhead{( m\AA)} &
787: \colhead{} &\colhead{} & \colhead{}
788: }
789: \startdata
790: 
791: \multicolumn{6}{l}{He {\tiny I} ($\frac{N_{{\rm He}}}{N_{{\rm tot}}}$ =0.072 $\pm$ 0.004 )}\\
792: 4471.498 &  169087.008 & \nodata&    0.052 &  0.070 &\\  
793: 4713.139 &  169086.864 &  5 &   -1.233 &  0.078 &Fe {\tiny II} \W4713.193 \\     
794: 4921.931 &  171135.000 &  8 &   -0.435 &  0.060 & \\
795: 5015.678 &  166277.546 &\nodata &   -0.820 &  0.078 &  Fe {\tiny II} \W5015.755\\
796: 6678.154 &  171135.000 &  5 &    0.329 &  0.070 & Fe {\tiny II} \W6677.306 \\
797: \\
798: \multicolumn{6}{l}{C {\tiny I} ($\log\frac{N_{{\rm C}}}{N_{{\rm tot}}}$ = -4.14 $\pm$0.04, [$N_{\rm{C}}$/$N_{\rm{tot}}$]\tablenotemark{b} = -0.62)}\\
799: 4770.021 &  60352.639 &  7  &   -2.052 & -4.16& \\ 
800: 4771.730 &  60393.148 & 25  &   -1.488 & -4.16&\\
801: 4775.889 &  60393.148 &  7  &   -2.013 & -4.16&\\
802: 4932.050 &  61981.818 & 16  &   -1.574 & -4.06&\\
803: \\
804: \multicolumn{6}{l}{O {\tiny I} ($\log\frac{N_{{\rm O}}}{N_{{\rm tot}}}$ = -3.32 $\pm$ 0.04, [$N_{\rm{O}}/N_{\rm{tot}}$] = -0.11)}\\
805: 
806: 5329.099 &   86625.757 & 34\tablenotemark{c}  &   -1.730 & -3.31&\\
807: 5329.690 &   86627.778 & \nodata&   -1.410 & -3.31& \\
808: 5330.741 &   86631.454 & 24 &   -1.120 & -3.31&  \\
809: 6046.438 &   88631.146 & 10 &   -1.675 & -3.26& \\
810: 6155.971 &   86625.757 & 77\tablenotemark{d}  &   -1.051 & -3.36& \\  
811: 6156.778 &   86627.778 & \nodata &   -0.731 & -3.36&  \\
812: 6158.187 &   86631.454 & 59  &   -0.441 & -3.36& \\
813: \\
814: \multicolumn{6}{l}{Mg {\tiny I} ($\log\frac{N_{{\rm Mg}}}{N_{{\rm tot}}}$ = -5.12 $\pm$ 0.05, [$N_{\rm{Mg}}/N_{\rm{tot}}$] = -0.66)}\\
815: 
816: 4702.991 &   35051.264 & 29  &   -0.666 & -5.06& \\
817: 5167.321 &   21850.405 & 81  &   -1.030 &-5.06 & Fe {\tiny I} \W5167.488\\
818: 5172.684 &   21870.464 &102  &   -0.402 & -5.16 &\\
819: 5183.604 &   21911.178 &119  &   -0.180 & -5.16&\\
820: 5528.405 &   35051.264 & 28  &   -0.620 & -5.16 &\\          
821: \\
822: \multicolumn{6}{l}{Mg {\tiny II} ($\log\frac{N_{{\rm Mg}}}{N_{{\rm tot}}}$ = -5.06 $\pm$ 0.04, [$N_{{\rm Mg}}/N_{\rm{tot}}$] = -0.6)}\\
823: 
824: 4427.994 &   80619.500 & \nodata &   -1.210 & -5.06 &  \\
825: 4433.988 &   80650.020 &\nodata  &   -0.910 & -5.11 &Fe {\tiny I} \W4433.782\\ 
826: 4481.126 &   71490.190 &\nodata  &    0.740 & -5.01 &\\       
827: \\
828: \multicolumn{6}{l}{Al {\tiny II} ($\log\frac{N_{{\rm Al}}}{N_{{\rm tot}}}$ = -6.22, [$N_{{\rm Al}}/N_{\rm{tot}}$] =  -0.65)} \\
829: 
830: 4663.046 &   85481.350 &\nodata  &   -0.284 & -6.22 &\\  
831: \\
832: \multicolumn{6}{l}{Si {\tiny II} ($\log\frac{N_{{\rm Si}}}{N_{{\rm tot}}}$ = -5.15 $\pm$ 0.05, [$N_{{\rm Si}}/N_{\rm{tot}}$] = -0.66)}\\
833: 
834: 4128.054 &   79338.500 & 32 &    0.316 & -5.19 & Mn {\tiny II} \W4128.129\\
835: 4130.872 &   79355.020 & 54 &   -0.824 & -5.19 & \\ 
836: 5055.984 &   81251.320 & 60 &    0.593 & -5.19 & \\
837: 6347.109 &   65500.470 &118 &    0.297 &-5.09 & Mg {\tiny II} \W6346.742 \\
838:          &             &    &          &      & Mg {\tiny II} \W6346.964\\
839: 6371.371 &   65500.470 & 82 &   -0.003 & -5.09 &\\   
840: \\
841: \multicolumn{6}{l}{S {\tiny I} ($\log\frac{N_{{\rm S}}}{N_{{\rm
842: tot}}}$ = -5.01, [$N_{{\rm S}}/N_{\rm{tot}}$] = -0.3)}\\
843: 
844: 6052.674 &   63475.051 &  7 &   -0.740 & -5.01 &\\
845: \\
846: \multicolumn{6}{l}{Ca {\tiny I} ($\log\frac{N_{{\rm Ca}}}{N_{{\rm tot}}}$ = -6.72 $\pm$ 0.12, [$N_{{\rm Ca}}/N_{\rm{tot}}$] = -1.04)}\\
847: 
848: 4226.728 &       0.000 &\nodata  &    0.243 & -6.73 &\\   
849: 4434.957 &   15210.063 &\nodata  &   -0.029 & -6.73&\\
850: 4585.865 &   20371.000 &  1  &   -0.386 & -6.68 &\\ 
851: 5588.749 &   20371.000 &  1  &    0.210 & -6.63 &\\   
852: 5594.462 &   20349.260 &  5  &   -0.050 & -6.63 &\\  
853: 5598.480 &   20335.360 &  4  &   -0.220 & -6.63 &Fe {\tiny I} \W5598.287\\
854: 6162.173 &   15315.943 &  9  &    0.100 & -6.98 &\\   
855: \\
856: \multicolumn{6}{l}{Sc {\tiny II}  ($\log\frac{N_{{\rm Sc}}}{N_{{\rm tot}}}$ = -9.97 $\pm$ 0.05, [$N_{{\rm Sc}}/N_{\rm{tot}}$] = -1.1)}\\
857: 
858: 4246.822 &    2540.950 &  5  &    0.320 &-10.02 &\\
859: 5526.79  &   14261.320 &  9  &    0.130 & -9.92 &\\  
860: \\
861: 
862: \multicolumn{6}{l}{Ti {\tiny II} ($\log\frac{N_{{\rm Ti}}}{N_{{\rm tot}}}$ = -7.65 $\pm$ 0.09, [$N_{{\rm Ti}}/N_{\rm{tot}}$] = -0.63)}\\
863: 4468.507 &    9118.260 & 70  &   -0.600 & -7.82&\\
864: 4529.474 &   12676.970 &  9  &   -1.830 & -7.69&\\ 
865: 4563.761 &    9850.900 & 57  &   -1.010 & -7.51& \\
866: 4589.958 &    9975.920 & 16  &   -1.790 & -7.61&Cr {\tiny II} \W4589.901\\
867: 4708.665 &    9975.920 &  3  &   -2.410 & -7.69 & \\     
868: 4779.985 &   16515.860 & 12  &   -1.420 & -7.59 &\\  
869: 4805.085 &   16625.110 & 21  &   -1.100 & -7.59 &\\        
870: 5336.771 &   12758.110 & 12  &   -1.700 & -7.72 &\\  
871: \\
872: 
873: \multicolumn{6}{l}{Cr {\tiny II} ($\log\frac{N_{{\rm Cr}}}{N_{{\rm tot}}}$ = -6.91 $\pm$ 0.1, [$N_{{\rm Cr}}/N_{\rm{tot}}$] = - 0.54)}\\
874: 4252.632 &   31117.390 &  6  &   -2.018 & -6.97&\\
875: 4261.847 &   25033.700 & 18  &   -3.004 & -6.92 &Cr {\tiny II} \W4261.913\\
876: 4554.988 &   32836.680 & 20  &   -1.430 & -6.87 &  \\
877: 4558.650 &   32854.310 & 61  &   -0.660 & -6.87 &  \\ 
878: 4565.740 &   32603.400 &  7  &   -1.910 & -7.07 &  \\ 
879: 4588.199 &   32836.680 & 48  &   -0.830 & -6.87 &  \\
880: 4592.049 &   32854.950 & 18  &   -1.420 & -6.87 &  \\
881: 4616.629 &   32844.760 & 16  &   -1.530 & -6.87 &\\  
882: 4618.803 &   32854.950 & 36  &   -1.070 & -6.87 &\\  
883: 4634.070 &   32844.760 & 29  &   -1.220 & -6.82 &\\   
884: 4812.337 &   31168.580 &  6  &   -1.930 & -7.07 &\\
885: 4824.127 &   31219.350 & 39  &   -1.220 & -6.72&\\
886: 5334.869 &   32844.760 & 10  &   -1.562 & -7.07 &\\  
887: \\    	 
888: 
889: \multicolumn{6}{l}{Mn {\tiny I} ($\log\frac{N_{{\rm Mn}}}{N_{{\rm tot}}}$ = -7.45, [$N_{{\rm Mn}}/N_{\rm{tot}}$] = -0.8)}\\
890: 4783.405 &    18531.663 &  2 &  0.042 & -7.45&\\   
891: \\
892: 
893: \multicolumn{6}{l}{Fe {\tiny I} ($\log\frac{N_{{\rm Fe}}}{N_{{\rm tot}}}$ = - 5.51 $\pm$ 0.1, [$N_{{\rm Fe}}/N_{\rm{tot}}$] =-0.97)}\\
894: 4132.058 &   12968.553 & 29 &   -0.650 & -5.54 &Fe {\tiny I} \W4131.935\\
895:          &             &    &          &       &Fe {\tiny I} \W4131.971\\
896: 4134.677 &   22838.321 &  8 &   -0.490 & -5.54 &Fe {\tiny I} \W4134.42	\\
897: 4136.998 &   27543.001 &  4 &   -0.540 & -5.54 &\\	 
898: 4250.119 &   19912.494 & 18 &   -0.405 & -5.49 &Fe {\tiny II} \W4250.437\\
899: 4250.787 &   12560.933 & 27 &   -0.710 & -5.49 &Fe {\tiny II} \W4250.437\\
900: 4260.474 &   19350.890 & 36 &   -0.020 & -5.39 &   \\
901: 4466.551 &   22838.321 &  9 &   -0.590 & -5.36 &   \\
902: 4476.019 &   22946.814 &  8 &   -0.570 & -5.79 &Fe {\tiny I} \W4476.076\\
903: 4528.614 &   17550.180 & 15 &   -1.072 & -5.51 &   \\
904: 4918.994 &   23110.937 & 16 &   -0.640 & -5.51 &   Fe {\tiny I} \W4918.954 \\
905: 4920.502 &   22845.867 & 27 &   -3.955 & -5.51 &   Cr {\tiny II} \W4920.23 \\
906: 5324.179 &   25899.987 & 12 &   -0.240 & -5.49 &   \\
907: 5586.756 &   27166.818 & 11 &   -0.210 & -5.59 &  Fe {\tiny II} \W5587.114\\
908: 5615.644 &   26874.548 & 15 &   -0.140 & -5.44 &  \\
909: \\
910: 
911: \multicolumn{6}{l}{Fe {\tiny II} ($\log \frac{N_{{\rm Fe}}}{N_{{\rm tot}}}$ = -5.12 $\pm$ 0.09, [$N_{{\rm Fe}}/N_{\rm{tot}}$] = -0.58)}\\
912: 4258.154 &   21812.055 & 14 &  -0.467 & -5.34 & Fe {\tiny II} \W4258.34\\
913: %4472.929 &   22939.358 & 11 &  -0.570 & -5.34 & \\			    
914: 4520.224 &   22637.205 & 45 &  -2.990 & -5.13 &\\			    
915: 4522.634 &   22939.358 & 69 &  -2.700 & -5.01 &\\			    
916: 4576.340 &   22939.358 & 23 &  -3.390 & -5.06 &  \\			    
917: 4582.835 &   22939.358 & 17 &  -3.570 & -5.06 &  \\			    
918: 4583.837 &   22637.205 & 88 &  -2.490 & -4.97 &  Fe {\tiny II} \W4583.999\\
919: 4596.015 &   50212.826 & \nodata&  -2.057 & -5.21 &  Fe {\tiny II} \W4595.682\\
920: 4620.521 &   22810.357 & 14 &  -3.650 & -5.16 &  \\			    
921: 4635.316 &   48039.090 & 13 &  -1.650 & -5.21 &  \\			    
922: 4656.981 &   23317.633 & 12 &  -3.950 & -5.11 & Ti {\tiny II} \W4657.206  \\
923: 4663.708 &   23317.633 &  5 &  -4.145 & -5.11 & \\   		    
924: 4666.758 &   22810.357 & 11 &  -3.700 & -5.01 & \\ 			    
925: 4670.182 &   20830.582 &  8 &  -4.350 & -5.11 &  Sc {\tiny II} \W4670.407\\
926: 4923.927 &   23317.633 &114 &  -1.820 & -5.06 & \\  			    
927: %5169.033 &   23317.633 &132 &  -0.459 & -5.54 &\\			    
928: 5534.847 &   26170.181 & 25 &  -2.930 & -5.09 &  \\  		    
929: 6147.741 &   31364.440 & 14 &  -2.721 & -5.24 &    \\		    
930: 6149.258 &   31368.450 & 13 &  -2.724 & -5.24 & \\			    
931: %6369.375 &   57493.321 &  5 &  -3.669 & -5.04 &\\		    
932: \\
933: \multicolumn{6}{l}{Ni {\tiny I} ($\log\frac{N_{{\rm Ni}}}{N_{{\rm tot}}}$ =  -6.79, [$N_{{\rm Ni}}/N_{\rm{tot}}$] = -1.0)}\\
934: 4714.417 &    27260.894 &  3  &    0.160 & -6.79 &\\   
935: \\
936: 
937: \multicolumn{6}{l}{Ba {\tiny II} ($\log \frac{N_{{\rm Ba}}}{N_{{\rm tot}}}$ =  -11.21, [$N_{{\rm Ba}}/N_{\rm{tot}}$] = -1.3)}\\
938: 4554.029 &       0.000 & 13 &   0.430 &-11.21 & \\
939: 4934.076 &       0.000 &  7 &  -0.150 &-11.21 & Fe {\tiny I} \W4934.005\\
940: \\
941: 
942: \enddata
943: \tablenotetext{a}{For helium abundance, $\frac{N_{{\rm He}}}{N_{{\rm tot}}}$ }
944: \tablenotetext{b}{[$N_{{\rm el}}/N_{\rm{tot}}$] = $\log\frac{N_{{\rm el}}}{N_{{\rm
945: tot}}} - \log{(\frac{N_{{\rm el}}}{N_{{\rm tot}}})}_{\odot}$ }
946: \tablenotetext{c}{The equivalent width is for the blend with O {\tiny I} \W 5629.690.}
947: \tablenotetext{d}{The equivalent width is for the blend with O {\tiny I} \W 6156.778. }
948: 
949: 
950: \end{deluxetable}
951: \clearpage
952: 
953: \begin{table}
954: \begin{center}
955: \caption{Comparisons with the previous abundance studies\label{tbl-2}}
956: \begin{tabular}{lcccccc}
957: \tableline\tableline
958: Atomic && \multicolumn{2}{c}{log M/H } & &$\log\frac{N_{{\rm el }}}{N_{{\rm
959: tot}}}\tablenotemark{c}$ & [$N_{{\rm el}}/N_{\rm{tot}}$]\tablenotemark{d}\\
960: \cline{3-4}\cline{6-7}
961: Species  & &SN\tablenotemark{a} &AG\tablenotemark{b} & & \multicolumn{2}{c}{This work} \\
962: \tableline
963: He \I  &&\nodata&-1.52  && -1.14 &-0.04   \\
964: C \I   &&\nodata& -3.81 && -4.14 & -0.62 \\
965: N \I\,\tablenotemark{e}&&\nodata&\nodata &&-4.53 & +0.09\\ 
966: O \I   &&\nodata&\nodata&& -3.32 & -0.11 \\
967: Mg \I  &&-4.61  & -5.07 &&-5.12  & -0.66 \\
968: Mg \II &&-4.96  &-5.11  &&-5.06  & -0.60 \\
969: Al \II &&\nodata&-6.33  && -6.22 & -0.65 \\
970: Si \II &&\nodata&\nodata&& -5.15 & -0.66 \\
971: S \I   &&\nodata&\nodata&&-5.01  & -0.30 \\
972: Ca \I  &&-6.11  &-6.21  &&-6.72  & -1.04 \\
973: Sc \II &&-9.42  &-9.62  && -9.97 & -1.10 \\
974: Ti \II &&-7.31  &-7.47  && -7.65 & -0.63  \\
975: Cr \II &&-6.90  &-6.76  && -6.91 &-0.54  \\
976: Mn \I  &&-6.87  &-7.16  &&-7.45  & -0.80 \\
977: Fe \I  &&-5.09  &-5.05  && -5.51 & -0.97 \\
978: Fe \II &&-5.09  &-5.12  && -5.12 & -0.61 \\
979: Ni \I  &&-5.94  &-6.38  && -6.79 & -1.00  \\
980: Ba \II &&-10.25 &-10.58 &&-11.21 & -1.30 \\
981: \tableline
982: \end{tabular}
983: \tablecomments{The definition of the abundances we use differs from 
984: that adopted by SN and AG. For the helium abundance found here, the SN
985:  and AG abundances will be systematically larger than ours by 0.03 dex.}
986: \tablenotetext{a}{\citet{Sadakane1981}}
987: \tablenotetext{b}{\citet{Adelman1990}}
988: \tablenotetext{c}{Abundances from Table \ref{tbl-1}}
989: \tablenotetext{d}{Solar abundances have been taken from \citet{Grevesse1998}}
990: \tablenotetext{e}{Abundance based on \citet{Venn1990} equivalent widths}
991: \end{center}
992: \end{table}
993: 
994: 
995: \end{document}
996: