1:                                                                                 

2: \documentclass[12pt,preprint]{aastex}

3:                                                                                 

4: \setcounter{totalnumber}{100}

5:                                                                                 

6: \newcommand{\eps}[1]{\mbox{log~$\epsilon$(#1)}}

7: \newcommand\iso[2]{$^{\rm #1}$#2}

8: \newcommand\wave[1]{\mbox{#1\,\AA}}

9: \def\cs22964{\mbox{CS~22964-161}}

10: \def\deg{{$^{\circ}$}}

11: \def\eg{\mbox{\it e.g.}}

12: \def\etal{\mbox{et al.}}

13: \def\ie{\mbox{i.e.}}

14: \def\kmsec{\mbox{km~s$^{\rm -1}$}}

15: \def\logg{\mbox{log~{\it g}}}

16: \def\Msun{\mbox{$M_{\odot}$}}

17: \def\teff{\mbox{$T_{\rm eff}$}}

18: \def\vturb{\mbox{$v_{\rm t}$}}

19: \def\rpro{\mbox{$r$-process}}

20: \def\spro{\mbox{$s$-process}}

21: \def\ncap{\mbox{$n$-capture}}

22:                                                                                 

23: \shorttitle{A Double-Lined Binary CEMP Star}

24: \shortauthors{Thompson et al.}

25:                                                                                 

26: \begin{document}

27:                                                                                 

28: \title{

29: CS22964-161: A Double-Lined Carbon- and s-Process-Enhanced 

30: Metal-Poor Binary Star\footnote{

31: This paper includes data gathered with the 6.5m Magellan and 2.5m 

32: du Pont Telescopes located at Las Campanas Observatory, Chile.}}

33:                                                                                 

34: \author{

35: Ian B. Thompson\altaffilmark{1},

36: Inese I. Ivans\altaffilmark{1,2}, 

37: Sara Bisterzo\altaffilmark{3},

38: Christopher Sneden\altaffilmark{4,1}, \\

39: Roberto Gallino\altaffilmark{3,5},

40: Sylvie Vauclair\altaffilmark{6},

41: Gregory S. Burley\altaffilmark{1}, \\

42: Stephen A. Shectman\altaffilmark{1},

43: George W. Preston\altaffilmark{1}

44: }

45:                                                                                 

46: \altaffiltext{1}{The Observatories of the Carnegie Institution of Washington,

47: 813 Santa Barbara St., Pasadena, CA 91101; ian,iii,burley,gwp@ociw.edu}

48:                                                                                 

49: \altaffiltext{2}{Princeton University Observatory, Peyton Hall, Princeton,

50: NJ 08544}

51: 

52: \altaffiltext{3}{Dipartimento di Fisica Generale, Universita' di Torino,

53: 10125 Torino, Italy; bisterzo,gallino@ph.unito.it}

54: 

55: \altaffiltext{4}{Department of Astronomy and McDonald Observatory, 

56: The University of Texas, Austin, TX 78712; chris@verdi.as.utexas.edu}

57: 

58: \altaffiltext{5}{Centre for Stellar and Planetary Astrophysics, Monash 

59: University, PO BOX 28M, Clayton, VIC 3800 Australia}

60: 

61: \altaffiltext{6}{Laboratoire d'Astrophysique de Toulouse et Tarbes 

62: - UMR 5572 - Universit\'e Paul Sabatier Toulouse III - CNRS, 

63: 14 Av. E. Belin, 31400 Toulouse, France; svcr@ast.obs-mip.fr}

64: 

65: 

66: 

67: 

68: 

69: \begin{abstract}

70: A detailed high-resolution spectroscopic analysis is presented for the 

71: carbon-rich low metallicity Galactic halo object \cs22964.

72: We have discovered that \cs22964\ is a double-lined spectroscopic

73: binary, and have derived accurate orbital components for the system.

74: From a model atmosphere analysis we show that both components are near 

75: the metal-poor main-sequence turnoff. 

76: Both stars are very enriched in carbon and in neutron-capture elements

77: that can be created in the \spro, including lead.

78: The primary star also possesses an abundance of lithium close to the 

79: value of the ``Spite-Plateau''.

80: The simplest interpretation is that the binary members seen today

81: were the recipients of these anomalous abundances from

82: a third star that was losing mass as part of its AGB evolution.

83: We compare the observed \cs22964\ abundance set with nucleosynthesis

84: predictions of AGB stars, and discuss issues of envelope stability

85: in the observed stars under mass transfer conditions, and consider

86: the dynamical stability of the alleged original triple star.

87: Finally, we consider the circumstances that permit survival of lithium, 

88: whatever its origin, in the spectrum of this extraordinary system.

89: \end{abstract}

90: 

91: \keywords{binaries: spectroscopic --- stars: Population II ---

92:           stars: abundances --- stars: individual (CS 22964-161) ---

93:           nuclear reactions, nucleosynthesis, abundances ---

94:           diffusion}

95: 

96: 

97: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

98: \section{INTRODUCTION\label{intro}}

99: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

100: 

101: The chemical memory of the Galaxy's initial elemental production in 

102: short-lived early-generation stars survives in  the present-day 

103: low-mass, low metallicity halo stars showing starkly contrasting 

104: abundance distributions.

105: Metal-poor stars have been found with order-of-magnitude differences in 

106: lithium contents, large ranges in $\alpha$-element abundances

107: (from [Mg,Si,Ca,Ti/Fe]\footnote{

108: We adopt the standard spectroscopic notation (Helfer, Wallerstein, 

109: \& Greenstein 1959\nocite{hel59}) that for elements A and B, 

110: \eps{A} $\equiv$ {\rm log}$_{\rm 10}$(N$_{\rm A}$/N$_{\rm H}$) + 12.0, and

111: [A/B] $\equiv$

112: {\rm log}$_{\rm 10}$(N$_{\rm A}$/N$_{\rm B}$)$_{\star}$ $-$

113: {\rm log}$_{\rm 10}$(N$_{\rm A}$/N$_{\rm B}$)$_{\odot}$.

114: Also, metallicity is defined as the stellar [Fe/H] value.} 

115: ~$<$ 0 to $\sim$+1), non-solar Fe-peak ratios,

116: and huge bulk variations in neutron-capture (\ncap) abundances.

117: 

118: One anomaly can be easily spotted in medium-resolution 

119: (R~$\equiv$ $\lambda/\Delta\lambda$~$\simeq$~2000)

120: spectroscopic surveys of metal-poor stars: large star-to-star 

121: variations in CH G-band strength, leading to similarly large carbon

122: abundance ranges.

123: Carbon-enhanced metal-poor stars (hereafter CEMP $\equiv$ [C/Fe]~$\gtrsim$~+1)

124: are plentiful at metallicities [Fe/H]~$<$~$-$2, with estimates

125: of their numbers ranging from $\simeq$14\% (Cohen \etal\ 2005\nocite{coh05})

126: to $\gtrsim$21\% (Lucatello \etal\ 2006\nocite{luc06}).

127: The large carbon overabundances are mainly (but not always) accompanied

128: by large \ncap\ overabundances, which are usually detectable only at higher

129: spectral resolution (R~$\gtrsim$~10,000).

130: In all but one of the CEMP stars discovered to date, the \ncap\ abundance

131: pattern has its origin in slow neutron-capture synthesis (the \spro).

132: The known exception is CS~22892-052 (\eg, Sneden \etal\ 2003a\nocite{sne03a} 

133: and references therein), which has [C/Fe]~$\sim$~+1 but an \ncap\ 

134: overabundance distribution clearly consistent with a rapid 

135: neutron-capture (the \rpro) origin.

136: Properties of CEMP stars have been summarized recently in a large-sample 

137: high-resolution survey by Aoki \etal\ (2007)\nocite{aok07}.

138: They conclude in part that the abundances of carbon and (\ncap)

139: barium are positively correlated (their Figure~6), pointing to 

140: a common nucleosynthetic origin of these elements in many CEMP stars.

141: 

142: \cs22964\ was first noted in the ``HK'' objective prism survey of low 

143: metallicity halo stars (Beers, Preston, \& Shectman 1992\nocite{bps92}, 

144: hereafter BPS92).

145: Using as metallicity calibration the strength of the \ion{Ca}{2} K~line,

146: BPS92 estimated [Fe/H]~$\simeq$~$-$2.62.

147: They also found \cs22964\ to be one of a small group of stars with

148: unusually strong CH G-bands (see their Table~8).

149: Recently, Rossi \etal\ (2005)\nocite{ros05} analyzed a moderate 

150: resolution spectrum of this star.

151: From the extant BVJK photometry and distance estimate they suggested

152: that \cs22964\ is a subgiant: \teff~=~5750~K and \logg~=~3.3.

153: Two metallicity estimates from the spectrum were in agreement

154: at [Fe/H]~=~$-$2.5, and three independent approaches to analysis of the

155: overall CH absorption strength suggested a large carbon abundance,

156: [C/Fe]~=~+1.1.

157: 

158: We observed \cs22964\ as part of a high resolution survey of candidate 

159: low-metallicity stars selected from BPS92.

160: When we discovered that the star shows very strong features of CH

161: and \ncap\ species \ion{Sr}{2} and \ion{Ba}{2}, similar to many 

162: binary blue metal-poor (BMP) stars (Sneden, Preston, \& Cowan

163: 2003b\nocite{sne03b}), we added it to a radial velocity monitoring program. 

164: Visual inspection of the next observation showed two sets of spectral 

165: lines, and so an intensive monitoring program was initiated on the du Pont 

166: and Magellan Clay telescopes at Las Campanas Observatory.

167: In this paper we present our orbital and abundance analyses of \cs22964.

168: Radial velocity data and the orbital solution are given in \S\ref{radvel},

169: and broadband photometric information in \S\ref{phot}.

170: We discuss the raw equivalent width measurements for the combined-light

171: spectra and the extraction of individual values for the \cs22964\ primary

172: and secondary stars in \S\ref{ewcalc}.

173: Determination of stellar atmospheric parameters is presented in 

174: \S\ref{model}, followed by abundance analyses of the individual stars

175: in \S\ref{abboth} and of the  \cs22964\ system in \S\ref{abtotal}.

176: Interpretation of the large lithium, carbon, and \spro\ abundances in

177: primary and secondary stars is discussed in \S\ref{history}.

178: Finally, we speculate on the nature of former asymptotic giant branch 

179: (AGB) star that we suppose was responsible for creation of the unique

180: abundance mix of the \cs22964\ binary in \S\ref{agb}.

181: 

182: 

183: 

184: 

185: 

186: 

187: 

188: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

189: \section{RADIAL VELOCITY OBSERVATIONS\label{radvel}}

190: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

191: 

192: We obtained spectroscopic observations of \cs22964\ with the Clay 6.5-m 

193: MIKE (Bernstein \etal\ 2003\nocite{ber03}) and du~Pont 2.5-m echelle 

194: (R~$\simeq$~25,000) spectrographs.

195: Properties of the two spectrographs are presented at the LCO

196: website\footnote{http://www.lco.cl}.

197: The Clay MIKE data have $R$~$\equiv$~$\lambda/\delta\lambda$ and continuous

198: spectral coverage for 3500~\AA~$\lesssim$~$\lambda$ $\lesssim$~7200~\AA.

199: The du~Pont data have R~$\simeq$~25,000, and out of the large spectral

200: coverage of those data we used the region

201: 4300~\AA~$\lesssim$~$\lambda$ $\lesssim$~4600~\AA for velocity measurements.

202: Exposure times ranged from 1245 to 3500 seconds on the Clay telescope and

203: 3000 to 4165 seconds on the du Pont telescope.

204: The observations generally consist of two exposures flanked by observations 

205: of a thorium-argon hollow-cathode lamp.

206: 

207: The Magellan observations were reduced with pipeline software written

208: by Dan Kelson following the approach of Kelson (2003, 2006).\nocite{kel03} 

209: \nocite{kel06} Post-extraction processing of the spectra

210: was done within the IRAF ECHELLE package.\footnote{IRAF is distributed by 

211: the National Optical Astronomy Observatories, which are operated by the 

212: Association of Universities for Research in Astronomy, Inc., under 

213: cooperative agreement with the NSF.} 

214: The du Pont observations were reduced completely with IRAF ECHELLE software. 

215: 

216: Velocities were initially measured with the IRAF FXCOR package using 

217: MIKE observations of HD~193901 as a template, and a preliminary 

218: orbit was derived. 

219: The three MIKE observations of \cs22964\ obtained at zero velocity 

220: crossing (phase~$\simeq$~0.11) were then averaged together to define a 

221: new template, hereafter called the syzygy spectrum. 

222: This spectrum has a total integration time of 10215~sec and 

223: S/N~$>$~160 at 4260~\AA.

224: We remeasured the velocities using this new template with the TODCOR 

225: algorithm (Zucker \& Mazeh 1994).\nocite{zuc94} 

226: The cross-correlations covered the wavelength interval 

227: 4130~\AA~$<$~$\lambda$~$<$~4300~\AA.

228: The syzygy spectrum was also used extensively in our abundance 

229: analysis (\S\ref{abboth}).

230: 

231: The radial velocity data were fit with a non-linear least squares solution.

232: We chose to fit to only the higher resolution and higher S/N Magellan data, 

233: using the du Pont data to confirm the orbital solution. 

234: The observations are presented in Table~\ref{tab1} which lists the 

235: heliocentric Julian Date (HJD) of mid-exposure, the velocities of the primary 

236: and secondary components, and the orbital phases of the observations. 

237: The adopted orbital elements are listed in Table~\ref{tab2} and 

238: the adopted orbit is plotted in Figure~\ref{f1}.

239: Of particular importance to later discussion in this paper are the

240: derived masses:  $M_{p}~sin^{3}~i$~= 0.773~$\pm~$0.009~$M_{\odot}$

241: and $M_{s}~sin^{3}~i~$~= 0.680~$\pm$~0.007~$M_{\odot}$; the orbital

242: inclination cannot be derived from our data.

243: 

244: We will discuss the velocity residuals to the orbital solution

245: further in \S\ref{agb}.

246: Hereafter, references to individual spectra will be by the HJD of

247: Table~\ref{tab1}, \eg, the spectrum obtained with Magellan MIKE on HJD 

248: 13817.90625 will be called ``observation 13817''.

249: 

250: 

251: 

252: 

253: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

254: \section{PHOTOMETRIC OBSERVATIONS\label{phot}}

255: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

256: 

257: Preston, Shectman, \& Beers (1991)\nocite{pre91} list $V$~=~14.41, 

258: $B-V$~=~0.488, and $U-B$~=~$-$0.171 for \cs22964\ based on a single 

259: photoelectric observation on the du Pont telescope. 

260: New CCD observations of this star were obtained with the du Pont 

261: telescope on UT 28 May 2006.

262: The data were calibrated with observations of the Landolt 

263: standard field Markarian A (Landolt 1992\nocite{lan92}). 

264: Observations of the standards and program object were taken at an airmass of 

265: $\sim$1.12, and we used standard extinction coefficients in the reductions. 

266: We derived $V$~=~14.43 and $B-V$~=~0.498 for \cs22964.

267: Old and new photometric data are consistent with the typical $\sim$1\%

268: observational uncertainties in magnitudes, so we adopt final values

269: of $V$~=~14.42 and $B-V$~=~0.49.

270: 

271: The ephemeris given in Table~\ref{tab2} predicts a primary eclipse at

272: HJD2454315.045 (2 August 2007, UT~13.08 hours) and a secondary eclipse at

273: HJD 2454344.453 (31 August 2007, UT~22.87 hours). \cs22964\ was

274: monitored on the nights of 2/3 August 2007 and 31 August/1 September

275: using the CCD camera on the Swope telescope at Las Campanas. Observations

276: were obtained approximately every hour with a $V$ filter. No

277: variations in the $V$ magnitude of \cs22964\ in excess of 0.02 

278: magnitudes were detected.

279: 

280: 

281: 

282: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

283: \section{EQUIVALENT WIDTH DETERMINATIONS\label{ewcalc}}

284: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

285: 

286: A double-line binary spectrum is a complex, time-variable mix of two 

287: sets of absorption lines and two usually unequal continuum flux levels.

288: Following Preston (1994)\nocite{pre94}\footnote{

289: A similar kind of analysis of the very metal-poor binary CS~22876-032

290: has been discussed by Norris, Beers, \& Ryan (2000\nocite{nor00}).}

291: we define {\it observed} equivalent 

292: widths $EW_o$ to be those measured in the combined-light spectra, 

293: and {\it true} equivalent widths $EW_t$ to be those of each star in the 

294: absence of its companion's contribution.

295: If the primary-secondary velocity separation is large, one can attempt

296: to measure observed $EW$s of primary and secondary stars independently.

297: If this can be accomplished then true $EW$s can be computed from

298: knowledge of the relative flux levels of the stars.

299: In practice the derivation of true $EW$s is complex and subject to large

300: uncertainties.

301: 

302: In Figure \ref{f2} we illustrate the difficulties in deducing observed

303: $EW$s for each star from the observed \cs22964\ composite spectra.

304: Recall that in the syzygy spectrum (the co-addition of three individual

305: observations), there is no radial velocity separation, or 

306: $\Delta V_R$~$\simeq$~0~\kmsec. 

307: The primary and secondary spectral lines coincide in wavelength, producing 

308: the simple spectrum shown at the bottom of Figure~\ref{f2}.

309: In contrast, the middle spectrum in this figure was generated from spectra 

310: in which the velocity separation was large enough for primary and secondary 

311: lines to be resolved (hereafter called velocity-split spectra).

312: We co-added three of these spectra that have similarly large velocity

313: differences, $\Delta V_R$~$\simeq$~60~\kmsec\ (observations 13580, 

314: 60.12~\kmsec; 13817, 58.09~\kmsec; and 13832, 62.40~\kmsec; 

315: Table~\ref{tab1}).

316: The weaker, redshifted secondary absorption lines are obvious from 

317: comparison of this and the syzygy spectrum.

318: 

319: Identification of the secondary absorption lines is clarified by 

320: inspection of the top spectrum in Figure \ref{f2}. 

321: We created this artificial spectrum to mimic the appearance of the

322: secondary by diluting the observed syzygy spectrum with addition of a 

323: constant three times larger than the observed continuum, and shifting 

324: that spectrum redward by 60~\kmsec.

325: In this line-rich wavelength domain, the velocity shift yields a few 

326: cleanly separated primary and secondary absorption features. 

327: For example, 4199.11~\AA\ \ion{Fe}{1} in the primary becomes 

328: 4199.95~\AA\ in the secondary, and observed $EW$s of both stars

329: can be measured.

330: More often however, secondary lines are shifted to wavelengths 

331: very close to other primary lines, destroying the utility of both

332: primary and secondary features.

333: An example of this is the \ion{Fe}{1} 4198.33~\AA\ line, which at 

334: $\Delta V_R$~$\simeq$~60~\kmsec\ becomes 4199.17~\AA\ for the secondary

335: star, which contaminates the \ion{Fe}{1} 4199.11~\AA\ line of the 

336: primary star. 

337: The various blending issues, combined with the intrinsic weakness of 

338: the secondary spectrum, yield few spectral features with clean

339: $EW$ values for both primary and secondary stars.

340: 

341: 

342: 

343: 

344: \subsection{Equivalent Widths from Comparison of Syzygy and Velocity-Split

345:             Spectra\label{ewsyzygy}}

346: 

347: In view of the blending issues outlined above, we derived the 

348: observed $EW$s through a spectrum difference technique. 

349: On the mean syzygy spectrum, we measured $EW_{o,tot} = EW_{o,p} + EW_{o,s}$,

350: where subscripts $o$ denotes an observed $EW$, and $p$ and $s$ denote

351: primary and secondary stars.  

352: On five spectra with large primary-secondary velocity separations (13580,

353: 13585, 13587, 13817, 13834) we measured the primary star lines, $EW_{o,p}$.

354: These five independent values were then averaged, and then the secondary

355: star's values were computed as $EW_{o,s} = EW_{o,tot} - <EW_{o,p}>$.

356: In this procedure we attempted to avoid primary lines that would have 

357: significant contamination by secondary lines in the velocity-split spectra.

358: 

359: The observed $EW$ values are given in Table~\ref{tab3}, along with the line

360: excitation potentials and transition probabilities.

361: In this table we also include the atomic data for lines from which 

362: abundances ultimately derived from synthetic spectrum rather than 

363: $EW$ computations.  

364: For one estimate of the uncertainty in our $EW$ measurement procedure, 

365: we computed standard deviations for the primary-star measurements 

366: of each line $\sigma EW_{o,p}$.

367: The mean and median of these values for the whole line data set were

368: 5.2 and 4.1~m\AA, respectively.

369: The uncertainties in $EW_{o,tot}$ are expected to be smaller because the

370: syzygy spectrum is the mean of three individual observations and is thus

371: of higher S/N.

372: Therefore we take 4$-$5~m\AA\ as an estimate of the uncertainty in $EW_{o,s}$

373: values determined in the subtraction procedure.

374: 

375: For a few very strong transitions the secondary lines are deep enough to

376: cleanly detect when they split away from the primary lines.

377: We have employed these to assess the reliability of the $EW$s derived

378: by the subtraction method described above.

379: In Figure~\ref{f3} we show a comparison of $EW_{o,s}$ values given

380: in Table~\ref{tab3} for seven lines that we measured on up to six 

381: velocity-split spectra (the five named in the previous paragraph 

382: plus 13832, which was not used in the subtraction procedure).

383: Taking the means of the $EW_{o,s}$ measurements of each line and comparing

384: them to the subtraction-based values of Table~\ref{tab3}, the average

385: difference for the seven lines is 0.8~m\AA\ with a scatter 

386: $\sigma$~=~3.5~m\AA\ (the median difference is 0.6~m\AA).

387: This argues that in general the individual $EW_{o,s}$ values agree with the 

388: $EW_{o,s}$ subtraction-based ones.

389: 

390: Derivation of true $EW$s depends on knowledge of the relative 

391: luminosities of primary and secondary stars.

392: Formally, from Preston's (1994)\nocite{pre94} equations 4$-$5, we have

393: $EW_{p,t} = EW_{p,o}(1 + l_s/l_p)$, and

394: $EW_{s,t} = EW_{s,o}(1 + l_p/l_s)$, where subscript $t$ represents the

395: true $EW$, and $l$ denotes an apparent luminosity.

396: The luminosity ratios will be wavelength-dependent if the two stars are

397: not identical in temperature.  

398: For the entire line data set, ignoring weaker primary lines (those

399: with $EW_{p,o}$~$<$~25.0~m\AA), we calculated a median equivalent 

400: width ratio ($EW_{p,o}/EW_{s,o}$)~=~5.2.

401: Inspection of the velocity-split spectra suggested that {\it relative}

402: line strengths in the secondary spectrum were not radically different 

403: from those of the primary spectrum.  

404: Therefore we concluded that the spectral types of the stars are not too 

405: dissimilar and therefore we concluded that to first approximation

406: $EW_{p,t}$~$\sim$~$EW_{s,t}$.

407: This assumption then leads to $EW_{p,o}/EW_{s,o}$~$\sim$ $l_p/l_s$~$\sim$~5.

408: 

409: To account roughly for the small derived temperature difference 

410: (see \S\ref{model}) we finally adopted $l_p/l_s$~=~5.0 in the 

411: photometric $V$ bandpass ($\lambda \simeq 5500$~\AA), and increased 

412: the ratio linearly by a small amount with decreasing wavelength. 

413: Thus in the $B$ bandpass ($\lambda \simeq 4400$~\AA) we used

414: $l_p/l_s$~=~5.6.

415: Final true $EW$ values using this prescription are given in Table~\ref{tab3}.

416: The correction factors between observed and true $EW$s were approximately 

417: 1.2 and and 6.0 for primary and secondary stars in the $V$ spectral region, 

418: and the disparity in these factors is larger at $B$.

419: Clearly the $EW_{p,t}$ values of Table~\ref{tab3} are much more reliably

420: determined than the $EW_{s,t}$ ones.

421: 

422: 

423: 

424: 

425: 

426: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

427: \section{STELLAR ATMOSPHERE PARAMETERS\label{model}}

428: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

429: 

430: 

431: \subsection{Derivation of Parameters\label{params}}

432: 

433: We used the observed $B-V$ color and minimum masses for the binary 

434: together with the Victoria-Regina stellar models (VandenBerg, 

435: Bergbusch, \& Dowler 2006)\nocite{van06}, to estimate initial model 

436: atmospheric parameters for the component stars of \cs22964.

437: We adopted $B-V$~=~0.49 from \S\ref{phot} and $E(B-V)$~=~0.07 (BPS92; 

438: a value we also estimate employing the dust maps of Schlegel, Finkbeiner, 

439: \& Davis 1998)\nocite{sch98} to obtain $(B-V)_{0}$ = 0.42. 

440: We assumed initially that $sin^{3}~i$~$\simeq$~1.0 for the binary orbit

441: and thus $M_{p}$~= 0.773~$M_{\odot}$ and $M_{s}$~= 0.680~$M_{\odot}$, 

442: as derived in \S\ref{radvel}.

443: We interpolated the Victoria-Regina models computed for [Fe/H]~=~$-$2.31,

444: [$\alpha$/Fe]~=~+0.3, and $Y$~=~0.24, to obtain evolutionary tracks

445: for these masses.

446: These tracks were used to derive $B-V$ colors and luminosity ratios for

447: the component stars.

448: We adopted starting values of \teff, \logg, and $l_p/l_s$ where

449: the tracks give $B-V$~=~0.42 for the combined system.

450: These values were \teff$_{,p}$~=~6050~K, \teff$_{,s}$~=~5950~K 

451: and \logg$_p$~=~3.6, \logg$_s$~=~4.2. 

452: The luminosity ratios using these parameters were $l_p/l_s~(B)$~=~8.65 

453: and $l_p/l_s~(V)$~=~8.39, somewhat larger than implied by our spectra.

454: 

455: Final model atmospheric parameters were found iteratively from the $EW$ 

456: data for the two stars.

457: We employed the LTE line analysis code MOOG\footnote{

458: Available at: http://verdi.as.utexas.edu/moog.html}

459: (Sneden 1973)\nocite{sne73} and

460: interpolated model atmospheres from the Kurucz 

461: (1998)\nocite{kur98}\footnote{ 

462: Available at: http://kurucz.harvard.edu/} grid computed with no convective 

463: overshoot (as recommended by Castelli, Gratton, \& Kurucz 1997\nocite{cas97},

464: and by Peterson, Dorman, \& Rood 2001\nocite{pet01}).  

465: 

466: We began by using standard criteria to estimate the model parameters of 

467: the \cs22964\ primary:  {\it (a)} for \teff, no trend of derived \ion{Fe}{1}

468: individual line abundances with excitation potential; {\it (b)} for

469: \vturb, no trend of \ion{Fe}{1} abundances with $EW$; {\it (c)} for

470: \logg, equality of mean \ion{Fe}{1} and \ion{Fe}{2} abundances (for no

471: other element could we reliably measure lines of the neutral and ionized

472: species); and {\it (d)} for model metallicity [M/H], a value roughly

473: compatible with the Fe and $\alpha$ abundances.

474: These criteria could be assessed reliably for the primary star because

475: it dominates the combined light of the two stars.

476: In the top panel of Figure~\ref{f4} we illustrate the line-to-line

477: scatter and (lack of) trend with wavelength of the primary's \ion{Fe}{1}

478: and \ion{Fe}{2} transitions.

479: With iteration among the parameters we derived

480: (\teff$_{,p}$, \logg$_p$, \vturb$_p$, [M/H]$_p$)~= 

481: (6050$\pm$100~K, 3.7$\pm$0.2, 1.2$\pm$0.3~\kmsec, $-$2.2$\pm$0.2).

482: 

483: To estimate model parameters for the secondary we assumed that 

484: derived [Fe/H] metallicities and abundance ratios of the lighter 

485: elements (Z~$\leq$~30) in the primary and secondary stars should be 

486: essentially identical if they were formed from the same interstellar cloud.

487: Iteration among several sets of (\teff$_{,s}$, \logg$_s$) pairs was done

488: until [Fe/H]$_s$~$\simeq$~[Fe/H]$_p$ and Fe ionization equilibrium

489: for the secondary was achieved.

490: Given the weakness of the secondary spectrum and the resulting large 

491: correction factors used to calculate $EW_{s,t}$ from $EW_{s,o}$ values,

492: it is not surprising that the spectroscopic constraints on the

493: secondary parameters were weak.

494: This is apparent from the adopted Fe abundances displayed in the 

495: bottom panel of Figure~\ref{f4}.

496: The $\sigma$ values for individual line abundances were about three times 

497: larger for the secondary than the primary (Table~\ref{tab4}), 

498: and number of transitions was substantially smaller (\eg, we measured 

499: seven \ion{Fe}{2} lines for the primary but only three for the secondary).

500: 

501: We adopted final model parameters of 

502: (\teff$_{,s}$, \logg$_s$, \vturb$_s$, [M/H]$_s$)~= 

503: (5850~K, 4.1, 0.9~\kmsec, $-$2.2).

504: Model uncertainties for the secondary were not easy to estimate and, 

505: of course, were tied to our opening assumption that the two stars have

506: identical overall metallicities and abundance ratios.

507: Thus with [M/H]$_p$~$\equiv$~[M/H]$_s$, the uncertainties in \teff,

508: \logg, and \vturb\ of the secondary are approximately double their

509: values quoted above for the primary.  

510: Thus caution is obviously warranted in interpretation of the model

511: parameters of the \cs22964\ secondary.

512:                                                                                 

513:                                                                                 

514: 

515: \subsection{Comparison to Evolutionary Tracks\label{evol}}

516:                                                                                 

517: The well-determined \teff$_{,p}$ and \logg$_p$ values can be used

518: with the mass and luminosity ratios of the stars to provide an

519: independent estimate of \teff$_{,s}$ and \logg$_s$.  

520: Standard relations $L \propto R^2\teff^4$ and $g \propto M/R^2$ lead to

521: $${\rm log}(L_s/L_p) = {\rm log}(l_s/l_p) = 

522: {\rm log}(M_s/M_p) + 4\times{\rm log}(\teff_{,s}/\teff_{,p}) - 

523: {\rm log}(g_s/g_p)$$

524: Taking an approximate average luminosity ratio to be 

525: $l_p/l_s$~$\simeq$~5.3, adopting $M_s/M_p$~=~1.15 (Table~\ref{tab2})

526: and assuming \teff$_{,p}$~=~6050~K and \logg$_p$~=~3.7 from above, 

527: we get a predicted temperature-gravity relationship 

528: \logg$_s$~$\simeq$~4$\times$log(\teff$_{,s}$)~$-$~10.8 .

529: 

530: In Figure~\ref{f5} we show the very metal-poor main sequence 

531: turnoff region of the H-R diagram in (log~\teff, \logg) units.

532: The Victoria-Regina evolutionary tracks (VandenBerg \etal\ 

533: 2006\nocite{van06}; [Fe/H]~=~$-$2.31, [$\alpha$/Fe]~=~+0.3, 

534: and $Y$~=~0.24) discussed in \S\ref{params} are again employed.

535: However, the unknown binary orbital inclination of \cs22964\ 

536: cannot be ignored here.

537: Therefore we have plotted the tracks for three pairs of masses 

538: corresponding to assumed $sin\ i$ values of 90\deg, 80\deg, and 75\deg.

539: Also plotted is a straight line representing the \cs22964\ secondary star

540: temperature-gravity equation derived above.

541: 

542: The \cs22964\ primary and secondary and (\teff, \logg) positions are

543: indicated with filled  circles in Figure~\ref{f5}.

544: We also add data indicated with open triangles from CEMP high 

545: resolution spectroscopic studies for C-rich stars of similar metallicity, 

546: taken here to be [Fe/H]~=~$-$2.4~$\pm$~0.4.

547: The studies include those of Aoki \etal\ (2002)\nocite{aok02},

548: Sneden \etal\ (2003b)\nocite{sne03b}, and

549: Cohen \etal\ (2006)\nocite{coh06}.

550: If \teff$_{,s}$~=~5850~K from the spectroscopic analysis then the 

551: temperature-gravity relationship from above predicts \logg$_s$~=~4.3 

552: (indicated by a filled square in the figure). 

553: Our spectroscopic value of \logg$_s$~=~4.1 lies well within the 

554: uncertainties of both estimates.

555: 

556: Given the apparently anomalous position of the \cs22964\ secondary

557: in Figure~\ref{f5}, it is worth repeating its abundance analysis 

558: using atmospheric parameters forced to approximately conform with the

559: evolutionary tracks.

560: This is equivalent to attempting a model near the high temperature,

561: high gravity end of its predicted relationship shown in the figure.

562: Therefore we computed abundances for a model with parameter set

563: (\teff$_{,s}$, \logg$_s$, [M/H]$_s$)~= (6300~K, 4.5, $-$2.2).

564: Assumption of a secondary microturbulent velocity

565: \vturb$_s$~= \vturb$_p$~=~ $-$1.2~\kmsec, yields \eps{Fe}~=~5.45 

566: and 5.27 from \ion{Fe}{1} and \ion{Fe}{2} lines, respectively.

567: These values are substantially larger than the mean abundances for

568: primary and secondary given in Table~\ref{tab4}, \eps{Fe}~=~5.11.

569: This is in agreement with expectations of a larger derived 

570: metallicity from the 350~K increase in \teff$_{,s}$ for this test.

571: However, the \ion{Fe}{1} line abundances for this hotter model also

572: exhibit an obvious trend with $EW$.

573: This problem could be corrected by increasing \vturb\ to 2.4~\kmsec, 

574: and then we get \eps{Fe}~=~5.18 and 5.15 from \ion{Fe}{1} and 

575: \ion{Fe}{2} lines, very close to our final adopted Fe abundances for 

576: the secondary.

577: However, it is difficult to reconcile such a large \vturb\ with

578: the much smaller value determined with more confidence in the primary,

579: as well as standard values determined in many literature studies of 

580: near-turnoff stars.

581: 

582: 

583: 

584:                                                                                

585: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

586: \section{ABUNDANCES OF THE INDIVIDUAL STARS\label{abboth}}

587: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

588: 

589: With the $EW$ data of Table~\ref{tab3} and the interpolated model 

590: atmospheres described in \S\ref{model}, we determined abundances of 

591: a few key elements whose absorption lines are detectable in both 

592: primary and secondary observed spectra.

593: These abundances are given in Table~\ref{tab4}.

594: They suggest that in general the abundance ratios of all elements in

595: the two stars agree to within the stated uncertainties. 

596: Both stars are relatively enriched in the $\alpha$ elements:

597: $<$[Mg,Ca,Ti/Fe]$_p>$~$\simeq$ +0.4 and $<$[Mg,Ca,Ti/Fe]$_s>$~$\simeq$ +0.3.

598: Both stars have solar-system Ni/Fe ratios, probably no substantial

599: depletions or enhancements of Na (special comment on this element

600: will be given in \S\ref{abtotal}), and large deficiencies of Al.

601: All these abundance ratios are consistent with expectations for 

602: normal metal-poor Population~II stars.

603: 

604: More importantly, we find very large relative abundances of C, Sr, 

605: and Ba ([X/Fe]~=~+0.5 to +1.5; Table~\ref{tab4}) in both primary 

606: and secondary stars of the \cs22964\ system.

607: For Sr and Ba abundances we first used the $EW$ subtraction technique, which 

608: suggested roughly equal abundances of these elements in both stars.

609: We confirmed and strengthened this result through synthetic spectrum 

610: computations of the strong \ion{Sr}{2} 4077.71, 4215.52~\AA\ and the 

611: \ion{Ba}{2} 4554.04~\AA\ lines in the six velocity-split spectra.

612: To produce the binary syntheses we modified the MOOG line analysis code 

613: to compute individual spectra for primary and secondary stars, then to 

614: add them after {\it (a)} shifting the secondary spectrum in wavelength 

615: to account for the velocity difference between the stars, and 

616: {\it (b)} weighting the primary and secondary spectra by the 

617: appropriate luminosity ratio.

618: 

619: In Figure~\ref{f6} we show the resulting observed/synthetic binary 

620: spectrum match for the \ion{Sr}{2} 4077.71~\AA\ line in observation 13817.

621: The absorption spectrum of the secondary star is shifted by +58.1~\kmsec\

622: (+0.79~\AA), in agreement with the observed feature in Figure~\ref{f6}.

623: The depth of the \ion{Sr}{2} line in the secondary is weak, as expected due

624: to the luminosity difference between the two stars ($l_p/l_s$~$\simeq$ 6)

625: at this wavelength.

626: Note also in Figure~\ref{f6} the relative insensitivity of the

627: feature to abundance changes for the secondary, even with the large 

628: ($\pm$0.5~dex) excursions in its assumed Sr content.

629: This occurs because the \ion{Sr}{2} line is as saturated in the secondary

630: as it is in the primary, but the central intensity of the unsmoothed 

631: spectrum is $\sim$0.2 of the continuum.  

632: Thus after the secondary's spectrum is diluted by the much larger flux of

633: the primary, a weak line that is relatively insensitive to abundance

634: changes ensues in the combined spectrum (and a naturally weaker line in the 

635: secondary simply becomes undetectable in the sum).

636: 

637: In Figure~\ref{f7} we illustrate the appearance of a small portion of

638: the CH G-band $A^{\rm 2}\Delta-X^{\rm 2}\Pi^+$ Q-branch bandhead in the 

639: observed combined-light spectra with a large velocity split.

640: This portion of the G-band has a sharp blue edge; the central wavelength 

641: of the first line is 4323.0~\AA. 

642: With $\Delta V_R$~= +58.1~\kmsec, the left edge of the secondary's 

643: bandhead begins at 4323.8~\AA, as indicated in the bottom panel of 

644: Figure~\ref{f7}.

645: In the top panel we show attempts to match the observed bandhead with

646: synthetic spectra that include only CH lines of the primary star.

647: When the spectral interval $\lambda$~$<$~4324.8~\AA\ is fit well, 

648: observed absorption is clearly missing at longer wavelengths in the 

649: synthetic spectrum.

650: This is completely solved by the addition of the secondary's CH bandhead,

651: at a comparable C abundance level to that of the primary, as shown in the 

652: bottom panel.

653: 

654: It is very difficult to derive reliable abundances in the \cs22964\

655: secondary even for the strong features illustrated here.

656: Nevertheless, it is clear that both components of this binary have

657: substantial overabundances of C and \ncap\ elements Sr and Ba.

658: Within the uncertainties of our analysis, the overabundance factors

659: for these elements appear to be the same.

660: Enhanced C accompanies \spro\ synthesis of \ncap\ elements during 

661: partial He-burning episodes of low/intermediate-mass stars, and the 

662: joint production of these elements is evident in the observed 

663: abundances of a number of BMP stars such as CS~29497-030 

664: (Sneden \etal\ 2003b\nocite{sne03b}; Ivans \etal\ 2005\nocite{iva05}).

665: However, a substantial C overabundance has also been seen in the

666: \rpro-rich star CS~22892-052 (Sneden \etal\ 2003a\nocite{sne03a}).

667: The Sr and Ba abundances determined to this point cannot distinguish between 

668: possible \ncap\ mechanisms that created the very heavy elements in \cs22964.

669: 

670: 

671: 

672: 

673: 

674: 

675: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

676: \section{ABUNDANCES FROM THE SYZYGY SPECTRUM\label{abtotal}}

677: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

678: 

679: To gain further insight into the \ncap\ element abundance distribution

680: we returned to the higher S/N mean syzygy spectrum.

681: Preston \etal\ (2006b)\nocite{pre06a} argued that the relative 

682: strengths of \ion{La}{2} and \ion{Eu}{2} lines can easily distinguish 

683: \spro\ dominance (stronger La features) from \rpro\ dominance (stronger 

684: Eu); see their Figure~1.

685: In panels (a) and (b) of Figure~\ref{f8} we show the same La and

686: Eu lines discussed by Preston \etal; the greater strength of the La feature

687: is apparent.

688: 

689: We computed a mean $EW$ from \ion{La}{2} lines at 3988.5, 3995.7,

690: 4086.7, and 4123.2~\AA, and a mean $EW$ from \ion{Eu}{2} lines at 

691: 3907.1, 4129.7, and 4205.1~\AA\ for a few warm metal-poor stars 

692: with \ncap\ overabundances.

693: The resulting ratio $<EW_{\rm La}>$/$<EW_{\rm Eu}>$~$\simeq$ 0.5 

694: for the \rpro-rich red horizontal-branch star CS~22886-043 

695: (Preston \etal\ 2006a)\nocite{pre06b}, $\simeq$~2.8 for the \spro-rich 

696: RR~Lyrae TY~Gruis (Preston \etal\ 2006a), and $\simeq$~2.7 for the 

697: $r$+$s$ BMP star CS~29497-030 (Ivans \etal\ 2005).\nocite{iva05}

698: For \cs22964\ we found $<EW_{\rm La}>$/$<EW_{\rm Eu}>$~$\simeq$~2.7, 

699: an unmistakable signature of an \spro\ abundance distribution.

700: 

701: For a more detailed \ncap\ element distribution for \cs22964\ we computed 

702: synthetic spectra of many transitions in the syzygy spectrum. 

703: We used the same binary synthesis version of the MOOG code that 

704: was employed for the velocity-split spectra shifted to 

705: $\Delta V_R$~=~0~\kmsec\ (as described in \S\ref{abboth}).

706: For these computations it was also necessary to assume that 

707: \eps{X}$_p$~=~\eps{X}$_s$ for all elements X.

708: We also derived abundances of a few lighter elements of interest

709: with this technique.

710: It should be noted that since the primary star is 4--6 times brighter than the 

711: secondary, abundances derived in this manner mostly apply to the primary.

712: 

713: In Table~\ref{tab5} we give the abundances for the \cs22964\ 

714: binary system.

715: For elements with abundances determined for the individual stars as 

716: discussed in \S\ref{abboth}, we also give estimates of their ``system''

717: abundances in this table.

718: These mean abundances were computed from the entries in 

719: Table~\ref{tab4}, giving both the abundances and their 

720: uncertainties ($\sigma$) of the primary star five times more 

721: weight than those of the secondary.

722: For elements with abundances derived from synthetic spectra of the syzygy

723: spectrum, the abundances and $\sigma$ values are means of the

724: results from individual lines, wherever possible.

725: For several of these elements only one transition was employed.

726: In these cases we adopted $\sigma$ = 0.20 or 0.25, depending on the

727: difficulties attendant in the synthetic/observed spectrum matches.

728: In the next few paragraphs we discuss the analyses of a few species 

729: that deserve special comment.

730: 

731: {\it Li I:} The resonance transition at 6707.8~\AA\ was easily 

732: detected in all \cs22964\ spectra, with $EW_{o,tot}$~$\simeq$~24.5~m\AA\ 

733: from the syzygy data. 

734: Our synthesis included only \iso{7}{Li}, but with its full 

735: hyperfine components.

736: The resolution and S/N combination of our spectra precluded any 

737: meaningful search for the presence of \iso{6}{Li}.

738: Reyniers \etal\ (2002)\nocite{rey02} have shown that the presence of a 

739: \ion{Ce}{2} transition at 6708.09~\AA\ can substantially contaminate 

740: the Li feature in \ncap-rich stars.

741: However, in our spectra the Ce line wavelength is too far from the 

742: observed feature, and our syntheses indicated that the Ce abundance 

743: would need to be about two orders of magnitude larger than our derived 

744: value (Table~\ref{tab5}) to produce measurable absorption in the 

745: \cs22964\ spectrum.

746: 

747: We attempted to detect the secondary \ion{Li}{1} line in two different ways.

748: First, we applied the subtraction technique (\S\ref{ewcalc}) to this

749: feature. 

750: However, the six velocity-split spectra yielded $EW_{o,p}$~=~23.5~m\AA\ 

751: with $\sigma$~=~3.5~m\AA\ (consistent with typical uncertainties for this

752: technique).

753: Thus the implied $EW_{o,s}$~=~1.0~m\AA\ is consistent with no detection

754: of the secondary's line.

755: Second, we co-added the velocity-split spectra after shifting them to the

756: rest system of the secondary.

757: A very weak ($EW_{o,s}$~$\simeq$~4~m\AA) line appears at 

758: $\lambda$~=~6707.9~\AA, about 0.1~\AA\ redward of the \ion{Li}{1} 

759: feature centroid.

760: We computed synthetic spectra assuming that the observed line is Li, and 

761: derived \eps{Li}~$\simeq$~+2.0 with an estimated uncertainty of $\pm$0.2

762: from the observed/synthetic fit.

763: This abundance is fortuitously close to the system value, but we do not

764: believe that the present data warrant a claim of Li detection in the

765: \cs22964\ secondary.

766: 

767: {\it Na I:} The 5682, 5688~\AA\ doublet, used in abundance analyses when the 

768: D-lines at 5890, 5896~\AA\ are too strong, is undetectably weak in \cs22964.

769: A synthetic spectrum match to the syzygy spectrum suggests that

770: \eps{Na}~$\lesssim$ 3.9.

771: The D~lines are strong, $\sim$40\% deep.  

772: Unfortunately, their profiles appear to be contaminated by telluric emission.

773: If we assume that the line centers and red profiles are unblended then

774: \eps{Na}~$\simeq$ 3.9.

775: 

776: Given the great strengths of the D-lines in the syzygy spectrum (dominated

777: by the primary star) we returned to the velocity-split spectra and 

778: searched for the secondary Na~D lines.

779: Absorptions a few percent deep at the expected wavelengths were seen

780: in all spectra.

781: Co-addition of these spectra after shifting to the rest velocity system of 

782: secondary yielded detection of both D-lines, and we estimated

783: \eps{Na}~$\simeq$ 3.8.

784: A problem with our analyses of the D-lines in all spectra was the lack of

785: telluric H$_{\rm 2}$O line cancellation; we did not observe suitable hot,

786: rapidly rotating stars for this purpose.

787: However, any unaccounted-for telluric features would drive the derived Na

788: abundances to larger values.  

789: Thus we feel confident that Na is not overabundant in \cs22964.

790: 

791: {\it CH:} Carbon was determined from CH G-band features in the 

792: wavelength range 4260--4330~\AA, with a large number of individual lines 

793: contributing to the average abundance.

794: Our CH line list was taken from the Kurucz (1998)\nocite{kur98} compendium.

795: The solar C abundance listed in Table~\ref{tab5} was determined with

796: the same line list (see Sneden \etal\ 2003b\nocite{sne03b}), rather than 

797: adopted from the recent revision of its abundance from other spectral

798: features by Allende Prieto, Lambert, \& Asplund (2002)\nocite{ale02}.

799: 

800: {\it Si I:} The only detectable line, at 3905.53~\AA, suffers potentially 

801: large  CH contamination, as pointed out by, \eg, Cohen \etal\ 

802: (2004)\nocite{coh04}.

803: These CH lines are very weak in ordinary warm metal-poor stars (Preston 

804: \etal\ 2006a)\nocite{pre06b}, but are strong in C-enhanced \cs22964.

805: We took full account of the CH in our synthesis of the \ion{Si}{1} line.

806: Note that the secondary's \ion{Si}{1} line can be detected in the 

807: velocity-split spectra, but it is always too blended with primary 

808: CH lines to permit a useful primary/secondary Si abundance analysis.

809: 

810: {\it Species with substructure:} Lines of \ion{Sc}{2}, \ion{Mn}{1}, 

811: \ion{Y}{2}, and \ion{La}{2} have significant hyperfine subcomponents,

812: which were explicitly accounted for in our syntheses (each of these 

813: elements have only one naturally-occurring isotope).

814: For \ion{Ba}{2} and \ion{Yb}{2} (whose single spectral feature at

815: 3694.2~\AA\ is shown in panel (c) of Figure~\ref{f8}, 

816: both hyperfine and isotopic substructure were included.

817: We assumed an \spro\ distribution of their isotopic fractions: 

818: $f$(\iso{134}{Ba})~=~0.038, $f$(\iso{135}{Ba})~=~0.015, 

819: $f$(\iso{136}{Ba})~=~0.107, $f$(\iso{137}{Ba})~=~0.080, 

820: $f$(\iso{138}{Ba})~=~0.758; and

821: $f$(\iso{171}{Yb})~=~0.180, $f$(\iso{172}{Yb})~=~0.219, 

822: $f$(\iso{173}{Yb})~=~0.191, $f$(\iso{174}{Yb})~=~0.226, 

823: $f$(\iso{176}{Yb})~=~0.185.

824: We justify the \spro\ mixture choice below by showing that the total

825: abundance distribution, involving 17 elements, follows an \spro-dominant

826: pattern.

827: 

828: {\it Pb I:} The 4057.8~\AA\ line was easily detected in the syzygy spectrum 

829: (panel (d) of Figure~\ref{f8}), with $EW_{o,tot}$~$\sim$~5~m\AA.

830: This line is very weak, but we confirmed its existence and approximate

831: strength through co-addition of the six velocity-split spectra.

832: The Pb abundance was derived from synthetic spectra in which the

833: isotopic and hyperfine substructure were taken into account following

834: Aoki \etal\ (2002)\nocite{aok02}, and using the \ion{Pb}{1} line data

835: of their Table~4.

836: Variations in assumed Pb isotopic fractions produced no change in the

837: derived elemental abundance.

838: As with other features in the blue spectral region, CH contamination

839: exists, but our syntheses suggested that it is only a small fraction

840: of the \ion{Pb}{1} strength.

841: 

842: 

843: 

844:                                                                                 

845:                                                                                 

846: 

847: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

848: \section{IMPLICATIONS OF THE \cs22964\ ABUNDANCE PATTERN\label{history}}

849: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

850: 

851: 

852: In the top panel of Figure~\ref{f9} we plot the abundances relative

853: to solar values for the 10 \ncap\ elements detected in the \cs22964\ 

854: syzygy spectrum, together with the \ncap\ abundances in the very 

855: \spro-rich BMP star CS~29497-030.  

856: These two stars have nearly the same overall [Fe/H] metallicity, 

857: as indicated by the horizontal lines in the panel.

858: The \ncap\ overabundance pattern of \cs22964\ clearly identifies it

859: as another member of C- and \spro-rich ``lead star'' family.

860: 

861: However, the relative \ncap\ abundance enhancements [X/Fe] of CS~29497-030 

862: are about 1~dex higher than they are in \cs22964.

863: This is emphasized by taking the difference between the two abundance sets, 

864: as is displayed in the lower panel of Figure~\ref{f9}.

865: These differences illustrate the trend toward weaker \spro\ enhancements 

866: of the heaviest elements in \cs22964\ compared to CS~29497-030.

867: 

868: Many studies have argued that these overall abundance anomalies, 

869: now seen in many low metallicity stars, must have originated from mass 

870: transfer from a (former) AGB companion to the stars observed today.

871: Whole classes of stars with large C and \spro\ abundances are dominated by

872: single-line spectroscopic binaries, including the high-metallicity 

873: ``Ba II stars'' (McClure, Fletcher, \& Nemec 1980\nocite{mcc80};

874: McWilliam 1988\nocite{mcw88}; McClure \& Woodsworth 1990\nocite{mcc90}), 

875: the low-metallicity red giant ``CH stars'' (McClure 1984\nocite{mcc84}),

876: the ``subgiant CH stars'' (McClure 1997)\nocite{mcc90}, and the

877: BMP stars (Preston \& Sneden 2000\nocite{pre00}; note that not all 

878: BMP stars share these abundance characteristics).

879: The case is especially strong for the subgiant CH and BMP stars, for these

880: are much too unevolved to have synthesized C and \spro\ elements in their

881: interiors and dredged these fusion products to the surfaces.

882: Our analysis has confirmed that primary and secondary \cs22964\ 

883: stars are on or near the main sequence. 

884: Therefore we suggest that a third, higher-mass star is now or 

885: once was a member of the \cs22964\ system.

886: During the AGB evolutionary phase of the third star, it transferred 

887: portions of its C- and \spro-rich envelope to the stars we now observe.

888: 

889: The very large Pb abundance of \cs22964\ strengthens the AGB 

890: nucleosynthesis argument.

891: Gallino \etal\ (1998\nocite{gal98}) and Travaglio \etal\ 

892: (2001\nocite{tra01}) predicted substantial Pb production

893: in \spro\ fusion zones of metal-poor AGB stars.

894: In metal-poor stars, the neutron-to-seed ratio is quite high, 

895: permitting the neutron-capture process to run through to Pb and Bi, 

896: the heaviest stable elements along the \spro\ path.  

897: Prior to this theoretical prediction, it was assumed that a $strong$ 

898: component of the \spro\ was required for the manufacture of half of the 

899: \iso{208}{Pb} in the solar system (Clayton \& Rassbach 1967\nocite{cr67}).

900: That the patterns of the abundances of the neutron-capture elements

901: in \cs22964\ and CS~29497-030 resemble each other so well is a

902: reflection of how easily low-metallicity AGB stars can synthesize

903: the heavy elements.

904: 

905: 

906: 

907: 

908: 

909: \subsection{Inferred Nucleosynthetic and Dilution 

910:                     History\label{nucleosynthesis}}

911: 

912: We explored the origin of the neutron-capture enhancements in \cs22964\ 

913: by comparing the derived photospheric abundances with predicted stellar 

914: yields from the \spro.

915: Employing FRANEC stellar evolutionary computations (see Straniero 

916: \etal\ 2003\nocite{straniero+03}; Straniero, Gallino, \& Cristallo 

917: 2006\nocite{sgc06}), we performed nucleosynthetic calculations following 

918: those of Gallino \etal\  (1998\nocite{gal98}; 2006a\nocite{gal06a}) 

919: and Bisterzo \etal\ (2006\nocite{bisterzo+06}).  

920: We then sought out good matches between the observed and predicted abundance 

921: pattern distributions with low-mass AGB progenitors of comparable 

922: metallicities and a range of initial masses. 

923: 

924: We employed AGB models adopting different \iso{13}{C}-pocket efficiencies 

925: and  initial masses to explore the nucleosynthetic history of the 

926: observed chemical compositions (Busso, Gallino \& Wasserburg 

927: 1999\nocite{busso+99}; Straniero \etal\ 2003\nocite{straniero+03}). 

928: Permitting some mixing of material between the \spro-rich 

929: contributions of the AGB donor, and the H-rich material of the 

930: atmosphere in which the contributions were deposited, we identified

931: those AGB model progenitors that would yield acceptable fits for the 

932: predicted yields of all \spro\ elements beyond Sr.

933: 

934: Ascertaining the best matches between the observed and predicted yields 

935: was performed in the following way.  

936: For a given initial AGB mass, we inspected the abundances of [Zr/Fe], 

937: [La/Fe], and [Pb/Fe] for various \iso{13}{C}-pocket efficiencies 

938: adopted in the calculations.  

939: The difference between our predicted [La/Fe] and that of the observed 

940: abundance ratio gave us a first guess to the dilution of material for a 

941: given \iso{13}{C}-pocket efficiency.

942: The dilution factor (dil) is defined as the logarithmic ratio

943: of the mass of the envelope of the observed star polluted with

944: AGB stellar winds and the AGB total mass transferred:

945: dil~$\equiv$~log$(M^{env}_{obs}/M^{AGB}_{transf})$.

946: The \iso{13}{C}-pocket efficiency is defined in terms of the 

947: \iso{13}{C} pocket mass involved in an AGB pulse that was adopted by 

948: Gallino \etal\ (1998)\nocite{gal98}.

949: Their \S2.2 states, ``The mass of the \iso{13}{C} pocket is 

950: 5.0$\times$10$^{-4}$~\Msun\ [here called ST], about 1/20 of the typical 

951: mass involved in a thermal pulse. 

952: It contains 2.8$\times$10$^{-6}$~\Msun\ of \iso{13}{C}.''

953: Thus our shorthand notation for \iso{13}{C}-pocket efficiency will

954: be ST/N, where ''N'' is the reduction factor employed in generating

955: a particular set of AGB abundance predictions.

956: Subtracting this dilution amount from the predicted [Pb/Fe], we then 

957: compared the result to the observed [Pb/Fe].  

958: The large range of \iso{13}{C}-pocket efficiencies was then narrowed down, 

959: keeping only those results that fit the abundances of both [La/Fe] 

960: and [Pb/Fe] within 0.2 dex.  

961: Similar iterations were performed employing the abundances of [Zr/Fe].

962: 

963: We repeated this process for a range of initial AGB mass choices: 1.3, 

964: 1.5, 2, 3, and 5\Msun.  

965: For AGB models of initial mass 3 and 5\Msun\ no good match was found 

966: because the light s-elements (ls~$\equiv$ Sr, Y, Zr) were predicted to

967: have too high abundances with respect to the heavy s-elements 

968: (hs~$\equiv$ Ba, La, Ce, Nd).

969: For AGB models of $M$~$\leq$~2\Msun\ a satisfactory solution was found for 

970: ls, hs, and Pb, provided a proper \iso{13}{C}-pocket efficiency and 

971: dilution factor were chosen.

972: In Figures~\ref{f10} and \ref{f11} we show 

973: the matches between predicted and observed abundances for the two

974: lowest-mass models, 1.3 and 1.5\Msun, respectively.

975: 

976: The abundances of the light elements Na and Mg further narrowed the

977: range of allowable AGB progenitor models.

978: Two independent channels are responsible for creating \iso{23}{Na}:

979: the \iso{22}{Ne}($p$,$\gamma$)\iso{23}{Na} reaction during H-shell burning,

980: and the neutron capture on \iso{22}{Ne} via the chain

981: \iso{22}{Ne}($n$,$\gamma$)\iso{23}{Ne}($\beta^-$$\nu$)\iso{23}{Na} in the

982: convective He flash (see Mowlavi 1999\nocite{mow99}; 

983: Gallino \etal\ 2006\nocite{gal06b}).

984: A large abundance of primary \iso{22}{Ne} derives from the primary 

985: \iso{12}{C} mixed with the envelope by previous third dredge up episodes, 

986: then converted to primary \iso{14}{N} by HCNO burning in the H-burning 

987: shell and followed by double $\alpha$ capture via the chain

988: \iso{14}{N}($\alpha$,$\gamma$)\iso{18}{F}($\beta^+$$\nu$)\iso{18}{O}($\alpha$,$\gamma$)\iso{22}{Ne}

989: during the early development of the convective thermal pulse. 

990: The next third dredge up episode mixes part of this primary 

991: \iso{22}{Ne} with the envelope. 

992: Finally, while the H-burning shell advances in mass, all the

993: \iso{22}{Ne} present in the H shell is converted to \iso{23}{Na} 

994: by proton capture, accumulating in the upper region of He intershell.

995: Note that in intermediate mass AGB stars suffering the so-called hot

996: bottom burning (HBB) in the deeper layers of their convective envelope,

997: efficient production of \iso{23}{Na} results from proton capture on 

998: \iso{22}{Ne} (Karakas \& Lattanzio 2003\nocite{kar03}). 

999: Furthermore, the marginal activation in the convective He flash of the 

1000: reactions \iso{22}{Ne}($\alpha$,$n$)\iso{25}{Mg} and

1001: \iso{22}{Ne}($\alpha$,$\gamma$)\iso{26}{Mg} would lead to enhanced Mg. 

1002: At the same time, most neutrons released by  

1003: \iso{22}{Ne}($\alpha$,$n$)\iso{25}{Mg} are captured by the very abundant 

1004: primary \iso{22}{Ne}, thus producing \iso{23}{Na} through the second 

1005: channel indicated above.

1006: 

1007: Comparison of the abundance data in Figures~\ref{f10} and 

1008: \ref{f11} shows that the observed [Na/Fe] argues

1009: for the exclusion of the 1.5\Msun\ AGB model, which over-predicts

1010: this abundance ratio by $\sim$0.5~dex.

1011: The same statement applies to the 2\Msun\ model, not illustrated here.

1012: The larger Na abundances produced in these models, relative to the 

1013: 1.3\Msun\ model, is related to the larger number of thermal pulses 

1014: (followed by third dredge up) that these stars experience.

1015: These arguments leave the 1.3\Msun\ AGB model as the only one 

1016: capable of providing a global good match to all observed elements.

1017: 

1018: In order to match these abundances, the predicted AGB yields 

1019: require an inferred dilution by H-rich material.  

1020: We do not know the distance between the AGB primary donor and the low 

1021: mass companion (actually, the binary system) that is now observed.  

1022: And, neither do we know precisely the mass loss rate of the AGB.  

1023: Dilution by about 1~dex of \spro-rich AGB material with the original 

1024: composition of the observed star, as we deduce on the basis of the above

1025: nucleosynthesis analysis, only fixes the ratio of the accreted matter with 

1026: the outer envelope of the observed star.  

1027: A more detailed discussion is taken up in \S\ref{dilution}.

1028: 

1029: A final remark concerns C, for which Figures~\ref{f10} and

1030: \ref{f11} indicate an $\sim$0.8~dex over-production

1031: compared to the observed [C/Fe].

1032: This mismatch can in principle be substantially reduced by the occurrence 

1033: of the so-called cool bottom process (CBB: see Nollett, Busso, \& Wasserburg 

1034: 2003\nocite{nol03}; Dom\'inguez et al. 2004a,b\nocite{dom04a,dom04b}). 

1035: This reduction would imply a substantial increase of predicted N abundance,

1036: from [N/Fe]~$\sim$ +0.5 shown in Figures~\ref{f10} and

1037: \ref{f11} to perhaps +1.5.

1038: We attempted without success to detect the strongest CN absorption at the

1039: 3883~\AA\ bandhead.

1040: Trial spectral syntheses of this wavelength region suggested that even

1041: [N/Fe]~$\sim$ +1.5 would not produce detectable CN.

1042: 

1043: 

1044: 

1045: 

1046: 

1047: \subsection{Mixing and Dilution in the Stellar Outer Layers\label{dilution}}

1048: 

1049: Very recently Stancliffe \etal\ (2007)\nocite{sta07}, extending the 

1050: earlier discussion of Theuns, Boffin, \& Jorissen (1996)\nocite{the96}, 

1051: report that extensive mixing due to thermohaline instability in a 

1052: metal-poor star on the main sequence may severely dilute material that 

1053: has been accreted from a companion.

1054: They computed the mixing time scale with the assumption that thermohaline 

1055: convection behaved as a simple diffusion process, and they concluded 

1056: that the new matter should rapidly be mixed down over about 90\% of the

1057: stellar mass.

1058:                                                                                 

1059: Treating thermohaline convection simply as diffusion does not take into 

1060: account the special nature of this process, which has extensively been 

1061: studied in oceanography (e.g. Veronis 1965\nocite{ver65}; 

1062: Kato 1966\nocite{kat66}; Turner 1973\nocite{tur73}; Turner \& Veronis 

1063: 2000\nocite{tur00}; Gargett \& Ruddick 2003\nocite{gar03}), and has

1064: also been considered in stars (Gough \& Toomre 1982\nocite{gou82}, 

1065: Vauclair 2004\nocite{vau04}). 

1066: Thermohaline convection occurs in the ocean when warm salty water comes 

1067: on top of cool fresh water. 

1068: In this case the stabilizing thermal gradient acts against the 

1069: destabilizing salt gradient. 

1070: If the stabilizing effect compensates the destabilizing one, the medium 

1071: should be stable, but it remains unstable due to double-diffusion: 

1072: when a blob begins to fall down, heat diffuses out of it more rapidly 

1073: than salt; then the blob goes on falling as the two effects no longer 

1074: compensate. 

1075: This creates the well known ``salt fingers'', thus is a very different 

1076: physical environment than ordinary convection.

1077:                                                                                 

1078: A similar situation occurs in stars when high-$\mu$ matter comes upon a

1079: lower-$\mu$ region, which is the case for hydrogen-poor accretion. 

1080: This has been studied in detail for the case of planetary accretion on

1081: solar-type stars (Vauclair 2004\nocite{vau04}). 

1082: As shown in laboratory experiments (Gargett \& Ruddick 2003\nocite{gar03}) 

1083: the fingers develop in a special layer with a depth related to the 

1084: velocity of the blobs and to the dissipation induced by hydrodynamical 

1085: instabilities at their edges. 

1086: This is a complicated process which may also be perturbed by other 

1087: competing hydrodynamical effects. 

1088: Depending on the situation, it is possible that the ``finger'' regime stops 

1089: before complete mixing of the high-$\mu$ material into the low-$\mu$ one. 

1090: In this case, we expect that the final amount of matter that remains 

1091: in the thin sub-photospheric convective zone depends only on the final 

1092: $\mu$-gradient, whatever the original amount accreted: more accreted 

1093: matter leads to more mixing so that the final result is the same. 

1094:                                                                                 

1095: Another very important point has to be mentioned. 

1096: At the epoch when accretion occurs on the dwarfs, these stars are 

1097: already about 3 to 4 Gyr old. 

1098: Gravitational settling of helium and heavy elements has already

1099: occurred and created a stabilizing $\mu$-gradient below the convective

1100: zone (Vauclair 1999\nocite{vau99}; Richard, Michaud, \& Richer 

1101: 2002\nocite{ric02}). 

1102: For example, as we illustrate in Figure~\ref{f12}, 

1103: in a 0.78\Msun\ star with [Fe/H]~=~$-$2.3, after 3.75~Gyr 

1104: the $\mu$-value is as low as 0.58 inside the convective zone while 

1105: it goes up to nearly 0.60 in deeper layers. 

1106: Thus the stabilizing $\Delta\mu / \mu$ is already of order 0.02.

1107: If the star accretes hydrogen-poor matter, thermohaline convection may begin

1108: but it remains confined by the $\mu$-barrier. The star can accrete as

1109: much high-$\mu$ matter as possible until this stabilizing $\mu$-gradient

1110: is flattened.

1111:                                                                                 

1112: In \S\ref{nucleosynthesis} we saw, from the observed abundances and 

1113: the AGB nucleosynthetic computations, that the dilution factor is of the 

1114: order of 1~dex for the AGB model of initial mass $\sim$1.3\Msun.

1115: Table~\ref{tab7} lists the predicted mass fraction of hydrogen ($X$),

1116: helium ($Y$), and heavier elements ($Z$) in the envelope of AGB 

1117: progenitor stars of 1.3 and 1.5\Msun\ for two cases.

1118: The first case describes the final abundance mix produced by the AGB at 

1119: the end of its nucleosynthetic lifetime (the last Third Dredge-Up).  

1120: The second case describes the mass average of the winds from the AGB 

1121: over its $s$-processing lifetime.  

1122: Also noted is the mean molecular weight of the material ($\mu$), assuming 

1123: fully ionized conditions.

1124: From Table~\ref{tab7}, the $\mu$-value in the wind lies in the range 

1125: 0.61 to 0.78. 

1126: From these computations, we find that, after dilution, the $\mu$-value 

1127: inside the convective zone should increase by 

1128: $\Delta\mu / \mu \simeq 0.1\%$ to $4\%$. 

1129: Clearly most of the inferred values for the dilution factor are compatible 

1130: with stellar physics. 

1131: For the lowest values it is possible that the accreted matter simply 

1132: compensates the effect of gravitational diffusion.

1133: For larger values, thermohaline convection can have larger effects but, 

1134: following Vauclair (2004\nocite{vau04}), if we accept an inverse 

1135: $\mu$-gradient of order $\Delta\mu / \mu$ = 0.02 below the convective 

1136: zone, such an accreted amount would still be possible.

1137: 

1138: 

1139: 

1140: 

1141: 

1142: \subsection{The Extraordinary Abundance of Lithium\label{lithium}}

1143: 

1144: Near main-sequence-turnoff CEMP stars display a variety of Li abundances.

1145: The majority have undetectable 6708~\AA\ \ion{Li}{1} lines, implying

1146: \eps{Li}~$<$~+1.

1147: A few have abundances similar to the Spite \& Spite 

1148: (1982)\nocite{spi82} ``plateau'': \eps{Li}~=~+2.10~$\pm$~0.09 

1149: (Bonifacio \etal\ 2007)\nocite{bon07}.

1150: These include our CEMP binary \cs22964, and at least CS~22898-027 

1151: (Thorburn \& Beers 1992\nocite{tho92}) and LP~706-7 

1152: (Norris, Ryan, \& Beers 1997\nocite{nor97}).

1153: A few more stars have Li abundances somewhat less than the Spite plateau

1154: value, \eg, \eps{Li}~$\simeq$~+1.7 in CS~31080-095 and CS~29528-041 

1155: (Sivarani \etal\ 2006\nocite{siv06}).

1156: This CEMP Li abundance variety stands in sharp contrast to the nearly 

1157: constant value exhibited by almost all normal metal-poor stars near the 

1158: main-sequence turnoff region.

1159: 

1160: Although \cs22964\ has a Spite plateau Li abundance, it is unlikely this

1161: is the same Li with which the system was born.

1162: Considering just the primary star, suppose that the C and \spro-rich 

1163: material that it received was Li-free. 

1164: If this material simply blanketed the surface of the primary,

1165: mixing only with the original Li-rich atmosphere and outer 

1166: envelope skin, then the resulting observed Li would be at least 

1167: less than the plateau value.

1168: Alternatively, if the transferred material induced more significant mixing 

1169: of the stellar envelope, the surface Li would be severely diluted

1170: below the limit of detectability in the primary's atmosphere.

1171: In order to produce the large observed Li abundance in the primary,

1172: freshly minted Li must have been transferred from the AGB donor star.

1173: 

1174: Li can be produced during the AGB phase both in massive AGB stars and

1175: in low-mass AGB stars via

1176: \iso{3}{He}($\alpha$,$\gamma$)\iso{7}{Be}($e^-,\nu_e$)\iso{7}{Li}, but hot 

1177: protons rapidly destroy it via \iso{7}{Li}($p$,$\alpha$)\iso{4}{He}.

1178: Cameron \& Fowler 1971\nocite{cam71}) proposed that in some circumstances

1179: the freshly minted \iso{7}{Be} can however be transported to cooler interior

1180: regions before decaying to \iso{7}{Li}.

1181: In massive AGB stars ( 4\Msun~$\lesssim$~$M$~$\lesssim$~7\Msun) nucleosynthesis 

1182: at the base of the convective envelope (``hot bottom burning''; Scalo, 

1183: Despain, \& Ulrich 1975\nocite{sdu75}) can produce abundances via the

1184: Cameron-Fowler mechanism as high as log~$\epsilon$(\iso{7}{Li})~$\sim$~4.5 

1185: (Sackmann \& Boothroyd 1992\nocite{sb92}).  

1186: However, the lifetimes of these massive AGB stars are far too short to be 

1187: considered as a likely source of Li in \cs22964.

1188: 

1189: In models of low-mass AGB stars (1\Msun~$\leq$~$M$~$\le$~3\Msun) of 

1190: [Fe/H]~=~$-$2.7, Iwamoto \etal\ (2004\nocite{iwamoto+04}) find that a 

1191: H-flash episode can occur subsequent the first fully developed He shell flash.  

1192: The convection produced by the He shell flash in this metal-poor star 

1193: model can reach the bottom of the over-lying H-layer and bring protons down 

1194: into the He intershell region, a region hot enough to induce a H flash.

1195: The high-temperature conditions of the H-flash can then produce 

1196: \iso{7}{Li} by the Cameron-Fowler mechanism.  

1197: Although the Iwamoto \etal\ 2.5\Msun\ model produced a weak H

1198: flash, the less massive models produced more energetic ones.

1199: In the lowest mass models, surface abundances of

1200: log~$\epsilon$(\iso{7}{Li})~$>$~3.2 were achieved.

1201: Note that there is a maximum metallicity close to [Fe/H]~= $-$2.5 beyond 

1202: which such an H-flash is not effective at producing \iso{7}{Li}.

1203: Another viable mechanism for producing a very high abundance of \iso{7}{Li}

1204: in low mass AGB stars of higher metallicities, as observed in the intrinsic 

1205: C(N) star Draco~461 ([Fe/H]~= $-$2.0~$\pm$~0.2), was advanced by 

1206: Dom\'inguez (2004a,b)\nocite{dom04a,dom04b} and is based on the operation

1207: of the CBP introduced in \S\ref{nucleosynthesis}.

1208:                                                                                 

1209: Although \spro\ calculations were not performed in the

1210: Iwamoto \etal\ study, we note that it would seem that the

1211: parameters of the best-fit model of \S\ref{nucleosynthesis} would

1212: produce sufficient Li to fit the observations of \cs22964.

1213: 

1214: 

1215: 

1216: 

1217: 

1218: 

1219: 

1220: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1221: \section{IMPLICATIONS FOR THE PUTATIVE AGB DONOR STAR\label{agb}}

1222: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1223: 

1224: The binary nature of \cs22964\ presents a unique opportunity to explore

1225: the nature of the AGB star presumably responsible for its abundance anomalies.

1226: The ingredients of the exploration are:  {\it (a)} the analysis of our

1227: radial velocities of \cs22964\ obtained over an interval of about 1100~days,

1228: {\it (b)} estimation of the orbit changes caused by accretion from an

1229: AGB companion, {\it (c)}  stability of the hierarchical triple system

1230: in its initial and final states (Donnison \& Mikulskis 1992\nocite{don92},

1231: hereafter DM92; Szebehely \& Zare 1977\nocite{sze77};

1232: Tokovinin 1997\nocite{tok97}; Kiseleva, Eggleton, \& Anolova 

1233: 1994\nocite{kis94}), and {\it (d)} application of an appropriate 

1234: initial-final mass relation for stars of intermediate mass (Weidemann 

1235: 2000\nocite{wei00}, and references therein).

1236: 

1237: 

1238:                                                                                 

1239:                                                                                 

1240: \subsection{Evidence for an AGB relic?\label{relic}}

1241:                                                                                 

1242: Velocity residuals calculated for our orbit solution show no trend

1243: with time over the nearly 1100-day interval of observation.

1244: This is illustrated by the plot of orbital velocity residuals versus

1245: Julian Date in Figure~\ref{f13}.

1246: Thus, we have yet to find evidence for a third component in the system.

1247: The Julian Date interval of our \cs22964\ data is modest compared to

1248: the longest orbital periods of the so-called Barium and CH giant stars

1249: (McClure \& Woodsworth 1990)\nocite{mcc90} and main-sequence/subgiant

1250: CH stars (McClure 1997)\nocite{mcc90} summarized in Table~\ref{tab6}.

1251: On the other hand the errors of our observations ($\sigma$~=~0.5~\kmsec)

1252: are small compared to typical velocity semi-amplitudes of these comparison

1253: C-rich stellar samples, so drift of the center-of-mass velocity by as much

1254: as 1 \kmsec\ due to the presence of a third star should be detectable.

1255: Accordingly, Figure~\ref{f13} encourages us to contemplate the

1256: possibility that \cs22964\ is not accompanied by a white dwarf relic of

1257: AGB evolution.

1258: Continuing observation will be required to test this notion.

1259:                                                                                 

1260: 

1261: 

1262: \subsection{Change in Orbit Dimensions Induced by Accretion\label{orbchange}}

1263:                                                                                 

1264: The consequences of mass accretion by a binary were first explored by

1265: Huang (1956)\nocite{hua56}, who calculated the change in semi-major axis

1266: of a binary embedded in a stationary interstellar cloud.

1267: Using McCrea's (1953)\nocite{mcc53} ``retarding force'' that followed

1268: from Bondi-Hoyle-Lyttleton accretion theory (Bondi \& Hoyle

1269: 1944)\nocite{bon44}, Huang found that mass accretion shrinks orbits, and

1270: he proposed that this process could produce close binaries.

1271: The problem of binary accretion has been revisited in recent years by

1272: numerical simulations of star formation in molecular clouds.

1273: Bate \& Bonnell (1997)\nocite{bat97} calculate the changes in binary

1274: separation due to accretion by an infalling circumstellar cloud.

1275: The change is very sensitive to the assumed specific angular momentum (SAM)

1276: carried by the accreted material.

1277: In the case of low SAM they recover Huang's result.

1278: For large SAM orbits the binaries expand.

1279: More recently Soker (2004)\nocite{sok04} has calculated the SAM of mass

1280: accreted by a binary system embedded in the wind of an AGB star,

1281: which is our case.

1282: For triple systems containing a distant AGB companion with approximately

1283: coplanar inner and outer orbital orbits (statistically most probable in the

1284: absence of a special orientation mechanism) he also recovers Huang's result.

1285: Finally, building on an earlier analysis by Davies \& Pringle

1286: (1980)\nocite{dav80}, Livio \etal\ (1986)\nocite{liv86} found that the

1287: rate of accretion of angular momentum from winds is significantly less

1288: than that deposited in the simple Bondi-Hoyle accretion model.

1289:                                                                                 

1290: Our gloss on all these analyses is that the most likely consequence of

1291: mass transfer to \cs22964\ from its AGB companion a long time ago

1292: was a decrease in separation, \ie, the initial separation was larger

1293: than its present value of $\sim$0.9~AU.

1294: Conservatively, we use 0.9 AU as a lower limit in the estimates that

1295: follow below.

1296:                                                                                 

1297: Empirical stability criteria for hierarchical triple systems are expressed

1298: as the ratio of the semi-major axis of the outer orbit to

1299: the semi-major axis of inner orbit (Heintz 1978\nocite{hei78};

1300: Szebehely \& Zare 1977\nocite{sze77}), and alternatively as the ratio

1301: of periods of the outer and inner orbits

1302: (Kiseleva \etal\ 1994\nocite{kis94}; Tokovinin 1997\nocite{tok97}).

1303: If these ratios exceed critical values, a triple system can be long-lived.

1304: Note, however, the cautionary remarks of Szebehely \& Zare  about the

1305: role of component masses in evaluation of stability for particular cases,

1306: and the relevance of these masses to the conclusions of DM92 about 

1307: the fate of our AGB relic.

1308:                                                                                 

1309: Let $a_{12}$ be the semi-major axis of the relative orbit of \cs22964\

1310: and $a_3$ be the relative semi-major axis that joins the center of mass

1311: of \cs22964\ to the 3$^{rd}$ (AGB-to-be) star.

1312: If we accept the Heintz (1978)\nocite{hei78} and Szebehely \& Zare

1313: (1977)\nocite{sze77} stability criterion $a_3/a_{12}$~$>$~8, then

1314: $a_3$~$>$~7.3~AU and the orbital period of the outer binary had to be

1315: $P$~$>$~4000~d.

1316: For Tokovinin's (1997)\nocite{tok97} more permissive criterion,

1317: $P_3/P_{12}$~$>$~10 the minimum period is $P$~$>$~2500~d and

1318: $a_3/a_{12}$~$>$~4.6.

1319: Under these conditions none of our candidate metal-poor AGB stars (see the

1320: discussion in \S\ref{nucleosynthesis}) with $M$~$<$~2\Msun\ fill the 

1321: Roche lobes (Marigo \& Girardi 2007)\nocite{mar07} of these minimum orbits, 

1322: so wind accretion must have been the mechanism of mass transfer.

1323: According to Theuns \etal\ (1996)\nocite{the96} a 3\Msun\ AGB star in a

1324: 3-AU circular orbit will transfer no more than $\sim$1\% of its wind mass

1325: to a 1.5 solar mass companion by wind accretion.

1326: Applying their result to our case, $M$(AGB)~$\leq$~1.5\Msun\ and AGB relic

1327: mass $\sim$0.5\Msun, the maximum mass transfer is about

1328: (1.5$-$0.5)$\times$0.015~$\simeq$~0.015\Msun.

1329:                                                                                 

1330: The remainder of the AGB wind mass is ejected from the system.

1331: We calculated the change in period and semi-major axis of the AGB orbit

1332: following Hilditch (2001)\nocite{hil01} for the case of a spherically

1333: symmetric wind.

1334: The orbit expands and the period lengthens according to

1335: $${\rm ln}(a_f/a_i) = {\rm ln}(1+u_i) - {\rm ln}(1+u_f)$$  and

1336: $${\rm ln}(P_f/P_i) = 2\times{\rm ln}(1+u_i) - 2\times{\rm ln}(1+u_f)$$

1337: in which initial and final values of $u$ are

1338: $u_i = M_{3i}/(M_1 + M_2)$ and $u_f = M_{3f}/(M_1 + M_2)$ .

1339: We have ignored small changes in $M_1$ and $M_2$.

1340: For $M_3$~=~1.3~\Msun\ and the Tokovinin (1997)\nocite{tok97} stability 

1341: condition ($a_3/a_{12}$~$>$~4.6), the minimum orbit expands from 4.2~AU 

1342: to 5.9~AU and the minimum period lengthens from 1900 days to 3700 days.

1343: For the more restrictive Heintz ($a_3/a_{12}$~$>$~8) stability condition

1344: the minimum orbit expands from 7.3~AU to 10.2~AU and the minimum period

1345: lengthens from 4300 days to 8400 days.

1346: The K-values for the center-of-mass of the \cs22964\ binary for the two

1347: cases are 4.3~\kmsec\ and 3.3~\kmsec, respectively, both readily

1348: detectable by conventional echelle spectroscopy.

1349:                                                                                 

1350:                                                                                 

1351:                                                                                 

1352:                                                                                 

1353:                                                                                 

1354: \subsection{Dynamical Stability of \cs22964\ as a Hierarchical Triple System

1355:              \label{dynamical}}

1356:                                                                                 

1357: The DM92 study investigated the stability of triple systems that

1358: consist of a close inner binary with masses ($M_1,M_2$) and semi-major

1359: axis $a1$ attended by a remote companion of mass $M_3$ at

1360: distance $q2$ from the center of mass of the inner binary.

1361: They use the parameter notation of Harrington's (1975)\nocite{har75}

1362: pioneering numerical simulations.

1363: DM92 investigate stability in three mass regimes, by integration of the

1364: equations of motion for various sets of masses, initial positions and

1365: velocities: \\

1366: \hspace*{1.5in} case {\it (i):}   $M_3$~$\leq$~minimum($M_1,M_2$) \\

1367: \hspace*{1.5in} case {\it (ii):}  $M_1$~$\geq$~$M_3$~$\geq$~$M_2$ \\

1368: \hspace*{1.5in} case {\it (iii):} $M_3$~$\geq$~maximum($M_1,M_2$) \\

1369: They pursued the integrations for at least 1000 orbits of the inner

1370: binary or until disruption, which ever happened first.

1371: Among the several results in \S5 of their paper one is of particular

1372: interest for us: in their words ``When $M_3$ was the least massive body in

1373: the system, the system invariably became unstable through the tendency

1374: of $M_3$ to escape from the system altogether.''

1375: Accepting the McClure (1984)\nocite{mcc84} paradigm, we suppose that

1376: \cs22964\ began its existence as a case {\it (iii)} hierarchical triple

1377: system in which a relatively low-mass AGB progenitor 

1378: (\S\ref{nucleosynthesis}) was orbited by a close binary with component 

1379: masses only slightly smaller than their present values, 

1380: $M_1$sin$^3i$~=~0.77\Msun\ and $M_2$sin$^3i$~=~0.68~\Msun.

1381: Evidently, the distance between $M_3$ and the close binary was

1382: large enough to insure stability during the lifetime of $M_3$.

1383: Curiously, although Kiseleva \etal\ (1994)\nocite{kis94} made numerical

1384: simulations apparently similar to those of DM92, they make no mention of the

1385: ejection of low-mass outer components that attracted our attention to DM92.

1386:                                                                                 

1387: We confine our attention to initial values of AGB mass $<$2.0\Msun,

1388: adopted in accordance with the nucleosynthetic calculations in 

1389: \S\ref{nucleosynthesis}.

1390: In the late stages of AGB evolution the system evolved from case

1391: {\it (iii)} to case {\it (i)} via mass loss due to the AGB superwind.

1392: For all acceptable models in \S\ref{nucleosynthesis} the mass of the WD 

1393: relic for an AGB initial mass model of 1.3\Msun\ is less than 

1394: 0.60\Msun\ (Weidemann 2000\nocite{wei00}).

1395: Therefore, according to DM92 $M_3$ will be ejected following this 

1396: ancient AGB evolution.

1397: Our extant radial velocity data suggest that the ejection already occurred.

1398: From the effect of the AGB relic on the $\gamma$ velocity of CS 22964-161 

1399: calculated above we believe that this conclusion can be tested by future 

1400: observations.

1401: 

1402: 

1403: 

1404: 

1405: 

1406: 

1407: 

1408: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1409: \section{CONCLUSIONS}

1410: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1411: 

1412: We have obtained high resolution, high S/N spectra over a three-year

1413: period for \cs22964, which was known to be metal-poor and C-rich from 

1414: previous low resolution spectroscopic studies.

1415: We discovered \cs22964\ to be a double-line spectroscopic binary, and

1416: have derived the binary orbital parameters ($P$~=~252~d, $e$~=0.66, 

1417: $M_{p}~sin^{3}~i$~= 0.77~\Msun, $M_{s}~sin^{3}~i$~= 0.68~\Msun).

1418: Both binary members lie near the metal-poor main-sequence turnoff

1419: (\teff$_{,p}$~= 6050~K, \logg$_p$~= 3.7, and \teff$_{,s}$~=5850~K,

1420: \logg$_s$~= 4.1).

1421: We derived similar overall metallicities for primary and secondary,

1422: [Fe/H]~= $-$2.4 and similar abundance ratios [X/Fe].

1423: In particular, both stars are similarly enriched in C and the \ncap\

1424: elements, with a clear \spro\ nucleosynthesis signature.

1425: The primary has a large Li content; the secondary's \ion{Li}{1} feature

1426: is undetectably weak, and does not usefully constrain its Li abundance.

1427: 

1428: The observed Li, C and \spro\ abundances of the \cs22964\ system must have 

1429: been produced by an AGB star whose relic, depending on its mass, may have 

1430: been ejected from the post-AGB hierarchical triple.  

1431: It seems AGB enrichment was produced by minor mass accretion in the AGB

1432: super-wind rather than Roche-lobe overflow. 

1433: As discussed in \S\ref{history}, and contrary to the thermohaline diffusion

1434: suggestion of Stancliffe \etal\ (2007\nocite{sta07}), such hydrogen-poor 

1435: accreted matter probably remained in the outer stellar layers, owing to the

1436: stabilizing $\mu$-gradient induced by helium gravitational settling.

1437: Strong thermohaline diffusion is difficult to reconcile with the observed

1438: Li:  about 9/10 of the stellar mass would be efficiently mixed on a time 

1439: scale of 1~Gyr, which would imply the complete destruction of \iso{7}{Li} 

1440: because of the very high temperature reached in the inner zones.

1441: It seems more likely that the moderate thermohaline mixing discussed 

1442: \S\ref{history}, related to the effect of gravitational settling in the 

1443: first 3 to 4~Gyr before mass accretion by the AGB donor, would save the 

1444: Li from destruction. 

1445: In the model of a low mass AGB of low metallicity calculated by 

1446: Dom\'inguez \etal\ (2004a,b\nocite{dom04a,dom04b}) with the inclusion of 

1447: ``cool bottom processing'' during the AGB phase, \eps{Li}~$>$ 3 is achieved

1448: in the envelope.

1449: A dilution factor of about 1~dex would bring the Li abundance into accord

1450: with the observed value.

1451: 

1452: 

1453: 

1454: 

1455: 

1456: 

1457: 

1458: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1459: \acknowledgments

1460: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1461: It is a pleasure to thank Guillermo Torres for sharing his TODCOR 

1462: and orbit solution source code with us, Oscar Straniero for helpful 

1463: discussions and continuous advice on AGB models of low-mass and very low

1464: metallicity,  and Noam Soker 

1465: for helpful discussions on wind accretion in binary stars.

1466: We are grateful to Olivier Richard for kindly providing one of his 

1467: population~II stars models, including consistently computed atomic diffusion.

1468: We thank Alan Boss, Janusz Kaluzny, Wojtek Krzeminski, and

1469: Kamil Zloczewski for obtaining photometric observations.

1470: Some of this work was accomplished while CS was a Visiting Scientist 

1471: at the Carnegie Observatories. The hospitality and support of the 

1472: observatory director is greatly appreciated.  

1473: This work has been supported by the U.S. National Science Foundation 

1474: through grants AST-0307495 and AST-0607708 to CS and AST-0507325 to IBT,

1475: and by by the Italian MIUR-PRIN06 Project ``Late phases of Stellar 

1476: Evolution: Nucleosynthesis in Supernovae, AGB stars, Planetary Nebulae''

1477: to RG.

1478: 

1479: 

1480: 

1481: 

1482: 

1483: 

1484: 

1485: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1486: %      HERE ARE REFERENCES

1487: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1488: \clearpage

1489: \begin{thebibliography}

1490: 

1491: \bibitem{aok07}

1492: Aoki, W., Beers, T.~C., Christlieb, N., Norris, J.~E., Ryan, S.~E.,

1493: \& Tsangarides, S. \ 2007, \apj, 655, 492

1494: 

1495: \bibitem{arl99}

1496: Arlandini, C., Kaeppeler, F., Wisshak, K., Gallino, R., Lugaro, M., 

1497: Busso, M., \& Straniero, O. \ 1999, \apj, 525, 886

1498: 

1499: \bibitem{aok02}

1500: Aoki, W., Ryan, S.~G., Norris, J.~E., Beers, T.~C., Ando, H., \& Tsangarides, S.\ 2002, \apj, 580, 1149

1501: 

1502: \bibitem{ale02}

1503: Allende Prieto, C., Lambert, D.~L., \& Asplund, M.\ 2002, \apjl, 573, L137

1504: 

1505: \bibitem{bat97}

1506: Bate, M.~R., \& Bonnell, I.~A. \ 1997, \mnras, 285, 33

1507: 

1508: \bibitem{bps92}

1509: Beers, T.~C., Preston, G.~W., \& Shectman, S.~A.\ 1992, \aj, 103, 1987 (BPS92)

1510: 

1511: \bibitem{ber03}

1512: Bernstein, R., Shectman, S.~A., Gunnels, S.~M., Mochnacki, S.,   

1513: \& Athey, A.~E. 2003, Instrument Design and Performance for

1514: Optical/Infrared Ground-based Telescopes. Edited by Iye, Masanori;

1515: Moorwood, Alan F. M. Proceedings of the SPIE, 4841, 1694

1516: 

1517: \bibitem[Bisterzo et al. (2006)]{bisterzo+06} 

1518: Bisterzo, S., Gallino, R., Straniero, O., Ivans, I.\ I., K\"appeler, F., 

1519: \& Aoki, W. 2006, Mem. Soc. Astron. It., 77, 985

1520: 

1521: \bibitem{bon44}

1522: Bondi, H., \& Hoyle, F. \ 1944, \mnras, 104, 273

1523:                                                                                 

1524: \bibitem{bon07}

1525: Bonifacio, P., \etal\ \ 2007, \aap, 462, 851

1526: 

1527: \bibitem[Busso et al. (2001)]{busso+01} 

1528: Busso, M., Gallino, M., Lambert, D.\ L., Travaglio, C., \& Smith, V.\ 

1529: 2001, \apj, 557, 802

1530: 

1531: \bibitem[Busso et al. (1999)]{busso+99} 

1532: Busso, M., Gallino, R., \& Wasserburg, G.\ J.\ 1999, \araa, 37, 239

1533: 

1534: \bibitem{bus04}

1535: Busso, M., Straniero, O., Gallino, R., \& Abia, C. \ 2004, 

1536: Origin and Evolution of the Elements, Carnegie Observatories Centennial 

1537: Symposia. eds A. McWilliam \& M. Rauch (Cambridge: Cambridge University

1538: Press), 67

1539: 

1540: \bibitem[Cameron \& Fowler 1971]{cam71} 

1541: Cameron, A.\ G.\ W., \& Fowler, W.\ A. 1971, \apj, 164 111

1542: 

1543: \bibitem{cas97}

1544: Castelli, F., Gratton, R.~G., \& Kurucz, R.~L.\ 1997, \aap, 318, 841

1545: 

1546: \bibitem{cr67}

1547: Clayton, D.\ D., \& Rassbach, M.\ E.\ 1967, \apj, 148, 69

1548: 

1549: \bibitem{coh04}

1550: Cohen, J.~G., \etal\ 2004, \apj, 612, 1107

1551: 

1552: \bibitem{coh05}

1553: Cohen, J.~G., \etal\ 2005, \apj, 633, L109

1554: 

1555: \bibitem{coh06} 

1556: Cohen, J.~G., \etal\ 2006, \aj, 132, 137

1557: 

1558: \bibitem{dav80}

1559: Davies, R.~E., \& Pringle, J.~E.\ 1980, \mnras, 191, 599

1560: 

1561: \bibitem{dom04a}

1562: Dom\'inguez, I., Abia, C., \& Straniero, O.\ 2004a, Memorie della 

1563: Societa Astronomica Italiana, 75, 601

1564: 

1565: \bibitem{dom04b}

1566: Dom\'inguez, I., Abia, C., Straniero, O., Cristallo, S., \& Pavlenko, Y.~V. 

1567: 2004b, \aap, 422, 1045

1568: 

1569: \bibitem{don92}

1570: Donnison, J.~R., \& Mikulskis, D.~F.\ 1992, \mnras, 254, 21 (DM92)

1571: 

1572: \bibitem{duq91} 

1573: Duquennoy, A., \& Mayor, M.\ 1991, \aap, 248, 485

1574: 

1575: \bibitem{gal98}

1576: Gallino, R., Arlandini, C., Busso, M., Lugaro, M., Travaglio, C., 

1577: Straniero, O., Chieffi, A., \& Limongi, M.\ 1998, \apj, 497, 388

1578: 

1579: \bibitem{gal06b} 

1580: Gallino, R., Bisterzo, S., Husti, L., K{\"a}ppeler, F., Cristallo, S., 

1581: \& Straniero, O.\ 2006b, International Symposium on Nuclear Astrophysics 

1582: - Nuclei in the Cosmos, Proc. of Science, 100.1

1583: 

1584: \bibitem{gal06a} 

1585: Gallino, R., Bisterzo, S., Staniero, O., Ivans, I.\ I.\ \& 

1586: K\"appeler, F., 2006a, Mem. Soc. Astron. It., 77, 985

1587: 

1588: \bibitem{gar03}

1589: Gargett, A, Ruddick, B. 2003, Progress in Oceanography, 56, 381

1590: 

1591: \bibitem{gou82}

1592: Gough, D.O., \& Toomre, J. 1982, J. Fluid Mechanics, 125, 75

1593: 

1594: \bibitem{har75}

1595: Harrington, R.~S. \ 1975, \aj, 80, 1081

1596: 

1597: \bibitem{hei78}

1598: Heintz, W. \ 1978, Double Stars (Dordreecht: Reidel), p. 66

1599: 

1600: \bibitem{hel59}

1601: Helfer, H.~L., Wallerstein, G., \& Greenstein, J.~L.\ 1959, \apj, 129, 700

1602: 

1603: %\bibitem{her04}

1604: %Herwig, F. \ 2004, \apjs, 155, 651

1605: 

1606: \bibitem{hil01}

1607: Hilditch, R. W. 2001, An Introduction to Close Binary Stars 

1608: (Cambridge: Cambridge U. Press), p 164

1609: 

1610: \bibitem{hua56}

1611: Huang, S.~S. \ 1956, \aj, 61, 49

1612: 

1613: %\bibitem{ibg06}

1614: %Ivans, I., Bisterzo, S., \& Gallino, R. 2006, Mem. Soc. Astron. It., 77, 979

1615: 

1616: \bibitem{iva05} 

1617: Ivans, I.~I., Sneden, C., Gallino, R., Cowan, J.~J., \& Preston, G.~W.\ 

1618: 2005, \apjl, 627, L145

1619: 

1620: \bibitem[Iwamoto et al. 2004]{iwamoto+04}

1621: Iwamoto, N., Kajino, T., Mathews, G.\ J., Fujimoto, M.\ Y., \& Aoki, W. 

1622: 2004, \apj, 602, 377

1623:                                                                                 

1624: %\bibitem{iwa03}

1625: %Iwamoto, N., Mathews, G.~J., Fujimoto, M.~Y., Kajino, T., \& Aoki, W. \ 

1626: %2003, Nuclear Physics A, 718, 193

1627: 

1628: \bibitem{jor00} 

1629: Jorissen, A., \& van Eck, S.\ 2000, The Carbon Star Phenomenon, 

1630: IAU Symp., 177, 259

1631: 

1632: \bibitem{kar03}

1633: Karakas, A. I., \& Lattanzio, J. C. 2003, Publ. Astron. Soc. Australia, 

1634: 20, 279

1635: 

1636: \bibitem{kat66}

1637: Kato, S.\ 1966, \pasj, 18, 374

1638: 

1639: \bibitem{kel03} 

1640: Kelson D.~D. 2003, \pasp, 115,  688

1641: 

1642: \bibitem{kel06} 

1643: Kelson D.~D. 2006, \aj, submitted

1644: 

1645: \bibitem{kis94}

1646: Kiseleva, L. G., Eggleton, P. P., \& Anolsova, J. P. 1994, \mnras, 267, 161

1647: 

1648: \bibitem{kur98}

1649: Kurucz, R.~L.\ 1998, IAU Symp.~189: Fundamental Stellar Properties, 189, 217

1650: 

1651: \bibitem{lan92}

1652: Landolt, A. 1992, \aj, 104, 340

1653: 

1654: \bibitem{lat98}

1655: Latham, D.~W., Stefanik, R.~P., Mazeh, T., Goldberg, D., Torres, G., 

1656: \& Carney, B.~W.\ 1998, Cool Stars, Stellar Systems, and the Sun, 

1657: ASP Conf. Ser., 154, 2129

1658: 

1659: \bibitem{liv86}

1660: Livio, M., Soker, N., de Kool, M., \& Savonije, G.~J.\ 1986, \mnras, 218, 593

1661: 

1662: \bibitem{luc06}

1663: Lucatello, S., Beers, T.~C., Christlieb, N., Barklem, P.~S., Rossi, S.,

1664: Marsteller, B., Sivarani, T., \& Lee, Y.~S. \ 2006, \apj, 652, L37

1665: 

1666: \bibitem{luc91} 

1667: Luck, R.~E., \& Bond, H.~E.\ 1991, \apjs, 77, 515

1668: 

1669: \bibitem{mar07}

1670: Marigo, P., \& Girardi, L. 2007, \aap, in press, astro-ph/0703139

1671: 

1672: \bibitem{mcc84}

1673: McClure, R.~D.\ 1984, \apjl, 280, L31

1674: 

1675: \bibitem{mcc97}

1676: McClure, R.~D.\ 1997, \pasp, 109, 536

1677: 

1678: \bibitem{mcc80} 

1679: McClure, R.~D., Fletcher, J.~M., \& Nemec, J.~M.\ 1980, \apjl, 238, L35

1680: 

1681: \bibitem{mcc90} 

1682: McClure, R.~D., \& Woodsworth, A.~W.\ 1990, \apj, 352, 709

1683: 

1684: \bibitem{mcc53}

1685: McCrea, W.~H.\ 1953, \mnras, 113, 162

1686: 

1687: \bibitem{mcw88}

1688: McWilliam, A.\ 1988, Ph.D.~Thesis, Univ. Texas at Austin

1689: 

1690: \bibitem{mow99}

1691: Mowlavi, N. 1999, \aap, 350, 73

1692: 

1693: \bibitem{nol03}

1694: Nollett, K. M., Busso, M., \& Wasserburg, G. J. 2003, \apj, 582, 1036

1695: 

1696: \bibitem{nor00}

1697: Norris, J.~E., Beers, T.~C., \& Ryan, S.~G. \ 2000, \apj, 540, 456

1698: 

1699: \bibitem{nor97}

1700: Norris, J.~E., Ryan, S.~G., \& Beers, T.~C.\ 1997, \apj, 488, 350

1701: 

1702: \bibitem{nor00b}

1703: North, P., Jorissen, A., \& Mayor, M.\ 2000, The Carbon Star Phenomenon, 

1704: IAU Symp., 177, 269

1705: 

1706: \bibitem{pet01}

1707: Peterson, R.~C., Dorman, B., \& Rood, R.~T.\ 2001, \apj, 559, 372

1708: 

1709: %\bibitem{pin01}

1710: %Pinsonneault, M.~H., DePoy, D.~L., \& Coffee, M. \ 2001, \apj, 556, L59

1711: 

1712: \bibitem{pre94}

1713: Preston, G.~W. 1994, \aj, 108, 2267

1714: 

1715: \bibitem{pre91}

1716: Preston, G~W., Shectman, S.~A., \& Beers, T.~C. 1991, \apjs, 76, 1001

1717: 

1718: %%%%%%%%%%%%%%%%GENERAL BMP SURVEY %%%%%%%%%%%%%%%%%%%%%%%%%%%%

1719: \bibitem{pre00}

1720: Preston, G.~W., \& Sneden, C.\ 2000, \aj, 120, 1014

1721: 

1722: %%%%%%%%%%%%%%%%%%% RHB SURVEY PAPER %%%%%%%%%%%%%%%%%%%%%%%%%%

1723: \bibitem{pre06b}

1724: Preston, G.~W., Sneden, C., Thompson, I.~B., Shectman, S.~A., \& 

1725: Burley, G.~S.\ 2006a, \aj, 132, 85

1726: 

1727: %%%%%%%%%%%%%%%%%%% TY GRUIS PAPER %%%%%%%%%%%%%%%%%%%%%%%%%%%%

1728: \bibitem{pre06a} 

1729: Preston, G.~W., Thompson, I.~B., Sneden, C., Stachowski, G., \& 

1730: Shectman, S.~A.\ 2006b, \aj, 132, 1714

1731: 

1732: \bibitem{rey02}

1733: Reyniers, M., Van Winckel, H., Bi{\'e}mont, E., \& Quinet, P.\ 2002, 

1734: \aap, 395, L35

1735: 

1736: \bibitem{ric02}

1737: Richard, O., Michaud, G., \& Richer, J. \ 2002, \apj, 580, 1100

1738: 

1739: \bibitem{ros05}

1740: Rossi, S., Beers, T.~C., Sneden, C., Sevastyanenko, T., Rhee, J., \& 

1741: Marsteller, B.\ 2005, \aj, 130, 2804

1742: 

1743: \bibitem[Sackmann & Boothroyd 1992]{sb92}

1744: Sackmann, I.-J., \& Boothroyd, A.\ I. 1992, \apj, 392, L71

1745: 

1746: \bibitem[Scalo et al. 1975]{sdu75}

1747: Scalo, J.\ M., Despain, K.\ H., \& Ulrich, R.\ K. 1974, \apj, 187, 555

1748: 

1749: \bibitem{sch98}

1750: Schlegel, D.~J., Finkbeiner, D.~P., \& Davis, M. 1998, \apj, 500, 525

1751: 

1752: \bibitem{siv06}

1753: Sivarani, T., \etal\ 2006, \aap, 459, 125

1754: 

1755: \bibitem{sne73}

1756: Sneden, C.\ 1973, \apj, 184, 839

1757: 

1758: %%%%%%%%%%%%%%%%%%%%%CS22892052 R_RICH STAR PAPER%%%%%%%%%%%%%%

1759: \bibitem{sne03a} 

1760: Sneden, C., \etal\ 2003a, \apj, 591, 936

1761: 

1762: %%%%%%%%%%%%%%%%%%%%%BMP BINARY DISCOVERY PAPER%%%%%%%%%%%%%%%%

1763: \bibitem{sne03b} 

1764: Sneden, C., Preston, G.~W., \& Cowan, J.~J.\ 2003b, \apj, 592, 504

1765: 

1766: \bibitem{sok04}

1767: Soker, N. \ 2004, \mnras, 350, 1366

1768: 

1769: \bibitem{spi82}

1770: Spite, F., \& Spite, M. \ 1982, \aap, 115, 357

1771: 

1772: \bibitem{sta07}

1773: Stancliffe, R.~J., Glebbeek, E., Izzard, R.~G., \& Pols, O.~R. 2007, 

1774: \aap, 464, L57

1775: 

1776: \bibitem[Straniero et al.(2003)]{straniero+03} 

1777: Straniero, O., Dom\'inguez, I., Cristallo, S., \& Gallino, R.\ 2003, 

1778: Pub. Astr. Soc. Aus., 20, 389

1779: 

1780: \bibitem[Straniero et al.(2006)]{sgc06} Straniero, O., Gallino, R., 

1781: \& Cristallo, S.\ 2006, Nucl.\ Phys.\ A, 777, 311

1782: 

1783: \bibitem{sze77}

1784: Szebehely, V., \& Zare, K. \ 1977, \aap, 58, 145

1785: 

1786: \bibitem{the96}

1787: Theuns, T., Boffin, H.~M.~J., \& Jorissen, A.\ 1996, \mnras, 280, 1264

1788: 

1789: \bibitem{tho92}

1790: Thorburn, J.~A., \& Beers, T.~C.\ 1992, \baas, 24, 1278

1791: 

1792: \bibitem{tok97}

1793: Tokovinin, A. 1997, \aaps, 124, 75

1794: 

1795: \bibitem{tra01} 

1796: Travaglio, C., Gallino, R., Busso, M., \& Gratton, R.\ 2001, \apj, 549, 346

1797: 

1798: \bibitem{tur73}

1799: Turner, J.S. 1973, Buoyancy effects in fluids, Cambridge: Cambridge 

1800: University Press

1801: 

1802: \bibitem{tur00}

1803: Turner, J.~S., \& Veronis, G.\ 2000, Journal of Fluid Mechanics, 405, 269

1804: 

1805: \bibitem{van06}

1806: VandenBerg, D.~A., Bergbusch, P.~A., \& Dowler, P.~D. 2006, 

1807: \apjs, 162, 375

1808: 

1809: \bibitem{vau99}

1810: Vauclair, S.\ 1999, in LiBeB Cosmic Rays, and Related X- and Gamma-Rays, 

1811: Astr. Soc. Pac. Conf. Ser., 171, 48

1812: 

1813: \bibitem{vau04} 

1814: Vauclair, S.\ 2004, \apj, 605, 874

1815: 

1816: \bibitem{ver65}

1817: Veronis, G.J. 1965, Marine Res., 21, 1

1818: 

1819: \bibitem{wal98} 

1820: Wallerstein, G., \& Knapp, G.~R.\ 1998, \araa, 36, 369

1821: 

1822: \bibitem{wei00} 

1823: Weidemann, V.\ 2000, \aap, 363, 647

1824: 

1825: \bibitem{zuc94}

1826: Zucker, S., \& Mazeh, T. 1994, \apj, 420, 806

1827: \end{thebibliography}

1828: 

1829: 

1830: 

1831: 

1832: 

1833: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1834: %      HERE ARE FIGURES

1835: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1836: \clearpage

1837: \begin{figure}

1838: \epsscale{1.0}

1839: \plotone{f1.eps}

1840: \caption{

1841: Top panel:  radial velocity observations of \cs22964 primary and 

1842: secondary stars, and best-fit orbital solutions from these velocities.

1843: Circles represent data taken with MIKE on the Magellan Clay telescope 

1844: and squares represent data taken with the echelle spectrograph on the 

1845: du~Pont telescope. 

1846: Filled symbols represent data for the primary and open symbols are for 

1847: the secondary. 

1848: The lines are fits to the Magellan data only.

1849: Bottom panel: differences of the observed velocities and the orbital

1850: solutions.

1851: \label{f1}}

1852: \end{figure}

1853: 

1854: \clearpage

1855: \begin{figure}

1856: \epsscale{0.9}

1857: \plotone{f2.eps}

1858: \caption{

1859: Observed composite absorption spectra of primary and secondary stars.  

1860: The bottom spectrum is the spectrum at syzygy ($\Delta V_R$~= 0~\kmsec); 

1861: this is the mean of three individual observations.

1862: The middle spectrum is the mean of three observations obtained

1863: at phases with $\Delta V_R$~= 60~\kmsec.

1864: The top spectrum represents the original syzygy spectrum after application

1865: of a velocity shift of 60~\kmsec\ and dilution by addition of a continuum 

1866: flux that yields secondary line depths approximately matching those

1867: seen in the middle spectrum.

1868: Dotted lines point to secondary positions of a few lines that are very

1869: strong in the primary spectrum.

1870: The flux scale of the syzygy spectrum is correct, and that other two spectra

1871: have been vertically shifted in the figure for display purposes.

1872: The rest wavelength scale is that of the primary star.

1873: \label{f2}}

1874: \end{figure}

1875:                                                                                 

1876: \clearpage

1877: \begin{figure}

1878: \epsscale{0.9}

1879: \plotone{f3.eps}

1880: \caption{

1881: Comparison of observed equivalent widths of the secondary star inferred 

1882: from the syzygy spectrum and those measured directly from the six spectra 

1883: with large velocity splits between primary and secondary stars.

1884: Measurements for each spectral feature are connected by vertical lines 

1885: and labeled with the feature wavelength.

1886: Medians of the $EW$s are displayed as $\times$ symbols.

1887: The dashed slanting line indicates equality of $EW$ values.

1888: \label{f3}}

1889: \end{figure}

1890:                                                                                 

1891: \clearpage

1892: \begin{figure}

1893: \epsscale{0.9}

1894: \plotone{f4.eps}

1895: \caption{

1896: Iron abundances plotted as a function of wavelength for primary (top panel) 

1897: and secondary (bottom panel) stars.  

1898: The symbols are defined in the figure legend.

1899: Dashed horizontal lines are drawn to indicate the mean \ion{Fe}{1}

1900: abundances of primary and secondary stars, \eps{Fe}~=~5.13 and 5.11,

1901: respectively.

1902: \label{f4}}

1903: \end{figure}

1904: 

1905: \clearpage

1906: \begin{figure}

1907: \epsscale{0.9}

1908: \plotone{f5.eps}

1909: \caption{

1910: Evolutionary tracks (VandenBerg \etal\ 2006) and CEMP data plotted in the 

1911: log~\teff~$-$~\logg\ plane near the metal-poor main-sequence turnoff.

1912: The chosen tracks correspond to three different inclinations of the binary;

1913: heavy and light lines are for the primary and secondary, respectively.

1914: A full track covers an age range of 4 to 14~Gyr.  

1915: The figure legend identifies the masses of the tracks for three different 

1916: \cs22964\ orbital $sin\ i$ choices, using the notation 

1917: ($sin\ i$: $M_{p}$, $M_{s}$).

1918: Filled circles represent the \cs22964\ primary and secondary 

1919: (log~\teff, \logg) values derived from the spectroscopic analysis 

1920: (\S\ref{params}).

1921: The plotted error bars for the primary are also from the spectroscopy.

1922: The square shows the secondary parameters from the dimensional analysis 

1923: described in \S\ref{evol},

1924: The line about this point corresponds to $\Delta$\teff~= $\pm$400~K. 

1925: The triangles represent CEMP stars from the literature; see \S\ref{evol} 

1926: for details.

1927: The triangles for three pairs of stars with identical temperatures and 

1928: gravities have been shifted apart from each other for display purposes. 

1929: \label{f5}}

1930: \end{figure}

1931: 

1932: \clearpage

1933: \begin{figure}

1934: \epsscale{0.9}

1935: \plotone{f6.eps}

1936: \caption{

1937: Observed and synthetic spectra of the \ion{Sr}{2} 4077.71~\AA\

1938: line in observation 13817 of \cs22964\ ($\Delta V_R$~$\simeq$~58.09~\kmsec).

1939: The top spectrum shows the full relative flux scale, and the bottom spectrum

1940: covers just the relative flux region above 0.75.

1941: The wavelength scale is at rest velocity for the primary star.

1942: The observed spectrum is depicted with open circles.

1943: The blue line represents a synthetic spectrum computed without any

1944: contribution from Sr.

1945: The black line shows the best overall fit to primary and secondary

1946: \ion{Sr}{2} features: \eps{Sr}~= 0.98 and 0.78, respectively.

1947: The red and green lines indicate syntheses for each star that are 0.5~dex 

1948: smaller and larger, respectively, than the best fits.

1949: \label{f6}}

1950: \end{figure}

1951:                                                                                 

1952: \clearpage

1953: \begin{figure}

1954: \epsscale{0.9}

1955: \plotone{f7.eps}

1956: \caption{

1957: Observed and synthetic spectra of the CH G-bandhead in \cs22964.

1958: The observation is 13817, for which $\Delta V_R$~= 58.09~\kmsec.

1959: Open circles represent the observed spectrum.

1960: In the top panel only the contribution of the primary star to this

1961: blend is considered. 

1962: The abundances of the synthetic spectra, in order of increasing CH

1963: strength, are \eps{C}$_p$~= $-\infty$, 7.24, 7.64, and 8.04 (the blue,

1964: red, black, and green lines, respectively).

1965: In the bottom panel, the abundance of the primary is fixed at 

1966: \eps{C}$_p$~= 7.64, and that of the secondary is, again in order of

1967: increasing CH strength, \eps{C}$_s$~= $-\infty$, 6.44, 7.44, and 8.44.

1968: An arrow indicates the wavelength offset between primary and secondary

1969: stars of this observation.

1970: \label{f7}}

1971: \end{figure}

1972:                                                                                 

1973: \clearpage

1974: \begin{figure}

1975: \epsscale{0.9}

1976: \plotone{f8.eps}

1977: \caption{

1978: Selected spectral features of \ncap\ species in the \cs22964\ 

1979: syzygy spectrum.

1980: \label{f8}}

1981: \end{figure}

1982: 

1983: \clearpage

1984: \begin{figure}

1985: \epsscale{0.9}

1986: \plotone{f9.eps}

1987: \caption{

1988: In the top panel, \ncap\ abundances relative to solar values [X/H] 

1989: from the \cs22964\ combined spectrum are plotted along with those of 

1990: the BMP very \spro-rich star CS~29497-030 (Ivans \etal\ 2005).

1991: Horizontal lines are drawn to indicate the general Fe-metallicity levels 

1992: of the two stars.

1993: In the bottom panel the abundance differences between these two stars

1994: are shown.

1995: \label{f9}}

1996: \end{figure}

1997: 

1998: \clearpage

1999: \begin{figure}

2000: \epsscale{0.75}

2001: \plotone{f10.eps}

2002: \caption{Comparison of the [X/Fe] abundances in \cs22964\ with predictions 

2003: from \spro\ calculations of a 1.3\Msun\ AGB star model.  

2004: The solid red line corresponds to the best match between the observed 

2005: and predicted abundance pattern.  

2006: The dotted blue and dashed green lines indicate the difference that 

2007: adopting models of $\pm$0.05\Msun\ would make (effectively, changing 

2008: the number of thermal pulses from 4 to 6, identified in the figure

2009: legend as n4, n5, n6).

2010: For the specific choice of the \iso{13}{C}-pocket (ST/12), and for the

2011: definition of the dilution factor ($dil$), see the text.

2012: \label{f10}}

2013: \end{figure}

2014: 

2015: \clearpage

2016: \begin{figure}

2017: \epsscale{0.75}

2018: \plotone{f11.eps}

2019: \caption{Comparison of the [X/Fe] abundances in \cs22964\ with predictions 

2020: from \spro\ calculations of a 1.5\Msun\ AGB star model.  

2021: The solid red line corresponds to the best match between the observed 

2022: and predicted abundance pattern.  

2023: The dotted magenta, long-dashed green, and short-dashed blue lines show 

2024: the difference made to the pattern by adopting different amounts 

2025: of dilution.

2026: For discussion of the the various choices of the \iso{13}{C}-pocket 

2027: efficiency (ST/N), and the dilution factor (dil), see the text.

2028: \label{f11}}

2029: \end{figure}

2030: 

2031: \clearpage

2032: \begin{figure}

2033: \epsscale{0.9}

2034: \plotone{f12.eps}

2035: \caption{Mean molecular weight ($\mu$) and temperature (T) profiles, as 

2036: functions of the fractional mass (mass above the considered layer), in a 

2037: 0.78\Msun\ star with [Fe/H]~=~--2.3, at an age of 3.75~Gyr. 

2038: Here $\mu$ is the real molecular weight, including partial ionization. 

2039: The fractional mass of the convective zone is 0.004.  

2040: The helium depletion due to gravitational settling, of about

2041: 20\% in this case, already leads to an important stabilizing

2042: $\mu$-gradient below the convective zone. 

2043: This model is one of those computed by Richard \etal\ (2002), including 

2044: pure atomic diffusion. 

2045: The full model was made available to us by Richard (private communication).

2046: \label{f12}}

2047: \end{figure}

2048:                                                                                 

2049: \clearpage

2050: \begin{figure}

2051: \epsscale{1.0}

2052: \plotone{f13.eps}

2053: \caption{

2054: Velocity residuals to the fitted orbit are plotted against the date of 

2055: observation for the Magellan data. 

2056: Filled circles represent the primary and open circles represent the

2057: secondary. 

2058: These residuals show no trend with time.

2059: \label{f13}}

2060: \end{figure}

2061:                                                                                 

2062:                                       

2063: 

2064: 

2065: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2066: %      HERE ARE TABLES

2067: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2068: 

2069: \clearpage

2070: \input{tab1}

2071: 

2072: \clearpage

2073: \input{tab2}

2074: 

2075: \clearpage

2076: \input{tab3}

2077: 

2078: \clearpage

2079: \input{tab4}

2080: 

2081: \clearpage

2082: \input{tab5}

2083: 

2084: \clearpage

2085: \input{tab6}

2086: 

2087: \clearpage

2088: \input{tab7}

2089: 

2090: 

2091: 

2092: \end{document}

2093: 

2094: 

2095: 

2096: