1: %\documentclass[manuscript]{aastex}
2: \documentclass[12pt,preprint]{aastex}
3: %% preprint2 produces a double-column, single-spaced document:
4: %\documentclass{emulateapj}
5:
6: %\documentclass[preprint2]{aastex}
7: %\usepackage[dvips]{epsfig}
8: %\usepackage[dvips]{rotating}
9: \bibpunct{(}{)}{;}{a}{}{,}
10:
11:
12: \slugcomment{May 20, 2008}
13: \shorttitle{New 3D$-$1D corrected abundances of HE~1327$-$232}
14: \shortauthors{A. Frebel et al.}
15:
16: \begin{document}
17: \title{HE~1327$-$2326, an unevolved star with $\mbox{[Fe/H]}<-5.0$.\\
18: II. New 3D$-$1D corrected abundances from a VLT/UVES spectrum\altaffilmark{1}}
19:
20: \author{
21: Anna Frebel\altaffilmark{2},
22: Remo Collet\altaffilmark{3},
23: Kjell Eriksson\altaffilmark{4},
24: Norbert Christlieb\altaffilmark{4}
25: and Wako Aoki\altaffilmark{5, 6}}
26:
27: \altaffiltext{1}{Based on observations collected at the European
28: Southern Observatory, Paranal, Chile (Proposal ID 075.D-0048).}
29:
30: \altaffiltext{2}{McDonald Observatory, The University of Texas at
31: Austin, 1 University Station, C1400, Austin, TX 78712-0259;
32: anna@astro.as.utexas.edu}
33:
34: \altaffiltext{3}{Max-Planck-Institut f{\"u}r Astrophysik, Karl-Schwarzschild-Str.~1,
35: Postfach 1317, D$-$85741 Garching b. M{\"u}nchen, Germany}
36:
37: \altaffiltext{4}{Department of
38: Astronomy and Space Physics, Uppsala University, Box 515, SE-751-20 Uppsala,
39: Sweden; norbert@astro.uu.se}
40:
41: \altaffiltext{5}{National Astronomical Observatory of Japan, 2-1-21 Osawa,
42: Mitaka, Tokyo, 181-8588 Japan}
43:
44: \altaffiltext{6}{Department of Astronomical Science, The Graduate
45: University of Advanced Studies, Mitaka, Tokyo, 181-8588 Japan}
46:
47: \begin{abstract}
48: We present a new abundance analysis of HE~1327$-$2326, the currently
49: most iron-poor star, based on observational data obtained with
50: VLT/UVES. We correct the 1D LTE abundances for 3D effects to provide
51: an abundance pattern that supersedes previous works, and should be
52: used to observationally test current models of the chemical yields of
53: the first-generation SNe. Apart from confirming the 1D LTE abundances
54: found in previous studies before accounting for 3D effects, we make
55: use of a novel technique to apply the 3D$-$1D corrections for CNO
56: which are a function of excitation potential and line strength for the
57: molecular lines that comprise the observable CH, NH, and OH
58: features. We find that the fit to the NH band at 3360\,{\AA} is
59: greatly improved due to the application of the 3D$-$1D
60: corrections. This may indicate that 3D effects are actually observable
61: in this star. We also report the first detection of several weak Ni
62: lines. The cosmologically important element Li is still not detected;
63: the new Li upper limit is extremely low, $A({\rm Li})<0.62$, and in
64: stark contrast with results not only from WMAP but also from other
65: metal-poor stars. We also discuss how the new corrected abundance
66: pattern of HE~1327$-$2326 is being reproduced by individual and
67: integrated yields of SNe.
68:
69:
70: \end{abstract}
71:
72: \keywords{Galaxy: halo --- stars: abundances --- stars: formation ---
73: stars: individual (\objectname{HE~1327$-$2326}) --- stars:
74: Population II --- early universe }
75:
76: \section{Introduction}
77: The interpretation of the chemical abundance patterns of the most
78: metal-poor Galactic stars provides important clues to our
79: understanding of the early Universe and the chemical evolution of the
80: Milky Way. To advance our knowledge about this early time the observed
81: stellar signatures have to be compared with theoretical models of the
82: first cosmic chemical enrichment events. Currently, several groups
83: \citep{UmedaNomotoNature,chieffi_limongi2002, meynet2002, meynet2005,
84: heger2008} are working on different models of Population\,III
85: (Pop\,III) supernovae (SNe). The elemental yields of these SN
86: calculations are used, among other things, for a comparison with
87: individual old metal-poor stars as well as groups of metal-poor stars
88: with similar chemical patterns. These stars are believed to have
89: formed from material exclusively enriched during these early SN
90: events. It can thus be tested whether just a single SN was responsible
91: for the chemical signature in what could be some of the first low-mass
92: stars that formed in the Universe.
93:
94: The recent discoveries of two stars with iron abundances of
95: $\mbox{[Fe/H]}\footnote{Where \mbox{[A/B]}$ = \log(N_{\rm A}/N_{\rm
96: B}) - \log(N_{\rm A}/N_{\rm B})_\odot$ for the number N of atoms of
97: element A and B and $\odot$ refers to the Sun}<-5$
98: \citep{HE0107_Nature, HE0107_ApJ, HE1327_Nature, Aokihe1327} have
99: already proved to be ideal test stars for this scenario because they
100: likely formed very early on in the Universe. \citet{UmedaNomotoNature}
101: first explained the abundance pattern of HE~0107$-$5240 with the
102: yields by one single SN undergoing a mixing and fallback mechanism.
103: Subsequent work by \citet{iwamoto_science} explained HE~1327$-$2326's
104: signature in the same manner. However, constructing such a SN model
105: requires as many observed abundances as possible-- it is a challenge
106: for this star, as it is extremely iron-deficient, and any spectral
107: lines are very weak (intrinsically, and because of the effective
108: temperature of 6180\,K) and difficult to detect.
109:
110: In this paper, we thus present new abundance measurements of
111: HE~1327$-$2326, the currently most iron-poor star known (Frebel et
112: al. 2005; Aoki et al. 2006). In 2005, a new VLT spectrum was
113: obtained. The initial discovery spectrum had been taken with the
114: Subaru telescope. The new data have higher $S/N$ and thus already
115: facilitated the determination of the O abundance of HE~1327$-$2326
116: from very weak near-UV OH lines (Frebel et al. 2006). The O
117: abundance provided a crucial ingredient for the interpretation of the
118: overall chemical abundance pattern (e.g., \citealt{iwamoto_science,
119: meynet2005}).
120:
121: We now use these new VLT data to i) re-determine the abundances of all
122: elements obtained from the Subaru spectrum, ii) to search for
123: (additional) lines of new and already detected elements in the star,
124: iii) to obtain tighter upper limits of existing estimates, and iv) to
125: correct all abundances arising from our 1D model stellar atmosphere
126: analysis for 3D effects \citep{collet06}. For some elements, these
127: effects are quite significant. Our measurements, and in particular the
128: new detection of Ni, will provide observational constraints on the
129: yields of the first generations of SNe. Hence, the enrichment scenario
130: of HE~1327$-$2326 can be further explored, as well as the origins of
131: other extremely metal-poor stars.
132:
133:
134: %==========================
135:
136:
137: \section{Observations}
138: A description of the spectroscopic observations can be found in Frebel et
139: al. (2006). We remind the reader that due to the choice of settings
140: the wavelength coverage of the new data is limited to three regions:
141: 3050--3870\,{\AA}, 4780--6805\,{\AA}, and 5720--9470\,{\AA}. The data
142: were reduced with the REDUCE package \citep{reduce} and subsequently
143: rebinned. The signal-to-noise ($S/N$) ratios of different wavelength
144: regions are listed in Table~\ref{Tab:SN}.
145:
146:
147: \section{Stellar Parameters and Model Atmospheres}
148: We adopt an effective temperature of $T_{\rm eff}=6180$\,K, as
149: determined in our previous analyses \citep{HE1327_Nature,
150: Aokihe1327}. Back then it was not possible, however, to distinguish whether
151: HE~1327$-$2326 would be a subgiant or a main-sequence star. In the
152: meantime, \citet{korn_fs3} carried out a new analysis of the Ca\,I/II
153: ionization equilibrium, which functions as a surface-gravity
154: indicator. As Ca\,I line formation is prone to departures from LTE
155: \citep{Mashonkina_ca}, detailed calculations have been necessary to
156: determine the magnitude of non-LTE effects. Based on the results of a
157: combined analysis of Ca\,I/II ionization equilibrium and Balmer lines,
158: \citet{korn_fs3} conclude that HE~1327$-$2326 is more likely a
159: subgiant. Hence, for the present work, we will consider the star to
160: be in this evolutionary stage ($\log g= 3.7$).
161:
162: For the 1D abundance analysis, we use a customary 1D LTE {\sc marcs}
163: model stellar atmosphere (B. Gustafsson et al. 2008, in prep.). The
164: model has been constructed with $T_{\rm eff}=6180$\,K, $\log g= 3.7$,
165: and $v_{micr}=2.0$\,km/s. The complete abundance pattern of the star
166: as derived by \citet{Aokihe1327} has been taken into account in the
167: actual modeling. Incidentally, \citet{Aokihe1327} also showed that
168: an analysis based on a Kuruzc model with a solar chemical composition
169: scaled down to $\mbox{[Fe/H]}=-5$ returns a very similar abundance
170: pattern for HE~1327$-$2326 (within 0.1-0.2\,dex). The mixing-length
171: theory (MLT) formulation for the {\sc marcs} model adopted here comes
172: from \citet{henyey06}.
173:
174: Recently, \citet{collet06} presented 3D$-$1D corrections to the
175: elemental abundances of HE~1327$-$2326 derived from individual atomic
176: as well as molecular lines. Such corrections were computed under the
177: assumption of local thermodynamic equilibrium (LTE) by means of a
178: differential 3D$-$1D analysis based on a 3D hydrodynamical and a 1D
179: {\sc marcs} model atmosphere with the same stellar parameters; the
180: authors then applied the results of their 3D$-$1D analysis to the
181: abundances derived by \citet{Aokihe1327} and \citet{o_he1327} with a
182: 1D analysis to obtain the 3D$-$1D corrected chemical composition of
183: {HE~1327$-$2326}. In the present work, we carry out a differential
184: 3D$-$1D LTE abundance analysis for the newly observed atomic lines
185: following the same procedure as in \citet{collet06}. With the
186: availability of such corrections, in this paper we focus on correcting
187: the abundances of HE~1327$-$2326 for 3D effects to provide the most
188: accurate as possible stellar abundances.
189:
190: \section{Differential 3D$-$1D abundance analysis}
191: We use convection simulation of a metal-poor turnoff star (T$_{\rm
192: eff}{\simeq}6200$~K, $\log g=4.04$~[cgs], and solar chemical
193: composition scaled down to $\rm{[Fe/H]}=-3$) previously generated by
194: \citet{asplund01} as a time-dependent 3D hydrodynamical model
195: atmosphere to study the formation of the atomic and molecular features
196: observed in HE~1327$-$2326 under the assumption of LTE. The surface
197: convection simulation was carried out with the 3D, time-dependent,
198: compressible, explicit, Eulerian, radiative-hydrodynamical code by
199: \citet{nordlund82,stein98}. The physical domain of the simulation is
200: a cubic portion of stellar surface about $21{\times}21{\times}7$~Mm in
201: size, large enough to cover about ten granules horizontally and twelve
202: pressure scales in depth. The domain is discretized on a mesh with
203: $100{\times}100{\times}82$ numerical resolution. In terms of
204: continuum optical depth at $\lambda=5000$~{\AA}, the simulation
205: extends from $\log{\tau_{\rm{5000}}}\la-5$ down to
206: $\log{\tau_{\rm{5000}}}\ga6$. Open boundaries are employed vertically
207: and periodic ones horizontally.
208:
209: The adopted gravity of $\log g=4.04$ in the 3D simulation is
210: intermediate between what until recently were considered the two best
211: estimates of HE~1327$-$2326's surface gravity, i.e., $\log g=3.7$
212: (subgiant solution) or $\log g=4.5$ (dwarf). In solar-like stars,
213: absolute elemental abundances derived from lines of minority species
214: are essentially insensitive to the exact value of surface gravity
215: \citep[see also Tab.~9 and~10 in][]{Aokihe1327}. On the contrary,
216: abundances inferred from lines of majority species (e.g., Fe~II) and
217: molecules are more surface gravity dependent. In practice, such
218: surface gravity sensitivity of the derived abundances also changes
219: depending on whether one considers a 3D or 1D model atmosphere for the
220: spectral line formation calculations. Therefore, rather than directly
221: determining the chemical composition of HE~1327$-$2326 with the
222: exclusive use of the above 3D model, we instead compute
223: \emph{differential} abundances with respect to the results of a 1D
224: abundance analysis based on a 1D {\sc marcs} model atmosphere
225: \citep{gustafsson75,asplund97} with identical stellar parameters,
226: input data, and chemical composition as the convection simulation.
227: Figure~\ref{optical_depth} shows the temperature structure of the 3D
228: model atmosphere adopted here as a function of optical depth; the
229: temperature stratification of the corresponding 1D {\sc marcs} model
230: atmosphere is also plotted for comparison. For relatively small
231: changes in terms of surface gravity, such \emph{differential} 3D$-$1D
232: should in fact be only very marginally sensitive to $\log g$.
233:
234: While both the 3D simulation and the 1D {\sc marcs} model atmosphere
235: are constructed for a solar chemical composition scaled down to
236: $\rm{[Fe/H]}=-3$, in the line formation calculations we assume the
237: chemical abundance pattern to be the same as for HE~1327$-$2326
238: \citep{Aokihe1327} when computing ionization and molecular equilibria
239: and continuous opacities; as a rule, only the abundances of the trace
240: elements are varied when calculating line opacities. In fact, using a
241: scaled solar chemical composition with $\rm{[Fe/H]}=-3$ would lead to
242: overestimate the abundance of elements with low ionization potentials
243: and, in turn, the electron density, therefore affecting ionization
244: balance, continuous opacities, and, ultimately, line strengths. In
245: addition, when computing background opacities and electron density, we
246: adopt C, N, and O abundances midway between the values derived by
247: \citet{Aokihe1327} and \citet{o_he1327} for the subgiant and dwarf
248: solutions: this is done in order to compensate for the particular
249: choice of surface gravity in the 3D simulation and corresponding 1D
250: model.
251:
252: We compute flux profiles for all the spectral lines from neutral and
253: singly-ionized metals presently observed in HE~1327$-$2326. We also
254: consider a set of ``fictitious'' CH, NH, and OH (see
255: e.g., \citealt{collet06,collet07}) lines with varying lower excitation
256: potentials ($0$ to $3.5$~eV) and $\log{gf}$ values within the range
257: typical for the observed molecular bands in this study. From the full
258: 3D simulation we select a 60~minutes long (stellar time) sequence of
259: 30 snapshots equidistant in time. We decrease the horizontal
260: resolution of the simulation from $100{\times}100$ to $50{\times}50$
261: to reduce the computational load and time of the 3D line formation
262: calculations; we also interpolate all simulation snapshots to a finer
263: depth-scale in order to increase the resolution of the atmospheric
264: layers ($\log{\tau_{\rm{5000}}}{\la}2.5$) to improve the numerical
265: accuracy of the radiative transfer solution. For each line we solve
266: the radiative transfer equation for all grid-points across the
267: horizontal plane and along 33 directions (4 $\mu$-angles, 8
268: $\phi$-angles, and the vertical), after which we perform a
269: disk-integration and a time-average over all selected 3D snapshots.
270: We use 70 wavelength points to resolve each line profile. The source
271: function for lines and continuum is approximated with the Planck
272: function at the local temperature and scattering is treated as true
273: absorption. Continuous opacities come from the Uppsala package
274: \citep[][and subsequent updates]{gustafsson75}, the equation-of-state
275: from \citet[][and subsequent updates]{mihalas88}, the partition data
276: for atoms and ions from \citet{irwin81} and for molecules from
277: \citet{sauval84}. The same radiative transfer solver and input data
278: are also used for spectral line formation calculations with the 1D
279: model atmosphere. In addition, for the 1D case, we adopt a
280: micro-turbulence of $\xi=1.6$\,km\,s$^{-1}$ for the turnoff star. We
281: stress that no micro- nor macro-turbulence parameters enter the 3D
282: spectral line synthesis calculations: only the velocity fields
283: predicted by the actual 3D simulation are used to reproduce
284: non-thermal line broadening and asymmetries associated with convective
285: Doppler shifts.
286:
287: For each line, we quantify the impact of the 3D model in the analysis,
288: by varying the abundance of the trace element
289: independently in the 1D and 3D line formation calculations
290: until the measured equivalent width is matched:
291: the difference between the 3D and 1D derived abundances then
292: defines the 3D$-$1D abundance correction for the given line.
293:
294:
295:
296:
297: Before proceeding, we caution that the actual 3D$-$1D abundance
298: corrections depend in part on the choice of 1D model atmosphere in the
299: differential abundance analysis. Ideally, one should compute the
300: differential corrections using a 1D model atmosphere that relies on
301: exactly the same input physics and treatment of radiative transfer as
302: the convection simulation. Although the input physics of the {\sc
303: marcs} model atmosphere adopted here does not depart substantially
304: from the one used in the 3D case, some differences in practice still
305: exist. In particular, contrary to the 3D hydrodynamical simulation,
306: scattering is treated correctly as such and not as true absorption in
307: the 1D {\sc marcs} model. Also, the two kinds of models rely on
308: slightly different equations-of-state. Finally, while the sources of
309: continuous and line opacities are namely the same
310: \citep{gustafsson75,kurucz92,kurucz93}, during the 3D simulation
311: opacities are grouped in four opacity bins to ease the computational
312: burden of the radiative transfer calculations. To estimate what the
313: effect of the particular choice of model atmosphere is on the 3D$-$1D
314: corrections, we also performed a 3D$-$1D differential abundance
315: analysis based on a 1D {\sc atlas} model atmosphere \citep{kurucz93}
316: with same stellar parameters and chemical composition as the 3D model.
317: Because of the slightly higher temperature of the 1D {\sc atlas} model
318: atmosphere in line formation regions, the resulting 3D$-$1D abundance
319: corrections for all elements are systematically shifted $0.05$ to
320: $0.1$\,dex downwards (i.e., toward more negative values) with respect
321: to the differential analysis based on the 1D {\sc marcs} model. We
322: finally note that there is no dependence of the 3D$-$1D corrections on
323: the adopted micro-turbulence of the 1D model (see
324: \citealt{Aokihe1327}), nor any significant dependency on
325: MLT. Molecular lines, for which the 3D$-$1D corrections are the largest,
326: form in the upper layers of the photosphere where the type of
327: mixing-length prescription is not expected to have an effect on the
328: atmospheric structure.
329:
330:
331: \section{Chemical Abundances}
332: We now describe the abundances of individual elements as far as the
333: new analysis is concerned. Details on the 3D$-$1D corrections and our
334: new technique to apply those to abundances derived from molecular
335: features are given where appropriate. The new abundance pattern is
336: summarized in Table~\ref{Tab:Abundances}, and in
337: Figure~\ref{abund_pattern} our results for HE~1327$-$2326 are compared
338: with those for HE~0107$-$5240, the other star with
339: $\mbox{[Fe/H]}<-5$. We also refer the reader to \citet{Aokihe1327};
340: they presented a detailed discussion of the HE~1327$-$2326's chemical
341: pattern as well as a comparison with other metal-poor stars. A
342: detailed uncertainty analysis for the abundances of the star was also
343: presented (their Table~10) and will not be repeated here.
344:
345: \subsection{Line Data and Analysis Techniques}
346: Atomic line data for the abundance analysis have been taken from
347: \citet{Aokihe1327} and references therein, substituted with lines from
348: \citet{HE0107_ApJ} and references therein, and from the Vienna Atomic Line
349: Database (VALD; \citealt{vald}). Molecular line data have been taken
350: from B. Plez (priv. communication) for CH and
351: \citet{gillis_ohlinelist} for OH. For NH we are using the linelist of
352: \citet{kurucz_nh}, but add a 0.4\,dex to the gf-values to fit the
353: solar spectrum (see \citealt{Aokihe1327} for details). For CH and OH we use
354: dissociation energies of 3.47\,eV and 4.39\,eV, respectively
355: \citep{CH_OH_diss_en}. For NH we employ 3.37\,eV \citep{NH_diss_en}.
356:
357:
358: Equivalent widths are measured from the spectrum by fitting Gaussian
359: profiles to the generally very weak lines. Upper limits ($3\sigma$)
360: for equivalent widths are derived from $\sigma=w n_{pix}^{1/2}/(S/N)$
361: (where $w$ is the pixel width, $n_{pix}$ is the number of pixels
362: across the line (e.g., where a weak line would reach the continuum if
363: a line was present), and $S/N$ is per pixel; \citealt{bohlin}). The
364: measurements and upper limits for HE~1327$-$2326 are presented in
365: Table~\ref{Tab:Eqw}. Apart from hydrogen and molecular lines of CH,
366: NH, and OH we now find 44 atomic lines in the spectrum of eight
367: elements (Na, Mg, Al, Ca, Ti, Fe, Ni, Sr) and are able to determine
368: upper limits of nine additional elements. The previous analysis by
369: \citet{Aokihe1327} identified 22 atomic lines.
370:
371: \subsection{Iron}
372:
373: In the Subaru spectrum seven weak Fe\,I lines were detected
374: \citep{Aokihe1327}. The strongest line was measured to have an
375: equivalent width of 6.8\,m{\AA}. As already reported in Frebel et
376: al. (2006) the new data confirm the presence of six of the seven lines
377: (the wavelength of one line is not covered by our VLT
378: spectrum). Additionally, we detect four new lines and report tentative
379: detections of another three. This brings the total of detected lines
380: to 11 Fe\,I lines from which the metallicity is deduced. We note that
381: the average that includes the abundances of the tentative Fe\,I lines
382: agrees within 0.03\,dex with the average from the 11 detected
383: lines only. Figure~\ref{feI_plot} shows the detected and tentatively
384: detected 13 Fe\,I lines in the VLT spectrum.
385:
386: Based on the new equivalent width measurements together with the one
387: from Aoki et al. (2006) for the line that is not covered, we find
388: HE~1327$-$2326 to have a 1D LTE Fe abundance\footnote{This value has
389: essentially already been published in \citet{o_he1327}. We simply give
390: more details on the derivation of this value, and also report the
391: detection of additional Fe\,I lines.} of $\mbox{[Fe/H]} =-5.66$, the
392: same as the previous result reported in \citet{Aokihe1327}.
393:
394:
395: In this work we are now applying 3D$-$1D LTE corrections to our 1D LTE
396: abundances. Within the LTE framework, the average 3D$-$1D Fe
397: abundance correction for Fe\,I lines is negative ($-0.24$;
398: \citealt{collet06}). Thus, the Fe\,I abundance of HE~1327$-$2326 is
399: reduced to $\mbox{[Fe/H]}=-5.90$. To illustrate why the 3D$-$1D Fe
400: abundance correction from Fe\,I lines is negative, in
401: Figure~\ref{feifraction} we compare the fraction of neutral to total
402: iron in the 3D and 1D model atmospheres. While in the 1D model iron
403: is essentially all ionized to Fe\,II, in 3D a significant fraction of
404: iron is in neutral form in the upper photospheric layers because of
405: the cooler surface temperature stratification predicted by the 3D
406: metal-poor convection simulation (see e.g., \citealt{asplund01}).
407: Hence, at a given Fe abundance and under the assumption of LTE, Fe\,I
408: lines appear stronger in the 3D line formation calculations than they
409: do in 1D. This also implies that the same equivalent width of a Fe\,I
410: line is matched by a lower Fe abundance in 3D than in 1D.
411: Alternatively, one can also show that the same Fe\,I line forms at
412: different depths in the 3D and 1D models. Figure~\ref{feicontrib}
413: shows the contribution function
414: $\cal{C}_{\rm{I}}(\tau_\lambda)=S_{\lambda}(\tau_\lambda)\exp{(-\tau_\lambda)}$
415: to the outgoing intensity at line center in the vertical direction for
416: the Fe\,I line at $\lambda=3859.9$~{\AA} in the two models. It is
417: apparent from the figure that, in 3D, the main contribution to the
418: outgoing vertical intensity for the Fe\,I line is given by layers
419: lying higher up in the atmosphere where the temperature differences
420: between 3D and 1D structures are larger.
421:
422: We note that any possible non-LTE effects present within the 3D Fe\,I
423: line formation calculations are currently unexplored, but likely to be
424: positive \citep{asplund_araa}. For completeness, we remind the reader
425: that in previous works the measured Fe\,I abundance was corrected by
426: +0.2\,dex to account for 1D non-LTE effects in Fe\,I
427: (\citealt{collet06} predict for Fe\,I a much stronger departure from
428: LTE). In principle, Fe\,II lines should be the preferred Fe abundance
429: indicator as they are usually expected to be less sensitive to
430: departures from LTE. However, even the strongest Fe\,II lines are
431: expected to be extremely weak in HE~1327$-$2326 given its stellar
432: parameters and composition. In fact, no Fe\,II line could be detected
433: in either the Subaru or in the new VLT spectrum. The 3D$-$1D corrected
434: upper limit is $\mbox{[Fe\,II/H]}<-5.40$. We can also place an upper
435: limit on the Fe\,I 3D non-LTE effect of $\lesssim0.5$. This is derived
436: under the assumption that there is an ionization equilibrium of Fe\,I
437: and Fe\,II if the surface gravity of $\log g=3.7$, as derived from an
438: isochrone, is correct, and that Fe\,II lines are not affected by
439: non-LTE. In the absence of Fe\,II lines, Ca is the only element
440: detectable in two ionization stages. As mentioned above, this offers
441: the possibility of using Ca\,I/II as a surface-gravity indicator
442: \citep{korn_fs3}.
443:
444: \subsection{CNO elements}
445:
446: Carbon, nitrogen and oxygen abundances for HE~1327$-$2326 were derived
447: by \citet{Aokihe1327} (C and N; from the Subaru spectrum) and then by
448: \citet{o_he1327} (C and O; from the VLT spectrum). For the latter 1D LTE
449: analysis, corrections were suggested for all three elements to account
450: for 3D model atmosphere effects. The estimated 3D$-$1D corrections were
451: mostly based on works described in \citet{asplund_araa} and references
452: therein.
453:
454:
455: Molecular features consist of many different lines with a variety of
456: excitation potentials. As 3D$-$1D abundance corrections depend in
457: general on the excitation potentials of the contributing lines, the
458: final 3D$-$1D correction to the 1D abundance of the trace elements in the
459: molecular features has to be an appropriate average of the 3D$-$1D
460: abundance corrections derived for individual lines. \citet{collet06}
461: computed 3D$-$1D corrections for a few fictitious lines of CH, NH, and
462: OH with selected excitation potentials. By using fictitious lines,
463: the 3D$-$1D corrections to the CNO abundances could be studied without
464: resorting to a full spectrum synthesis with a 3D model atmosphere.
465: Here we extend the study by \citet{collet06} to consider a more
466: comprehensive range of lower excitation potentials and $\log{gf}$
467: values of the CH, NH, and OH lines. Figures~\ref{ch_oh_corr}
468: and~\ref{nh_corr} show the 3D$-$1D LTE corrections to the CNO
469: abundances derived from the molecular lines as a function of line
470: strength and excitation potential. In the case of CH and OH lines,
471: 3D$-$1D LTE C and O abundances are also plotted as a function of the
472: corrections to oxygen and carbon abundances, respectively. The
473: magnitudes of abundance corrections derived from molecular lines
474: depend in fact not only on the different photospheric structures of
475: the 3D and 1D model atmospheres but also on the overall chemical
476: composition and details of the molecular equilibrium. In particular,
477: the line strengths of CH and OH lines are sensitive to both the carbon
478: \emph{and} oxygen abundances assumed in the spectral line formation
479: calculations because of the competing formation of CO molecules.
480: Temperatures in the upper photospheric layers of the metal-poor 3D
481: model atmosphere adopted here are on average considerably lower than
482: in the corresponding 1D model. At low temperatures (${\la}3700$~K),
483: significant fractions of C and O become locked in CO molecules. Thus,
484: for instance, if one were to increase slightly the \emph{oxygen}
485: abundance in the line formation calculations with the metal-poor 3D
486: model, that would then lead to the formation of more CO and therefore
487: reduce the amount of \emph{carbon} available for other molecules such
488: as CH. Hence, at a fixed carbon abundance, but with higher oxygen
489: abundance, CH lines would become \emph{weaker}, and the associated
490: 3D$-$1D LTE C abundance corrections would in turn be \emph{larger}
491: (i.e., more negative). The same argument holds for the variation of
492: the equivalent widths of OH lines as a function carbon abundance.
493:
494:
495: In order to apply the 3D$-$1D corrections to the CNO abundances in a
496: more suitable way, we developed a new technique. We first interpolate
497: in between the 3D$-$1D abundance corrections given as a function of
498: excitation potential and line-strength to obtain appropriate
499: corrections for all individual lines that comprise the molecular
500: feature of interest. The abundance corrections so computed (in
501: \emph{dex}) are then \emph{subtracted} (the negative corrections) from
502: the $\log{gf}$ values for all lines in our linelist. The observed
503: spectrum is then re-synthesized with the new ``3D-adjusted'' linelist.
504: We call this procedure ``3D-aided spectrum synthesis''.
505:
506: Before we proceed further, however, a number of tests have to be made:
507: i) We re-determine the ``1D'' abundances from the new VLT spectrum for
508: a comparison with previously published values. ii) We synthesize the
509: spectrum with the new 3D-adjusted linelist. iii) We compare the
510: synthetic fit derived from the original ``1D'' linelist with the one
511: computed with the new 3D-adjusted linelist to explore the influence of
512: the different 3D$-$1D corrections on different excitation potentials. The
513: results are shown in Figures~\ref{CH_comp}, \ref{NH_comp}, and
514: \ref{OH_comp}.
515:
516:
517: We also caution the reader that the here computed
518: average 3D$-$1D abundance corrections are only valid for the
519: considered wavelength range (because of, e.g., the specific combination of
520: lower excitation potentials of the contributing lines, the
521: continuous opacities at these wavelengths, etc.),
522: and may thus not apply in general to other molecular features
523: in different regions of the spectrum.
524: Finally, we also remind that any possible 3D non-LTE
525: or non-equilibrium chemistry effects remain unexplored at this point in time,
526: but are expected to be positive \citep{asplund_araa} and therefore
527: reduce (i.e., make less negative) the effective 3D$-$1D
528: correction to CNO abundances.
529:
530:
531: \subsubsection{Carbon}
532:
533: The 1D C abundance was already determined from the VLT spectrum by
534: \citet{o_he1327}. In the meantime, however, a new CH linelist has
535: become available (B. Plez, priv communication, but see
536: \citealt{he1523} for some technical details) which yields much
537: improved syntheses for metal-poor stars. Hence, we use the new
538: linelist, and derive the C abundance from the G-band head at
539: $\sim4313$\,{\AA}. The newly derived 1D abundance\footnote{Where $\log
540: \epsilon ({\rm X}) = \log(N_{\rm X}/N_{\rm H}) + 12.0$} of $\log
541: \epsilon = 6.90$ agrees very well with previous values of $\log
542: \epsilon = 6.90$ \citep{HE1327_Nature}, $\log \epsilon = 6.99$
543: \citep{Aokihe1327} and $\log \epsilon = 6.86$ (from various CH
544: features; \citealt{o_he1327}).
545:
546:
547:
548: The excitation potentials of the participating lines in the G-band
549: spectral region are of a similar level (top panel in the figure), that
550: results in a almost constant 3D$-$1D correction across that region.
551: Hence, the variation in abundance between the 1D and 3D-aided
552: synthetic spectra does not exceed $\sim0.05$\,dex (see the second
553: panel from the top in Figure~\ref{CH_comp}). The fit with the
554: 3D-adjusted linelist to the individual lines of the G-band is,
555: however, slightly improved due to the changes in line strengths
556: (compare with second panel from the bottom in the figure). The
557: overall fit to the data is excellent, and makes it easy to accurately
558: determine a mean 3D$-$1D correction of C of $-0.69$ (bottom panel of
559: figure). The final C abundance is $\mbox{[C/Fe]}=3.78$\,dex. Finally,
560: concerning the $^{12}$C/$^{13}$C ratio, unfortunately the VLT data do
561: not cover appropriate wavelengths to re-measure this important ratio.
562:
563:
564: \subsubsection{Nitrogen}
565: The 1D N abundance was determined from the NH band at 3360\,{\AA} for
566: a comparison with the value obtained from the Subaru discovery
567: spectrum. Our new fit is much improved due to increased $S/N$ in this
568: spectral region (now $\sim200$ red-ward of the band). The newly
569: derived 1D abundance of $\log \epsilon (N)=6.79$ agrees with the
570: values of 6.68 and 6.83 presented in
571: \citet{HE1327_Nature}\footnote{For NH, a dissociation energy of 3.47
572: was used.If a value of 3.37 is employed the abundances becomes $\log
573: \epsilon (N)=6.78$.} and \citet{Aokihe1327}.
574:
575: Figure~\ref{NH_comp} shows the excitation potentials of the molecular
576: lines that comprise the NH fit. There is a gradual change to higher
577: excitation potentials from 3357 to 3365\,{\AA}. From the comparison of
578: the 1D and 3D-aided synthetic spectra it can be seen that the change
579: in excitation potentials results in the 3D-aided synthetic spectrum
580: having slightly stronger lines blue-ward ($\sim+0.1$\,dex) of the main
581: NH feature at 3360\,{\AA} and slightly weaker ($\sim-0.2$\,dex) lines
582: on the red side (see second panel from top in Figure~\ref{NH_comp}).
583: Irrespective of any resulting abundance, this effect causes an
584: improvement of the new fit to the observed NH spectral region compared
585: with the original 1D linelist fit.
586:
587: This effect can not be ``fixed'' with a simple abundance offset since
588: it would be the same for all participating lines irrespective of
589: excitation potential. We note that this apparent improvement can not
590: be accounted to continuum normalization issues since the continuum
591: points of the synthetic spectra match the observed spectrum very
592: well. If the continuum of the observed spectrum were determined
593: incorrectly, those points should not be reproduced by the fit. We
594: checked the Subaru spectrum, and as can be seen in Figure~7 of Aoki et
595: al. (2006), the effect of the fit significantly underproducing the
596: observed spectrum on the blue side, and very slightly overproducing it
597: on the red side (their best fit is aimed at reproducing the red side)
598: is present there as well. From this comparison we conclude that it is
599: not a data reduction effect. We also tested if that effect would
600: result from the dependence of the 3D$-$1D correction on
601: line strength, but found it to be negligible. We then investigated the
602: influence of the effective temperature on relative line strengths,
603: which in turn depend on their excitation potentials. We changed the
604: effective temperature of the model atmosphere by $-400$\,K, and fitted
605: the blue side of the NH band again. On the red side the new fit did
606: almost match the fit with the original model atmosphere. The
607: difference on the red side was only about one third of the difference
608: between the 1D and 3D-aided fits. It follows that the apparent
609: improvement of the fit to the data is still present, and can not be
610: explained with a temperature effect only.
611:
612: It may thus be possible that the 3D effect can be directly seen in the
613: NH band, due to a fortuitous distribution of excitation potentials. The
614: overall 3D$-$1D correction over the NH range is $-0.69$\,dex. A larger, or
615: smaller, correction would, again, result in a less optimal
616: reproduction of the observed data, thus suggesting that such a large 3D
617: correction is correct and necessary to obtain a more accurate N
618: abundance for the star. The final N abundance is $\mbox{[N/Fe]}=4.28$.
619:
620:
621: We note though, that we cannot exclude possible systematic effects in
622: the present 3D$-$1D differential analysis due to neglected departures from LTE and
623: from chemical equilibrium at the local temperature. Hypothetically, if
624: such departures affected all 3D$-$1D corrections for individual molecular
625: lines (with different excitation potentials) by an approximately
626: constant factor, then a re-scaling of all gf-values in the 3D-adjusted
627: linelist by the inverse of the same factor would still lead to a very
628: good fit of the molecular band but with a different final 3D
629: correction. The success in the present 3D analysis thus lies in
630: the fact that the relative line-to-line scatter in the abundances derived from
631: individual lines in the molecular feature is very much reduced
632: compared with the 1D analysis.
633:
634: Often, synthetic fits to observed molecular bands are not
635: fully satisfactory. This is always accounted for in terms of uncertainties in the
636: molecular line data used to generate the synthetic spectra. While this
637: may still be true in the case of NH, the uncertainties would be the same for both the
638: original 1D and the 3D-adjusted linelists, and only affects our conclusions in
639: the sense that the molecular NH data, i) either is better than
640: previously thought, or ii) a different, yet unexplored, effect is
641: responsible for the better fit to the observed data. The application
642: of 3D$-$1D corrections in the same way as presented here should be further
643: investigated for other metal-poor stars to test whether this is a more
644: general effect or limited to the extreme case of HE~1327$-$2326. We
645: note that while the 3D$-$1D corrections for CNO elements HE~0107$-$5240 are
646: similarly large, the $S/N$ ratio of the data at $\sim3360$\,{\AA} is
647: not yet sufficient to infer a similar effect for this object (N. Christlieb
648: et al. 2008, in prep.).
649:
650: \subsubsection{Oxygen}
651:
652: The O abundance determination from the VLT data was already described
653: in detail by \citet{o_he1327}. Apart from the offset in abundance
654: caused by the application of the 3D$-$1D correction, there generally is
655: good agreement between the two fits (see the second panel from the top
656: in Figure~\ref{OH_comp}). The overall 3D$-$1D correction for OH as
657: determined in this work is $-0.72$\,dex, and thus a slightly lower than what was
658: estimated ($-0.9$\,dex) in \citet{o_he1327}. The final O abundance is
659: $\mbox{[O/Fe]}=3.42$.
660:
661:
662: \subsection{Lithium and Beryllium}
663: The Li\,I doublet at 6707 {\AA} is {\it still} not detected in this
664: relatively unevolved star. From the extremely high $S/N$ ($\sim 600$
665: per pixel) data we are able to derive a new $3\sigma$ upper limit of
666: $\log \epsilon ({\rm Li}) = A({\rm Li})<0.70$ from our 1D analysis;
667: this is further reduced to $A({\rm Li})<0.62$ by application of the 3D
668: correction. Figure~\ref{li_plot} shows the Li spectral regions
669: overplotted with synthetic spectra of different abundances. This is
670: significantly lower (i.e., by 1\,dex) than the previous limit of
671: $\log\epsilon ({\rm Li})=1.6$ \citep{Aokihe1327}. Other unevolved
672: metal-poor stars with $\mbox{[Fe/H]}\sim -3.5$ have Li values around
673: $\log\epsilon ({\rm Li})\sim2.0$ (e.g.,
674: \citealt{ryan_postprim}). These objects have been used to infer the
675: primordial Li abundance, which can be compared to the value expected
676: from the baryon-to-photon ratio inferred from WMAP data \citep{WMAP},
677: $\log\epsilon=2.6$. The discrepancy between the WMAP result and the
678: observed Li abundance in metal-poor stars has amply been established,
679: so we shall not further elaborate on it here. We only note that even
680: though the Li \,I lines, in general, are prone to non-LTE effects,
681: they are likely to be far too small ($\sim0.1$\,dex) to bridge the gap
682: between the Li abundance of HE~1327$-$2326 and that of other
683: metal-poor star as well as the WMAP-inferred value. Also,
684: gravitational settling of lithium as a possible solution to the
685: discrepancy is not sufficient \citep{korn_fs3} It is thus very
686: surprising that the upper limit of HE~1327$-$2326 is not only lower
687: than the WMAP result but also considerably lower (by more than 1\,dex)
688: than values derived from other metal-poor star with similar
689: evolutionary status. This means that either all of this star's Li was
690: destroyed during its lifetime on the main-sequence, or the star was
691: born from a Li-depleted material.
692:
693: \citet{piau} developed a picture where a large fraction of the early
694: interstellar medium (ISM) is ``recycled'' through massive rotating
695: Pop\,III stars with the consequence that Li depleted material is
696: deposited back into the ISM through stellar winds. The first low-mass
697: stars would primarily form in the wake of those Li-deficient, but CNO
698: rich winds (\citealt{meynet2005}, but also see above). The new lower
699: limit of Li (together with high CNO) in HE~1327$-$2326 is in line with
700: this theory. If this scenario were correct, any (future) observations
701: of yet-to-be-found dwarfs with metallicities of
702: $\mbox{[Fe/H]}\lesssim-4.0$ should indeed reveal strong
703: Li-depletion. Other options for the Li deficiency have been explored
704: in \citet{Aokihe1327}, and include rapid rotation and HE~1327$-$2326
705: being a member of a binary system. As for the binary system, continued
706: radial velocity monitoring does not (yet) suggest the presence of any
707: companion.
708:
709: Beryllium is another light, fragile element observed in very
710: metal-poor turn-off stars. Standard Big-Bang nucleosynthesis models do
711: not predict a significant yield of this element. Be is destroyed at
712: $T=3.5 \times 10^{6} $\,K, and is thus slightly more robust than
713: Li. Hence, if the Be abundance, or a strong upper limit of the
714: abundance, is determined for HE~1327$-$2326, that could provide
715: important clues as to why the Li abundance in the star is so extremely
716: low.
717:
718: In Figure~\ref{be_plot} we show the spectral region of the two
719: strongest Be\,II lines in the near UV. As can be seen there is a
720: (noise) peak at the right wavelength of the $3131$\,{\AA}
721: line. However, the $S/N$ in this spectral region is low, so that we
722: cautiously adopt an upper limit of $A({\rm Be})<-1.2$. This value is
723: as low as the Be abundances measured for extremely metal-poor stars
724: (e.g., \citealt{boesgaard_novicki}). Thus, we can not derive any
725: definitive conclusion from the upper limit of the Be abundance for the
726: depletion of Li in this object.
727:
728:
729: \subsection{Other elements}
730: For the lines that were previously measured by \citet{Aokihe1327},
731: there generally is good agreement between their and our equivalent
732: widths (there is no significant offset). Figure~\ref{eqw_comp} shows a
733: comparison of the two measurements for the lines in common. We detected
734: many more lines in the spectrum which confirm the abundances as
735: derived from the Aoki et al. lines.
736: We can also report the first detection of four Ni lines in HE~1327$-$2326
737: (see Figure~\ref{Ni_plot}). Previously only an upper limit could be
738: derived, but now four weak lines are detected. The equivalent width
739: measurements and upper limits are presented in Table~\ref{Tab:Eqw},
740: while the resulting abundances are given in
741: Table~\ref{Tab:Abundances}.
742:
743: We also re-determined upper limits for Cr, Co, and Zn. The limits are
744: much tighter (by $\sim$0.4 to 0.8\,dex) than obtained from the Subaru
745: spectrum, due to the increased $S/N$ ratio of the VLT data
746: particularly in the blue spectral region. New upper limits for Be, Sc
747: and V could also be derived. Our new upper limits, together with the
748: detection of Ni, are of interest when comparing the overall abundance
749: pattern with theoretical calculations of SN yields. We will return to
750: this point in \S~\ref{disc}.
751:
752: Unfortunately, our spectrum does not cover suitable lines of Mn, Si,
753: S, Sr, Ba and Eu. That leaves us with no information about Si and S,
754: and only the previous values from the Subaru spectrum in the case of
755: Mn, Sr and Ba. For Mn, the second strongest line yields the same upper
756: limit as has been previously derived from the strongest line at
757: 4020\,{\AA}. The same is the case for Ba. A weaker line at 4943\,{\AA}
758: line is covered by the red setting. Figure~\ref{ba_plot} shows the Ba
759: spectral regions overplotted with synthetic spectra of different Ba
760: abundances. The inferred upper limit yielded almost the same as
761: was reported in \citet{Aokihe1327} based on the resonance line at
762: 4554\,{\AA}. The resulting ratio is $\mbox{[Sr/Ba]}_{\rm 3D}>-0.29$.
763:
764:
765: This limit is already higher than what is found in typical r-process
766: enhanced metal-poor stars ($-0.4$ to $-0.5$;
767: \citealt{Sneden1998,Hilletal:2002, he1523}). The value is certainly
768: inconsistent with ratios found in s-process-rich stars, which have
769: much lower [Sr/Ba] values. Large excesses of light neutron-capture
770: elements are also found in extremely metal-poor stars, whose abundance
771: ratios of light to heavy neutron-capture elements, such as Sr/Ba, are
772: significantly higher than the ratios of the known r-process yields
773: (e.g., \citealt{aoki05}). The detailed abundance patterns of such
774: stars have been recently determined by \citet{honda06,honda07}. If the
775: Sr of HE~1327$-$2326 were related to the process that provided
776: extremely metal-poor stars with light neutron-capture elements, the Sr
777: overabundance could be a key to understanding the nucleosynthesis
778: mechanism of the progenitor object (e.g., \citealt{froehlich}). Such
779: process is associated with explosive nucleosynthesis, and is thus
780: compatible with the pre-enrichment scenario by a previous generation
781: SN as explanation for the overall abundance pattern rather than
782: through mass transfer across a binary system (see below). As for Eu,
783: since the strongest line at 4129\,{\AA} is not covered, only a
784: meaningless upper limit of $\mbox{[Eu/Fe]}<4.1$ could be derived from
785: a line at 6645\,{\AA}.
786:
787: For completeness, we also provide the upper limits derived from the
788: O\,I triplet and forbidden O\,I lines, as presented in
789: \citet{o_he1327}. Upper limits on the equivalent width and abundances
790: are listed in Tables~\ref{Tab:Eqw}~and~\ref{Tab:Abundances}.
791:
792:
793: \section{Discussion and Conclusions}\label{disc}
794: We have presented an abundance analysis of the currently most Fe-poor
795: star, HE~1327$-$2326, based on the currently best available
796: observational data obtained with VLT/UVES. In an attempt to obtain the
797: best possible abundances from those data, we have combined our 1D
798: abundances with state-of-the-art 3D$-$1D model atmosphere
799: corrections. Hence, these abundances supersede our previous
800: measurements. We recommend using these new 3D$-$1D corrected LTE
801: abundances for future modeling of the abundance pattern of
802: HE~1327$-$2326.
803:
804: Figure~\ref{abund_summ} summarizes all abundances measurements for the
805: star published to date. We plot the ``best available'' values at the
806: time, i.e., 1D abundances with non-LTE corrections where available in
807: the older cases, and LTE abundances with differently determined 3D
808: corrections in the newer studies. The Fe abundance differs in these
809: works, so that we compare [X/H] values which leave Fe as ``free
810: parameter''. The 3D$-$1D corrected abundances are somewhat shifted
811: towards lower values compared with the 1D abundances. The biggest
812: differences are found among the CNO elements ($\Delta\sim-0.7$)
813: whereas for other elements the differences between 1D and 3D analyses
814: are less pronounced. We note that when comparing [X/Fe] measurements,
815: because of the different [Fe/H] values, some of those differences in
816: [X/H] are canceled out. This fact should be kept in mind when
817: comparing the overall abundance pattern with theoretically derived
818: abundances predictions based on SN yields.
819:
820: Several groups of authors have already employed the (in the most cases
821: the old) abundances to test current scenarios for the chemical yields
822: of the first-generation SNe. In order to compare how well all these
823: predictions reproduce our new 3D$-$1D corrected abundances presented
824: here, we plot them all in Figure~\ref{models}. We note that all the
825: above interpretations assume that the star is not a member of a binary
826: system. This is indeed supported by ongoing radial velocity
827: measurements of HE~1327$-$2326. Two measurements, one from 2007
828: (D. Lai, priv. comm.) and one from 2008 agree with the previous data
829: \citep{o_he1327} and does not indicate any significant change of the
830: radial velocity.
831:
832: \citet{iwamoto_science} were the first authors to compare the 1D
833: abundances of HE~1327$-$2326 \citep{HE1327_Nature} with their yields
834: of a 25\,M$_{\odot}$ Pop\,III hypernova, for given mixing and fallback
835: parameters. Particularly, the high CN abundance together with the
836: elevated Mg abundance were well reproduced by their yields. An updated
837: fit was presented in \citet{tominaga07} after the O abundance (with
838: estimates for the 3D$-$1D correction) of HE~1327$-$2326 had been determined
839: \citep{o_he1327}. Hence, this model now particularly well fits our new CNO
840: abundances. Na, Mg, Al, Ca and Fe abundances are well reproduced,
841: too. The new abundance of Ni (3D$-$1D corrected) also agrees with
842: their predicted Ni yield.
843:
844: \citet{meynet2005} investigated mass loss by a massive
845: (60\,M$_{\odot}$) rotating Pop\,III star, and the chemical yields of
846: the subsequent SN explosion. They find that from the combined yields
847: of the stellar wind and the SN, the highly elevated 1D CNO pattern of
848: HE~1327$-$2326 \citep{HE1327_Nature} and that of other metal-poor
849: stars can qualitatively be reproduced. In their ``wind-only'' model, a
850: high N abundance can be achieved, although all other elements
851: considered (except Al) are produced in insufficient quantities to
852: reproduce the observed abundance pattern.
853:
854: \citet{heger2008} have computed new chemical yields for a range
855: (10--100\,M$_{\odot}$) of Pop\,III SNe. They are fitting the observed
856: abundances of HE~1327$-$2326 \citep{Aokihe1327} with either the yield
857: of a single SN or the integrated yields of a range of SNe for a given
858: initial mass function. For various progenitor masses (mass ranges),
859: explosion energies, and mixing parameters the abundance pattern can be
860: roughly reproduced. The best fit model for a single SN requires a
861: 21.5\,M$_{\odot}$ progenitor, while the best fit model for integrated
862: yields stems from the range 15-35\,M$_{\odot}$ of
863: progenitors. Overall, this is a similar progenitor mass as
864: \citet{iwamoto_science} suggested. In Figure~\ref{models} we show the
865: ``best single stars'' fit (Panel A in their Figure~14;
866: \citealt{heger2008}) in comparison with our new abundances. Overall,
867: the fit is in reasonable agreement with the data and provides an
868: interesting counterpart to the work of \citet{tominaga07}.
869:
870:
871: Recently, \citet{dtrans} used C and O abundances of metal-poor stars
872: for a comparison with theoretical predictions for the critical
873: metallicity that is required to facilitate low-mass star formation in
874: the early Universe. Based on the theory of fine-structure line
875: cooling, it was predicted that all stars with
876: $\mbox{[Fe/H]}\lesssim-4.0$ should have elevated C and/or O abundances
877: and have $D_{\rm trans}=\log(10^{\rm{[C/H]}}+10^{\rm{[O/H]}}) >-3.5$\,dex. The
878: new 3D$-$1D corrected abundances of HE~1327$-$2326 presented here lead
879: to $D_{\rm trans}=-2.18$\,dex, well above the critical value.
880:
881:
882: \section{Outlook}\label{outlook}
883:
884: To improve our understanding of the conditions of the early Universe,
885: and to put all the above theories to further critical observational
886: tests, clearly, more stars are needed with metallicities of
887: $\mbox{[Fe/H]}\lesssim-4$. \citet{he0557} recently reported the
888: discovery of a giant with $\mbox{[Fe/H]}=-4.75$. In particular, more
889: giants should be sought at the lowest metallicities because their
890: (lower) temperature, at a given abundance, allows for stronger lines
891: that can be detected more easily. The Fe lines in the subgiant
892: HE~1327$-$2326 are only barely detectable because of its hotter
893: temperature. Further ultra-metal-poor stars to be discovered in the
894: future will likely be fainter than HE~1327$-$2326 ($B\sim14.0$\,mag)
895: and even HE~0107$-$5240 ($B\sim15.6$\,mag), which imposes yet another
896: challenge when it comes to obtaining the necessary high $S/N$ spectra to
897: measure very weak spectral lines particularly in the blue spectral
898: region. Detailed studies of larger samples of stars at
899: $\mbox{[Fe/H]}<-5.0$ will therefore require 30\,m class telescopes,
900: which will make possible such difficult but extremely important
901: observations.
902:
903:
904: \acknowledgements
905: We thank Martin Asplund for discussions on abundance
906: analyses based on 3D model stellar atmospheres and an anonymous
907: referee for valuable suggestions regarding our new 3D$-$1D abundance
908: correction technique. A.~F. thanks Uppsala Astronomical Observatory
909: for its hospitality during parts of the write-up of this paper. She
910: acknowledges support through the W.~J.~McDonald Fellowship of the
911: McDonald Observatory. N.~C. acknowledges financial support by Deutsche
912: Forschungsgemeinschaft through grants Ch~214/3 and Re~353/44. He is a
913: Research Fellow of the Royal Swedish Academy of Sciences supported by
914: a grant from the Knut and Alice Wallenberg Foundation.
915:
916: \textit{Facilities:}
917: %\facility{VLT:Kueyen (UVES)}
918: VLT:Kueyen (UVES)
919:
920: \begin{thebibliography}{34}
921: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
922:
923: \bibitem[{{Aoki} {et~al.}(2006){Aoki}, {Frebel}, {Christlieb}, {Norris},
924: {Beers}, {Minezaki}, {Barklem}, {Honda}, {Takada-Hidai}, {Asplund}, {Ryan},
925: {Tsangarides}, {Eriksson}, {Steinhauer}, {Deliyannis}, {Nomoto}, {Fujimoto},
926: {Ando}, {Yoshii}, \& {Kajino}}]{Aokihe1327}
927: {Aoki}, W., {Frebel}, A., {Christlieb}, N., {Norris}, J.~E., {Beers}, T.~C.,
928: {Minezaki}, T., {Barklem}, P.~S., {Honda}, S., {Takada-Hidai}, M., {Asplund},
929: M., {Ryan}, S.~G., {Tsangarides}, S., {Eriksson}, K., {Steinhauer}, A.,
930: {Deliyannis}, C.~P., {Nomoto}, K., {Fujimoto}, M.~Y., {Ando}, H., {Yoshii},
931: Y., \& {Kajino}, T. 2006, ApJ, 639, 897
932:
933: \bibitem[{{Aoki} {et~al.}(2005){Aoki}, {Honda}, {Beers}, {Kajino}, {Ando},
934: {Norris}, {Ryan}, {Izumiura}, {Sadakane}, \& {Takada-Hidai}}]{aoki05}
935: {Aoki}, W., {Honda}, S., {Beers}, T.~C., {Kajino}, T., {Ando}, H., {Norris},
936: J.~E., {Ryan}, S.~G., {Izumiura}, H., {Sadakane}, K., \& {Takada-Hidai}, M.
937: 2005, \apj, 632, 611
938:
939: \bibitem[Asplund et al.(1997)]{asplund97}
940: Asplund, M., Gustafsson, B., Kiselman, D., \& Eriksson, K.
941: 1997, \aap, 318, 521
942:
943: \bibitem[Asplund \& Garc{\'\i}a P{\'e}rez(2001)]{asplund01}
944: Asplund, M., \& Garc{\'\i}a P{\'e}rez, A. E.
945: 2001, \aap, 372, 601
946:
947: \bibitem[{{Asplund}(2005)}]{asplund_araa}
948: {Asplund}, M. 2005, ARA\&A, 43, 481
949:
950: \bibitem[{{Asplund} {et~al.}(2005){Asplund}, {Grevesse}, \&
951: {Sauval}}]{solar_abund}
952: {Asplund}, M., {Grevesse}, N., \& {Sauval}, A.~J. 2005, in ASP Conf. Ser. 336:
953: Cosmic Abundances as Records of Stellar Evolution and Nucleosynthesis, 25
954:
955: \bibitem[{{Bauschlicher} \& {Langhoff}(1987)}]{NH_diss_en}
956: {Bauschlicher}, Jr., C.~W. \& {Langhoff}, S.~R. 1987, Chem. Phys. Lett., 135, 67
957:
958: \bibitem[{{Boesgaard} \& {Novicki}(2006)}]{boesgaard_novicki}
959: {Boesgaard}, A.~M. \& {Novicki}, M.~C. 2006, \apj, 641, 1122
960:
961: \bibitem[{{Bohlin} {et~al.}(1983){Bohlin}, {Jenkins}, {Spitzer}, {York},
962: {Hill}, {Savage}, \& {Snow}}]{bohlin}
963: {Bohlin}, R.~C., {Jenkins}, E.~B., {Spitzer}, L., {York}, D.~G., {Hill}, J.~K.,
964: {Savage}, B.~D., \& {Snow}, T.~P. 1983, \apjs, 51, 277
965:
966: \bibitem[{{Chieffi} \& {Limongi}(2002)}]{chieffi_limongi2002}
967: {Chieffi}, A. \& {Limongi}, M. 2002, \apj, 577, 281
968:
969: \bibitem[{{Christlieb} {et~al.}(2002){Christlieb}, {Bessell}, {Beers},
970: {Gustafsson}, {Korn}, {Barklem}, {Karlsson}, {Mizuno-Wiedner}, \&
971: {Rossi}}]{HE0107_Nature}
972: {Christlieb}, N., {Bessell}, M.~S., {Beers}, T.~C., {Gustafsson}, B., {Korn},
973: A., {Barklem}, P.~S., {Karlsson}, T., {Mizuno-Wiedner}, M., \& {Rossi}, S.
974: 2002, Nature, 419, 904
975:
976: \bibitem[{{Christlieb} {et~al.}(2004){Christlieb}, {Gustafsson}, {Korn},
977: {Barklem}, {Beers}, {Bessell}, {Karlsson}, \& {Mizuno-Wiedner}}]{HE0107_ApJ}
978: {Christlieb}, N., {Gustafsson}, B., {Korn}, A.~J., {Barklem}, P.~S., {Beers},
979: T.~C., {Bessell}, M.~S., {Karlsson}, T., \& {Mizuno-Wiedner}, M. 2004, \apj,
980: 603, 708
981:
982: \bibitem[{{Collet} {et~al.}(2006){Collet}, {Asplund}, \&
983: {Trampedach}}]{collet06}
984: {Collet}, R., {Asplund}, M., \& {Trampedach}, R. 2006, ApJ, 644, L121
985:
986: \bibitem[{{Collet} {et~al.}(2007){Collet}, {Asplund}, \&
987: {Trampedach}}]{collet07}
988: {Collet}, R., {Asplund}, M., \& {Trampedach}, R. 2007, \aap, 469, 687
989:
990:
991: \bibitem[{{Frebel} {et~al.}(2005){Frebel}, {Aoki}, {Christlieb}, {Ando},
992: {Asplund}, {Barklem}, {Beers}, {Eriksson}, {Fechner}, {Fujimoto}, {Honda},
993: {Kajino}, {Minezaki}, {Nomoto}, {Norris}, {Ryan}, {Takada-Hidai},
994: {Tsangarides}, \& {Yoshii}}]{HE1327_Nature}
995: {Frebel}, A., {Aoki}, W., {Christlieb}, N., {Ando}, H., {Asplund}, M.,
996: {Barklem}, P.~S., {Beers}, T.~C., {Eriksson}, K., {Fechner}, C., {Fujimoto},
997: M.~Y., {Honda}, S., {Kajino}, T., {Minezaki}, T., {Nomoto}, K., {Norris},
998: J.~E., {Ryan}, S.~G., {Takada-Hidai}, M., {Tsangarides}, S., \& {Yoshii}, Y.
999: 2005, Nature, 434, 871
1000:
1001: \bibitem[{{Frebel} {et~al.}(2006){Frebel}, {Christlieb}, {Norris}, {Aoki}, \&
1002: {Asplund}}]{o_he1327}
1003: {Frebel}, A., {Christlieb}, N., {Norris}, J.~E., {Aoki}, W., \& {Asplund}, M.
1004: 2006, ApJ, 638, L17
1005:
1006: \bibitem[{{Frebel} {et~al.}(2007{\natexlab{a}}){Frebel}, {Christlieb},
1007: {Norris}, {Thom}, {Beers}, \& {Rhee}}]{he1523}
1008: {Frebel}, A., {Christlieb}, N., {Norris}, J.~E., {Thom}, C., {Beers}, T.~C., \&
1009: {Rhee}, J. 2007{\natexlab{a}}, \apjl, 660, L117
1010:
1011: \bibitem[{{Frebel} {et~al.}(2007{\natexlab{b}}){Frebel}, {Johnson}, \&
1012: {Bromm}}]{dtrans}
1013: {Frebel}, A., {Johnson}, J.~L., \& {Bromm}, V. 2007{\natexlab{b}}, \mnras, 380,
1014: L40
1015:
1016: \bibitem[{{Fr{\"o}hlich} {et~al.}(2006){Fr{\"o}hlich},
1017: {Mart{\'{\i}}nez-Pinedo}, {Liebend{\"o}rfer}, {Thielemann}, {Bravo}, {Hix},
1018: {Langanke}, \& {Zinner}}]{froehlich}
1019: {Fr{\"o}hlich}, C., {Mart{\'{\i}}nez-Pinedo}, G., {Liebend{\"o}rfer}, M.,
1020: {Thielemann}, F.-K., {Bravo}, E., {Hix}, W.~R., {Langanke}, K., \& {Zinner},
1021: N.~T. 2006, Physical Review Letters, 96, 142502
1022:
1023: \bibitem[{{Gillis} {et~al.}(2001){Gillis}, {Goldman}, {Stark}, \&
1024: {Rinsland}}]{gillis_ohlinelist}
1025: {Gillis}, J.~R., {Goldman}, A., {Stark}, G., \& {Rinsland}, C.~P. 2001, Journal
1026: of Quantitative Spectroscopy \& Rad. Transfer, 68, 225
1027:
1028: \bibitem[Gustafsson et al.(1975)]{gustafsson75}
1029: Gustafsson, B., Bell, R. A., Eriksson, K., \& Nordlund, \AA.
1030: 1975, \aap, 42, 407
1031:
1032: \bibitem[{{Heger} \& {Woosley}(2008)}]{heger2008}
1033: {Heger}, A. \& {Woosley}, S.~E. 2008, ApJ, submitted, astro-ph/0803.3161,
1034:
1035: \bibitem[{{Henyey} {et~al.}(1965){Henyey}, {Vardya}, \&
1036: {Bodenheimer}}]{henyey06}
1037: {Henyey}, L., {Vardya}, M.~S., \& {Bodenheimer}, P. 1965, \apj, 142, 841
1038:
1039: \bibitem[{Hill {et~al.}(2002)Hill, Plez, Cayrel, Nordstr{\" o}m, Andersen,
1040: Spite, Spite, Barbuy, Bonifacio, Depagne, Fran{\c c}ois, \&
1041: Primas}]{Hilletal:2002}
1042: Hill, V., Plez, B., Cayrel, R., Nordstr{\" o}m, T. B.~B., Andersen, J., Spite,
1043: M., Spite, F., Barbuy, B., Bonifacio, P., Depagne, E., Fran{\c c}ois, P., \&
1044: Primas, F. 2002, A\&A, 387, 560
1045:
1046: \bibitem[{{Honda} {et~al.}(2007){Honda}, {Aoki}, {Ishimaru}, \&
1047: {Wanajo}}]{honda07}
1048: {Honda}, S., {Aoki}, W., {Ishimaru}, Y., \& {Wanajo}, S. 2007, \apj, 666, 1189
1049:
1050: \bibitem[{{Honda} {et~al.}(2006){Honda}, {Aoki}, {Ishimaru}, {Wanajo}, \&
1051: {Ryan}}]{honda06}
1052: {Honda}, S., {Aoki}, W., {Ishimaru}, Y., {Wanajo}, S., \& {Ryan}, S.~G. 2006,
1053: \apj, 643, 1180
1054:
1055: \bibitem[{{Huber} \& {Herzberg}(1979)}]{CH_OH_diss_en}
1056: {Huber}, K. P. \& {Herzberg}, G. 1979, Molecular Spectra and Molecular
1057: Structure, Vol. IV, Constants of Diatomic Molecules (New York: Van
1058: Nostrand Reinhold)
1059:
1060: \bibitem[{{Irwin}(1981)}]{irwin81}
1061: {Irwin}, A.~W. 1981, \apjs, 45, 621
1062:
1063: \bibitem[{{Iwamoto} {et~al.}(2005){Iwamoto}, {Umeda}, {Tominaga}, {Nomoto}, \&
1064: {Maeda}}]{iwamoto_science}
1065: {Iwamoto}, N., {Umeda}, H., {Tominaga}, N., {Nomoto}, K., \& {Maeda}, K. 2005,
1066: Science, 309, 451
1067:
1068: \bibitem[{{Korn} {et~al.}(2008){Korn}, {Mashonkina}, {Richard}, {Frebel}, {Aoki},\&
1069: {Christlieb}}]{korn_fs3}
1070: {Korn}, A. J., {Mashonkina}, L., {Richard}, O., {Frebel}, A., {Aoki},W., \&
1071: {Christlieb}, N. 2008, in AIP Conf. Proc. 990: First Stars III, 167
1072:
1073: \bibitem[{{Kupka} {et~al.}(1999){Kupka}, {Piskunov}, {Ryabchikova}, {Stempels},
1074: \& {Weiss}}]{vald}
1075: {Kupka}, F., {Piskunov}, N., {Ryabchikova}, T.~A., {Stempels}, H.~C., \&
1076: {Weiss}, W.~W. 1999, \aaps, 138, 119
1077:
1078: \bibitem[{{Kurucz}(1992)}]{kurucz92}
1079: {Kurucz}, R.~L. 1992, Rev. Mex. Astron. Astrofis., 23, 181
1080:
1081: \bibitem[{Kurucz(1993)}]{kurucz93}
1082: Kurucz, R.~L. 1993, Kurucz CD-ROMs, Vol. 2--12, {Opacities for Stellar
1083: Atmospheres} (Cambridge, Mass.: SAO)
1084:
1085: \bibitem[{{Kurucz}(1993)}]{kurucz_nh}
1086: {Kurucz}, R.~L. 1993, {Kurucz CD-ROM 15, Diatomic Molecular Data for Opacity
1087: Calculations}, Cambridge: SAO
1088:
1089: \bibitem[{{Mashonkina} {et~al.}(2007){Mashonkina}, {Korn}, \&
1090: {Przybilla}}]{Mashonkina_ca}
1091: {Mashonkina}, L., {Korn}, A.~J., \& {Przybilla}, N. 2007, \aap, 461, 261
1092:
1093: \bibitem[{{Meynet} {et~al.}(2006){Meynet}, {Ekstr{\"o}m}, \&
1094: {Maeder}}]{meynet2005}
1095: {Meynet}, G., {Ekstr{\"o}m}, S., \& {Maeder}, A. 2006, A\&A, 447, 623
1096:
1097: \bibitem[{{Meynet} \& {Maeder}(2002)}]{meynet2002}
1098: {Meynet}, G. \& {Maeder}, A. 2002, \aap, 390, 561
1099:
1100: \bibitem[Mihalas et al.(1988)]{mihalas88}
1101: Mihalas, D., D{\"a}ppen, W., \& Hummer, D. G.
1102: 1988, \apj, 331, 815
1103:
1104: \bibitem[Nordlund(1982)]{nordlund82}
1105: Nordlund, \AA.
1106: 1982, \aap, 107, 1
1107:
1108: \bibitem[{{Norris} {et~al.}(2007){Norris}, {Christlieb}, {Korn}, {Eriksson},
1109: {Bessell}, {Beers}, {Wisotzki}, \& {Reimers}}]{he0557}
1110: {Norris}, J.~E., {Christlieb}, N., {Korn}, A.~J., {Eriksson}, K., {Bessell},
1111: M.~S., {Beers}, T.~C., {Wisotzki}, L., \& {Reimers}, D. 2007, \apj, 670, 774
1112:
1113: \bibitem[{{Piau} {et~al.}(2006){Piau}, {Beers}, {Balsara}, {Sivarani},
1114: {Truran}, \& {Ferguson}}]{piau}
1115: {Piau}, L., {Beers}, T.~C., {Balsara}, D.~S., {Sivarani}, T., {Truran}, J.~W.,
1116: \& {Ferguson}, J.~W. 2006, ApJ, 653, 300
1117:
1118: \bibitem[{{Piskunov} \& {Valenti}(2002)}]{reduce}
1119: {Piskunov}, N.~E. \& {Valenti}, J.~A. 2002, \aap, 385, 1095
1120:
1121: \bibitem[{{Ryan} {et~al.}(1999){Ryan}, {Norris}, \& {Beers}}]{ryan_postprim}
1122: {Ryan}, S.~G., {Norris}, J.~E., \& {Beers}, T.~C. 1999, ApJ, 523, 654
1123:
1124: \bibitem[{{Sauval} \& {Tatum}(1984)}]{sauval84}
1125: {Sauval}, A.~J. \& {Tatum}, J.~B. 1984, \apjs, 56, 193
1126:
1127: \bibitem[{{Sneden} {et~al.}(1998){Sneden}, {Cowan}, {Burris}, \&
1128: {Truran}}]{Sneden1998}
1129: {Sneden}, C., {Cowan}, J., {Burris}, D., \& {Truran}, J. 1998, ApJ, 496, 235
1130:
1131: \bibitem[{{Spergel} {et~al.}(2007){Spergel}, {Bean}, {Dor{\'e}}, {Nolta},
1132: {Bennett}, {Dunkley}, {Hinshaw}, {Jarosik}, {Komatsu}, {Page}, {Peiris},
1133: {Verde}, {Halpern}, {Hill}, {Kogut}, {Limon}, {Meyer}, {Odegard}, {Tucker},
1134: {Weiland}, {Wollack}, \& {Wright}}]{WMAP}
1135: {Spergel}, D.~N., {Bean}, R., {Dor{\'e}}, O., {Nolta}, M.~R., {Bennett}, C.~L.,
1136: {Dunkley}, J., {Hinshaw}, G., {Jarosik}, N., {Komatsu}, E., {Page}, L.,
1137: {Peiris}, H.~V., {Verde}, L., {Halpern}, M., {Hill}, R.~S., {Kogut}, A.,
1138: {Limon}, M., {Meyer}, S.~S., {Odegard}, N., {Tucker}, G.~S., {Weiland},
1139: J.~L., {Wollack}, E., \& {Wright}, E.~L. 2007, ApJS, 170, 377
1140:
1141: \bibitem[Stein \& Nordlund(1998)]{stein98}
1142: Stein, R. F., \& Nordlund, \AA.
1143: 1998, \apj, 499, 914
1144:
1145: \bibitem[{{Tominaga} {et~al.}(2007){Tominaga}, {Maeda}, {Umeda}, {Nomoto},
1146: {Tanaka}, {Iwamoto}, {Suzuki}, \& {Mazzali}}]{tominaga07}
1147: {Tominaga}, N., {Maeda}, K., {Umeda}, H., {Nomoto}, K., {Tanaka}, M.,
1148: {Iwamoto}, N., {Suzuki}, T., \& {Mazzali}, P.~A. 2007, \apjl, 657, L77
1149:
1150: \bibitem[{Umeda \& Nomoto(2003)}]{UmedaNomotoNature}
1151: Umeda, H. \& Nomoto, K. 2003, Nature, 422, 871
1152:
1153: \end{thebibliography}
1154:
1155:
1156: \clearpage
1157: \begin{deluxetable}{lclcc}
1158: \tablecolumns{9}
1159: \tablewidth{0pt}
1160: \tabletypesize{\small}
1161: \tablecaption{\label{Tab:SN} Measured $S/N$ ratios in different wavelength regions}
1162: \tablehead{
1163: \colhead{$\lambda$ ({\AA}) } &
1164: \colhead{$S/N$ (pixel)} &
1165: \colhead{Setting} &
1166: \colhead{Pixel size (km\,s$^{-1}$)} &
1167: \colhead{FWHM of Th-Ar lines (pixel)} }
1168: \startdata
1169:
1170: $\sim 3500$ &$\sim 170$ & BLUE 346\,nm& $\sim 2.7$ & $\sim 3.8$ \\
1171: $\sim 3800$ &$\sim 250$ & BLUE 346\,nm& $\sim 2.4$ & $\sim 3.7$ \\
1172: $\sim 5000$ &$\sim 470$ & RED 580\,nm & $\sim 2.3$ & $\sim 3.2$ \\
1173: $\sim 6000$ &$\sim 750$ & RED 580\,nm & $\sim 2.4$ & $\sim 3.5$ \\
1174: $\sim 6700$ &$\sim 600$ & RED 580\,nm & $\sim 2.1$ & $\sim 3.5$ \\
1175: $\sim 8500$ &$\sim 200$ & RED 760\,nm & $\sim 2.3$ & $\sim 3.4$
1176: \enddata
1177: \end{deluxetable}
1178:
1179: \clearpage
1180: \begin{deluxetable}{lccrrrrrl}
1181: \tabletypesize{\footnotesize}
1182: \tabletypesize{\tiny}
1183: \tablewidth{0pc}
1184: \tablecaption{\label{Tab:Abundances} VLT/UVES abundances of HE~1327$-$2326}
1185: \tablehead{
1186: \colhead{Species} &
1187: \colhead{$\log\epsilon (\mbox{X})_{\odot}$} & \colhead{$N_{\mbox{\scriptsize lin.}}$\tablenotemark{a}} &
1188: \colhead{$\log\epsilon (\mbox{X})_{\rm 1D}$} &
1189: \colhead{$\log\epsilon (\mbox{X})_{\rm 3D}$} &
1190: \colhead{[X/H]$_{\rm 3D}$} & \colhead{[X/Fe]$_{\rm 3D}$} }
1191: \startdata
1192: % eps1D eps 3D
1193: C\,(CH)& $8.39$ &syn & $6.90 $ & $6.21$ & $-2.18$ & $3.78$ \\
1194: N\,(NH)& $7.78$ &syn & $6.79 $ & $6.10$ & $-1.68$ & $4.28$ \\
1195: O\,(OH)& $8.66$ &syn & $6.84 $ & $6.12$ & $-2.54$ & $3.42$ \\
1196: %
1197: Na\,I & $6.17$ &2 & $2.99 $ & $2.94$ & $-3.23$ & $2.73$ \\
1198: Mg\,I & $7.53$ &4 & $3.54 $ & 3.54 & $-3.99$ & $1.97$ \\
1199: Al\,I & $6.37$ &1 & $1.90 $ & 1.87 & $-4.50$ & $1.46$ \\
1200: Ca\,I & $6.31$ &1 & $0.88 $ & 0.79 & $-5.52$ & $0.44$ \\
1201: Ca\,II & $6.31$ &4 & $1.34 $ & 1.26 & $-5.05$ & $0.91$ \\
1202: Ti\,II & $4.90$ &15 & $-0.09$ &$-0.15$ & $-5.05$ & $0.91$ \\
1203: Fe\,I & $7.45$ &12 & $ 1.79$ & 1.49 & $-5.96$ &\nodata \\
1204: Ni\,I & $6.23$ &4 & $ 0.73$ & $ 0.45$ & $-5.78$ & $0.18$ \\
1205: Sr\,II & $2.92$ &2 & $-1.76$ & $-1.87$ & $-4.79$ & $1.17$ \\
1206: \hline
1207: Li\,I & $1.05$ & 6707 & $<0.70$ & $ <0.62$ & \nodata &\nodata \\
1208: Be\,II & $1.38$ & 3131 & $<-1.20$& $<-1.15$ & \nodata &\nodata \\
1209: Sc\,II & $3.05$ & 3613 & $<-1.68$& $<-1.80$ & $<-4.85$ & $<1.11$ \\
1210: V\,I & $4.00$ & 3184 & $<1.50$ & $ <1.36$ & $<-2.64$ & $<3.32$ \\
1211: Cr\,I & $5.64$ & 3579 & $<0.45$ & $ <0.26$ & $<-5.38$ & $<0.58$ \\
1212: Mn\,I & $5.39$ & 3439 & $<0.84$ & $ <0.55$ & $<-4.84$ & $<1.12$ \\
1213: Fe\,II & $7.45$ & 3227 & $<1.99$ & $ <2.05$ & $<-5.40$ & $<0.56$ \\
1214: Co\,I & $4.92$ & 3454 & $<0.58$ & $ <0.31$ & $<-4.61$ & $<1.35$ \\
1215: Zn\,I & $4.60$ & 4810 & $<1.61$ & $ <1.65$ & $<-2.95$ & $<3.01$ \\
1216: Ba\,II & $2.17$ & 4554 & $<-2.14$& $<-2.39$ & $<-4.56$ & $<1.40$ \\
1217: Eu\,II & $0.52$ & 6645 & $<-0.76$& $<-0.80$ & $<-1.32$ & $<4.64$ \\
1218: \enddata
1219: \tablecomments{
1220: All [X/Fe] 3D ratios are computed with $\mbox{[Fe/H]}=-5.96$. Solar abundances
1221: have been taken from \citet{solar_abund}.}
1222:
1223: \tablenotetext{b}{``Syn'' indicates the use of spectrum synthesis for
1224: the abundance determination. For upper limits the wavelength of the
1225: employed line is given.}
1226: \tablenotetext{b}{3D CNO corrections have been determined from
1227: spectrum synthesis, see text for details.}
1228:
1229: \end{deluxetable}
1230:
1231:
1232: \clearpage
1233: \begin{deluxetable}{lllrrrrr}
1234: \tablecolumns{9}
1235: \tablewidth{0pt}
1236: \tabletypesize{\small}
1237: \tablecaption{\label{Tab:Eqw} Atomic data and measured equivalent widths}
1238: \tablehead{
1239: \colhead{} & \colhead{} & \colhead{} & \colhead{} & \multicolumn{2}{c}{$W_{\lambda}$}\\
1240: \cline{5-6}\\
1241: \colhead{} & \colhead{$\lambda$} & \colhead{$\chi$} & \colhead{$\log gf$} &
1242: \colhead{This work} & \colhead{Aoki et al.} \\
1243: \colhead{Species} & \colhead{({\AA})} & \colhead{(eV)} & \colhead{(dex)} &
1244: \colhead{(m{\AA})} & \colhead{(m{\AA})} &\colhead{$\log \epsilon$(1D)} &\colhead{$\log \epsilon$(3D)}
1245: }
1246: \startdata
1247: % lambda excit loggf this work WA abund
1248: Na\,I & $5889.951$ & $0.00$& $ 0.12$ & 53.8 & $ 48.9$ & 2.99 &$2.93$ \\
1249: Na\,I & $5895.924$ & $0.00$& $-0.18$ & 33.6 & $ 31.5$ & 2.99 &$2.94$ \\
1250: Mg\,I & $3829.355$ & $2.71$& $-0.21$ & 23.9 & $ 22.6$ & 3.48 &$3.50$ \\
1251: Mg\,I & $5167.321\tablenotemark{a}$ & $2.71$& $-1.03$ & 8.4 & $ 9.8$ & 3.68 &$3.68$ \\
1252: Mg\,I & $5172.684$ & $2.71$& $-0.40$ & 20.9 & $ 21.1$ & 3.50 &$3.49$ \\
1253: Mg\,I & $5183.604$ & $2.72$& $-0.18$ & 30.4 & $ 30.1$ & 3.50 &$3.49$ \\
1254: Al\,I & $3961.529$ & $0.01$& $-0.34$ & b& $ 11.0$ & 1.92 &$1.87$ \\
1255: Ca\,I & $4226.728$ & $0.00$& $ 0.24$ & b& $ 2.7$ & 0.89 &$0.79$ \\
1256: Ca\,II& $3933.663$ & $0.00$& $ 0.11$ & b& $128.9$ & 1.31 &$1.16$ \\
1257: Ca\,II& $8498.023$ & $1.69$& $-1.42$ &$ 10.0$& \nodata & 1.31 &$1.28$ \\
1258: Ca\,II& $8542.091$ & $1.70$& $-0.46$ &$ 54.7$& \nodata & 1.38 &$1.29$ \\
1259: Ca\,II& $8662.141$ & $1.69$& $-0.72$ &$ 39.1$& \nodata & 1.36 &$1.31$ \\
1260: Ti\,II& 3072.984 & 0.00 & $-0.36$ & 5.4 &\nodata &$ 0.20$ &$ 0.11$ \\
1261: Ti\,II& 3088.037 & 0.05 & $ 0.25$ & 7.2 &\nodata &$-0.24$ &$-0.32$ \\
1262: Ti\,II& 3229.198 & 0.00 & $-0.55$ & 3.5 &\nodata &$ 0.17$ &$ 0.08$ \\
1263: Ti\,II& $3234.520$ & $0.05$& $ 0.43$ & 10.1 & $ 7.2$ &$-0.27$ &$-0.34$ \\
1264: Ti\,II& 3236.578 & 0.03 & $ 0.23$ & 6.8 &\nodata &$-0.29$ &$-0.38$ \\
1265: Ti\,II& 3239.044 & 0.01 & $ 0.06$ & 7.5 &\nodata &$-0.08$ &$-0.17$ \\
1266: Ti\,II& 3241.994 & 0.00 & $-0.05$ & 3.9 &\nodata &$-0.30$ &$-0.39$ \\
1267: Ti\,II& 3322.941 & 0.15 & $-0.09$ & 3.8 &\nodata &$-0.13$ &$-0.22$ \\
1268: Ti\,II& $3349.408$ & $0.05$& $ 0.59$ & 15.3 & $ 16.6$ &$-0.23$ &$-0.33$ \\
1269: Ti\,II& 3372.800 & 0.01 & $ 0.27$ & 13.5 &\nodata &$-0.02$ &$-0.12$ \\
1270: Ti\,II& 3380.279 & 0.05 & $-0.57$ & 4.8 &\nodata &$ 0.36$ &$ 0.27$ \\
1271: Ti\,II& 3383.768 & 0.00 & $ 0.14$ & 9.1 &\nodata &$-0.10$ &$-0.20$ \\
1272: Ti\,II& 3387.846 & 0.03 & $-0.43$ & 5.1 &\nodata &$ 0.23$ &$ 0.13$ \\
1273: Ti\,II& $3759.296$ & $0.61$& $ 0.27$ & 4.9 & $ 5.9$ &$-0.04$ &$-0.08$ \\
1274: Ti\,II& $3761.323$ & $0.57$& $ 0.17$ & 2.7 & $ 4.7$ &$-0.24$ &$-0.28$ \\
1275: %============================================================================
1276: Fe\,I & $3440.606$ & $0.00$& $ -0.67$&$ 5.1$ &\nodata & 1.86 &$1.48$ \\
1277: Fe\,I & 3440.989 & 0.05 & $ -0.96$&$ 3.0$ &\nodata & 1.80 &$1.43$ \\
1278: Fe\,I & $3581.193$ & $0.86$& $ 0.42$ &$ 6.2$ & $ 5.9$ & 1.75 &$1.56$ \\
1279: Fe\,I & $3618.768$ & $0.99$& $ 0.00$ & 1.2: &\nodata & 1.53:\tablenotemark{c}&$1.36$ \\
1280: Fe\,I & $3719.935$ & $0.00$& $ -0.43$&$ 7.9$ &\nodata & 1.85 &$1.48$ \\
1281: Fe\,I & $3737.131$ & $0.05$& $-0.57$ &$ 4.5$ & $ 3.9$ & 1.76 &$1.41$ \\
1282: Fe\,I & $3745.561$ & $0.09$& $-0.77$ &$ 3.9$ & $ 4.8$ & 1.82 &$1.48$ \\
1283: Fe\,I & $3748.262$ & $0.11$& $ -1.02$&$ 1.9$ &\nodata & 1.86 &$1.53$ \\
1284: Fe\,I & $3758.233$ & $0.96$& $-0.01$ &$ 2.0$ & $ 5.1$ & 1.69 &$1.54$ \\
1285: Fe\,I & $3820.425$ & $0.86$& $ 0.16$ &$ 3.4$ & $ 2.5$ & 1.66 &$1.49$ \\
1286: Fe\,I & $3825.881$ & 0.92 & $-0.04$ & 1.8: &\nodata & 1.58:\tablenotemark{c}&$1.43$ \\
1287: Fe\,I & 3856.372 & 0.05 & $-1.29$ & 1.1: &\nodata & 1.79:\tablenotemark{c}&$1.43$ \\
1288: Fe\,I & $3859.912$ & $0.00$& $-0.71$ & 4.8 & $ 6.8$ & 1.83 &$ 1.45$ \\
1289: Fe\,I & $4045.812$ & $1.49$& $ 0.29$ & b & $ 1.9$ & 1.78 &$ 1.68$ \\
1290: %============================================================================
1291: Ni\,I & $3414.761$ & $0.03$& $-0.03$ & $ 2.9$& $<7.5$ &$ 0.70$ &$ 0.40$ \\
1292: Ni\,I & 3515.049 & 0.11 &$-0.21 $ & 2.1 & \nodata & 0.81 &$ 0.53$ \\
1293: Ni\,I & $3524.535$ & $0.03$& $ 0.01$ & $2.4$& \nodata & 0.57 &$ 0.26$ \\
1294: Ni\,I & 3619.386 & 0.42 & 0.04 & 2.0 & \nodata & 0.83 &$ 0.61$ \\
1295: Sr\,II& $4077.724$ & $0.00$& $ 0.16$ & b& $ 7.3$ &$-1.76$ &$-1.87$ \\
1296: Sr\,II& $4215.540$ & $0.00$& $-0.16$ & b& 3.8 &$-1.75$ &$-1.86$ \\
1297: %============================================================================
1298: \hline
1299: Li\,I & 6707.761 & 0.00 & $-0.01$& syn\tablenotemark{d}&\nodata&$<0.70$ &$0.62$ \\
1300: Li\,I & 6707.912 & 0.00 & $-0.31$& syn\tablenotemark{d}&\nodata&$<0.70$ &$0.62$ \\
1301: Be\,II& 3130.420 & 0.00 & $-0.17$& syn\tablenotemark{d}&\nodata&$<-1.20$&$-1.15$ \\
1302: Be\,II& 3131.065 & 0.00 & $-0.47$& syn\tablenotemark{d}&\nodata&$<-1.20$&$-1.15$ \\
1303: \mbox{[O\,I]}& 6300.304 & 0.00 & $-9.82$& $<1.8$&\nodata & $<8.01$ &$< 8.07$ \\
1304: O\,I & 7771.944 & 9.15 & $ 0.32$& $<2.0$&\nodata & $<6.19$ &$< 6.21$ \\
1305: O\,I & 7774.166 & 9.15 & $ 0.17$& $<2.0$&\nodata & $<6.34$ &$< 6.36$ \\
1306: O\,I & 7775.388 & 9.15 & $-0.05$& $<2.0$&\nodata & $<6.56$ &$< 6.58$ \\
1307: Sc\,II& $3572.526$ & $0.02$& $ 0.27$ &$<1.1$& \nodata& $<-1.57$ &$<-1.68$ \\
1308: Sc\,II& $3613.829$ & 0.02 & 0.42 &$<1.0$& \nodata& $<-1.68$ &$<-1.80$ \\
1309: Sc\,II& $3642.784$ & $0.00$& $ 0.13$ &$<1.2$& \nodata& $<-1.45$ &$<-1.54$ \\
1310: V\,I & 3183.410 & 0.02 & 0.46 &$<2.6$& \nodata& $<1.60$ &$< 1.46$ \\
1311: V\,I & 3183.970 & 0.04 & 0.58 &$<2.6$& \nodata& $<1.50$ &$< 1.36$ \\
1312: V\,I & 3184.013 & 0.00 & 0.34 &$<2.6$& \nodata& $<1.70$ &$< 1.54$ \\
1313: Cr\,I & $3578.684$ & $0.00$& $ 0.41$ &$<1.2$& \nodata& $<0.45$ &$< 0.26$ \\
1314: Cr\,I & $3593.481$ & $0.00$& $ 0.31$ &$<1.3$& \nodata& $<0.59$ &$< 0.40$ \\
1315: Cr\,I & $4254.332$ & $0.00$& $-0.11$ & b& $<2.0$ & $<1.07$ &$< 0.88$ \\
1316: Mn\,I & $4030.753$ & $0.00$& $-0.47$ & b& $<2.0$ & $<0.84$ &$< 0.55$ \\
1317: Fe\,II& $3227.742$ & $1.67$& $-1.10$ &$<2.2$& \nodata& $<1.99$ &$< 2.05$ \\
1318: Fe\,II& $5018.450$ & $2.89$& $-1.22$ &$<0.8$& $<2.0$ & $<2.61$ &$< 2.66$ \\
1319: Co\,I & $3453.514$ & $0.43$& $ 0.38$ &$<1.5$& $<6.3$ & $<0.58$ &$< 0.31$ \\
1320: Zn\,I & $4810.528$ & $4.08$& $-0.15$ &$<1.0$& $<2.0$ & $<1.61$ &$< 1.65$ \\
1321: Ba\,II& $4554.029$ & $0.00$& $ 0.17$ & b & $<1.8$ & $<-2.14$ &$<-2.39$ \\
1322: Ba\,II& $4934.100$ & $0.00$& $-0.16$ &$<1.1$& \nodata& $<-2.04$ &$<-2.23$ \\
1323: Eu\,II& $6645.064$ & $1.38$& $ 0.21$ &$<1.1$& \nodata& $<-0.76$ &$<-0.80$
1324: \enddata
1325: \tablenotetext{a}{The blend with an Fe line is negligible.}
1326: \tablenotetext{b}{These lines are not covered by the new VLT data.}
1327: \tablenotetext{c}{These lines are not included in the final abundance estimate.}
1328: \tablenotetext{d}{Abundance limit was derived from spectrum synthesis.}
1329: \end{deluxetable}
1330: %===================================
1331:
1332: \clearpage
1333: \begin{figure}
1334: \begin{center}
1335: \includegraphics[height=9cm, clip=true,bbllx=0, bblly=0, bburx=540, bbury=360]{f1.ps}
1336: \figcaption{ \label{optical_depth}
1337: Temperature structure the upper layers of the 3D hydrodynamical
1338: simulation adopted in the present 3D$-$1D differential
1339: abundance analysis.
1340: \textit{Gray shaded area}: temperature distribution as a function of optical
1341: depth at $5000$~{\AA} in the 3D model.
1342: Darker areas indicate temperature values with higher probability.
1343: \textit{Dashed line}: mean temperature stratification of the
1344: simulation (averaged over surfaces of constant optical depth at
1345: $5000$~{\AA}). \textit{Solid line}: temperature stratification
1346: of the 1D {\sc marcs} model constructed for the same stellar parameters.}
1347: \end{center}
1348: \end{figure}
1349:
1350: \clearpage
1351: \begin{figure}
1352: \begin{center}
1353: \includegraphics[height=9cm, clip=true,bbllx=60, bblly=417, bburx=480, bbury=711]{f2.ps}
1354: \figcaption{\label{abund_pattern} Abundance pattern of
1355: HE~1327$-$2326 (\textit{filled circles}) in comparison with that of
1356: HE~0107$-$5240 (\textit{open squares}). Typical $1\,\sigma$ errors of
1357: $0.2$\,dex are shown. Upper limits are indicated by an arrow. The 3D
1358: abundances for HE~0107$-$5240 are taken from \citet{collet06}. }
1359: \end{center}
1360: \end{figure}
1361:
1362:
1363:
1364: \clearpage
1365: \begin{figure}
1366: \begin{center}
1367: \includegraphics[height=15cm, clip=true,bbllx=30, bblly=248, bburx=540, bbury=744]{f3.ps} \figcaption{\label{feI_plot} All
1368: detected and tentatively detected Fe\,I lines in the VLT/UVES
1369: spectrum of HE~1327$-$2326. Equivalent width measurements can be
1370: found in Table~\ref{Tab:Eqw}.}
1371: \end{center}
1372: \end{figure}
1373:
1374:
1375: \clearpage
1376: \begin{figure}
1377: \begin{center}
1378: \includegraphics[height=9cm, clip=true,bbllx=0, bblly=0, bburx=504, bbury=360]{f4.ps}
1379: \figcaption{ \label{feifraction}
1380: Ratio of neutral to total iron number densities
1381: ($n_{\rm{Fe\,I}}/n_{\rm{Fe}}$) as a function of
1382: optical depth at $5000$\,{\AA}.
1383: \textit{Gray shaded area}: distribution of $n_{\rm{Fe\,I}}/n_{\rm{Fe}}$
1384: values in the 3D model atmosphere adopted here;
1385: darker areas indicate values with higher probability.
1386: Over-plotted are the curves for the mean 3D stratification
1387: (\textit{dashed line}) and for the corresponding 1D {\sc marcs}
1388: model atmosphere (\textit{solid line}). }
1389: \end{center}
1390: \end{figure}
1391:
1392: \clearpage
1393: \begin{figure}
1394: \begin{center}
1395: \includegraphics[height=9cm, clip=true,bbllx=0,bblly=0,bburx=504,bbury=360]{f5.ps}
1396: \figcaption{ \label{feicontrib}
1397: Contribution function to outgoing intensity in the vertical direction
1398: at line centre for the Fe\,I 3859.9\,{\AA} line
1399: as a function of optical depth.
1400: \textit{Gray shaded area}: distribution of the contribution function values
1401: in the 3D model atmosphere adopted here;
1402: darker areas indicate values with higher probability.
1403: Over-plotted (\textit{solid line}) is the contribution function
1404: computed with the corresponding 1D {\sc marcs} model.
1405: The reported 3D and 1D Fe abundances are the ones reproducing
1406: the measured equivalent width of the line.}
1407: \end{center}
1408: \end{figure}
1409:
1410:
1411: \clearpage
1412: \begin{figure*}
1413: \begin{center}
1414: \resizebox{\hsize}{!}{
1415: \includegraphics{f6a.ps}
1416: \includegraphics{f6b.ps} }
1417: \figcaption{ \label{ch_oh_corr}
1418: 3D$-$1D LTE corrections to the C (\textit{left panel}) and O abundances
1419: (\textit{right panel}) as derived from fictitious CH and OH lines.
1420: Carbon and oxygen abundance corrections are plotted
1421: as a function of line strength for selected values of the
1422: lower excitation potential ($0$ to $1.5$\,eV) and for different values of
1423: 3D O and C abundances respectively. }
1424: \end{center}
1425: \end{figure*}
1426:
1427: \clearpage
1428: \begin{figure}
1429: \begin{center}
1430: \includegraphics{f7.ps}
1431: \figcaption{ \label{nh_corr}
1432: 3D$-$1D LTE corrections to the N abundance as derived from
1433: fictitious NH lines. The corrections are plotted
1434: as a function of line strength for different values of the
1435: lower excitation potential.}
1436: \end{center}
1437: \end{figure}
1438:
1439:
1440: \clearpage
1441: \begin{figure}
1442: \begin{center}
1443: \includegraphics[height=16cm, clip=true,bbllx=50, bblly=164,
1444: bburx=526, bbury=714]{f8_color.ps}
1445: \figcaption{\label{CH_comp} CH G-band spectral region of
1446: HE~1327$-$2326. In the top panel, the excitation potentials of the
1447: molecular CH lines are presented. In the second panel from the top,
1448: the flux difference between the ``1D'' and ``3D-aided'' synthetic spectra
1449: (from the panel below) is given. A comparison of the ``1D''
1450: synthetic spectrum with a synthetic spectrum generated from the
1451: 3D-adjusted linelist is shown in the second panel from the bottom. The
1452: abundance of the ``3D-aided'' synthetic spectrum was varied to reproduce
1453: the ``1D'' spectrum, and so that the average difference of the two
1454: fluxes seen in the panel above becomes zero over the given
1455: wavelength range. Abundances are as given. In the bottom panel we
1456: show the observed Subaru/HDS spectrum (\textit{dots}) overplotted
1457: with ``3D-aided'' synthetic spectra with C abundances of
1458: $\mbox{[C/Fe]}=3.78$ (\textit{solid line}), 3.58, and 3.98
1459: (\textit{dotted lines}). The average 3D$-$1D correction over this
1460: wavelength range for CH is $-0.69$\,dex.}
1461: \end{center}
1462: \end{figure}
1463:
1464: \clearpage
1465: \begin{figure}
1466: \begin{center}
1467: \includegraphics[height=18cm, clip=true,bbllx=55, bblly=164,
1468: bburx=526, bbury=714]{f9_color.ps}
1469: \figcaption{\label{NH_comp} Same as for Figure~\ref{CH_comp}, but
1470: for the NH spectral region. The observed VLT/UVES spectrum (\textit{dots}) in
1471: the bottom panel is overplotted with ``3D-aided'' synthetic spectra with N
1472: abundances of $\mbox{[N/Fe]}=4.28$ (\textit{solid line}), 4.08, and
1473: 4.48 (\textit{dotted lines}). The average 3D$-$1D correction over this
1474: wavelength range for NH is $-0.69$\,dex.}
1475: \end{center}
1476: \end{figure}
1477:
1478: \clearpage
1479: \begin{figure}
1480: \begin{center}
1481: \includegraphics[height=20cm, clip=true,bbllx=52, bblly=84, bburx=526,
1482: bbury=800]{f10_color.ps}
1483: \figcaption{\label{OH_comp} Same as for Figure~\ref{CH_comp}, but
1484: for several spectral region with OH lines. The observed VLT/UVES spectrum
1485: (\textit{dots}) in the two panels at the bottom is overplotted with
1486: ``3D-aided'' synthetic spectra with O abundances of $\mbox{[O/Fe]}=3.42$
1487: (\textit{solid line}), 3.22, and 3.62 (\textit{dotted lines}). The
1488: average 3D$-$1D correction over this wavelength range for OH is
1489: $-0.72$\,dex.}
1490: \end{center}
1491: \end{figure}
1492:
1493: \clearpage
1494: \begin{figure}
1495: \begin{center}
1496: \includegraphics[height=8cm, clip=true,bbllx=40, bblly=330, bburx=540,
1497: bbury=626]{f11_color.ps} \figcaption{\label{li_plot}
1498: Spectral region (\textit{connected dots}) of the Li\,I doublet at
1499: 6707\,{\AA} in HE~1327$-$2326. Synthetic spectra are overplotted
1500: (\textit{solid lines}) for three Li abundances. The upper limit is
1501: $A({\rm Li})<0.62$.}
1502: \end{center}
1503: \end{figure}
1504:
1505: \clearpage
1506: \begin{figure}
1507: \begin{center}
1508: \includegraphics[height=8cm, clip=true,bbllx=53, bblly=478, bburx=386,
1509: bbury=677]{f12_color.ps} \figcaption{\label{be_plot} Spectral
1510: region (\textit{connected dots}) of the two beryllium lines at
1511: $\sim3130$\,{\AA} in HE~1327$-$2326. Synthetic spectra are
1512: overplotted (\textit{solid lines}) for three Be abundances. The
1513: upper limit derived from both lines is $\log \epsilon ({\rm Be})<-1.2$.}
1514: \end{center}
1515: \end{figure}
1516:
1517:
1518: \clearpage
1519: \begin{figure}
1520: \begin{center}
1521: \includegraphics[height=8cm, clip=true,bbllx=64,
1522: bblly=342, bburx=270, bbury=540]{f13.ps}
1523: \figcaption{\label{eqw_comp} Comparison of the equivalent width
1524: measurements between this work and \citet{Aokihe1327}.}
1525: \end{center}
1526: \end{figure}
1527:
1528:
1529:
1530: \clearpage
1531: \begin{figure}
1532: \begin{center}
1533: \includegraphics[height=4cm, clip=true,bbllx=30,
1534: bblly=340, bburx=540, bbury=460]{f14.ps}
1535: \figcaption{\label{Ni_plot} Three of the four detected Ni\,I lines in the VLT/UVES spectrum of
1536: HE~1327$-$2326.}
1537: \end{center}
1538: \end{figure}
1539:
1540: \clearpage
1541: \begin{figure}
1542: \begin{center}
1543: \includegraphics[height=8cm, clip=true,bbllx=45, bblly=338, bburx=540,
1544: bbury=626]{f15_color.ps} \figcaption{\label{ba_plot}
1545: Spectral region (\textit{connected dots}) of the barium line at
1546: 4937\,{\AA} in HE~1327$-$2326. Synthetic spectra are overplotted
1547: (\textit{solid lines}) for three Ba abundances. The upper limit for
1548: this line is $\log \epsilon ({\rm Ba})<-2.04$.}
1549: \end{center}
1550: \end{figure}
1551:
1552: \clearpage
1553: \begin{figure}
1554: \begin{center}
1555: \includegraphics[height=12cm, clip=true,bbllx=45, bblly=88, bburx=540,
1556: bbury=426]{f16.ps} \figcaption{\label{abund_summ} Comparison of
1557: previously published abundances [X/H] of HE~1327$-$2326.}
1558: \end{center}
1559: \end{figure}
1560:
1561: \clearpage
1562: \begin{figure}
1563: \begin{center}
1564: \includegraphics[height=12cm, clip=true,bbllx=45, bblly=408, bburx=540, bbury=726]{f17_color.ps}
1565: \figcaption{\label{models}
1566: Comparison of the new 3D$-$1D corrected abundances [X/Fe] of
1567: HE~1327$-$2326 with predictions from various models based on
1568: chemical yields of the first-generation SNe. See text for
1569: discussion.}
1570: \end{center}
1571: \end{figure}
1572:
1573:
1574:
1575: \end{document}
1576: