1: \documentclass[12pt,preprint]{aastex}
2: \usepackage{flushrt,float}
3:
4: \slugcomment{Revised version submitted to the Astrophysical Journal Letters}
5:
6: \shortauthors{Limongi, Chieffi, Bonifacio} \shorttitle{HE0107-5240}
7:
8:
9: \begin{document}
10:
11: \title{On the origin of HE0107-5240, the most iron deficient star presently known}
12:
13: \author{Marco Limongi\altaffilmark{1,2,3}, Alessandro Chieffi\altaffilmark{4,2,3} and Piercarlo Bonifacio\altaffilmark{5}}
14:
15: \altaffiltext{1}{INAF - Osservatorio Astronomico di Roma, Via Frascati 33, I-00040, Monteporzio Catone, Italy;
16: marco@mporzio.astro.it}
17:
18: \altaffiltext{2}{School of Mathematical Sciences, P.O. Box, 28M, Monash University, Victoria 3800, Australia}
19:
20: \altaffiltext{3}{Centre for Astrophysics and Supercomputing, Swinburne University of Technology, Mail Number 31,
21: P.O. Box 218, Hawthorn, Victoria 3122, Australia}
22:
23: \altaffiltext{4}{Istituto di Astrofisica Spaziale e Fisica Cosmica (CNR), Via Fosso del Cavaliere, I-00133, Roma, Italy;
24: achieffi@rm.iasf.cnr.it}
25:
26: \altaffiltext{5}{INAF - Osservatorio Astronomico di Trieste, Via Tiepolo 11, I-34131, Trieste, Italy;
27: bonifaci@ts.astro.it}
28:
29:
30: \begin{abstract}
31:
32: We show that the "puzzling" chemical composition observed in the extremely metal
33: poor star HE0107-5240 may be naturally explained by the concurrent pollution of at
34: least two supernovae. In the simplest possible model a supernova of quite low mass
35: ($\rm \sim 15~M_\odot$), underwent a "normal" explosion and ejected $\rm \sim
36: 0.06~M_\odot$ of $\rm ^{56}Ni$ while a second one was massive enough ($\rm \sim
37: 35~M_\odot$) to experience a strong fall back that locked in a compact remnant all
38: the carbon-oxygen core. In a more general scenario, the pristine gas clouds were
39: polluted by one or more supernovae of relatively low mass (less than $\rm \sim
40: 25~M_\odot$). The successive explosion of a quite massive star experiencing an
41: extended fall back would have largely raised the abundances of the light elements
42: in its close neighborhood.
43:
44: \end{abstract}
45:
46:
47: \keywords{nuclear reactions, nucleosynthesis, abundances -- stars: evolution -- stars: interiors -- stars: supernovae }
48:
49: %********************************************* INTRODUCTION **************************************************
50:
51: \section{Introduction}
52:
53: Extremely metal poor stars (EMPS) are formed in the very early epochs of Galaxy
54: formation by gas clouds chemically enriched only by the very first stellar
55: generation. Hence they preserved up to the present time the fingerprints of this
56: primordial population. Up to now, more than a dozen of stars having $\rm [Fe/H]\leq
57: -3$ (here $\rm [A/B] = Log_{10} (X_A/X_B) - Log_{10} (X_A/X_B)_\odot$, where $\rm
58: X_A$ and $\rm X_B$ are the abundances of elements A and B respectively) have been
59: discovered, for which a quite large number of element abundance ratios have been
60: reliably determined (Bonifacio et al. 1998, Norris, Ryan \& Beers 2001, Depagne et
61: al. 2002, Christlieb et al. 2002, Aoki et al. 2002, Carretta et al. 2002, Fran{\c
62: c}ois et al. 2003). The picture that emerged from the available data prior to the
63: work of Fran{\c c}ois et al. (2003) was the existence of a quite large star to star
64: [el/Fe] scatter among the most metal poor stars (McWilliam et al., 1995; Ryan,
65: Norris and Beers, 1996); this was interpreted as an evidence of large
66: inhomogeneities in the first chemical enrichment of the pristine material. On the
67: contrary the new high resolution spectra of four EMPS analyzed by Fran{\c c}ois et
68: al. (2003) have shown that there is no evidence of a strong star to star scatter
69: for the elements Na to Ti in these stars. This would imply a more "homogeneous"
70: enrichment of the pristine material. By combining all the available data (and
71: including the C rich stars!), it is possible to identify essentially two groups of
72: EMPS. A first one includes stars sharing a) a quite similar abundance pattern for
73: all the observed elements from C to Ni and b) (el/Fe) ratios not largely deviating
74: from the solar ones: these stars (like, e.g., $\rm CD - 38^{o}245$, $\rm CD -
75: 24^{o}17504$) are naturally interpreted in {terms of
76: the ejecta of a single supernova}
77: (Chieffi \& Limongi 2002). Stars belonging to the second group (like e.g. CS22949-
78: 037, CS22498-043 and CS22957-027) show a) large enhancements of the light elements
79: relative to iron, b) significant star to star abundance scatter and c) an abundance
80: pattern of the heavier elements (Si and above) very similar to the one shown by the
81: stars belonging to the first group. These stars cannot be explained by the ejecta
82: of a "standard" single Supernova.
83:
84: HE0107-5240 is the most iron deficient star presently known (Christlieb et al.
85: 2002). Beyond its iron deficiency (1/200000 that of the Sun, i.e. [Fe/H]=-5.3) it
86: shows enormous C, N and Na enhancements (by a factor of $10^4$ , $10^{2.3}$ and 10,
87: respectively, relative to iron) while Mg, usually found enhanced in the EMPS, is
88: almost at the level of iron. This star is the most extreme example of stars
89: belonging to the second group and its understanding is of vital importance to
90: shed light on the nature of this variegate ensemble of "atypical" stars that are
91: preserving, up to the present time, clues about some still unknown characteristic,
92: dynamical and/or evolutionary, of the very first stellar generation.
93:
94: A discussion of all the possible scenarios that can produce the chemical composition
95: observed on the surface of this star is beyond the purposes of the present paper.
96: However we wish to briefly comment on the self-pollution scenario that has been
97: invoked (Christlieb et al. 2002 and references therein) to explain the large
98: overabundances of C and N observed in HE0107- 5240. In this scenario, the observed
99: overabundances of C and N are due to the penetration of the He convective shell
100: that forms at the He-flash in the H rich envelope. At the beginning the
101: protons engulfed in the He convective shell react with the C produced by the
102: $3~\alpha$ (producing therefore a large amount of primary N). Then the energy
103: released by the burning of these protons induces a large expansion and cooling of
104: all the H-rich mantle (down to the ignition point of the He flash{\bf ).} As a
105: result of such an expansion, the He convective shell and the convective envelope
106: merge so that a strong dredge up of C and N to the surface occurs. A paper fully
107: devoted to the analysis of this scenario is in preparation and it is based on an
108: extended set of new and detailed computations extending from the pre main sequence
109: up to the asymptotic giant branch (Picardi et al. 2003, in preparation).
110: Here we can anticipate that this (self-
111: pollution) scenario is {\it incompatible} with the observed chemical composition
112: of HE0107- 5240. There are at least four different and independent reasons for
113: this: 1) the models predict (very robustly) a C/N ratio of the order of 1 while the
114: observed ratio is of the order of 140, 2) the predicted $\rm ^{12}C/^{13}C$ ratio
115: is 4 while the observed lower limit is 30, 3) the time spent by the star on the Red
116: Giant Branch (RGB) after the dredge-up (second RGB) is 1/100 of the time spent on
117: the RGB before the dredge-up (this means that one expects roughly 100 RGB zero
118: metallicity stars not showing these huge enhancements for each star similar to
119: HE0107-5240: they are not observed), 4) the surface gravity of this star is
120: incompatible (too large) with the range of values it should have if it were on the
121: second RGB.
122:
123: The interpretation of the abundance pattern of the EMPS in terms of ejecta provided
124: by a single "standard" primordial core collapse supernova (SNII) fails miserably in
125: the case of HE0107-5240 because of two major inconsistencies that show up when
126: comparing the models to the observed abundances. The first one can be easily
127: understood by reminding that Fe (mainly produced as the unstable $\rm ^{56}Ni$
128: nucleus) is synthesized by both the incomplete explosive Si burning (at
129: temperatures in the range $\rm 4<T/10^{9}<5~K$) and by the deeper complete
130: explosive Si burning (at temperatures larger than $\rm 5\cdot 10^{9}~K$), while Ti
131: and Ni are synthesized only by the complete explosive Si burning. The observed
132: abundances of Ti and Ni in HE0107-5240 clearly point towards a mass cut (the mass
133: coordinate that separates the ejected matter from the compact remnant) deep enough
134: to allow the ejection of at least part of the products of the complete explosive Si
135: burning. Conversely, the large overabundance of, e.g., C relative to Fe definitely
136: would point towards a mass cut so external that all the layers exposed to the
137: complete explosive Si burning were locked in the remnant as well as most of the
138: matter exposed to the incomplete explosive Si burning. Figure \ref{inconsistencies}
139: clearly shows the inconsistency between the models and the observed abundances of
140: the light elements, once the mass cut has been chosen to match the iron peak
141: nuclei.
142:
143: The second inconsistency concerns the relative abundances among the light elements
144: C, Na and Mg. The [C/Mg] and [Na/Mg] ratios observed in HE0107-5240 are +3.8 and
145: +0.6 respectively and are incompatible with any mass cut internal to the CO core.
146: In other words C, Na and Mg are produced in the CO core in ratios completely
147: different from those seen in this star. Figure \ref{trendlight} shows, as an
148: example, the [C/Mg] and [Na/Mg] ratios (thick solid and dashed lines) in a $\rm
149: 35~M_\odot$ progenitor star as a function of the mass of the remnant. The thin
150: lines in Figure \ref{trendlight} show as a reference, in the background, the
151: profiles of He, C, Na and Mg within the exploding star. It is evident that the
152: values observed in HE0107-5240 are never attained within the CO core.
153:
154: The only way to get rid of the CO core is to let it collapse into the remnant but
155: in this way obviously all the elements produced in the CO core would be lost.
156: Extended tests done by changing the mass of the precursor and/or some basic
157: uncertainties connected to the computation of the evolution of these stars (like
158: the $\rm ^{12}C(\alpha,\gamma)^{16}O$ reaction rate) simply confirm these basic
159: inconsistencies.
160:
161: The observed ratios of [C/Mg], [N/Mg] and [Na/Mg] are compatible with those
162: obtained in a He convective shell engulfing some amount of fresh protons. Indeed,
163: the protons ingested in the He shell activate a sequence of reactions that lead to
164: the production of N, Na and Mg in the right proportions relative to the C produced
165: by the $\rm 3\alpha$ reactions. Such a partial mixing between the He convective
166: shell and the overlying H rich mantle is a quite common occurrence in stellar
167: models of initial zero metallicity (Woosley \& Weaver 1982, Chieffi et al. 2001,
168: Fujimoto et al. 1990) because of the low entropy barrier that develops at the H-He
169: interface in these stars.
170:
171: Recently Bonifacio, Limongi \& Chieffi (2003) suggested that the chemical
172: composition of HE0107-5240 is the result of the superposition of the ejecta of two
173: primordial SNII: a first of quite low mass would have been responsible for the
174: observed iron peak nuclei while an "almost failed" explosion of a quite massive
175: star would have been responsible for the observed light element abundances. In this
176: paper we will discuss this model more quantitatively and we will show how it can be
177: considered as a specific example of a much wider scenario.
178:
179:
180: \section{A possible interpretation of the element abundance pattern of HE0107-5240}
181:
182: A thorough analysis of the primordial core collapse supernova models and their associated
183: explosive yields presented by
184: Chieffi \& Limongi (2002) and Limongi \& Chieffi (2002) shows that primordial stars
185: of mass of the order of $\rm 35~M_{\odot}$ experience a partial mixing between the
186: He convective shell and the H rich envelope and that they produce C, N, Na and Mg
187: in this convective shell in relative proportions similar to the ones observed in
188: HE0107-5240 (see Figure \ref{trendlight}). A further interesting property of the
189: stars in this mass range is that they lie quite close to the mass limit above which
190: the explosion almost fails and a very extended fall back onto the remnant occurs
191: (Heger \& Woosley, 2002). Hence a primordial core collapse supernova of mass of the
192: order of $\rm 35~M_{\odot}$ meets all the requirement necessary to be the star that
193: provided the light elements we observe in HE0107-5240.
194:
195: Since the theoretical [el/Fe] ratios of the elements Si to Ni do not show a clear
196: dependence on the initial mass of the precursor, it is not possible to firmly
197: identify the star that produced the heavy elements observed in HE0107-5240. This is
198: a disadvantage but also an advantage as we will show below. Here, for sake of
199: simplicity, let us choose the star that provided the elements produced by the
200: explosive burnings. It is quite natural to think that the star responsible for the
201: production of the elements Si to Ni had to be significantly less massive than the
202: other one. There are two reasons for this: first of all the ejecta of this star
203: must not provide enough of light elements (C to Mg) to modify appreciably their
204: abundance ratios as provided by the other star. Second, it is reasonable to think
205: that the star with the more internal mass cut was also the one with the lowest
206: binding energy of the mantle. Both arguments point towards a core collapse
207: supernova of quite small mass. The $\rm 15~M_\odot$ star shows up as a good candidate
208: for being the star that provided the heavy elements we see in HE0107-5240. In fact,
209: by combining the yields provided by these two primordial core collapse supernovae a
210: very good fit to HE0107-5240 may be obtained (see Figure \ref{fitfinal}).
211:
212: Which was hence the sequence of events that provided the chemical composition we
213: observe in HE0107-5240?
214:
215: First, a star of $\rm 15~M_\odot$ explodes and pollutes the surrounding cloud. The
216: fit to the observed [Ca/Fe] requires the ejection of $\rm 5.6\cdot 10^{-2}~M_\odot$
217: of iron, and this fixes the mass of the remnant at $\rm 1.7~M_\odot$. Since the Fe
218: mass fraction observed in HE0107-5240 is $\rm 6.7\cdot 10^{-9}$, the ejecta
219: provided by the $\rm 15~M_\odot$ star must be diluted in roughly $\rm 7.6\cdot
220: 10^{6}~M_\odot$ of pristine material in order to obtain the right Fe abundance.
221: Within the cloud already polluted by this star, a $\rm 35~M_\odot$ explodes,
222: suffers an extended fall back and ejects just the mass above $\rm 9.4~M_\odot$. The
223: precise location of the mass cut for this second star is determined by the
224: requirement that [C/Mg]=3.8. A last quantity which must be fixed is the mass size
225: of the cloud polluted by this second supernova. Since the observed N mass fraction
226: in HE0107-5240 is $\rm 1.2\cdot 10^{-6}$ while the supernova ejects $\rm 3.4\cdot
227: 10^{-4} M_{\odot}$ of N, the ejecta of this star must be diluted with roughly
228: $\rm 2.6\cdot 10^{2}~M_\odot$ of gas previously enriched by the smaller supernova.
229: Such a "small" mixing of the ejecta produced by this explosion is the natural
230: consequence of the extended fall back and of the low velocity of the ejecta. One
231: could argue that a $\rm 35~M_\odot$ star evolves faster than a $\rm 15~M_\odot$ star and
232: therefore that it should have exploded first. This would be true only in the idealized
233: case in which all stars form simultaneously. In practice the lifetime of all these
234: stars is short enough (few millions of years) that stochastic inhomogeneities in
235: their formation process could have easily led to the explosion of a $\rm
236: 15~M_\odot$ star before that of a $\rm 35~M_\odot$ star.
237:
238: It is worth stressing that the final chemical composition predicted by this model
239: (shown in Figure \ref{fitfinal}) does not depend critically on the adopted mass cut
240: but, instead, it changes slowly and continuously with a change in either of the two
241: mass cuts and/or the dilution factors. For example, if the fall back in the $\rm
242: 35~M_\odot$ star would extend up to $\rm 9~M_\odot$ (instead of 9.4) the predicted
243: [C/Fe], [N/Fe] and [Na/Fe] would be 4.0, 2.3 and 0.7 respectively, while a fall
244: back extending up to $\rm 11~M_\odot$ (instead of 9.4) would imply a [C/Fe], [N/Fe]
245: and [Na/Fe] equal to 3.6, 2.3 and 0.21 respectively. This means that, if this
246: scenario is correct, one would expect a wide variety of overabundances of, e.g.,
247: [C/Fe] and [N/Fe], depending on the amount of fall back experienced by the massive
248: star. At variance with respect to this slow change of the resulting chemical
249: composition with varying mass cut in our model, in a scenario like the one
250: presented by Umeda and Nomoto (2003), even a tiny change in the mass of the remnant
251: would completely change the predicted [el/Fe] ratios by orders of magnitude.
252:
253: All the discussion presented above naturally points towards a more general scenario. The
254: abundance pattern of the elements produced by the explosive burnings does not show
255: a strong dependence on the initial mass of the supernova progenitor and hence the
256: observed abundances of these elements may well be the result of a single supernova
257: explosion but also the combination of two or more supernovae explosions: it is not
258: possible to distinguish between these two cases. This also means that the
259: similarity in the pattern of the heavy elements observed in the EMPS is neither
260: necessarily the consequence of a well mixed environment nor the result of a quite
261: constant IMF through the various clouds but it would simply descend from the fact
262: that the initial mass does not produce a firm signature on the ejecta of the
263: explosive burnings. In these clouds more or less similarly polluted in heavy
264: elements, every now and then a quite massive star explodes but, due to the large
265: binding energy, suffers an extended fall back and most of the matter (and of the
266: energy) remains locked in the remnant. Around these almost failed explosions, low
267: mass stars largely enhanced in the light elements may form. Since the ejecta of
268: these almost failed supernovae vary slowly with the mass of the remnant, this model
269: predicts also a wide spread of possible overabundances with respect to the solar
270: values. It is tempting to suggest that at least part of the C-rich
271: very metal poor stars, which comprise as many as 15\%-25\% of the stars
272: with [Fe/H]$<-2.5$ (Beers, 1999), are formed in this way.
273: It goes without saying that this scenario naturally explains the bulk of
274: "standard" EMPS as well, because these stars would simply be the ones born far
275: from these almost "failed" explosions. They would always have (like they do have) a
276: quite similar pattern of the heavy elements but much lower overabundances (and star
277: to star scatter) of the light elements relative to Fe.
278:
279: Figure \ref{fitfinal} shows all the abundances predicted by the specific model
280: described above; the measure of any of them in principle will constitute a possible
281: test of the present scenario. However, since the predicted element ratios do not
282: depend exclusively on the scenario itself but reflect (to some extent) also all the
283: assumptions adopted to compute both the pre explosive evolution of the stars (input
284: physics and the like) as well as the explosive nucleosynthesis, it would be wise to
285: measure as many element abundances as possible to really constrain any scenario.
286:
287: Let us eventually note that this model also predicts an initial He abundance in
288: HE0107-5240 0.015 dex higher than the primordial value and a ratio $\rm
289: ^{12}C/^{13}C=240$, a value compatible with the lower limit of 30 determined by
290: Christlieb et al (2002).
291:
292:
293:
294: \begin{references}
295:
296: \reference{} Aoki, W., Norris, J.E., Ryan, S.G., Beers, T.C., and Ando, H. 2002, ApJ, 576, L141
297: \reference{} Beers, T.~C.\ 1999, \apss, 265, 547
298: \reference{} Bonifacio, P., Limongi, M., and Chieffi, A. 2003, Nature, 422, 834
299: \reference{} Bonifacio, P., Molaro, P., Beers, T.C., and Vladilo, G. 1998, \aap, 332, 672
300: \reference{} Carretta, E., Gratton, R., Cohen, J.G., Beers, T.C., and Christlieb, N. 2002, AJ, 124, 481
301: \reference{} Caughlan, G.R. and Fowler, W.A., 1988, A.D.N.D.T., 40, 283
302: \reference{} Caughlan, G.R., Fowler, W.A., Harris M.J. and Zimmerman, B.A., 1985, A.D.N.D.T., 32, 197
303: \reference{} Chieffi, A., Dominguez, I., Limongi, M., and Straniero, O. 2001, ApJ, 554, 1159
304: \reference{} Chieffi, A., and Limongi, M. 2002, ApJ, 577, 281
305: \reference{} Chieffi, A., Limongi, M., and Straniero, O. 1998, ApJ, 502, 737
306: \reference{} Christlieb, N. et al. 2002, Nature, 419, 904
307: \reference{} Depagne, E. et al. 2002, A\&A, 390, 187
308: \reference{} Fran{\c c}ois, P. et al. 2003, \aap, 403, 1105
309: \reference{} Fujimoto, M.Y., Iben, I., and Hollowell, D. 1990, ApJ, 349, 580
310: \reference{} Heger, A. and Woosley, S.E. 2002, ApJ, 567, 532
311: \reference{} Imbriani, G., Limongi, M. Gialanella, L., Terrasi, F., Straniero, O., and Chieffi, A 2001,
312: ApJ, 558, 903
313: \reference{} Limongi, and Chieffi, A. 2002, PASA, 19, 1
314: \reference{} Limongi, M., Straniero, O., and Chieffi, A. 2000, ApJS, 129, 625
315: \reference{} McWilliam, A., Preston, G.W., Sneden, C., and Searle, L. 1995, \aj, 109, 2757
316: \reference{} Norris, J.E., Ryan, S.G., and Beers, T.C. 2001, ApJ, 561, 1034
317: \reference{} Picardi, I. et al. 2003, in preparation
318: \reference{} Ryan, S.G., Norris, J.E., and Beers, T.C. 1996, \apj, 471, 254
319: \reference{} Umeda, H. and Nomoto, K., 2003, Nature, 422, 871
320: \reference{} Woosley, S.E., and Weaver, T.A. 1982, in Supernovae: A Survey of Current Research,
321: ed. M.J. Rees \& R.J. Stoneham (Dordrecth: Reidel), 79
322:
323: \end{references}
324:
325:
326: \begin{figure} % fig.1
327: \plotone{f1.eps}
328: \caption{
329: Comparison between the [X/Fe] observed in HE0107-5240 (filled squares) and the ones
330: predicted by the explosion of stellar models of different masses:
331: $\rm 15~M_\odot$ ({\em open triangles}),
332: $\rm 20~M_\odot$ ({\em open hexagons}),
333: $\rm 25~M_\odot$ ({\em open circles}),
334: $\rm 35~M_\odot$ ({\em filled triangles}),
335: $\rm 50~M_\odot$ ({\em filled exagons}) and
336: $\rm 80~M_\odot$ ({\em filled circles}).
337: For each model the mass cut has been chosen to fit the observed [Ca/Fe].
338: \label{inconsistencies}}
339: \end{figure}
340:
341: \begin{figure}
342: \plotone{f2.eps}
343: \caption{ Trends of [C/Mg] ({\em thick solid line}), [N/Mg] ({\em thick dotted
344: line}) and [Na/Mg] ({\em thick dashed line}) as a function of the mass of the
345: remnant for the $\rm 35~M_\odot$ model. The internal profile of He ({\em solid}), C
346: ({\em dotted}), Na ({\em short dashed}) and Mg ({\em long dashed}) are shown as a
347: reference in background as thin lines.
348: \label{trendlight}}
349: \end{figure}
350:
351: \begin{figure} % fig.3
352: \plotone{f3.eps}
353: \caption{
354: Comparison between the element abundance ratios ({\em open circles}) predicted by the
355: two supernova model and those observed in HE0107-5240 ({\em filled squares}).
356: \label{fitfinal}}
357: \end{figure}
358:
359:
360: \end{document}
361: