1: \documentclass[12pt,preprint]{aastex}
2: %\documentclass[manuscript]{aastex}
3:
4: %\documentclass[preprint2]{aastex}
5:
6:
7:
8: \newcommand{\vdag}{(v)^\dagger}
9: \newcommand{\myemail}{skywalker@galaxy.far.far.away}
10:
11:
12: \slugcomment{Draft version November 13, 2007.}
13:
14:
15: \shorttitle{Fluorine in RCBs}
16: \shortauthors{Pandey et al.}
17:
18:
19: \begin{document}
20:
21:
22: \title{Fluorine in R Coronae Borealis Stars}
23:
24: \author{Gajendra Pandey}
25: \affil{Indian Institute of Astrophysics;
26: Bangalore, 560034 India}
27: \email{pandey@iiap.res.in}
28:
29: \author{David L.\ Lambert}
30: \affil{The W.J. McDonald Observatory, University of Texas at Austin; Austin,
31: TX 78712-1083}
32: \email{dll@astro.as.utexas.edu}
33:
34: \and
35:
36: \author{N. Kameswara Rao}
37: \affil{Indian Institute of Astrophysics;
38: Bangalore, 560034 India}
39: \email{nkrao@iiap.res.in}
40:
41: \begin{abstract}
42: Neutral fluorine (F\,{\sc i}) lines are identified in the optical
43: spectra of several R\,Coronae Borealis stars (RCBs) at maximum light.
44: These lines provide the first measurement of the fluorine
45: abundance in these stars. Fluorine is enriched in some RCBs by factors of 800 to 8000
46: relative to its likely initial abundance.
47: The overabundances of fluorine are evidence for the synthesis
48: of fluorine. These results are discussed in the light of the scenario that
49: RCBs are formed by accretion of an He white dwarf by a C-O white dwarf.
50: Sakurai's object (V4334\,Sgr), a final He-shell flash product, shows no
51: detectable F\,{\sc i} lines.
52: \end{abstract}
53:
54:
55: %\clearpage
56: \keywords{stars: abundances --
57: stars: chemically peculiar -- stars: evolution}
58:
59:
60: \section{Introduction}
61:
62: R Coronae Borealis (RCB) stars comprise a sequence of hydrogen-deficient
63: supergiants with effective temperatures from about $T_{\rm eff} = 3500$ K,
64: as represented by Z\,UMi and DY\,Per, to about 19,500 K, as represented
65: by DY\,Cen.
66: The characteristic of H-deficiency is shared by the H-deficient cool
67: carbon (HdC) stars at low temperatures and by the extreme helium
68: (EHe) stars at high temperatures. A common assumption is that the
69: sequence HdC - RCB - EHe in the ($T_{\rm eff},\log g$) plane
70: reflects a close evolutionary connection. In this sequence, the
71: RCBs are distinguished from HdC and EHe stars by their second principal defining
72: characteristic: their propensity to fade, in visual light, unpredictably as a
73: cloud of carbon dust obscures the star. This propensity is not universally shared:
74: XX\,Cam has yet to be observed below maximum light. In addition, DY\,Cen might be
75: considered as an EHe star known to experience RCB-like fadings.
76:
77: If HdC, RCB, and EHe stars share a common heritage, the expectation
78: is that their atmospheric compositions should show some common
79: features \citep{pan2004a,rao2005}. It is through the compositions that one hopes to test
80: theoretical ideas about the origins of these extremely rare
81: stars; just five HdC, about 40 RCB \citep{zanie2005}, and 21 EHe
82: stars are known in the Galaxy. Currently, two
83: scenarios remain in contention to account for these
84: H-poor high luminosity stars. In the first, a final He-shell flash in a
85: post-AGB star on the white dwarf cooling track creates a H-poor luminous
86: star. This is dubbed the `final flash' (FF) scenario. In the second, the H-poor
87: star is formed from a merger of a He white dwarf with a C-O white dwarf.
88: In a close binary system, accretion of the He white dwarf by the C-O white dwarf
89: may lead to a H-poor supergiant with the C-O white dwarf as its core. This is
90: called the `double degenerate' (DD) scenario.
91:
92: Products of both the FF and DD scenarios may be presumed to exist.
93: A determination of which scenario provided which star rests in large
94: part on the observed chemical composition of a star's atmosphere and theoretical
95: predictions about the FF and DD products. Evidence from
96: elemental abundances, especially the H, C, N, and O abundances, suggests that
97: the RCB and EHe stars evolved via the
98: DD rather than the FF route \citep{asp00,pan2001,saio02,pan2006}.
99: Convincing, essentially incontrovertible, evidence
100: that the DD scenario led to the HdCs and some cool RCBs was
101: presented by \citet{clay2007} with their discovery that
102: the $^{18}$O was very abundant in their atmospheres. This usually
103: rare isotope of oxygen was attributed to nucleosynthesis occurring
104: during and following accretion of the He-rich material onto the C-O white dwarf.
105:
106: Determination of the oxygen isotopic ratios demands
107: a cool star with the CO vibration-rotation bands in its spectrum.
108: The majority of RCBs and all of the EHes are too hot for CO to contribute
109: to their spectra \citep{tenen2005}.\footnote{A possibility exists that
110: circumstellar CO may be detectable for the RCBs. Cool gas containing
111: C$_2$ molecules has been seen for V854\,Cen, R\,CrB, and V\,CrA when the
112: stars were in decline \citep{raolamb2000,raolambsh2006,raolamb2007}.}
113: An alternative tracer of nucleosynthesis during a merger may
114: be provided by the fluorine abundance. Considerable enrichment
115: of EHe stars with F was discovered by \citet{pana2006} from detection and
116: analysis of about a dozen F\,{\sc i} lines in optical spectra of
117: cool EHe stars. Clayton et al.'s (2007) calculations suggest that F synthesis is
118: possible in the DD scenario. In this paper, we report on a search for
119: F\,{\sc i} lines in spectra of RCBs and discuss the F abundances in
120: light of the results for EHes and the expectations for the DD and FF
121: scenarios.
122:
123: \section{Observations}
124:
125: High-resolution optical spectra of RCBs at maximum light obtained
126: at the W.J. McDonald Observatory and the Vainu Bappu Observatory
127: were examined for the presence of F\,{\sc i} lines. A key requirement
128: of the spectra for inclusion in the study was that the absorption
129: line profiles be symmetric; an asymmetry and a doubling of the lines
130: is attributable to a pulsation of the atmosphere that may induce a
131: shock wave. An additional requirement was that the S/N ratio be
132: adequate for detection of weak lines.
133:
134: Spectra from the
135: McDonald Observatory were obtained with the 2.7 meter Harlan J.
136: Smith telescope and the coud\'{e} cross-dispersed
137: echelle spectrometer \citep{tull95} at a resolving power
138: $R = \lambda/\Delta\lambda = 60,000$ with a
139: bandpass covering much of the 3800 -- 10000 \AA\ interval.
140:
141: At the Vainu Bappu Observatory (VBO), spectra were acquired with the Vainu Bappu
142: Telescope with a fiber-fed cross-dispersed echelle spectrometer \citep{rao04,rao05b}.
143: These spectra were acquired with two different grating
144: settings to cover the wavelength regions of the crucial F\,{\sc i} lines and
145: are at a resolving power $R \simeq 30,000$
146: with a wavelength coverage of 4860\AA\ to 7900\AA.
147:
148: All spectra were reduced in standard fashion using the Image Reduction
149: and Analysis Facility (IRAF) software.\footnote{The IRAF software is
150: distributed by the National Optical Astronomy Observatories under contract
151: with the National Science Foundation}
152: The reduction procedures usually included the removal of
153: the telluric absorption lines through an interactive process (task
154: {\it telluric}) using a spectrum of an early-type rapidly rotating
155: star.
156:
157: Suitable McDonald spectra of 13 RCBs were available for analysis. In
158: several cases, spectra of the same star from different epochs were
159: available. The stars are listed in Table 1.
160: %{\bf This can be the table in
161: %which we list stellar parameters and the F abundances.}
162: At VBO, spectra were obtained of UW\,Cen, a star inaccessible from
163: the McDonald Observatory, and also of XX\,Cam, R\,CrB, RY\,Sgr, and VZ\,Sgr.
164: The UW\,Cen spectrum was complemented by a CTIO spectrum from 1992 (see
165: Asplund et al. 2000).
166:
167: \begin{deluxetable}{llcccccr}
168: \tabletypesize{\scriptsize}
169: \tablewidth{0pt}
170: \tablecolumns{7}
171: %\tablewidth{0pc}
172: \setcounter{table} {0}
173: \tablecaption{The analyzed RCBs, their stellar parameters,
174: and fluorine abundances from individual F\,{\sc i} lines. Sakurai's object, a final He-shell flash product, is also listed.}
175: \tablehead{
176: \colhead{Star} & \colhead{($T_{\rm eff}$, $\log g$, $\xi$)} & \multicolumn{6}{c}{log $\epsilon(\rm F)$}\\
177: \cline{3-8} \\
178: \colhead{} & \colhead{} & \colhead{7398.68\AA} & \colhead{7754.69\AA} & \colhead{6902.47\AA} & \colhead{7425.64\AA} & \colhead{6834.26\AA} & \colhead{Mean}}
179: %\colhead{} & \colhead{} & \colhead{}}
180: \startdata
181: V3795\,Sgr$^a$ & (8000, 1.0, 10.0) & \nodata & 6.60 & 6.65 & 6.70 & 6.70 & 6.66$\pm$0.05(4)\\
182: &&&&&&&\\
183: UW\,Cen & (7500, 1.0, 12.0) & 7.20 & \nodata & 7.00 & 7.10 & 7.20 & 7.1$\pm$0.1(4)\\
184: &&&&&&&\\
185: RY\,Sgr & (7250, 0.75, 6.0) & \nodata & \nodata & 6.80 & \nodata & 7.10 & 6.95$\pm$0.2(2) \\
186: &&&&&&&\\
187: XX\,Cam & (7250, 0.75, 9.0) & \nodata & \nodata & $<$5.6 & \nodata & $<$5.6 & $<$5.6 \\
188: &&&&&&&\\
189: UV\,Cas & (7250, 0.5, 7.0) & \nodata & \nodata & 6.20 & \nodata & \nodata & 6.2(1)\\
190: &&&&&&&\\
191: UX\,Ant & (7000, 0.5, 5.0) & \nodata & \nodata & $<$6.2 & \nodata & $<$6.2 & $<$6.2\\
192: &&&&&&&\\
193: VZ\,Sgr$^a$ & (7000, 0.5, 8.0) & \nodata & \nodata & 6.30 & \nodata & 6.50 & 6.4$\pm$0.1(2)\\
194: &&&&&&&\\
195: R\,CrB & (6750, 0.5, 7.0) & \nodata & \nodata & 6.85 & \nodata & 7.00 & 6.9$\pm$0.1(2)\\
196: &&&&&&&\\
197: V2552\,Oph & (6750, 0.5, 7.0) & \nodata & \nodata & 6.60 & \nodata & \nodata & 6.6(1)\\
198: &&&&&&&\\
199: V854\,Cen$^a$ & (6750, 0.0, 6.0) & \nodata & \nodata & \nodata & \nodata & $<$5.7 & $<$5.7\\
200: &&&&&&&\\
201: SU\,Tau & (6500, 0.5, 7.0) & \nodata & \nodata & 6.90 & \nodata & 7.00 & 6.95$\pm$0.1(2)\\
202: &&&&&&&\\
203: V\,CrA$^a$ & (6500, 0.5, 7.0) & \nodata & \nodata & 6.5: & \nodata & \nodata & 6.5:(1)\\
204: &&&&&&&\\
205: V482\,Cyg & (6500, 0.5, 4.0) & \nodata & \nodata & \nodata & \nodata & 6.6: & 6.6:(1)\\
206: &&&&&&&\\
207: GU\,Sgr & (6250, 0.5, 7.0) & \nodata & \nodata & \nodata & \nodata & 7.2: & 7.2:(1)\\
208: &&&&&&&\\
209: FH\,Sct & (6250, 0.25, 6.0) & \nodata & \nodata & \nodata & \nodata & 7.2: & 7.2:(1)\\
210: &&&&&&&\\
211: Sakurai's object & (7500, 0.0, 8.0) & \nodata & \nodata & \nodata & \nodata & $<$5.4 & $<$5.4\\
212: &&&&&&&\\
213: \enddata
214: \tablenotetext{a}{Minority RCBs}
215: %\tablenotetext{b}{See text}
216: \end{deluxetable}
217:
218: %{\bf I'd suggest that we introduce Sakurai's object at an appropriate
219: %point in the discussion and not mention it in this section.}
220:
221: \clearpage
222:
223: \begin{figure}
224: \epsscale{1.00}
225: \plotone{f1.eps}
226: \caption{The spectral region from 6900 -- 6928 \AA\ is shown for seven RCBs with the
227: hottest star at the top and the coolest star at the bottom. The F\,{\sc i}
228: line at 6902.47 \AA\ labelled below each spectrum is obviously
229: present in V3795\,Sgr and UW\,Cen (the hottest two stars), and absent from
230: XX\,Cam but a likely contributor to the spectra of the other stars. A
231: weaker F\,{\sc i} line at 6909.81 \AA\ is also labelled but is either
232: lost in the wing of a stronger line and/or masked by a telluric O$_2$
233: line. Some other atomic lines are identified that apply to all spectra. \label{fig1}}
234: \end{figure}
235:
236: \begin{figure}
237: \epsscale{1.00}
238: \plotone{f2.eps}
239: \caption{The spectral region from 6830 -- 6860 \AA\ is shown for seven RCBs with the
240: hottest star at the top and the coolest star at the bottom. The F\,{\sc i}
241: line at 6834.26 \AA\ labelled below each spectrum is obviously
242: present in V3795\,Sgr and UW\,Cen (the hottest two stars), and absent from
243: XX\,Cam but a likely contributor to the spectra of the other stars. A stronger
244: F\,{\sc i} line at 6856.03 \AA\ is also labeled.
245: This line is clearly present for V3795\,Sgr and UW\,Cen. In cooler
246: stars, the F\,{\sc i} contribution is blended severely with a Si\,{\sc i}
247: and a Fe\,{\sc i} line.
248: Some other atomic lines are identified that apply to all spectra. \label{fig2}}
249: \end{figure}
250:
251: \clearpage
252:
253: \section{F\,{\sc i} Lines}
254:
255: In the atmosphere of an RCB star, fluorine is present as neutral
256: atoms. The leading lines of F\,{\sc i} in optical spectra come from excited levels
257: with lower excitation potentials of 12.7 eV or higher. The
258: $2p^4(^3P)3s$ -- $2p^4(^3P)3p$ transition array provides 27 lines from
259: six multiplets with lower excitation potentials 12.70 -- 13.02 eV.
260: The adopted $gf$-values come from \citet{musie99}
261: who measured relative
262: transition probabilities for lines in the $3s-3p$ (and $3p-3d$) array
263: using emission lines from a wall-stabilized arc and used Opacity Project
264: calculations of selected strong lines to place results on an absolute
265: scale. The $gf$-values are estimated to be accurate to 15\% to 20\%.
266:
267: The predicted equivalent widths ($W_\lambda$) of F\,{\sc i} in RCB stars are
268: sensitive to the effective temperature ($T_{\rm eff}$).
269: This sensitivity is highlighted with a simple calculation of the
270: predicted $W_\lambda$ for the
271: F\,{\sc i} 6902.47\AA\ line, the fourth strongest line in the transition
272: array, with a $gf$-value of 1.51 for models with
273: a number density ratio C/He =1\%. The $W_\lambda$ varies from 75 m\AA\ at
274: $(T_{\rm eff},\log g$) = (8000 K, 1.0) to 35 m\AA\ at
275: (7000 K, 0.5) to 12 m\AA\ at (6250 K, 0.5) where the
276: ($T_{\rm eff},\log g$) combinations are representative of V3795\,Sgr,
277: VZ\,Sgr, and GU\,Sgr, respectively, and the F abundance chosen is representative
278: of values reported for EHe stars \citep{pana2006}.
279: These predictions suggest that opportunities for detection of
280: F\,{\sc i} lines will be best in the hottest RCB stars.
281: The high $T_{\rm eff}$ sensitivity, perhaps suggested by the excitation
282: potential of 13 eV, is moderated by the excitation potential (about
283: 9 -- 11 eV) of the high levels of the carbon atom providing the continuous
284: opacity by photoionization.
285: Note that the sensitivity of predicted $W_\lambda$ to $\log g$
286: at a given $T_{\rm eff}$ is not appreciable.
287: The calculations also show that the weakest lines in the transition array (say,
288: $gf$ $< 0.3$) will not be detectable in cooler RCBs.
289:
290: %\clearpage
291: %\begin{center}\small{Table 2} \\
292: %F\,{\sc i} lines from $2s^22p^4$($^3$P)$3s$ -- $2s^22p^4$($^3$P)$3p$ transition array detected in the
293: %spectra of the analyzed stars. The F\,{\sc i} lines used in abundance
294: %determination for all the analyzed RCBs are shown in bold\\
295: %\begin{tabular}{ccccl}
296: %\hline
297: %\hline
298: %Multiplet & $\lambda$ & $\chi$ & log $gf$ & Contributors \\
299: %(No.) & (\AA) & (eV) & & \\
300: %\hline
301: %$3s$ $^4$P -- $3p$ $^4$P$^{\circ}$ & 7398.68&12.70& $+$0.24 & F\,{\sc i}, in the gaps of the McDonald spectra\\
302: %%(1) & 7482.72&12.73& $-$0.66 & F\,{\sc i}, C\,{\sc i} $\lambda$7483.445\AA\ (red wing), Fe\,{\sc ii} $\lambda$7482.777\AA\ (weak)\\
303: %% &7514.92&12.75& $-$0.96 & F\,{\sc i} \\
304: %(1) & 7331.96&12.70& $-$0.11 & F\,{\sc i}, N\,{\sc i} $\lambda$7332.055\AA, Fe\,{\sc ii} $\lambda$7332.115\AA\ (weak)\\
305: % & {\bf 7425.64}&12.73& $-$0.19 & F\,{\sc i}, Fe\,{\sc ii} $\lambda$7425.095\AA \\
306: % &7552.23&12.73& $-$0.34 & F\,{\sc i} \\
307: % &7573.39&12.75& $-$0.34 & F\,{\sc i} \\
308: % & & & & \\
309: %$3s$ $^4$P -- $3p$ $^4$D$^{\circ}$ & 6856.03&12.70& $+$0.44 & F\,{\sc i}, C\,{\sc i} $\lambda$6855.553\AA, Ti\,{\sc ii} $\lambda$6855.732\AA, Fe\,{\sc i} \\
310: %(2) &{\bf 6902.47}&12.73& $+$0.18 & F\,{\sc i}, CN (weak), unknown blend (red wing) \\
311: % & 6909.81&12.75& $-$0.23 & F\,{\sc i}, severely blended \\
312: %% & 6773.98&12.70& $-$0.40 & F\,{\sc i}, Fe\,{\sc ii} $\lambda$6774.146\AA, Fe\,{\sc ii} $\lambda$6774.473\AA\ (red wing) \\
313: % & {\bf 6834.26}&12.73& $-$0.21 & F\,{\sc i}, La\,{\sc ii} $\lambda$6834.099\AA\ (very weak), CN(7,3) (weak), N\,{\sc i} (weak)\\
314: % &6870.22&12.73& $-$0.25 & F\,{\sc i}, severely blended by telluric lines \\
315: %% &6795.52&12.73& $-$1.09 & F\,{\sc i} \\
316: % & & & & \\
317: %$3s$ $^4$P -- $3p$ $^4$S$^{\circ}$ & 6239.65&12.70& $-$0.17 & F\,{\sc i}, Si\,{\sc ii} \\
318: %(3) & 6348.51&12.73& $-$0.25 & F\,{\sc i}, Si\,{\sc ii} $\lambda$6347.09\AA\ (red wing) \\
319: %% & 6413.65&12.75& $-$0.47 & F\,{\sc i}, C\,{\sc i} \\
320: % & & & & \\
321: %$3s$ $^2$P -- $3p$ $^2$D$^{\circ}$ & {\bf 7754.69}&12.98& $+$0.24 & F\,{\sc i}, Sm\,{\sc ii} $\lambda$7755.072\AA\ (red wing), CN (weak) \\
322: %(4) & 7800.21&13.03& $-$0.04 & F\,{\sc i}, Si\,{\sc i} $\lambda$7799.996\AA, Fe\,{\sc ii} $\lambda$7801.235\AA\ (red wing) \\
323: % & & & & \\
324: %$3s$ $^2$P -- $3p$ $^2$S$^{\circ}$ & 7311.02&12.98& $-$0.14 & F\,{\sc i}, Fe\,{\sc ii} \\
325: %(5) & & & & \\
326: % & & & & \\
327: %$3s$ $^2$P -- $3p$ $^2$P$^{\circ}$ & 7037.46&12.98& $+$0.10 & F\,{\sc i}, unknown blend (blue wing) most likely C\,{\sc i}\\
328: %(6) &{\bf 7127.89}&13.03& $-$0.12 & F\,{\sc i}, blends? \\
329: %% &6966.35&12.98& $-$1.01 & F\,{\sc i} \\
330: %% &7202.36&13.03& $-$0.33 & F\,{\sc i}, C\,{\sc i} $\lambda$7202.27\AA \\
331: %\hline
332: %\end{tabular}
333: %\end{center}
334: %%$^a$ This paper for the model ($T_{\rm eff}$,$\log g, \xi$) $\equiv$
335: %% (18300, 2.2, 16.0)\\
336: %%$^b$ Jeffery (1993) for the mo (18300, 2.2, 16.0)\\
337: %\clearpage
338:
339: The following F\,{\sc i} lines were identified as the principal or
340: leading contributor to a stellar line: 7398.68, 7754.69, 6902.47, 7425.65,
341: and 6834.26\AA.
342: Figures 1 and 2 illustrate the windows that represent a couple of these F\,{\sc i}
343: lines, where the RCBs are ordered from top to bottom in order
344: of decreasing effective temperature.
345: All spectra are aligned to the rest wavelengths of well known lines that fall in
346: the wavelength regions.
347:
348: For all F\,{\sc i} lines, a search was made for blending lines.
349: Databases examined included the Kurucz database\footnote{http://kurucz.harvard.edu},
350: tables of spectra of H, C, N, and O \citep{moor93} and
351: Si \citep{moore65,moore67}, Vienna Atomic Line Database (VALD), and
352: the new Fe\,{\sc i} multiplet table \citep{nave94}.
353: The following remarks describe the quintet of useful lines in order to decreasing
354: $gf$-value:
355:
356:
357:
358: {\bf 7398.68\AA} ($\log gf = +0.24$). This line is in
359: an inter-order gap for most McDonald spectra.
360: The exceptions are McDonald spectra of XX\,Cam, UV\,Cas,
361: and SU\,Tau and the VBT spectrum of UW\,Cen. The line is blended with
362: the N\,{\sc i} 7398.63\AA\ line in XX\,Cam,
363: UV\,Cas, and SU\,Tau but the F\,{\sc i} line's contribution is obtainable
364: for UW\,Cen.
365: The CN (4,1) 7398.473\AA\ may be contributing to the blue wing of this line for
366: cooler RCBs.
367: % there is a suggestion of this CN line in SU\,Tau's spectrum.
368:
369: {\bf 7754.69\AA} ($\log gf = +0.24$). The major contributors to the red wing of this line are
370: Sm\,{\sc ii} 7755.072\AA, Mg\,{\sc i} 7755.236\AA, and Fe\,{\sc ii} 7755.563\AA.
371: CN(5,2) 7753.767\AA\ may be contributing to the blue wing of this line for cooler RCBs.
372: The F\,{\sc i} line's contribution for V3795\,Sgr is measureable.
373:
374: {\bf 6902.47\AA} ($\log gf = +0.18$). This region is crossed by telluric absorption (O$_2$) lines.
375: From the available database of RCB spectra at maximum light, only those spectra were
376: used in which the 6902.47\AA\ stellar feature avoids these
377: telluric absorption lines (see Figure 1). Also, further care has been taken to remove these
378: telluric absorptions by ratioing these spectra with a spectrum of an
379: early-type rapidly rotating star.
380: This line is present in V3795\,Sgr and UW\,Cen with a depth
381: comparable to that of the 7754.69\AA\ line of similar $gf$-value.
382: There is obviously a line present to the red that dominates in XX\,Cam and other
383: Fe-rich RCBs. The suspected lines are Ni\,{\sc i} at 6902.791\AA\ (VALD), and
384: Fe\,{\sc i} at 6902.956\AA\ (Kurucz's database).
385: Through spectrum synthesis (see Section 4 for details), the line provides an estimate of or an upper limit
386: to the F abundance.
387:
388: For the spectrum synthesis of the region,
389: we take Fe\,{\sc i} to be the significant blend
390: in the RCB stars' spectra instead of the Ni\,{\sc i} line.
391: XX\,Cam's spectra clearly shows this Fe\,{\sc i} 6902.956\AA\ line with little
392: or no contribution from the F\,{\sc i} 6902.47\AA\ line (Figure 1).
393: We estimate the $gf$-value
394: of this Fe\,{\sc i} 6902.956\AA\ line by inversion using the XX\,Cam's spectrum; the
395: Fe abundance is taken from \citet{asp00}. Figure 3 shows the spectrum synthesis
396: of this F\,{\sc i} line region for several RCBs.
397:
398: {\bf 7425.64\AA} ($\log gf = -0.19$) This line is detected in V3795\,Sgr and UW\,Cen but not
399: in other RCBs for which the upper limit to the F abundance is consistent with
400: estimates from other lines.
401:
402: {\bf 6834.26\AA} ($\log gf = -0.21$). This line contributes in V3795\,Sgr and UW\,Cen, and shows its
403: presence in other RCBs (Figures 2 and 4).
404: This line is used for
405: abundance determination after accounting for all the known blends.
406: Blends include La\,{\sc ii} 6834.009\AA,
407: CN(7,3) 6834.236\AA\ and
408: N\,{\sc i} 6834.074\AA.
409: Kurucz's database also lists a La\,{\sc ii} 6837.904\AA\ line which is about
410: the same strength as the
411: La\,{\sc ii} 6834.009\AA\ line. However, the
412: La\,{\sc ii} 6837.904\AA\ line is not detected even in the $s$-process enriched
413: RCBs.
414: For the N\,{\sc i} line,
415: we take the $gf$-value from
416: Kurucz's database. XX\,Cam's spectrum essentially has little or no contribution
417: from the F\,{\sc i} 6834.26\AA\ line (Figure 2). The predicted N\,{\sc i} profile at 6834.074\AA\
418: reproduces the observed profile for XX\,Cam; N abundance is from \citet{asp00}.
419: Also, the equivalent widths of the common N\,{\sc i} lines measured in our RCB stars'
420: spectra are in agreement with the \citet{asp00} measurements within the
421: expected uncertainties. We adopt the N abundances from \citet{asp00} for
422: the analyzed RCBs except for SU\,Tau's N abundance that is from \citet{pan2004b}.
423: Figure 4 shows the spectrum synthesis of this F\,{\sc i} line region for several RCBs.
424:
425: Where they contribute, CN Red System lines are included with their $gf$-values
426: taken from Kurucz's database.
427: Note that CN lines listed in Kurucz's database with
428: predicted or extrapolated energies are not included in the
429: spectrum synthesis. The CN strengths are determined from the
430: strongest CN lines observed across the wavelength region of the
431: analyzed RCBs. CN contributes to these F\,{\sc i} lines only in the cooler members of
432: the RCBs. Note that the CN (2,0) and CN (3,0) bandheads are clearly present in the
433: cooler members of the RCBs; CN bands are typically strong in visible spectra
434: of cool RCB stars (e.g., Clayton 1996 and references therein).
435:
436: A search was made for all other lines in the transition array but with the
437: exception of clear indications of a F\,{\sc i} contribution for
438: V3795\,Sgr and UW\,Cen, the F\,{\sc i} line is masked by blending lines in the
439: other RCBs.
440:
441: \section{Abundance analysis}
442:
443: The fluorine abundances were determined from the
444: F\,{\sc i} lines by adopting the procedure described
445: in \citet{pan2006}.
446: Uppsala H-deficient model atmospheres \citep{asp97a} are used with the Uppsala spectral synthesis LTE code BSYNRUN
447: to compute either the equivalent width of a F\,{\sc i} line or
448: a synthetic spectrum for a selected spectral window. The synthetic spectrum was
449: convolved with a Gaussian profile to account for the combined effects of the stellar
450: macroturbulence, rotational broadening, and the instrumental profile, before matching
451: with the observed spectrum. The fact that several F\,{\sc i} lines are blended with other lines
452: requires that the fluorine abundance be extracted by spectrum synthesis.
453: Although F\,{\sc i} 6902.47\AA\ and F\,{\sc i} 6834.26\AA\ lines are blended
454: in most of the analyzed RCBs, we could
455: account for the contributing blending lines and extract the F\,{\sc i}
456: component just from these lines.
457: For the two hottest RCBs, fluorine was also measured from the F\,{\sc i} 7425.64\AA\
458: and F\,{\sc i} 7398.68\AA\ lines in UW\,Cen, and in V3795\,Sgr from
459: the F\,{\sc i} 7425.64\AA\ and F\,{\sc i} 7754.69\AA\ lines.
460: The number of usable F\,{\sc i} lines declines from four for V3795\,Sgr and
461: UW\,Cen, the hottest RCBs, to two or zero in the coolest RCBs. This
462: trend reflects the decline of the predicted equivalent width for constant F
463: abundance with decreasing $T_{\rm eff}$. As a reminder, we note that
464: 11 to 17 F\,{\sc i} lines were usable for the EHes with $T_{\rm eff}$'s
465: from 8750 K to 12750 K.
466:
467:
468:
469: %\clearpage
470: %\begin{center}\small{Table 3} \\
471: %Fluorine abundances from individual F\,{\sc i} lines for
472: %the analyzed RCBs \\
473: %\begin{tabular}{lccccc}
474: %\hline
475: %\hline
476: % & & & log $\epsilon$(F) & & \\
477: %\cline{2-6}
478: % & 6902.47\AA\ & 6834.26 \AA\ & 7754.69\AA\ & 7127.89\AA\ & 7425.64\AA\ \\
479: %\hline
480: %V3795\,Sgr & 6.65 & 6.70 & 6.60 & $<$6.94 & $<$6.80\\
481: %&&&&&\\
482: %UW\,Cen & 7.00 & 7.20 & 7.45 & \nodata & \nodata \\
483: %&&\\
484: %RY\,Sgr & 6.80 & 7.10 & 7.00 & \nodata & $<$7.50 \\
485: %&&\\
486: %XX\,Cam & $<$5.50 & $<$5.50 & 6.30 & \nodata & $<$5.60 \\
487: %&&\\
488: %UV\,Cas & 6.20 & 6.20 & 6.50 & \nodata & $<$5.60 \\
489: %&&\\
490: %UX\,Ant & $<$6.20 & $<$6.20 & $<$6.70 & \nodata & $<$6.80\\
491: %&&\\
492: %VZ\,Sgr & 6.30 & 6.50 & 6.50 & $<$7.10 & $<$6.10\\
493: %&&\\
494: %R\,CrB & 6.85 & 7.00 & 7.05 & \nodata & $<$7.20\\
495: %&&\\
496: %V2552\,Oph & 6.60 & 6.70 & 6.90 & \nodata & $<$7.10 \\
497: %&&\\
498: %V854\,Cen & \nodata & $<$5.70 & 6.30 & \nodata & $<$5.70 \\
499: %&&\\
500: %SU\,Tau & 6.90 & 7.00 & 7.40 & \nodata & $<$7.30 \\
501: %&&\\
502: %V\,CrA & 6.50: & 6.90: & \nodata & \nodata & \nodata \\
503: %&&\\
504: %V482\,Cyg & 6.70: & 6.60: & 7.00: & \nodata & \nodata \\
505: %&&\\
506: %GU\,Sgr & 6.90: & 7.20: & 6.70: & \nodata & \nodata \\
507: %&&\\
508: %FH\,Sct & 6.90: & 7.20: & 6.90: & \nodata & \nodata \\
509: %&&\\
510: %Sakurai's object & $<$4.80 & $<$5.20 & \nodata & \nodata & \nodata \\
511: %\hline
512: %\end{tabular}
513: %\end{center}
514: %%$^a$ This paper for the model ($T_{\rm eff}$,$\log g, \xi$) $\equiv$
515: %% (18300, 2.2, 16.0)\\
516: %%$^b$ Jeffery (1993) for the mo (18300, 2.2, 16.0)\\
517: %\clearpage
518:
519: \clearpage
520:
521: \begin{figure}
522: \epsscale{1.00}
523: \plotone{f3.eps}
524: \caption{Observed F\,{\sc i} 6902.47\AA\ line profiles (solid lines) of
525: several RCB stars. Synthetic spectra are shown for three fluorine
526: abundances, as shown on the figure. \label{fig3}}
527: \end{figure}
528:
529: \begin{figure}
530: \epsscale{1.00}
531: \plotone{f4.eps}
532: \caption{Observed F\,{\sc i} 6834.26\AA\ line profiles (solid lines) of
533: several RCB stars. Synthetic spectra are shown for three fluorine
534: abundances, as shown on the figure. \label{fig4}}
535: \end{figure}
536:
537: \clearpage
538:
539: The adopted stellar parameters: effective temperature ($T_{\rm eff}$),
540: surface gravity ($\log g$), and microturbulent velocity ($\xi$),
541: are from \citet{asp00} for most of the analyzed RCBs. The stellar parameters adopted
542: for V2552\,Oph, and V\,CrA are from \citet{raolamb2003}, and \citet{raolamb2007},
543: respectively.
544: We adopt the Uppsala model atmospheres \citep{asp97a} with a carbon-to-helium abundance ratio C/He
545: of 1\% for the analyzed RCBs. The C/He ratio is not directly
546: determinable for the RCBs. The choice of C/He of 1\% is provided by the measured
547: C/He of EHe stars, the close relatives of RCBs.
548: The analyzed EHe stars have a mean C/He of 0.8\% (Pandey et al. 2006 and references
549: therein, Pandey \& Reddy 2006), which suggests that the adopted C/He ratio of 1\%
550: for the Uppsala model atmospheres is a reasonable choice.
551:
552: %The $gf$-values of the blending
553: %lines Si\,{\sc i}, C\,{\sc i}, and rest of the lines (see Table 1), are from the
554: %compilations by R. E. Luck (private communication), NIST database, and Kurucz's database,
555: %respectively.
556: For several of the analyzed RCBs, the best fitting profiles for two of the studied
557: F\,{\sc i} wavelength regions are illustrated in Figures 3 and 4.
558: The measured mean fluorine abundances (Table 1) are given as
559: log $\epsilon(\rm F)$, normalized such that log $\Sigma$$\mu_i \epsilon(i)$ = 12.15
560: where $\mu_i$ is the atomic weight of element $i$.
561:
562: \clearpage
563:
564: \begin{figure}
565: \epsscale{1.00}
566: \plotone{f5.eps}
567: \caption{Observed F\,{\sc i} 6834.26\AA\ line profile (solid line) of
568: Sakurai's object. Synthetic spectra are shown for three fluorine
569: abundances, as shown on the figure. \label{fig5}}
570: \end{figure}
571:
572: \clearpage
573:
574: The errors of the derived F abundances
575: given in Table 1 are the line-to-line scatter. The number of lines used is given within
576: brackets. The F abundances derived for the RCBs cooler or about effective temperature 6500 K
577: are uncertain and are marked with a colon in Table 1. However, the F abundances so marked
578: are determinations and not upper limits.
579: An upper limit to the F abundance for XX\,Cam, V854\,Cen, and the Sakurai's object is obtained
580: by comparing the synthetic and the observed profiles of the F\,{\sc i} line at 6834.26\AA\ (Table 1).
581: The effective error in F abundances due to uncertainty in the
582: adopted stellar parameters, that are typically: $\Delta$$T_{\rm eff}$ = $\pm$250 K,
583: $\Delta$$\log g$ = $\pm$0.5 cgs and $\Delta$$\xi$ = $\pm$2 km s$^{-1}$, is about 0.3 dex.
584: These uncertainties are similar across the sample of analyzed RCBs.
585: Note that the derived fluorine abundances in LTE represent only the first step defining the
586: absolute F abundance in these stars; more reliable values should, in principle, come from
587: full non-LTE calculations.
588:
589:
590:
591: A spectrum of Sakurai's object (V4334\,Sgr), a final He-shell flash product,
592: obtained on 5 May 1996 was available for analysis. The nondetection of the
593: F\,{\sc i} line at 6834.26\AA\ provides with an upper limit to the F abundance (Figure 5).
594: Other potential F\,{\sc i} lines in this (then) quite hot star were blended or
595: in inter-order gaps, e.g., the 6902 \AA\ line is masked by overlying
596: telluric lines.
597: Note that the upper limit to the F abundance for the Sakurai's object in Table 1 is for the
598: model's adopted C/He ratio of 1\%; a model for a C/He ratio of 10\% returns an
599: upper limit to the F abundance of about 6.4 dex. The stellar parameters
600: are from \citet{asp97b}.
601:
602:
603:
604: \section{Discussion}
605:
606: The analyzed RCBs (excluding stars with upper limits to the F abundances and,
607: uncertain F abundances) have a mean F abundance of 6.7 which is the same
608: for the analyzed EHes \citep{pana2006}. Thus, the F abundances are similar across
609: an effective temperature range from about 6500 K to 14000 K, an indication
610: that non-LTE effects are possibly small.
611: The F abundances of the analyzed RCB and
612: EHe stars show no obvious trend with their abundances of other elements.
613: In Figure 6, the F abundance is shown as a function of the Fe abundance for
614: the RCBs from Table 1 and the EHes from \citet{pana2006}. The F abundances from
615: RCBs and EHes are in good agreement and independent or almost so of the Fe
616: abundance which spans a range of nearly two dex. In contrast, the N abundance
617: is well correlated with the Fe abundance: the N
618: abundances are consistent with the sum of the initial C, N, and O
619: abundances suggesting that CNO-cycling has controlled the N
620: abundances. Among EHes and RCBs, the O
621: abundance shows a significant scatter at a given Fe abundance. In Figure 7,
622: the F abundance is shown as a function of the O abundance: the
623: F abundances are uncorrelated with the O abundance.
624:
625: \clearpage
626:
627: \begin{figure}
628: \epsscale{1.00}
629: \plotone{f6.eps}
630: \caption{The F abundance for majority RCBs (unfilled circles), minority RCBs (unfilled squares), and EHes (filled circles)
631: as a function of the Fe abundance. The solid line shows the possible
632: initial F abundances assuming the solar F and Fe abundances (large cross)
633: and [F/Fe] = 0.
634: \label{fig6}}
635: \end{figure}
636:
637: \begin{figure}
638: \epsscale{1.00}
639: \plotone{f7.eps}
640: \caption{The F abundance for majority RCBs (unfilled circles), minority RCBs (unfilled squares), and EHes (filled circles)
641: as a function of the O abundance.
642: Solar F and O abundances are represented by the large cross.
643: \label{fig7}}
644: \end{figure}
645:
646: \clearpage
647:
648: \subsection{The setting}
649:
650: Knowledge of fluorine abundance in astronomical objects is limited but,
651: nonetheless, helpful in interpreting these measurements for
652: the RCBs (this paper) and EHes \citep{pana2006}. The solar system
653: abundance as measured from
654: meteorites is $\log\epsilon$(F) = 4.46$\pm0.06$ \citep{lod2003,asp2005}
655: with rather imprecise confirmation from the detection of HF lines in
656: sunspots. Measurements of HF lines in K and M Galactic
657: giants \citep{jori92,cunha2003} give a mean F abundance of $\log\epsilon$(F) = 4.62$\pm0.12$
658: for a sample with a mean Fe abundance ([Fe/H]$=-0.10\pm0.16$) not
659: significantly different from solar. The trend of F with Fe is quite uncertain.
660: The sole Galactic star with a (slightly) sub-solar Fe abundance is
661: Arcturus with [Fe/H]=$-$0.6 and a F abundance ($\log\epsilon$(F) = $4.10$ -
662: Cunha et al. 2003) suggesting that [F/Fe] $\sim 0$ but observations of
663: additional metal-poor giants are required to confirm this hint.
664:
665: The F vs Fe trend is one important datum for inferring the RCBs initial
666: F abundance from their present Fe abundances.
667: The mean Fe abundance
668: from \citet{asp00} and \citet{raolamb2003} for 14 majority
669: RCBs is $\log\epsilon$(Fe) = $6.5$ with a spread of 0.8 dex; three
670: minority RCBs (V\,CrA, VZ\,Sgr, and V3795\,Sgr) of very low Fe
671: abundance are not included here, however, these three minority RCBs
672: are not different in F abundances when compared with the majority RCBs.
673: To this mean Fe abundance must
674: be attached the qualification that the analyses assume a C/He ratio of
675: 1\% by number but the analyses do not return the input C abundance but a
676: value about 0.6 dex lower, a discrepancy termed `the carbon problem' by
677: Asplund et al. Analyses of EHe stars (Pandey et al. 2006 and references
678: therein) which are immune to the carbon problem give a mean Fe
679: abundance $\log\epsilon$(Fe) = $6.7$ and a spread of 1.1 dex with two
680: minority EHes excluded. The similar Fe abundances
681: of EHes and RCBs suggests that the carbon problem is not
682: adversely affecting the determination of the Fe abundance.
683: The mean C/He ratio found for the EHes is slightly less than the assumed
684: C/He = 1\% for the RCBs. Adoption of the C/He ratio of the EHes would
685: reduce the mean Fe abundance for the RCBs by about 0.15 dex.
686: The assumption is made that that the Fe abundance is not
687: affected by the stars' evolutionary history.
688: It is striking that the EHes and RCBs are quite Fe-poor
689: relative to the Sun. This might occasion concerns that the initial iron
690: abundance has been reduced in the process of evolution to a H-poor
691: star. In this connection, we note that abundance ratios [X/Fe],
692: especially those of the $\alpha$-elements -- Mg, Si, S and Ca --
693: are those expected of metal-poor stars and not those of solar
694: abundance stars \citep{asp00,pan2006}. In the
695: following discussion, we assume that the present abundances of iron
696: are the initial values.
697:
698: Although, as noted above, the observational
699: data are sparse in the extreme, it seems
700: safe to assume that [F/Fe] $\leq 0$ for metal-poor ([Fe/H]$<0$) stars.
701: Galactic chemical evolution models confirm this expectation \citep{renda2004}.
702: In estimating the F enrichment for EHes and
703: RCBs, we assume [F/Fe]=0 and the solar F abundance
704: $\log\epsilon$(F)$ = 4.46$. Thus, a representative RCB with
705: $\log\epsilon$(Fe)$ = 6.5$ or [Fe/H]= $-1.0$ is assumed to have
706: an initial F abundance of $\log\epsilon$(F)$ \leq 3.5$.
707: More interestingly, the F overabundances are extremely large:
708: enhancements of about 800, 2500, and 8000 at $\log\epsilon$(Fe)$ = 7.5$, 6.5, and 5.5,
709: respectively. The fact that the overabundances are very large is qualitatively
710: true irrespective of all plausible values for [F/Fe] as a function of [Fe/H].
711:
712:
713: \subsection{The DD Scenario and Fluorine}
714:
715: In the `cold' (i.e., no nucleosynthesis during the merger) version of the
716: DD scenario, the He white dwarf is stripped and accreted by the
717: C-O white dwarf. During the active phase of the merger, the accreted
718: material is mixed with the outer layer of the C-O white dwarf
719: which is the residue of the He-intershell of an AGB star for
720: which the C-O white dwarf was the electron degenerate core. Mixing
721: may also penetrate the outer layers of the C-O core. In the `hot' DD scenario, nucleosynthesis
722: occurs during accretion. Following both cold and hot mergers, the hot luminous
723: H-poor star evolves to the red before returning to the blue and descending the
724: white dwarf cooling track. Along this track, He-shell flashes are likely to
725: occur and dredge-ups are likely to alter the surface chemical evolution.
726: These alterations have not been extensively studied theoretically.
727: Limited predictions for a slow rate of accretion have been reported
728: by \citet{saio02}.
729:
730: Given the principal features of the expected compositions of the
731: He white dwarf and the He-shell around the C-O white dwarf, one
732: simple recipe that provides an atmosphere with a RCB- and EHe-like
733: composition is as follows:
734: a 0.3$M_\odot$ He white dwarf is mixed with 0.03$M_\odot$ of
735: He-intershell material. This cold recipe does not consider nucleosynthesis
736: to occur during or following the merger.
737: The relative masses of He-intershell and He white dwarf
738: determine the final C/He; the He white dwarf is
739: rich in He but highly deficient
740: in C and O and the He-intershell is relatively rich in C and O. The N abundance
741: is provided primarily by the He white dwarf.
742: A H-rich surface layer on
743: the He and/or the C-O white dwarf provides the residual H for the RCB and EHe.
744: For discussion of such a recipe and how it may match observed
745: compositions of RCBs and EHes, see \citet{saio02} and \citet{pan2006}.
746:
747: For elements greatly enriched in the He-intershell, their
748: abundance in the RCB and EHe is diluted by the
749: material from the He white dwarf. A dilution by a factor of order
750: ten is necesssary in light of the expected abundances of C and He in the
751: He-intershell. In the case of F, we may appeal to theoretical
752: predictions of F abundances in the He-intershell of AGB stars
753: \citep{lug2004} and to observations of F in PG1159 stars.
754: PG1159 stars are H-deficient very hot ($T_{\rm eff} = 75,000 - 200,000$ K)
755: post-AGB stars whose surfaces `offer the unique possibility of
756: studying intershell abundances {\it directly}' \citep{werner2006}.
757: The PG1159 stars are considered to be products of the FF scenario
758: and to involve a `very late thermal pulse' (VLTP) (see Werner \& Herwig [2006]
759: for the difference between a very late thermal pulse and a late thermal
760: pulse). The C/He ratio of PG1159 stars is too high for EHe and RCB stars to
761: be identified with the FF scenario. In the DD scenario, the high C/He ratio
762: is diluted to a lower ratio when the FF-like layer of the C-O white dwarf
763: is mixed with the He white dwarf.
764:
765: Fluorine in PG1159 stars shows a range of abundances \citep{werner2005,werner2006}
766: from solar to 250 times solar.
767: As Werner \& Herwig discuss, this
768: range is not out of line with theoretical predictions for the He-intershell. In addition,
769: the theoretical predictions account fairly well for the observed enrichments
770: of F in (H normal) giant stars known to be affected by the third dredge-up that
771: brings material from the He-intershell into the envelope \citep{jori92}.
772: But in the cold DD scenario, the He-intershell material
773: is diluted, according to the canonical recipe, by a factor of about ten and,
774: then, overabundances of up to 25 times solar for the
775: RCBs and EHes are predicted. The observed overabundances of F range upward of 1000 times.
776: Unless the VLTP experienced by the PG1159 stars drastically
777: reduced the F abundance of the He-intershell or the nucleosynthesis
778: experienced by the He-intershell of the C-O white dwarf differs
779: markedly from that experienced by the PG1159 stars, it would appear
780: that the cold DD scenario can not account for the observed F
781: abundances of RCBs and EHes.
782:
783:
784: The upper limit for the F abundance of Sakurai's object is consistent
785: with but does not demand that ingestion of H and the final
786: He-shell flash leads to a complete conversion of
787: $^{14}$N to $^{22}$Ne. The temperatures are so
788: high that $^{18}$O, which is the seed for $^{19}$F, is destroyed.
789: Note that the infrared spectra of Sakurai's object obtained
790: during 1997$-$1998 when it had strong CO overtone bands showed no
791: evidence for $^{12}$C$^{18}$O \citep{geballe2002}.
792: Therefore, Sakurai's object and the other FF stars, when compared
793: with most of the RCB and HdC stars, probably have different origins.
794:
795: A `hot' DD scenario is fuelled by a rapid deposition of accreted
796: He-rich material on the surface of the C-O white dwarf.
797: Observational evidence that suggests
798: a hot DD scenario is provided by Clayton et al.'s (2005, 2007) discovery
799: of huge overabundances of $^{18}$O in HdC and some cool RCBs through
800: medium-resolution spectroscopy of the first-overtone vibration-rotation
801: bands of the CO molecule near 2.3 microns. High-resolution
802: spectroscopy confirms that $^{16}$O/$^{18}$O ratios are low:
803: \citet{garc2007} find $^{16}$O/$^{18}$O $=0.3$
804: to 0.5 for the three HdCs with CO bands and 16 for the cool RCB S\,Aps.
805:
806: \citet{clay2007} show via qualitative arguments about temperature, density, and
807: timescale that the merger leads to temperatures at the base of the
808: accreted material approaching 200 million K with conditions
809: ripe for nucleosynthesis lasting several years.
810: Their calculations with a reaction network show that the
811: plentiful supply of $^{14}$N in the He white dwarf may be converted
812: to $^{18}$O by $\alpha$-capture.
813: Without this conversion, the DD scenario cannot account for the $^{18}$O
814: overabundances; the expected
815: $^{18}$O abundance in the He white dwarf and in the He-intershell
816: of the C-O white dwarf are very low. The present measurements of
817: $^{18}$O in HdC and RCB stars indicate that no more than a minor fraction
818: of the $^{14}$N is converted to $^{18}$O.
819: (`No overproduction of $^{18}$O is
820: expected in the FF scenario' state Clayton et al. [2007].)
821:
822: The possibilities for nucleosynthesis in the hot DD scenario are
823: more extensive if the He white dwarf has a H-rich outer layer.
824: Since many HdC, RCB, and EHe stars have retained some H, H
825: was likely a participant in the nucleosynthesis. \citet{clay2007}
826: note that the presence of protons in $^{18}$O-rich
827: gas promotes the synthesis of
828: $^{19}$F. The reaction $^{18}$O$(p,\gamma)^{19}$F provides
829: $^{19}$F directly.
830: % with an additional channel involving
831: %$^{18}$O(p,$\alpha$)$^{15}$N and $^{15}$N($\alpha$,$\gamma$)$^{19}$F.
832: The $^{18}$O is converted about 150 times faster by proton-captures to $^{12}$C:
833: $^{18}$O(p,$\alpha$)$^{15}$N(p,$\alpha$)$^{12}$C.
834: Clayton et al.'s (2007, their Figure 6) sample single-zone
835: calculation for a mixture with H added
836: and for an arbitrary temperature history and initial composition
837: showing effects of some He-burning
838: shows that
839: the $^{19}$F abundance increased by about a dex at the cessation of
840: nucleosynthesis from an initial value about a factor of three below
841: solar to a factor of two above solar. At its peak, F was briefly
842: about 100 times above its solar abundance before $^{19}$F($\alpha$,p)$^{22}$Ne
843: took its toll.
844: Clearly, although $^{19}$F synthesis is demonstrated
845: by these calculations, a remaining challenge is to show that the parameter
846: space covered by the hot DD scenario includes the possibility of
847: robustly increasing the F abundance not by a factor of 100 but
848: to the observed levels of 1000 to 8000
849: times over solar across the metallicity range [Fe/H] of $-0.5$ to $-2.0$.
850:
851: Fluorine abundances appear to demand that H be a key ingredient in
852: the hot DD scenario. This is suggested by the
853: F/N ratios. The N abundances of the
854: EHe stars are equal to the sum of the initial C, N, and O abundances,
855: as expected for He-rich material exposed to CNO-cycling \citep{pan2006}.
856: The RCBs give a similar result. The observed N/F ratios are about
857: 80, 60, and 5 at [Fe/H] $= -0.5$, $-1.0$, and $-2.0$, respectively.
858: Given that about 150 times more $^{18}$O is diverted to $^{12}$C than is
859: converted to $^{19}$F, the observed N/F abundances imply that
860: total conversion of the $^{14}$N from the He white dwarf to $^{18}$O
861: cannot provide the observed $^{19}$F abundances and this total
862: conversion certainly cannot account for the fact that the $^{14}$N
863: across the sample of EHe and RCB stars
864: is consistently close to that
865: expected from CNO-cycling for the He white dwarf.
866: The conclusion seems to be that hot accretion must involve
867: addition of H to the extent that N and $^{18}$O are regenerated by H-burning.
868: This and other
869: nucleosynthetic consequences are illustrated by
870: Clayton et al. (2007, their Figure 6). Their
871: calculation if representative would explain the star-to-star
872: variation in O abundances but the fact that the observed abundances
873: are close to the sum of the initial C, N, and O abundances has then to
874: be regarded as fortuitous. Obviously, challenges remain in accounting
875: for the F abundances. One challenge arises from incorporating Clayton et al.'s
876: single-zone calculations into a full model of the accreting star and
877: its subsequent evolution
878: with attendant shell flashes and dredge-ups.
879:
880: Another challenge involving nucleosynthesis concerns the abundance of
881: Li. Lithium is present in four of the analyzed 15 majority RCBs \citep{raolamb94}
882: and in one of the five HdCs \citep{raolamb96} with
883: an appreciable abundance $\log\epsilon$(Li) $\sim 3$. The only
884: conceivable reservoir of Li in the two white dwarfs ahead of their
885: merger is the H-rich layer around the He white dwarf but the mass of this
886: is too small to result in the above Li abundances, even if the H-rich
887: layer had retained its initial Li abundance. Lithium synthesis is
888: presumed to involve $^3$He and the reactions
889: $^3$He($\alpha$,$\gamma$)$^7$Be(e,$\nu$)$^7$Li.
890: The $^3$He may be present in the H-rich layer as
891: the product of the
892: first steps of the $pp$-chain and as an initial ingredient of the star.
893: In a hot DD scenario in which $^{14}$N is converted to $^{18}$O, it
894: seems inconceivable that $^7$Be or $^7$Li can avoid thorough
895: destruction by $\alpha$-capture; enrichment of Li in parallel with
896: enrichment of $^{18}$O and F seems impossible in nucleosynthesis
897: accompanying the DD scenario \citep{clay2007}.
898: One supposes that Li may be synthesized in `warm' DD scenarios in which
899: mild nucleosynthesis occurs with temperatures too low for $^{18}$O
900: production. Lithium synthesis is possible in the FF scenario.
901: Most irritatingly, the
902: HdC and RCB stars for which $^{18}$O has been detected are not those
903: in which Li is present; note that the possibility of $^{18}$O
904: enrichment in the Li-rich HdC star HD\,148839 is not discarded
905: by \citet{garc2007}.
906: However, the Li-rich RCB stars are not
907: distinguishable from other RCBs and all have the He/C/N/O ratios
908: of a DD scenario and incompatible with a FF scenario.
909: It is intriguing that the Li-rich stars (UW\,Cen, R\,CrB, and SU\,Tau) are among
910: the stars with the highest F abundances. The detection of high F in both
911: minority and majority RCBs is also very interesting since $^{18}$O
912: observations do not include any of these stars as these are too hot.
913: This adds the minority RCBs as well as the Li-rich RCB stars to the
914: stars that might have large amounts of $^{18}$O.
915:
916: \section{Concluding remarks}
917:
918: Lithium and fluorine are not the sole or even the principal
919: concerns for the DD scenario.
920: One wonders why the C/He $\sim 0.01$ ratio, as measured for the EHes, spans
921: a narrow range. The He is contributed primarily by the He white
922: dwarf and the C by the He-intershell of the C-O white dwarf. What sets
923: the mass ratio of these two regions rather precisely?
924:
925: The evolutionary connections between the H-deficient
926: stars remains unproven but with obvious potential
927: links. Both possible sequences --
928: HdC $\rightarrow$ RCB $\rightarrow$ EHe and
929: EHe $\rightarrow$ RCB $\rightarrow$ HdC are possible
930: because the DD scenario leads to a very hot star
931: evolving to lower temperatures before the evolutionary track reverses
932: direction.
933: An extension EHe $\rightarrow$ sdO $\rightarrow$ O(He)
934: to higher temperatures and higher surface gravities seems
935: plausible \citep{rauch2006,werner2006}.
936: The stars in this sequence are all candidates for an origin via the
937: DD scenario, i.e., they have the low C/He ratio and not the high
938: ratio characteristic of the the FF scenario; a similar
939: sequence for the FF scenario exists linking objects like FG\,Sge and
940: V4334\,Sgr (Sakurai's object) through the [WC] to the PG1159 stars.
941:
942:
943: Continued study of the compositions of stars along these
944: sequences will serve to sharpen understanding of the evolutionary
945: links and the entry points for stars into the sequence.
946: New data such as that for F and $^{18}$O stimulate continued
947: study. The F abundances of RCB and EHe stars are quite similar in
948: value and spread and thus support the presumed evolution from
949: RCB to EHe. On the other hand, there is a hint that the
950: $^{16}$O/$^{18}$O ratios of the HdC and RCB stars may be
951: different with the ratio being systematically higher in the
952: (cool) RCBs than in the three HdCs. Is the difference merely
953: a distraction caused by small number statistics or is it
954: symptomatic of a change introduced as a HdC evolves to a cool RCB or
955: does it reflect that the DD scenario can place stars on the evolutionary
956: sequence at different starting points with different compositions?
957:
958:
959:
960: We thank the referee Geoff Clayton for the nice report.
961: GP thanks the VBT staff for their assistance.
962: This research has been supported in part by the Robert A. Welch Foundation
963: of Houston, Texas.
964:
965:
966: \clearpage
967:
968: \begin{thebibliography}{}
969: \bibitem[Asplund et al.(1997a)]{asp97a} Asplund, M., Gustafsson, B., Kiselman, D.,
970: \& Eriksson, K. 1997a, \aap, 318, 521
971: \bibitem[Asplund et al.(1997b)]{asp97b} Asplund, M., Gustafsson, B., Lambert, D. L.,
972: \& Rao, N. K. 1997b, \aap, 321, L17
973: \bibitem[Asplund et al.(2000)]{asp00} Asplund, M., Gustafsson, B., Lambert, D. L.,
974: \& Rao, N. K. 2000, \aap, 353, 287
975: \bibitem[Asplund et al.(2005)]{asp2005} Asplund, M., Grevesse, N.,
976: \& Sauval, A. J. 2005, ASP Conf. Ser. 336, 25
977: \bibitem[Clayton(1996)]{clay96} Clayton, G. C. 1996, \pasp, 108, 225
978: \bibitem[Clayton et al.(2005)]{clay2005} Clayton, G. C., Herwig, F., Geballe, T. R., Asplund, M.,
979: Tenenbaum, E. D., Engelbracht, C. W., \& Gordon, K. D. 2005, \apjl, 623, L141
980: \bibitem[Clayton et al.(2007)]{clay2007} Clayton, G. C., Geballe, T. R., Herwig, F., Fryer, C.,
981: \& Asplund, M. 2007 \apj, 662, 1220
982: \bibitem[Cunha et al.(2003)]{cunha2003} Cunha, K., Smith, V. V., Lambert, D. L.,
983: \& Hinkle, K. H. 2003, \aj, 126, 1305
984: \bibitem[Geballe et al.(2002)]{geballe2002} Geballe, T. R., Evans, A., Smalley, B.,
985: Tyne, V. H., Eyres, S. P. S. 2002, Ap\&SS, 279, 39
986: \bibitem[Guerrero et al.(2004)]{gue04} Guerrero, J.,
987: Garc\'{\i}a-Berro, E., \& Isern, J. 2004, \aap, 413, 257
988: \bibitem[Garc\'{\i}a-Hern\'{a}ndez et al.(2007)]{garc2007} Garc\'{\i}a-Hern\'{a}ndez, D.A.,
989: Hinkle, K.H., Lambert, D.L., \& Eriksson, K. 2007, \apj, submitted
990: \bibitem[Jorissen et al.(1992)]{jori92} Jorissen, A., Smith, V. V., \&
991: Lambert, D. L. 1992, \aap, 261, 164
992: \bibitem[Lambert \& Rao(1994)]{raolamb94} Lambert, D. L., \& Rao, N. K. 1994, JAA, 15, 47
993: \bibitem[Lodders(2003)]{lod2003} Lodders, K. 2003, \apj, 591, 1220
994: \bibitem[Lugaro et al.(2004)]{lug2004} Lugaro, M., Ugalde, C., \&
995: Karakas, A., I. 2004, \apjl, 615, L934
996: \bibitem[Moore(1965)]{moore65} Moore, Ch. E. 1965, Selected Tables of Atomic Spectra,
997: NSRDS--NBS 3, Section 2, Washington
998: \bibitem[Moore(1967)]{moore67} Moore, Ch. E. 1967, Selected Tables of Atomic Spectra,
999: NSRDS--NBS 3, Section 1, Washington
1000: \bibitem[Moore(1972)]{moor72} Moore, Ch. E. 1972, A Multiplet Table of Astrophysical
1001: Interest, NSRDS--NBS, Washington
1002: \bibitem[Moore(1993)]{moor93} Moore, Ch. E. 1993, Tables of Spectra of Hydrogen, Carbon,
1003: Nitrogen, and Oxygen Atoms and Ions, Editor: Jean W. Gallagher, CRC Series in Evaluated Data
1004: in Atomic Physics, CRC press
1005: \bibitem[Musielok et al.(1999)]{musie99} Musielok, J., Pawelec, E., Griesmann, U., \&
1006: Wiese, W. L. 1999, \pra, 60, 947
1007: \bibitem[Nave et al.(1994)]{nave94} Nave, G., Johansson, S., Learner, R. C. M.,
1008: Thorne, A. P., \& Brault, J. W. 1994, \apjs, 94, 221
1009: \bibitem[Pandey et al.(2001)]{pan2001} Pandey, G., Rao, N. K., Lambert, D. L.,
1010: Jeffery, C. S., \& Asplund, M. 2001, \mnras, 324, 937
1011: \bibitem[Pandey et al.(2004a)]{pan2004a} Pandey, G., Lambert, D. L., Rao, N. K.,
1012: \& Jeffery, C. S. 2004a, \apjl, 602, L113
1013: \bibitem[Pandey et al.(2004b)]{pan2004b} Pandey, G., Lambert, D. L., Rao, N. K.,
1014: Gustafsson, B., Ryde, N., \& Yong, D. 2004b, \mnras, 353, 143
1015: \bibitem[Pandey et al.(2006)]{pan2006} Pandey, G., Lambert, D. L.,
1016: Jeffery, C. S., \& Rao, N. K. 2006, \apj, 638, 454
1017: \bibitem[Pandey \& Reddy(2006)]{panred2006} Pandey, G., \& Reddy, B. E. 2006,
1018: \mnras, 369, 1677
1019: \bibitem[Pandey(2006)]{pana2006} Pandey, G. 2006, \apjl, 648, L143
1020: \bibitem[Rao \& Lambert(1996)]{raolamb96} Rao, N. K., \& Lambert, D. L. 1996,
1021: ASP Conf. Ser. 96, 43
1022: \bibitem[Rao \& Lambert(2000)]{raolamb2000} Rao, N. K., \& Lambert, D. L. 2000,
1023: \mnras, 313, L33
1024: \bibitem[Rao \& Lambert(2003)]{raolamb2003} Rao, N. K., \& Lambert, D. L. 2003,
1025: \pasp, 115, 1304
1026: \bibitem[Rao et al.(2004)]{rao04} Rao, N. K., Sriram, S., Gabriel, F., Prasad,
1027: B. R., Samson, J. P. A., Jayakumar, K., Srinivasan, R., Mahesh, P. K.,
1028: \& Giridhar, S. 2004, Asian Journal of Physics, 13, 267
1029: \bibitem[Rao(2005)]{rao2005} Rao, N. K. 2005, ASP Conf. Ser. 336, 185
1030: \bibitem[Rao et al.(2005)]{rao05b} Rao, N. K., Sriram, S., Jayakumar, K.,
1031: \& Gabriel, F. 2005, JAA, 26, 331
1032: \bibitem[Rao et al.(2006)]{raolambsh2006} Rao, N. K., Lambert, D. L.,
1033: \& Shetrone, M. D. 2006, \mnras, 370, 941
1034: \bibitem[Rao \& Lambert(2007)]{raolamb2007} Rao, N. K., \& Lambert, D. L. 2007,
1035: \mnras, in press
1036: \bibitem[Rauch et al.(2006)]{rauch2006} Rauch, T., Reiff, E., Werner, K., Herwig, F.,
1037: Koesterke, L., \& Kruk, J. W. 2006, ASP Conf. Ser. 348, 194
1038: \bibitem[Renda et al.(2004)]{renda2004} Renda, A., et al. 2004, \mnras, 354, 575
1039: \bibitem[Saio \& Jeffery(2002)]{saio02} Saio, H., \& Jeffery, C. S. 2002, \mnras, 333, 121
1040: \bibitem[Tenenbaum et al.(2005)]{tenen2005} Tenenbaum, E. D., Clayton, G. C.,
1041: Asplund, M., Engelbracht, C. W., Gordon, K. D., Hanson, M. M., Rudy, R. J.,
1042: Lynch, D. K., Mazuk, S., Venturini, C. C., \& Puetter, R. C. 2005, \aj, 130, 256
1043: \bibitem[Tull et al.(1995)]{tull95} Tull, R. G., MacQueen P. J., Sneden, C., \&
1044: Lambert, D. L. 1995, \pasp, 107, 251
1045: \bibitem[Werner et al.(2005)]{werner2005} Werner, K., Rauch, T., \&
1046: Kruk, J. W. 2005, \aap, 433, 641
1047: \bibitem[Werner \& Herwig(2006)]{werner2006} Werner, K., \& Herwig, F. 2006,
1048: \pasp, 118, 183
1049: \bibitem[Zaniewski et al.(2005)]{zanie2005} Zaniewski, A., Clayton, G. C., Welch, D. L.,
1050: Gordon, K. D., Minniti, D., \& Cook, K. H. 2005, \aj, 130, 2293
1051:
1052:
1053: \end{thebibliography}
1054: \end{document}
1055:
1056:
1057:
1058: