0805.2436/ms.tex
1: % mn2esample.tex
2: %
3: % v2.1 released 22nd May 2002 (G. Hutton)
4: %
5: % The mnsample.tex file has been amended to highlight
6: % the proper use of LaTeX2e code with the class file
7: % and using natbib cross-referencing. These changes
8: % do not reflect the original paper by A. V. Raveendran.
9: %
10: % Previous versions of this sample document were
11: % compatible with the LaTeX 2.09 style file mn.sty
12: % v1.2 released 5th September 1994 (M. Reed)
13: % v1.1 released 18th July 1994
14: % v1.0 released 28th January 1994
15: 
16: \documentclass[useAMS,usenatbib]{mn2e}
17: 
18: % If your system does not have the AMS fonts version 2.0 installed, then
19: % remove the useAMS option.
20: %
21: % useAMS allows you to obtain upright Greek characters.
22: % e.g. \umu, \upi etc.  See the section on "Upright Greek characters" in
23: % this guide for further information.
24: %
25: % If you are using AMS 2.0 fonts, bold math letters/symbols are available
26: % at a larger range of sizes for NFSS release 1 and 2 (using \boldmath or
27: % preferably \bmath).
28: %
29: % The usenatbib command allows the use of Patrick Daly's natbib.sty for
30: % cross-referencing.
31: %
32: % If you wish to typeset the paper in Times font (if you do not have the
33: % PostScript Type 1 Computer Modern fonts you will need to do this to get
34: % smoother fonts in a PDF file) then uncomment the next line
35: % \usepackage{Times}
36: 
37: %%%%% AUTHORS - PLACE YOUR OWN MACROS HERE %%%%%
38: 
39: \usepackage{graphicx}
40: 
41: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
42: 
43: \title[Abundances of lithium, sodium, and potassium in Vega]
44: {Abundances of lithium, sodium, and potassium in Vega}
45: \author[Y. Takeda]{Y. Takeda\thanks{E-mail:
46: takeda.yoichi@nao.ac.jp}\footnotemark[0]\thanks{Based on observations
47: carried out at Okayama Astrophysical Observatory (Okayama, Japan).}\\
48: National Astronomical Observatory of Japan, \\
49: 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan\\
50: }
51: \begin{document}
52: 
53: \date{Accepted 2008 May 12. Received 2008 May 8; in original form 2007 December 14}
54: 
55: %\pagerange{\pageref{firstpage}--\pageref{lastpage}} \pubyear{2002}
56: 
57: \maketitle
58: 
59: \label{firstpage}
60: 
61: \begin{abstract}
62: Vega's photospheric abundances of Li, Na, and K were determined 
63: by using considerably weak lines measured on the very high-S/N spectrum, 
64: while the non-LTE correction and the gravity-darkening correction 
65: were adequately taken into account. 
66: It was confirmed that these alkali elements are mildly 
67: underabundant ([Li/H] $\simeq -0.6$, [Na/H] $\simeq -0.3$, and 
68: [K/H] $\simeq -0.2$) compared to the solar system values, 
69: as generally seen also in other metals.
70: Since the tendency of Li being more deficient than Na and K is 
71: qualitatively similar to what is seen in typical interstellar 
72: cloud, the process of interstellar gas accretion may be related 
73: with the abundance anomaly of Vega, as suspected in the case of 
74: $\lambda$ Boo stars.
75: \end{abstract}
76: 
77: \begin{keywords}
78: stars: abundances -- stars: atmospheres -- stars: early-type 
79: -- stars: individual: Vega.
80: \end{keywords}
81: 
82: \section{Introduction}
83: 
84: While Vega (= $\alpha$~Lyr = HR~7001 = HD~172167 = HIP~91262; A0V)
85: plays an important role as the fundamental photometric standard,
86: its photospheric chemical composition is known to be sort of
87: anomalous. Namely, most elements are mildly underabundant
88: (by $\sim -0.5$ dex on the average) in spite of its definite population I
89: nature, though the extent of deficiency is inconspicuous for several
90: volatile elements of low condensation temperature ($T_{\rm c}$) such
91: as C, N, O, and S. Since this is the tendency more manifestly shown 
92: by a group of A--F main-sequence stars known as ``$\lambda$~Boo-type 
93: stars'' (see, e.g., Paunzen 2004 and the references therein), 
94: Vega's moderate abundance peculiarities have occasionally been argued
95: in connection with this $\lambda$~Boo phenomenon (e.g., Baschek \& Slettebak 1998;
96: St\"{u}renburg 1993; Holweger \& Rentzsch-Holm 1995; Iliji\'{c} et al. 1998).
97: 
98: Regarding the origin of unusual surface compositions of $\lambda$~Boo 
99: stars, various models have been propounded so far, such as the 
100: diffusion/mass-loss model, accretion/diffusion model, or binary model 
101: (see the references quoted in Paunzen 2004). Among these, the 
102: interpretation that the anomaly was built-up by the accretion of 
103: interstellar gas (where refractory metals of high $T_{\rm c}$ are 
104: depleted because of being condensed into dust while volatile species 
105: of low $T_{\rm c}$ hardly suffer this process), such as the interaction model 
106: between a star and the diffuse interstellar cloud proposed by 
107: Kamp \& Paunzen (2002), appears to be particularly promising,
108: in view of Paunzen et al.'s (2002) recent finding that the [Na/H] 
109: values of $\lambda$ Boo stars (showing a large diversity) are closely 
110: correlated with the [Na/H]$_{\rm ISM}$ values (sodium abundance of
111: interstellar matter) in the surrounding environment (cf. Fig.~7 therein).
112: 
113: Then, according to the suspected connection between Vega and $\lambda$~Boo 
114: stars, it is natural to examine the photospheric Na abundance of Vega.
115: Besides, the abundances of Li and K may also be worth particular attention 
116: in a similar analogy, since these three alkali elements play significant roles 
117: in discussing the physical state of the cool interstellar gas through their 
118: interstellar absorption (resonance) lines of Na~{\sc i} 5889/5896 
119: (D$_{1}$/D$_{2}$), Li~{\sc i} 6708, and K~{\sc i} 7665/7699. 
120: 
121: It is somewhat surprising, however, that reliable determinations of Vega's Na 
122: abundance are barely available. To our knowledge, although a number of abundance 
123: analyses of Vega have been published so far over the past half century, Na abundance 
124: derivations were done only in four studies (among which original $W_{\lambda}$ 
125: measurements were tried only two of these) according to our literature survey:
126: Hunger (1955; Na~{\sc i} 5889/5896), the reanalysis of Hunger's $W_{\lambda}$
127: data by Strom, Gingerich \& Strom (1966),  Qiu et al. (2001; Na~{\sc i} 5896), 
128: and the reanalysis of Qiu et al's $W_{\lambda}$ data by Saffe \& Levato (2004). 
129: Unfortunately, none of these appear to be sufficiently credible as viewed from 
130: the present knowledge, because of neglecting the non-LTE effect for the strong
131: resonance D$_{1}$/D$_{2}$ lines which were invoked in all these
132: studies (see below in this section). 
133: 
134: The situation is even worse for Li and K, for which any determination of
135: their abundances in Vega has never been reported.
136: Gerbaldi, Fraggiana \& Castelli (1995) once tried to measure 
137: the $W_{\lambda}$(Li~{\sc i} 6708) value of Vega and derived 
138: a tentative value of 2\,m$\rm\AA$, 
139: which was not used however, since they did not regard it 
140: as real under the estimated ambiguity of 4\,m$\rm\AA$ (cf. their Table 6).
141: Regarding the K~{\sc i} 7665/7699 lines, even a trial of measuring their 
142: strengths has never been undertaken for Vega, as far as we know. 
143: 
144: Several reasons may be enumerated concerning this evident scarcity of 
145: investigations on Li, Na, and K (which actually holds for early-type stars 
146: in general):\\
147: --- First of all, since the strengths of Li~{\sc i}, Na~{\sc i}, and K~{\sc i} 
148: lines quickly fade out as $T_{\rm eff}$ becomes higher owing to the enhanced
149: ionization of these alkali atoms (only one valence electron being
150: weakly bound), one has to contend with the difficult task of measuring
151: very weak lines (e.g., $W_{\lambda}$ value even as small as $\sim$\,1m$\rm\AA$),
152: which requires a spectrum of considerably high quality.\\
153: --- Second, these lines situate in the longer wavelength region of
154: yellow--red to near-IR, where many telluric lines due to H$_{2}$O or O$_{2}$
155: generally exist to cause severe blending with stellar lines,
156: which often prevent us from reliable measurements of equivalent widths.
157: (see, e.g., Sect. 6.2 of Qiu et al. 2001, who remarked the large uncertainty
158: caused by this effect in their measurements of Na~{\sc i} D lines).\\
159: --- Third, the resonance lines in the longer wavelength region 
160: are generally known to subject to considerable non-LTE corrections
161: especially for the case of Na~{\sc i} 5889/5896 lines showing considerable
162: strengths. Unfortunately, since detailed non-LTE studies of these alkali 
163: elements have been rarely published as far as early-type (A--F) dwarfs are 
164: concerned,\footnote{Few available at present may be only two studies on neutral 
165: sodium lines: Takeda \& Takada-Hidai's (1994) non-LTE analysis of various 
166: Na~{\sc i} lines in $\alpha$ CMa (in addition to Procyon and A--F supergiants), 
167: and Andrievsky et al.'s (2002) non-LTE study of Na~{\sc i} D$_{1}$ and 
168: D$_{2}$ lines for late-B to early-F type dwarfs of $\lambda$~Boo candidates.}
169: lack of sufficient knowledge on non-LTE corrections may have hampered
170: investigating their abundances. 
171: 
172: Given this situation, we decided to spectroscopically establish Vega's 
173: photospheric abundances of Li, Na, and K scarcely investigated so far, 
174: while overcoming the problems mentioned above, in order to see how they are 
175: compared with the compositions of other elements in Vega as well as 
176: with those of interstellar gas. This is the purpose of this study.
177: 
178: This attempt was originally motivated by our recent work (Takeda, Kawanomoto 
179: \& Ohishi 2007) of publishing a digital atlas of Vega's high-resolution 
180: ($R\sim 100000$) and high-S/N ($\sim$\,1000--3000) spectrum along with that of 
181: Regulus (a rapid rotator to be used for a reference of contaminating telluric lines), 
182: by which the confronted main obstacles (the necessity of measuring very weak 
183: lines as well as removing telluric blends) may be adequately conquered.
184: Besides, regarding the technical side of abundance determinations, we are 
185: ready to evaluate not only the necessary non-LTE corrections for Li, Na, and K 
186: (according to our recent experiences of statistical equilibrium calculations 
187: on these elements for F--G--K stars), but also the gravity-darkening 
188: corrections which may be appreciable especially for very weak lines as revealed 
189: from our recent modeling of rapidly-rotating Vega (Takeda, Kawanomoto \& Ohishi 
190: 2008). Therefore, this would make a timely subject to address.
191: 
192: \section{Observational data}
193: 
194: The basic observational data of Vega used for this study are the high-S/N 
195: ($\sim$~1000--3000) and high resolution ($R\sim 100000$) spectra, 
196: which are based on the data obtained with the HIgh Dispersion Echelle 
197: Spectrograph (HIDES) at the coud\'{e} focus of the 188~cm reflector of Okayama 
198: Astrophysical Observatory.  These spectral data of Vega, along with those of 
199: Regulus (a rapid rotator serving as a reference of telluric lines) were already 
200: published as a digital atlas by Takeda et al. (2007)\footnote{ 
201: While the spectrum atlas is presented as the electronic tables of this paper,
202: the same material is available also at the anonymous FTP site of the
203: Astronomical Data Center of National Astronomical Observatory of
204: Japan: $\langle$ftp://dbc.nao.ac.jp/DBC/ADACnew/J/other/PASJ/59.245/$\rangle$.
205: } which may be consulted for more details.
206: 
207: The following 11 lines were selected as the target lines used for
208: abundance determinations: Na~{\sc i} 5890/5896, 8183/8195, 6154/6161, 
209: 5683/5688, K~{\sc i} 7665/7699, and Li~{\sc i} 6708 (cf. Table 1).
210: Since most of these wavelength regions (except for those of Na~{\sc i} 6154/6161 
211: and Li~{\sc i} 6708) are contaminated by telluric lines, these features were 
212: removed by dividing Vega's spectra by the reference spectra of Regulus 
213: (with the help of IRAF\footnote{IRAF is distributed 
214: by the National Optical Astronomy Observatories, which is operated by the 
215: Association of Universities for Research in Astronomy, Inc. under cooperative 
216: agreement with the National Science Foundation, USA.} task {\tt telluric}) 
217: which are also included in Takeda et al. (2007) along with the airmass information. 
218: Though this removal process did not always work very successfully (e.g., for the 
219: case of very strong saturated lines such as the O$_{2}$ in the K~{\sc i} 7665/7699 
220: region and the H$_{2}$O lines in the Na~{\sc i} 8183/8195 region), we could 
221: manage to recover the intrinsic stellar lines more or less satisfactorily, 
222: since none of these target lines fortunately coincided with the deep core 
223: positions of such very strong telluric features.
224: 
225: While most of the lines were confidently identified by comparing them with 
226: the theoretically simulated spectra, the detection of the extremely weak 
227: Li~{\sc i} 6708 line was still uncertain in the original spectrum even at
228: its very high S/N ratio of $\sim 2000$. Therefore, an extra smoothing
229: was applied to the spectrum portion in the neighborhood of this line
230: by taking a running mean over 9 pixels, by which the feature became
231: more distinct (cf. Fig.~6) thanks to the further improved S/N ratio 
232: by a factor of 3.\footnote{Though this results in a blurring of 
233: $\sim$~10~km~s$^{-1}$ corresponding to the width of 9-pixel boxcar function 
234: (1 pixel $\sim$ 0.03 $\rm\AA$), this does not cause any serious problem 
235: because it is smaller than the intrinsic line width of $\sim$~20--30 km~s$^{-1}$.} 
236: 
237: The measurements of the equivalent widths (with respect to the local 
238: continuum level specified by eye-inspection) were done by the Gaussian-fitting 
239: technique, and the resulting $W_{\lambda}$ values are given in Table 1. 
240: The relevant spectra at each of the wavelength regions are shown in 
241: Figs. 1 (Na\,{\sc i} 5890/5896), 2 (Na~{\sc i} 8183/8195), 
242: 3 (Na~{\sc i} 6154/6161), 4 (Na~{\sc i} 5683/5688), 
243: 5 (K~{\sc i} 7665/7699), and 6 (Li~{\sc i} 6708).
244: 
245: \begin{figure}
246: \includegraphics[width=9.0cm]{fig1.eps}
247: \caption{Spectra of 5885--5900\,$\rm\AA$ region comprising
248: Na\,{\sc i} 5890 and 5896 lines. Theoretical simulation (LTE,
249: broadened with $v_{\rm e} \sin i = 22$\,km\,s$^{-1}$),
250: Vega/Regulus (telluric lines removed), raw Vega, and raw Regulus 
251: are arranged from top to bottom with appropriately chosen offsets. 
252: The scale of the ordinate corresponds to the second-top spectrum,
253: for which equivalent-widths of these lines were measured by the 
254: Gaussian fitting method (as overplotted in solid lines on this 
255: spectrum). }
256: \label{fig1}
257: \end{figure}
258: 
259: \begin{figure}
260: \includegraphics[width=9.0cm]{fig2.eps}
261: \caption{Spectra of 8180--8200\,$\rm\AA$ region comprising
262: Na\,{\sc i} 8183 and 8195 lines. Theoretical simulation,
263: Vega/Regulus (telluric lines removed), raw Vega, and raw Regulus 
264: are arranged from top to bottom with appropriately chosen offsets.
265: Otherwise, the same as in Fig.~1.}
266: \label{fig2}
267: \end{figure}
268: 
269: \begin{figure}
270: \includegraphics[width=9.0cm]{fig3.eps}
271: \caption{Spectra of 6151--6164\,$\rm\AA$ region comprising
272: Na\,{\sc i} 6154 and 6161 lines. Theoretical simulation (upper)
273: and raw Vega (lower) are placed with an appropriate offset.
274: This region is not contaminated by telluric lines.
275: Otherwise, the same as in Fig.~1.}
276: \label{fig3}
277: \end{figure}
278: 
279: \begin{figure}
280: \includegraphics[width=9.0cm]{fig4.eps}
281: \caption{Spectra of 5680--5690\,$\rm\AA$ region comprising
282: Na\,{\sc i} 5683 and 5688 lines. Theoretical simulation,
283: Vega/Regulus (telluric lines removed), raw Vega, and raw Regulus 
284: are arranged from top to bottom with appropriately chosen offsets.
285: Otherwise, the same as in Fig.~1.}
286: \label{fig4}
287: \end{figure}
288: 
289: \begin{figure}
290: \includegraphics[width=9.0cm]{fig5.eps}
291: \caption{Spectra of 7660--7705\,$\rm\AA$ region comprising
292: K\,{\sc i} 7665 and 7699 lines. Theoretical simulation,
293: Vega/Regulus (telluric lines removed), raw Vega, and raw Regulus 
294: are arranged from top to bottom with appropriately chosen offsets.
295: Otherwise, the same as in Fig.~1.}
296: \label{fig5}
297: \end{figure}
298: 
299: \begin{figure}
300: \includegraphics[width=9.0cm]{fig6.eps}
301: \caption{Spectra of 6700--6716\,$\rm\AA$ region comprising
302: Li\,{\sc i} 6708 merged doublet. Theoretical simulation, smoothed Vega spectrum 
303: by convolving the 9-pixel boxcar function, and raw Vega spectrum 
304: are arranged from top to bottom with appropriately chosen offsets.
305: This region is hardly contaminated by telluric lines.
306: Otherwise, the same as in Fig.~1.}
307: \label{fig6}
308: \end{figure}
309: 
310: \section{Abundance determinations}
311: 
312: Regarding the fiducial atmospheric model of Vega, Kurucz's (1993) ATLAS9 model 
313: with the parameters of $T_{\rm eff} = 9550$~K (effective temperature), 
314: $\log g$ (cm~s$^{-2}$) = 3.95 (surface gravity), [X/H] = $-0.5$ (metallicity), 
315: and $v_{\rm t}$ = 2~km~s$^{-1}$ (microturbulence) was adopted for this study 
316: as in Takeda et al. (2007), which appears to be the best as far as within 
317: the framework of classical plane-parallel ATLAS9 models according to 
318: Castelli \& Kurucz (1994).
319: 
320: As the basic strategy, the LTE abundance ($A_{\rm LTE}$) was first derived
321: from the measured $W_{\lambda}$ by using the WIDTH9 program,\footnote{
322: This is a companion program to the ATLAS9 model atmosphere program written 
323: by R. L. Kurucz (Kurucz 1993), though it has been considerably modified by 
324: Y. T. (e.g., for treating multiple component lines or including non-LTE effects 
325: in line formation}, to which the non-LTE correction ($\Delta_{\rm NLTE}$) and 
326: the gravity-darkening correction ($\Delta_{\rm GD}$) were then applied to 
327: obtain the final abundance ($A_{\rm NLTE+GD}$). 
328: 
329: In this step-by-step way,\footnote{Complete simulations (where the 
330: non-LTE effect and the gravity-darkening effect were simultaneously 
331: taken into account in a rigorous manner) show that the finally derived 
332: abundances in this semi-classical approach are fairly reliable. 
333: This consistency check is separately described in Appendix B.}
334: one can quantitatively judge the importance/unimportance 
335: of each effect by comparing two abundance corrections, and
336: such factorized information may be useful in application to
337: other different situations (e.g., the case of slow rotator where only 
338: $\Delta_{\rm LTE}$ is relevant; cf. Appendix A).
339: Also, the use of ``relative correction'' (instead of directly
340: approaching the absolute abundance solution)
341: has a practical merit of circumventing the numerical-precision problem
342: involved in the spectrum computation on the gravity-darkened model 
343: (especially seen for the case of the extremely weak Li~{\sc i} 6708 line
344: composed of several components; cf. footnote 9). 
345: 
346: The non-LTE calculations for evaluating $\Delta_{\rm NLTE}$ were carried out 
347: in the same manner as described in Takeda et al. (2003; Na~{\sc i}), 
348: Takeda \& Kawanomoto (2005; Li~{\sc i}), and Takeda et al. (2002; K~{\sc i}).
349: Besides, all the relevant atomic data ($gf$ values, damping constants, etc.) 
350: for abundance determinations were so adopted as to be consistent with 
351: these three studies.
352: 
353: The gravity-darkening corrections ($\Delta_{\rm GD}$) were derived based on
354: the equivalent-width intensification factor computed by the program
355: CALSPEC under the assumption of LTE ($W_{4}/W_{0}$, where $W_{4}$ 
356: and $W_{0}$ correspond to the gravity-darkened rapid rotator model and 
357: the classical rigid-rotation model, respectively; and the abundance
358: was adjusted to make $W_{4}$ consistent with $W_{\lambda}^{\rm obs}$), 
359: the details about which are explained in Takeda et al. (2008). 
360: Practically, the relation 
361: $\Delta_{\rm GD} = -\log (W_{4}/W_{0})$ was assumed for the very weak lines
362: ($W_{\lambda} < 15$~m$\rm\AA$; i.e., Na~{\sc i} 6154/6161, Na~{\sc i} 5683/5688, 
363: K~{\sc i} 7665/7699, and Li~{\sc i} 6708) being guaranteed to locate on the
364: linear part of the curve of growth. The $\Delta_{\rm GD}$ values for the 
365: remaining four lines (Na~{\sc i} 5890/5896 and Na~{\sc i} 8183/8195) 
366: were directly obtained from the abundance difference 
367: between those derived from the true $W_{\lambda}^{\rm obs}$ and the 
368: perturbed $W_{\lambda}^{\rm obs}/(W_{4}/W_{0})$. 
369: 
370: The results of the abundances and the abundance corrections are summarized
371: in Table 1, where the finally adopted (average) values of 
372: $\langle A_{\rm NLTE+GD} \rangle$ and the abundances relative to the
373: standard solar system compositions 
374: ([X/H] $\equiv \langle A_{\rm NLTE+GD}^{\rm X} \rangle- A^{\odot}$) 
375: are also presented. As can be seen from this table, the non-LTE corrections
376: are always negative (corresponding to the non-LTE line-intensification) 
377: with appreciably $W_{\lambda}$-dependent extents (ranging from 0.1--0.7 dex).
378: Similarly, the gravity-darkening corrections are also negative with extents 
379: of 0.1--0.3 dex\footnote{It may be worth noting that these moderate 
380: $\Delta_{\rm GD}$ values correspond to Takeda et al.'s (2008) model No. 4 
381: ($v_{\rm e} = 175$~km~s$^{-1}$) which they concluded to be the best.
382: If higher-rotating models with larger gravity-darkening
383: were to be used, the corrections would naturally become more enhanced.
384: For example, if we adopt Takeda et al.'s (2008) model No. 8
385: with $v_{\rm e} = 275$~km~s$^{-1}$, which is near to the solution
386: suggested from interferometric observations (Peterson et al. 2006,
387: Aufdenberg et al. 2006), the extents of $\Delta_{\rm GD}$ turn out
388: to increase significantly by a factor of $\sim 3$ (i.e., $-0.22$, $-0.18$, 
389: $-0.31$, $-0.23$, $-0.42$, $-0.43$, $-0.36$, $-0.32$, $-0.34$, $-0.37$,
390: and $-0.82$, for Na~{\sc i} 5890, 5896, 8183, 8195, 6154, 6161, 5683, 5688,
391: K~{\sc i} 7664, 7699, and Li~{\sc i} 6708, respectively).}
392: (especially important for the weakest Li~{\sc i} 6708 line) 
393: reflecting the line-strengthening caused by rotation-induced $T$-lowering.
394: 
395: Regarding errors in the resulting abundances, those due to uncertainties 
396: in the atmospheric parameters (e.g., in $T_{\rm eff}$ or $\log g$) may  
397: not be very significant in the present comparatively well-established
398: case (an abundance change is only $\sim 0.1$~dex for a change either 
399: in $T_{\rm eff}$ by $\sim 200$~K or in $\log g$ by 0.2; cf. Table 5 in 
400: Takeda \& Takada-Hidai 1994 for the case of $\alpha$ CMa).
401: Rather, a more important source of error would be the ambiguities in 
402: the measurement of $W_{\lambda}$. While the photometric accuracy of
403: $W_{\lambda}$ estimated by applying Cayrel's (1988) formula 
404: (FWHM of $w \sim 0.5\rm\AA$, pixel size of $\delta x \sim 0.03 \rm\AA$, 
405: and photometric error of $\epsilon \sim 1/2000$) is only on the order of 
406: $\sim 0.1$~m$\rm\AA$ and inconsiderable, uncertainties in specifying the 
407: continuum level can be a more serious problem, especially for the case of 
408: extremely weak lines where the relative depression is only on the order of 
409: $\sim 10^{-3}$. Presumably, errors in $W_{\lambda}$ would amount to 
410: several tens of \% in the case of very faint lines (e.g., Na~{\sc i} 6154 
411: or Li~{\sc i} 6708) or the case of insufficient spectrum quality due to 
412: incompletely removed telluric lines (e.g., Na~{\sc i} 8183 or K~{\sc i} 7665). 
413: Accordingly, the abundance results derived from such problematic lines 
414: may be uncertain by up to $\sim~$0.2 dex, which should particularly be 
415: kept in mind in discussing the final abundances of Li and K being based on 
416: only 1 or 2 lines.
417: 
418: \section{Results and Discussions}
419: 
420: \subsection{Na abundance of Vega}
421: 
422: The abundances of sodium could be derived from 8 lines of different strengths 
423: (2 resonance lines of D$_{1}$+D$_{2}$ and 6 subordinate lines originating 
424: from the excited level of $\chi_{\rm low} = 2.21$~eV).
425: The three kinds of abundances ($A_{\rm LTE}$, $A_{\rm NLTE}$, $A_{\rm NLTE+GD}$; 
426: cf. Table 1) for each of the 8 lines are plotted against $W_{\lambda}$ in 
427: Fig.~7. We can read from this figure that the serious inconsistency
428: seen in the LTE abundances is satisfactorily removed by the non-LTE
429: corrections, which are significantly $W_{\lambda}$-dependent from 
430: $-0.7$ dex (for the strongest Na~{\sc i} 5890) to $-0.1$ dex (for the
431: weakest-class Na~{\sc i} 6154/6161/5683/5688 lines of $W_{\lambda} < 10$~m$\rm\AA$).
432: This may suggest that the applied non-LTE corrections are fairly 
433: reliable and the adopted microturbulence of $v_{\rm t}$ = 2~km~s$^{-1}$ 
434: is adequate (which affects the abundances of stronger Na~{\sc i} 5890/5896 
435: lines; cf. Table 5 of Takeda \& Takada-Hidai 1994).
436: The final sodium abundance of Vega, obtained by averaging $A_{\rm NLTE+GD}$ 
437: values of 8 lines, is 6.01 with the standard deviation of $\sigma = 0.10$, 
438: indicating a mildly subsolar composition ([Na/H] = $-0.3$) compared to
439: the solar abundance of $A_{\odot} = 6.31$.
440: 
441: As mentioned in Sect. 1, a few previous determinations based on D$_{1}$+D$_{2}$ 
442: lines are available for the Na abundance of Vega. Hunger (1955) tentatively 
443: derived $A \sim 6.7$ (coarse analysis) and $A \sim 7.8$ (fine analysis) 
444: from $W_{\lambda}$(5890/5896) = 195/115 m$\rm\AA$, though commenting 
445: that these abundances are very uncertain. Thereafter, based on 
446: Hunger's $W_{\lambda}$ data, Strom et al.'s (1966) model atmosphere analysis 
447: concluded $A \sim 7.7 (\pm 0.4)$ ($T_{\rm eff} = 9500$~K).
448: Then, after a long blank, Qiu et al. (2001) obtained $A = 6.45$ from
449: their measured $W_{\lambda}$(5896) value of 94~m$\rm\AA$, though they also
450: remarked that this result is quite uncertain because of blending with
451: water vapor lines. Soon after, Saffe \& Levato (2004) derived $A = 6.37$
452: using Qiu et al.'s $W_{\lambda}$ data. Accordingly, all these previous
453: work reported supersolar or near-solar Na abundances for Vega, contradicting
454: the conclusion of this study. It is evident that they failed to
455: obtain the correct abundance from the Na~{\sc i} 5890/5896 lines
456: because of neglecting the non-LTE effect, which is substantially important
457: for these strong resonance lines.
458: 
459: \begin{figure}
460: \includegraphics[width=9.0cm]{fig7.eps}
461: \caption{Na abundances derived from each of the 8 lines
462: plotted against the equivalent widths, based on the results
463: in Table 1. Squares, triangles, and circles correspond to
464: the LTE abundances, the non-LTE abundances, and the final abundances
465: (including both the non-LTE and gravity-darkening corrections),
466: respectively. The horizontal dotted line indicates the average
467: of the final abundances (6.01).}
468: \label{fig7}
469: \end{figure}
470: 
471: \subsection{Composition characteristics of Li, Na, and K}
472: 
473: According to Table 1, while potassium is only mildly subsolar ([K/H] 
474: $\simeq -0.2$) to an extent similar to sodium ($\sim -0.3$), lithium is 
475: more manifestly underabundant as [Li/H] $\simeq -0.6$. Therefore, we may 
476: conclude that the deficiency in Li is markedly different from that of Na 
477: and K, even considering the possible ambiguities of $\sim 0.2$\,dex 
478: (cf. Sect. 3). How should we interpret these results? Do they have something 
479: to do with the element depletion (due to dust condensation) in interstellar 
480: gas and its accretion? Since the quantitative amount of such a depletion 
481: is different from case to case depending on the physical condition
482: (e.g., see the variety of [Na/H]$_{\rm ISM}$ in Fig.~7 of Paunzen et al. 
483: 2002), it is not much meaningful to discuss the ``absolute'' values of [X/H] 
484: here. Rather, we should pay attention to their ``relative'' behaviors 
485: with each other or in comparison to those of various other elements.
486: 
487: The [X/H]$_{\rm Vega}$ values (those for Li, Na, and K are from this study,
488: while those for other elements were taken from various literature)
489: are plotted against the condensation temperature ($T_{\rm c}$) in Fig.~8 
490: (filled circles), where the [X/H]$_{\rm ISM}$ results of typical interstellar 
491: gas in the direction of $\zeta$ Oph (taken from Table 5 of Savage \& Sembach 1996; 
492: see also Fig.~4 therein) are also shown by open circles.
493: The following characteristics can be recognized from this figure:\\
494: --- (1) The [X/H]$_{\rm Vega}$ values, falling on a rather narrow range between 
495: $\sim -1$ and $\sim 0$, tend to decrease with $T_{\rm c}$, a qualitatively
496: similar trend to that seen in [X/H]$_{\rm ISM}$.\\  
497: --- (2) The run of [X/H]$_{\rm Vega}$ with $T_{\rm c}$ appears to be slightly 
498: discontinuous at $T_{\rm c} \sim 1000$\,K 
499: (i.e., [X/H]$_{\rm Vega} \sim -0.2$ to $-0.3$ at $T_{\rm c} \la 1000$\,K and
500: [X/H]$_{\rm Vega} \sim -0.5$ to $-0.6$ at $T_{\rm c} \ga 1000$\,K).\\
501: --- (3) Interestingly, the decreasing [X/H]$_{\rm ISM}$ with $T_{\rm c}$ also 
502: shows a discontinuity around this critical $T_{\rm c}$ ($\sim 1000$\,K) 
503: in the sense that the slope of $|d$[X/H]$_{\rm ISM}/dT_{\rm c}|$
504: becomes evidently steeper on the $T_{\rm c} \ga 1000$\,K side.\\
505: --- (4) Specifically, the tendency of [Li/H]$_{\rm Vega}$ ($-0.6$) being 
506: more deficient than [Na/H]$_{\rm Vega}$ ($-0.3$) and [K/H]$_{\rm Vega}$ 
507: ($-0.2$) is qualitatively similar to just what is seen in ISM 
508: ([Li/H]$_{\rm ISM}<$ [Na/H]$_{\rm ISM} \simeq$ [K/H]$_{\rm ISM}$).
509: 
510: All these observational facts may suggest the existence of some kind of 
511: connection between Vega's photospheric abundances and those of interstellar 
512: cloud, which naturally implies that an accretion/contamination of interstellar 
513: gas is likely to be responsible (at least partly) for the abundance 
514: peculiarity of Vega, such as being proposed for explaining the $\lambda$ Boo 
515: phenomenon. This is the conclusion of this study.
516: 
517: 
518: \begin{figure}
519: \includegraphics[width=9.0cm]{fig8.eps}
520: \caption{[X/H] values (relative abundances in comparison with
521: the solar-system values of Grevesse \& Noels 1993) of various elements 
522: plotted against the condensation temperature ($T_{\rm c}$). 
523: Open circles indicate the compositions of interstellar gas in the 
524: direction of $\zeta$~Oph (cool diffuse clouds), which were taken from
525: Savage \& Sembach (1996; cf. their Table 5 and Fig.~4). 
526: The [Na/H], [K/H], and [Li/H] for Vega derived in this study
527: are shown by larger filled circles. Besides, Vega's [X/H] values
528: for other elements are plotted by smaller filled circles for a reference, 
529: which were taken from from various sources:  Przybilla \& Butler 
530: (2001) (for C, N, and O; cf. their Table 7); Takada-Hidai \& Takeda (1996) 
531: (for S); Adelman \& Gulliver (1990) (for Mg, Ca, Ti, Cr, Mn, Fe, and Ni);
532: Qiu et al. (2001) (for Si and V).
533: [Note. Since these literature [X/H] values of Vega were derived in a 
534: conventional manner based on classical atmospheric model, further 
535: corrections for the gravity-darkening effect ($\Delta_{\rm GD}$) are 
536: to be expected. 
537: According to Takeda et al. (2008), however, the extents of (negative) 
538: $\Delta_{\rm GD}$ (becoming appreciable only for very weak lines and 
539: for specific elements/stages) are only $\sim 0.2$~dex at most (e.g., 
540: Adelman \& Gulliver's [Ca/H] value derived from very weak Ca~{\sc i} 
541: lines had better be reduced by $\sim 0.2$~dex) while $\la 0.1$~dex 
542: in many typical cases, which are thus unlikely to cause any significant 
543: change in the general pattern of [X/H] shown here.]
544: }
545: \label{fig8}
546: \end{figure}
547: 
548: 
549: 
550: \section*{Acknowledgments}
551: 
552: The author thanks S. Kawanomoto and N. Ohishi for their help
553: in the observations of Vega, based on the data from which 
554: this study is based.
555: 
556: 
557: \begin{thebibliography}{99}
558: 
559: \bibitem[]{}
560:   Adelman S.J., Gulliver A.F., 1990, ApJ, 348, 712
561: \bibitem[]{}
562:   Andrievsky S.M., et al., 2002, A\&A, 396, 641
563: \bibitem[]{}
564:   Aufdenberg J.P., et al. 2006, ApJ, 645, 664 (erratum: 651, 617)
565: \bibitem[]{}
566:   Baschek B., Slettebak A., 1988, A\&A, 207, 102
567: \bibitem[]{}
568:   Burkhart C., Coupry M.F., 1991, A\&A, 249, 205
569: \bibitem[]{}
570:   Castelli F., Kurucz R.L., 1994, A\&A, 281, 817
571: \bibitem[]{}
572:   Cayrel R. 1988, in Cayrel de Strobel G., Spite M., eds,
573:   The Impact of Very High S/N Spectroscopy on Stellar Physics, 
574:   Proc. IAU Symp. 132, Kluwer, Dordrecht, p.345
575: \bibitem[]{}
576:   Coupry M.F., Burkhart C., 1992, A\&AS, 95, 41
577: \bibitem[]{}
578:   Gerbaldi M., Faraggiana R., Castelli F., 1995, A\&AS, 111, 1
579: \bibitem[]{}
580:   Grevesse N., Noels A., 1993, in Prantzos N., Vangioni-Flam E., 
581:   Cass\'{e} M., eds, Origin and evolution of the elements,
582:   Cambridge University Press, Cambridge, p.15
583: \bibitem[]{}
584:   Gulliver A.F., Adelman S.J., \& Friesen T.P. 2004, A\&A, 413, 285,
585: \bibitem[]{}
586:   Holweger H., Rentzsch-Holm I., 1995, A\&A, 303, 819
587: \bibitem[]{}
588:   Hunger K., 1955, Zs. Ap., 36, 42
589: \bibitem[]{}
590:   Iliji\'{c} S., Rosandi\'{c} M., Dominis D., Planini\'{c} M.,
591:   Pavlovski K., 1998, Contrib. Astron. Obs. Skalnat\'{e} Pleso, 27, 467
592: \bibitem[]{}
593:   Kamp I., Paunzen E., 2002, MNRAS, 335, L45
594: \bibitem[]{}
595:   Kurucz R.L., 1993, Kurucz CD-ROM, No. 13, Harvard-Smithsonian Center
596:   for Astrophysics, http://kurucz.harvard.edu/cdroms.html
597: \bibitem[]{}
598:   Paunzen E., 2004, in Zverko J., 
599:   $\check{\rm Z}$i$\check{\rm{z}}$$\check{\rm{n}}$ovsk$\acute{\rm y}$ J., 
600:   Adelman S.J., Weiss W.W., eds, The A-Star Puzzle, Proc. IAU Symp. 224, 
601:   Cambridge University Press, Cambridge, p.443
602: \bibitem[]{}
603:   Paunzen E., Iliev I.Kh., Kamp I., Barzova I.S., 2002, MNRAS, 336, 1030
604: \bibitem[]{}
605:   Peterson D.M., et al. 2006, Nature, 440, 896
606: \bibitem[]{}
607:   Przybilla N., Butler K., 2001, A\&A, 379, 955
608: \bibitem[]{}
609:   Qiu H.M., Zhao G., Chen Y.Q., Li Z.W., 2001, ApJ, 548, 953
610: \bibitem[]{}
611:   Saffe C., Levato H., 2004, A\&A, 418, 1083
612: \bibitem[]{}
613:   Savage B.D., Sembach K.R., 1996, ARA\&A, 34, 279
614: \bibitem[]{}
615:   Strom S.E., Gingerich O., Strom K.M., 1966, ApJ, 146, 880
616: \bibitem[]{}
617:   St\"{u}renburg S., 1993, A\&A, 277, 139
618: \bibitem[]{}
619:   Takada-Hidai M., Takeda Y., 1996, PASJ, 48, 739
620: \bibitem[]{}
621:   Takeda Y., Kawanomoto S., 2005, PASJ, 57, 45
622: \bibitem[]{}
623:   Takeda Y., Takada-Hidai M., 1994, PASJ, 46, 395
624: \bibitem[]{}
625:   Takeda Y., Kawanomoto S., Ohishi N., 2007, PASJ, 59, 245
626: \bibitem[]{}
627:   Takeda Y., Kawanomoto S., Ohishi N., 2008, ApJ, 678, 446
628: \bibitem[]{}
629:   Takeda Y., Zhao G., Chen Y.-Q., Qiu H.-M., Takada-Hidai M., 2002, 
630:   PASJ, 54, 275
631: \bibitem[]{}
632:   Takeda Y., Zhao G., Takada-Hidai M., Chen Y.-Q., Saito Y.-J.,
633:   Zhang H.-W., 2003, ChJAA, 3, 316
634: 
635: \end{thebibliography}
636: 
637: \setcounter{table}{0}
638: \begin{table*}
639: \begin{minipage}{130mm}
640: \small
641: \caption{Atomic data, equivalent widths, and abundance results.}
642: \begin{center}
643: \begin{tabular}{cccccrccccc}\hline
644: Species & RMT & $\lambda$ & $\chi_{\rm low}$ & $\log gf$ & $W_{\lambda}$ & $A_{\rm LTE}$ & 
645: $\Delta_{\rm NLTE}$ & $\Delta_{\rm GD}$ & $A_{\rm NLTE+GD}$ & [X/H] \\ 
646:  &  &  ($\rm\AA$) & (eV) &  &  (m$\rm\AA$) &  &  &  &  &  \\
647: \hline
648: Na\,{\sc i} & 1  & 5889.95  & 0.00  & +0.12  & 133.3  & 6.82  & $-$0.69  & $-$0.09  & 6.03  &  \\ 
649: Na\,{\sc i} & 1  & 5895.92  & 0.00  & $-$0.18  & 104.2  & 6.58  & $-$0.49  & $-$0.08  & 6.01  &  \\
650: Na\,{\sc i} & 4  & 8183.26  & 2.10  & +0.22  &  19.7  & 6.16  & $-$0.22  & $-$0.12  & 5.82  &  \\
651: Na\,{\sc i} & 4  & 8194.82  & 2.10  & +0.52  &  45.1  & 6.39  & $-$0.28  & $-$0.10  & 6.01  &  \\
652: Na\,{\sc i} & 5  & 6154.23  & 2.10  & $-$1.56  &   0.8  & 6.36  & $-$0.10  & $-$0.12  & 6.13  &  \\
653: Na\,{\sc i} & 5  & 6160.75  & 2.10  & $-$1.26  &   1.6  & 6.36  & $-$0.10  & $-$0.14  & 6.12  &  \\
654: Na\,{\sc i} & 6  & 5682.63  & 2.10  & $-$0.67  &   4.3  & 6.19  & $-$0.09  & $-$0.13  & 5.97  &  \\
655: Na\,{\sc i} & 6  & 5688.21  & 2.10  & $-$0.37  &   8.4  & 6.20  & $-$0.09  & $-$0.12  & 5.99  &  \\
656:  &  &  &  &  &  &  &  &  & {\it 6.01}  & {\it $-$0.30}  \\ 
657: \hline
658: K\,{\sc i} & 1  & 7664.91  & 0.00  & +0.13  & 13.3  & 5.23  & $-$0.28  & $-$0.13  & 4.82  \\ 
659: K\,{\sc i} & 1  & 7698.97  & 0.00  & $-$0.17  & 10.7  & 5.42  & $-$0.27  & $-$0.14  & 5.01  \\ 
660:  &  &  &  &  &  &  &  &  & {\it 4.91}  & {\it $-$0.22} \\
661: \hline
662: Li\,{\sc i} & 1  & 6707.756  & 0.00  & $-$0.43  & 0.7  & 3.15  & $-$0.15  & $-$0.26  & 2.74  &  \\ 
663:  &  & 6707.768  &  & $-$0.21  &  &  &  &  &  &  \\ 
664:  &  & 6707.907  &  & $-$0.93  &  &  &  &  &  &  \\
665:  &  & 6707.908  &  & $-$1.16  &  &  &  &  &  &  \\
666:  &  & 6707.919  &  & $-$0.71  &  &  &  &  &  &  \\
667:  &  & 6707.920  &  & $-$0.93  &  &  &  &  &  &  \\ 
668:  &  &  &  &  &  &  &  &  & {\it 2.74}  & {\it $-$0.57} \\
669: \hline
670: \end{tabular}
671: \end{center}
672: Following the atomic data of spectral lines (species, multiplet No., wavelength,
673: lower excitation potential, and logarithmic $gf$ value) in columns 1--5,
674: column 6 gives the measured equivalent width. The results of the abundance analysis are 
675: given in column 7 (LTE abundance; in the usual normalization of H=12.00),
676: column 8 (non-LTE correction), column 9 (gravity-darkening correction),
677: and column 10 (non-LTE as well as gravity-darkening corrected abundance).  
678: The finally adopted average abundance and the corresponding [X/H] value 
679: ($\equiv A_{\rm X}^{\rm Vega} - A_{\rm X}^{\odot}$) are also given at the bottom of 
680: the section (in italics; columns 10 and 11),
681: where Grevesse \& Noels's (1993) values of 6.31 (Na), 3.31 (Li), and 5.13 (K) were used 
682: as the standard solar-system abundances ($A_{\rm X}^{\odot}$).
683: Note that the analysis of $W_{\lambda}$(Li\,{\sc i}~6708) was done by synthesizing the six 
684: component lines of $^{7}$Li (neglecting the contribution of $^{6}$Li; cf. Takeda \& 
685: Kawanomoto 2005), 
686: while all the remaining lines of Na and K were analyzed by the single-line treatment
687: as was done by Takeda et al. (2003; Na) and Takeda et al. (2002; K).
688: \end{minipage}
689: \end{table*}
690: 
691: \appendix
692: 
693: \section{Li abundance in $o$~Peg}
694: 
695: The enhanced deficiency of Li (compared to Na and K) was an important 
696: key result in deriving the conclusion of this study (cf. Sect. 4.2),
697: since we interpreted it as a manifestation of ISM compositions.
698: In this connection, it may be worth mentioning another remarkable 
699: very sharp-lined A1~IV star, $o$~Peg, for which we could also get 
700: information of the Li abundance. 
701: As the only available measurement of the Li~{\sc i} 6708 line for early 
702: A-type stars thus far to our knowledge, Coupry \& Burkhart (1992) derived 
703: $W_{\lambda}$(6708) = 1.3~m$\rm\AA$ for this star, which has atmospheric 
704: parameters ($T_{\rm eff}$ = 9650~K, $\log g = 3.6$) quite similar to 
705: those of Vega except for its near-normal metallicity ([Fe/H] 
706: $\simeq +0.1$; Burkhart \& Coupry 1991). 
707: 
708: Now, their $W_{\lambda}$($o$~Peg) value twice as large as that of Vega 
709: (0.7~m$\rm\AA$) would raise the Li abundance by +0.3 dex. Furthermore, since 
710: any flat-bottomed shape has never been reported in the spectral lines 
711: of $o$~Peg (as we can confirm from the high-quality spectrum atlas\footnote{
712: Also available at \\
713: http://www.brandonu.ca/physics/gulliver/ccd\_atlases.html .}
714: of this star published by Gulliver, Adelman, \& Friesen 2004), the 
715: application of the gravity-darkening correction ($\sim -0.3$\,dex for Vega) 
716: may not be relevant here, which again leads to an increase of 0.3\,dex. 
717: Hence, we have [Li/H]$_{\rm o Peg} (\simeq$~[Li/H]$_{\rm Vega}$ + 0.6) 
718: $\sim 0.0$; an interesting result that the photospheric Li abundance of 
719: $o$~Peg almost coincides with that of the solar-system composition.
720: This result, that [Li/H] (as well as [Fe/H]) is deficient/normal in 
721: Vega/$o$~Peg, may suggest that the mechanism responsible for producing the 
722: underabundance of these elements in Vega is irrelevant for $o$~Peg. Thus, 
723: according to our interpretation, $o$~Peg would not have suffered any 
724: pollution due to accretion of interstellar gas depleted in volatile
725: elements of higher $T_{\rm c}$. 
726: 
727: Yet, we had better keep in mind another possibility that the mechanism 
728: causing this Li deficiency in Vega might different from that of other metals.
729: For example, this underabundance could be attributed to some process of 
730: envelope mixing (e.g., meridional circulation or shear-induced 
731: turbulence which are supposed to be more significant as a star rotates
732: faster), because Li atoms are burned and destroyed when they are 
733: conveyed into the hot stellar interior ($T \ga 2.4 \times 10^{6}$\,K).
734: Namely, since we know that Vega rotates rapidly (as fast as 
735: $v_{\rm e} \sim 200$~km~s$^{-1}$) while $o$~Peg does not so much 
736: (at least the gravity-darkening effect is not so significant as in Vega), 
737: the underabundance of Li in the atmosphere of Vega might stem from the 
738: rotation-induced mixing (which is not expected for slowly rotating $o$~Peg).
739: If this is the case, however, the rough similarity in the extent of 
740: deficiency in Li as well as other metals has to be regarded as a mere
741: coincidence, which makes us feel this possibility as rather unlikely.
742: 
743: \section{Check for the final abundances: Complete spectrum synthesis}
744: 
745: Our basic strategy for deriving the abundances of Na, K, and Li in 
746: Sect. 3 was as follows. \\
747: --- (1) First, the $A_{\rm LTE}$ was derived from 
748: $W_{\lambda}^{\rm obs}$ based on the classical plane-parallel model 
749: atmosphere. \\
750: --- (2) Next, the non-LTE correction ($\Delta_{\rm NLTE}$) was 
751: derived in the conventional way by using this classical model. \\
752: --- (3) Then, the gravity darkening correction ($\Delta_{\rm GD}$)
753: was evaluated from the line-intensification factor 
754: $W_{4}^{\rm LTE}/W_{0}^{\rm LTE}$ (where the abundance was so adjusted
755: as to satisfy $W_{4}^{\rm LTE} \simeq W_{\lambda}^{\rm obs}$), which 
756: was computed by applying the CALSPEC program (Takeda et al. 2008) 
757: with the assumption of LTE to the gravity-darkened model~4 and 
758: the rigid-rotation model~0. \\
759: --- (4) Finally, $A_{\rm NLTE+GD}$ was
760: obtained as $A_{\rm LTE} + \Delta_{\rm NLTE} + \Delta_{\rm GD}$. 
761: 
762: Actually, such a phased approach has a distinct merit 
763: of clarifying the importance/contribution of two different effects 
764: (the non-LTE effect and the gravity-darkening effect).
765: Besides, from a practical point of view, the necessary amount of 
766: calculations to arrive at the final abundance solution (such that 
767: reproducing the observed spectrum) can be considerably saved.
768: 
769: However, there is some concern about whether such a step-by-step approach
770: (treating the two effects separately) really yield sufficiently correct 
771: results, because both are actually related with each other (e.g., how 
772: does the largely variable non-LTE corrections over the gravity-darkened 
773: stellar surface differing in $T$ or $g$ play roles? How reliable is the 
774: gravity-darkening correction derived by neglecting the non-LTE effect?). 
775: Therefore, it may be worth checking the validity of the finally derived 
776: abundances ($A_{\rm NLTE+GD}$) by carrying out
777: a complete spectrum synthesis including both NLTE and GD effects 
778: simultaneously. 
779: For this purpose, the CALSPEC program was modified
780: so as to allow inclusion of the non-LTE departure coefficients
781: (corresponding to the different conditions at each of the points 
782: over the gravity-darkened stellar surface), and the NLTE+GD
783: profiles ($R_{4}^{\rm NLTE}$) were computed for each of the 9 lines 
784: (Na~{\sc i} 5890, 5896, 8183, 8195, 6154, 6161, 5683, 5688,
785: K~{\sc i} 7664, 7699, and Li~{\sc i} 6708) by using the already known
786: $A_{\rm NLTE+GD}$ values (6.03, 6.01, 5.82, 6.01, 6.13, 6.12, 5.97,
787: 5.99, 4.82, 5.01, and 2.74, respectively). 
788: Such obtained $R_{4}^{\rm NLTE}$ profiles are depicted (in thick solid 
789: lines) in Fig. 9, where the relevant three kinds of line profiles 
790: [$R_{0}^{\rm NLTE}$ (thick dashed line), $R_{4}^{\rm LTE}$ (thin solid 
791: line), and $R_{0}^{\rm LTE}$ (thin dashed line)] are also shown for 
792: comparison.
793: 
794: \begin{figure}
795: \includegraphics[width=8.0cm]{figB1.eps}
796: \caption{
797: Theoretical profiles of the Na, K, and Li lines computed 
798: by the CALSPEC program (developed for synthesizing the flux spectrum 
799: for a given rotationally-distorted gravity-darkened stellar model; 
800: cf. Takeda et al. 2008), which has been modified to allow inclusion 
801: of the non-LTE effect. The $A_{\rm NLTE+GD}$ values given in Table 1 
802: were assumed as the abundances for each of the lines.
803: Results for the gravity-darkened model and the classical 
804: rigid-rotation model (model No. 4 and No. 0 in Takeda et al. 2008)
805: are discriminated by the line type (solid and dashed lines, 
806: respectively), and those for NLTE and LTE are by the line thickness
807: (thick and thin lines, respectively). 
808: As a result, four profiles are shown for each of the 9 lines: 
809: $R_{4}^{\rm NLTE}$ (thick solid line)
810: $R_{0}^{\rm NLTE}$ (thick dashed line), 
811: $R_{4}^{\rm LTE}$ (thin solid line), and 
812: $R_{0}^{\rm LTE}$ (thin dashed line).
813: Shown in the ordinate is the normalized flux divided by the theoretical 
814: (pure) continuum; therefore, the local continuum level sometimes 
815: turns out to be slightly less than unity because of the extended wings 
816: of H lines. (Note that the theoretical equivalent widths 
817: $W_{4}^{\rm NLTE}$ corresponding to $R_{4}^{\rm NLTE}$ profiles 
818: discussed in Appendix B were calculated with respect to the local continuum, 
819: i.e. the maximum level in the neighborhood of the line profiles, 
820: irrespective of the scales in the ordinate.)
821: (a) Na~{\sc i} 5890, 
822: (b) Na~{\sc i} 5896, (c) Na~{\sc i} 8183, (d) Na~{\sc i} 8195,
823: (e) Na~{\sc i} 6154, (f) Na~{\sc i} 6161, (g) Na~{\sc i} 5683,
824: (h) Na~{\sc i} 5688, (i) K~{\sc i} 7665, (j) K~{\sc i} 7699, and
825: (k) Li~{\sc i} 6708. }
826: \label{figB1}
827: \end{figure}
828: 
829: The resulting theoretical equivalent widths ($W_{4}^{\rm NLTE}$) 
830: computed by integrating $R_{4}^{\rm NLTE}(\lambda)$ are
831: 129.7, 101.3, 18.4, 43.2, 0.8, 1.5, 4.0, 7.8, 12.8, 10.3, and 0.4 m$\rm\AA$
832: for these 9 lines, respectively. Comparing these values with the
833: observed equivalent widths ($W_{\lambda}^{\rm obs}$) given in Table 1,
834: we can see a satisfactory agreement between $W_{4}^{\rm NLTE}$ and
835: $W_{\lambda}^{\rm obs}$ (typically to within several percent except for 
836: Li~{\sc i} 6708\footnote{We see a rather large discrepancy for this 
837: Li line ($W_{4}^{\rm NLTE}$ = 0.4~m$\rm\AA$, while $W_{\lambda}^{\rm obs}$ 
838: = 0.7~m$\rm\AA$). This must be due to the inevitable numerical 
839: errors in $R_{4}^{\rm NLTE}$ (Li~{\sc i} 6708) computed by the CALSPEC 
840: program (synthesizing the flux spectrum by integrating the contributions 
841: from a number of finite surface elements; each $1^{\circ} \times 1^{\circ}$ 
842: in latitude and longitude), which becomes appreciable especially for this case of 
843: Li~{\sc i} 6708 line involving synthesis of several component lines. 
844: As a matter of fact, we can see a spurious wavy pattern in the continuum
845: level outside of this line (see Fig. B1k) which indicates the existence
846: of numerical problems (even if small). It may be worth stressing that 
847: the gravity-darkening correction $\Delta_{\rm GD} \simeq -\log (W_{4}/W_{0})$ 
848: is not affected by this numerical error in $W$ because it is canceled
849: by taking the ratio of $W$. In this sense, our semi-classical approach of 
850: establishing the final abundance (application of two corrections to the 
851: classical LTE abundance solution; cf. Sect. 3) is surely advantageous 
852: as far as this Li~{\sc i} 6708 line is concerned.}), by which we may conclude 
853: that the approach adopted in Sect. 3 is practically validated.
854: 
855: \end{document}
856: