Possible signature of hypernova nucleosynthesis in a beryllium rich halo dwarf
ABSTRACT As part of a large survey of halo and thick disc stars, we found one halo star, HD 106038, exceptionally overabundant in beryllium. In spite of its low metallicity, [Fe/H] = -1.26, the star has log(Be/H) = -10.60, which is similar to the solar meteoritic abundance, log(Be/H) = -10.58. This abundance is more than ten times higher the abundance of stars with similar metallicity and cannot be explained by models of chemical evolution of the Galaxy that include the standard theory of cosmic-ray spallation. No other halo star exhibiting such a beryllium overabundance is known. In addition, overabundances of Li, Si, Ni, Y, and Ba are also observed. We suggest that all these chemical peculiarities, but the Ba abundance, can be simultaneously explained if the star was formed in the vicinity of a hypernova. Comment: Accepted for publication in the MNRAS letters
arXiv:0801.0965v1 [astro-ph] 7 Jan 2008
Mon. Not. R. Astron. Soc. 000, 1–5 (2007) Printed 2 February 2008(MN LATEX style file v2.2)
Possible signature of hypernova nucleosynthesis in a
beryllium rich halo dwarf⋆
R. Smiljanic1,2†, L. Pasquini2†, F. Primas2, P. A. Mazzali3,4,
D. Galli5, and G. Valle6
1Universidade de S˜ ao Paulo, IAG, Dpt. de Astronomia, Rua do Mat˜ ao 1226, S˜ ao Paulo-SP 05508-090, Brazil
2European Southern Observatory,Garching bei M¨ unchen, Germany
3Max-Planck-Institut f¨ ur Astrophysik, Garching bei M¨ unchen, Germany
4INAF - Osservatorio Astronomico de Trieste, Trieste, Italy
5INAF - Osservatorio Astrofisico di Arcetri, Firenze, Italy
6Dipartamento di Fisica, Universit´ a di Pisa, largo Pontecorvo 3, Pisa 56127, Italy
Accepted 2007. Received 2007; in original form 2007
As part of a large survey of halo and thick disc stars, we found one halo star, HD
106038, exceptionally overabundant in beryllium. In spite of its low metallicity, [Fe/H]
= −1.26, the star has log(Be/H) = −10.60, which is similar to the solar meteoritic
abundance, log(Be/H) = −10.58. This abundance is more than ten times higher the
abundance of stars with similar metallicity and cannot be explained by models of chem-
ical evolution of the Galaxy that include the standard theory of cosmic-ray spallation.
No other halo star exhibiting such a beryllium overabundance is known. In addition,
overabundances of Li, Si, Ni, Y, and Ba are also observed. We suggest that all these
chemical peculiarities, but the Ba abundance, can be simultaneously explained if the
star was formed in the vicinity of a hypernova.
Key words: stars: abundances – stars: chemically peculiar – stars: individual: HD
The single stable isotope of beryllium,9Be, is a pure prod-
uct of cosmic-ray spallation of heavy (mostly CNO) nuclei
(Reeves, Fowler & Hoyle 1970). Analyses of Be abundances
in metal poor stars (Molaro et al. 1997; Boesgaard et al.
1999) have found a relationship between [Fe/H]1
log(Be/H) with slope close to one, and between [O/H] and
log(Be/H) with slope between 1 and 1.5, depending on the
oxygen indicator used. Independently of the behaviour of
the [O/Fe] ratio at lower metallicities, these results suggest
a primary production of Be in the early Galaxy (King 2001).
As a primary element, and assuming cosmic-rays to be
globally transported across the early Galaxy, Be may show
a smaller scatter than the products of stellar nucleosynthe-
sis (Suzuki & Yoshii 2001) at a given time, suggesting its
potential use as a cosmochronometer.
⋆Based on observations made with ESO VLT, at Paranal Ob-
servatory, under programmes 076.B-0133 and on data obtained
from the ESO/ST-ECF Science Archive Facility.
† E-mail: firstname.lastname@example.org (RS); email@example.com (LP)
1[A/B] = log [N(A)/N(B)]⋆ - log [N(A)/N(B)]⊙
So far, the linear relations appear to be very tight, show-
ing a surprisingly low scatter comparable to the measure-
ment errors. This picture, however, might change with the
increase of the samples analysed, as hinted by the results of
Boesgaard & Novicki (2006). Nevertheless, turn-off stars of
the globular clusters NGC 6397 and NGC 6752 were found
(Pasquini et al. 2004, 2007) to have the same Be abundance
of field stars of the same metallicity. This strongly sup-
port the production of Be to be a global process. Ages de-
rived from these abundances, in a comparison with a model
of the evolution of Be with time (Valle et al. 2002), show
an excellent agreement with ages derived from theoretical
isochrones, supporting the use of Be as a cosmochronome-
ter. Moreover, Pasquini et al. (2005) showed that Be abun-
dances could be used to study the differences in the time
scales of star formation in the halo and the thick disc of the
In this letter, we report the discovery of an extremely
Be enriched halo star, HD 106038, with an abundance 1.2
dex higher than stars of similar metallicity. This unique star
deviates considerably from the observed relations of Be with
Fe and O. It was identified during the analysis of a large
2 R. Smiljanic et al.
Figure 1. Comparison between the spectra of HD 106038 (solid
line) and of HIP 7459 (dashed line), a star with close atmospheric
parameters and similar metallicity, in the Be region. The domi-
nating element of the nearby blended features are also indicated.
The V and Ti features have the same strength in the two stars
while some difference in the Zr line is noted.
sample containing near to one hundred halo and thick disc
stars (Smiljanic et al. 2008, in preparation).
Neither the standard scenario for Be production, in-
volving spallation of cosmic-rays on nuclei of the inter-
stellar medium (Valle et al. 2002), nor the superbubbles
(SBs) scenario (Parizot 2000) seem to be able to pro-
duce such Be enriched objects. The SBs model predict a
scatter in the Be abundance (Parizot & Drury 2000) that
may explain the stars found by Boesgaard et al. (1999) and
Boesgaard & Novicki (2006) which have similar atmospheric
parameters but Be abundances differing by ∼ 0.5 dex. The
very high Be abundance in HD 106038, however, would re-
quire an extremely poor mixing of the SNe ejecta with the
ISM which seems to be difficult to justify (Parizot 2000).
To the best of our knowledge, there is only one other
case of extremely Be enhanced star in the literature.2The
star J37 of the open cluster NGC 6633 was found by
Ashwell et al. (2005) to have log (Be/H) = −9.0 ± 0.5. The
chemical peculiarities of star J37 might be best explained
by the accretion of rocky material similar to chondritic me-
teorites (Ashwell et al. 2005). As we shall see below, the
accretion of such a material is unlikely for our population II
2 DATA AND ANALYSIS
The science raw data and calibration files of HD 106038 were
retrieved from the ESO science archive facility. The spectra
were originally obtained in 2000 April 12 with the UVES
2We exclude from the discussion the chemically peculiar A or F
stars with enhanced Be lines. The peculiar abundances of these
stars are thought to be caused by effects of diffusion. As shown by
Richard, Michaud & Richer (2002), these effects do not result in
overabundances in stars with similar temperature and metallicity
as HD 106038.
Figure 2. The abundance of Be as a function of [Fe/H]. The
stars HD 106038 and HIP 7459 (filled squares) are compared
to the linear relation defined by the stars from Boesgaard et al.
(1999) (open circles). The starred symbols are the stars from
Boesgaard et al. (1999) and Boesgaard & Novicki (2006). Two of
them deviate from the linear relation by ∼ 0.50 dex.
spectrograph (Dekker et al. 2000) of the ESO VLT at Cerro
Paranal, Chile. The data of HIP 7459 (CD−61 282), a halo
star used as comparison in this work, were obtained in 2005
September 22 with the same instrument. The spectra have
R ∼ 45000 and a final S/N ∼ 70 in the Be region.
For both stars, we adopt the atmospheric parameters
derived by Nissen & Schuster (1997). The parameters were
calculated with the standard spectroscopic approach using
Fe I and Fe II lines (Table 1). We refer the reader to the
original work for more details.
Abundances (Table 1) were derived through the syn-
thesis of the spectrum around the Be II resonance lines
at 3131.065 ˚ A and 3130.420 ˚ A. The codes for calculat-
ing synthetic spectra are described in Coelho et al. (2005).
Grids of model atmospheres without overshooting calcu-
lated by the ATLAS9 program (Castelli & Kurucz 2003)
were adopted. The list of atomic lines is that compiled by
Primas et al. (1997) and the molecular line list is described
in Coelho et al. (2005). A solar Be abundance was derived
using the UVES solar spectrum. We estimate the total un-
certainty from atmospheric parameters, continuum place-
ment, and synthetic fit affecting the Be abundance to be σ
= ± 0.13 dex. Zero point errors might also be present due,
for example, to the adopted model atmosphere. However,
we are conducting a differential analysis and these should
cancel out in a comparison between similar stars.
A comparison between the spectra of the two stars is
shown in Figure 1, in which the extreme enhancement of the
Be lines of HD 106038 when compared to the normal HIP
7459 is clear, confirming its very high Be overabundance.
We show in Figures 2 and 3 these two stars in the [Fe/H]
vs. log (Be/H) and [O/H] vs. log (Be/H) diagrams, respec-
tivelly, together with the stars from Boesgaard et al. (1999)
and Boesgaard & Novicki (2006). Of particular interest are
the two stars from Boesgaard & Novicki (2006) that deviate
from the linear trend. The anomalous position of HD 106038
clearly stands out.
A beryllium rich halo dwarf3
Figure 3. The abundance of Be as a function of [O/H]. Symbols
are as in Fig. 2. The stars of Boesgaard & Novicki (2006) are not
shown since O abundances were not derived by these authors.
Table 1. The adopted atmospheric parameters and beryllium
abundances derived using synthetic spectra for HD 106038, HIP
7459, a comparison star, and for the Sun.
2.1 Chemical abundances in the literature
Information on other chemical abundances might help un-
derstanding the origin of the Be overabundance. An over-
abundance of CNO elements, for example, would offer a good
explanation for the enhancement, since these elements are
dominant in the production of Be by spallation processes.
Asplund et al. (2006) determined a lithium abundance
of A(7Li) = 2.48 and claimed a detection of6Li,6Li/7Li =
0.031, compatible with the other 8 detections out of a sample
of 26 stars. The high7Li abundance, however, results in a
high6Li abundance, A(6Li) = 0.97, while the mean for the
other detections is 0.80.
Its7Li is particularly remarkable since it is larger than
the Spite plateau (Spite & Spite 1982). The plateau as found
by Asplund et al. (2006) is at 2.22, which means HD 106038
has an abundance excess of ∆A(7Li) = 2.13. Given that
lithium is also expected to be produced by cosmic-ray spal-
lation it seems safe to conclude that both7Li and Be over-
abundances have the same origin.
Abundances of other elements available in the litera-
ture are shown in Figure 4. Abundances of Na, Ca, Mg, Si,
Ti, Cr, Ni, Y, and Ba are from Nissen & Schuster (1997).
The carbon abundance is taken from Fabbian et al. (2006)
and oxygen from Mel´ endez et al (2006). Abundances of Mn
and Sc are from Nissen et al. (2000) and of S and Zn from
Nissen et al. (2007). In the same figure we also show the
mean abundances of the samples analysed in these same
works for comparison.
Figure 4. Elemental abundances, [X/Fe], of HD 106038 deter-
mined in the literature. The abundances are represented by the
full squares. In this case the error bar denotes the actual uncer-
tainty quoted by the original work. The starred symbols indicate
the mean abundance of that element for the remaining sample as
found in the same work. In this case the error bar indicates the
range of abundances for that given element in the original work.
HD 106038 is clearly overabundant in Si, Ni, Y, and Ba, and show
slightly larger abundances of C, S, Mg, and Zn when compared
to the mean of the original samples.
In addition to Be and Li, the star also shows clear en-
hanced abundances of Si, Ni, and of the neutron capture el-
ements Y and Ba. An enrichment in s-process elements may
explain the enhanced Zr line in Figure 1. We also note the
possibly larger amounts of C, S, Mg, and Zn, though these
remain marginally compatible with the mean abundances of
Since the standard scenario for cosmic-ray spallation does
not explain the enhancement of Be in HD 106038, a pecu-
liar and/or rare event may be related to its formation. A
combination of two or more rare events to produce the ob-
served features is unlikely, we therefore concentrate on single
To reproduce the very particular chemical pattern of
HD 106038, a nucleosynthetic site must be able to over-
produce Si and Ni without overproducing other α and iron
group elements. Elements in normal halo stars with the same
metallicity as HD106038 come mostly from SNe II. It is
therefore unlikely that the same SNe II may produce the
observed enhanced [Si/Fe] and [Ni/Fe] ratios. Moreover, it
has about 16 times more Be than what models involving SN
II predict for its metallicity (Valle et al. 2002). SNe Ia pro-
duce large amounts of Fe, Ni, and other iron group elements
but are not expected to produce large amounts of O and
Si (Iwamoto et al. 1999). Therefore, in case of a contribu-
tion from SNe Ia, ratios such as [O/Fe] and [Si/Fe] would
fall below normal halo stars, contrary to what is observed.
Moreover, SNe Ia are expected to produce about one order of
magnitude less spallation products than SN II (Fields et al.
4 R. Smiljanic et al.
All these features, however, may possibly be found in
the ejecta of a hypernova (HNe), which can be enriched in
both intermediate mass elements (as S and Si) as in iron
group elements (Nakamura et al. 2001; Podsiadlowski et al.
2002). Moreover, HNe can produce large amounts of Be and
Li by spallation (Fields et al. 2002; Nakamura & Shigeyama
2004). Therefore, we suggest that the material which formed
this star was probably contaminated by the nucleosynthetic
products of a HNe.
Hypernovae are core-collapse SNe (usually of type Ic) with
exceptionally large kinetic energy production, resulting in
spectra dominated by very broad absorption line blends
(Mazzali et al. 2000). The energy released in the explosion
can be one order of magnitude larger than that of normal
core-collapse SNe (Iwamoto et al. 2003). Some hypernovae,
typically the most massive and energetic events, are linked
to Gamma-Ray bursts (Iwamoto et al. 1998).
Fields et al. (2002) and Nakamura & Shigeyama (2004)
calculated the yields of spallation products resulting from
HNe explosions. While Nakamura & Shigeyama (2004) cal-
culate the energy distribution of the ejecta with a hydro-
dynamic code and solve the cosmic-ray transfer equation,
Fields et al. (2002) use an empirical formula for the energy
distribution and do not solve the transfer equation but adopt
an approximation to have the mass fraction of the ejecta that
produces the spallation products. Nakamura & Shigeyama
(2004) claim the simplifications adopted by Fields et al.
(2002) to overestimate the yields by a factor ∼ 3.
The yield of Be per HNe can be one or two orders
of magnitude larger than the one per SNe II (Fields et al.
2002; Nakamura & Shigeyama 2004). However, as a rare
event, they are not major contributors of Be in the Galaxy.
Fields et al. (2002) predict
5.6 × 10−7(both ratios by number). The calculations by
Nakamura & Shigeyama (2004) predict7Li/9Be ∼ 4.2, also
by number. Both predictions are close to what is observed in
HD 106038; the ratio between the observed excess of7Li and
9Be is7Li/9Be = 5.6 (while the HNe is expected to have pro-
duced all the observed Be abundance, it is responsible only
for the excess of Li with respect to the primordial plateau).
We cannot estimate the contribution of the possible HNe on
the observed oxygen abundance, thus only a lower limit can
be placed, Be/O >1.9×10−7, given by the assumption that
all the observed oxygen has been produced by the HNe.
Both models, however, predict much more6Li than ob-
served,7Li/6Li ∼ 1.9 by Fields et al. (2002) and7Li/6Li ∼
1.2 by Nakamura & Shigeyama (2004), by number. The ob-
served ratio between the excess of7Li and the6Li abundance
is7Li/6Li ? 15. We consider this ratio an upper limit since,
given its fragility, some6Li has probably been destroyed in
previous evolutionary phases. The production of Be with-
out a corresponding production of6Li would be extremely
difficult to understand.
Nucleosynthetic calculations by Nakamura et al. (2001)
find the ejecta of HNe to have smaller amounts of C and O
and larger amounts of Si, S, and Ar, when compared to nor-
mal SNe. Nakamura et al. (2001) and Nomoto et al. (2006)
also note larger [(Zn,Co)/Fe] and smaller [(Mn,Cr)/Fe] ra-
tios. An overabundance of Zn, which is not observed, can be
7Li/9Be ∼ 8.6 and Be/O ∼
avoided with a deeper mas cut3, which would also result in
a larger [Ni/Fe] (Nomoto et al. 2006). These are in qualita-
tively agreement with the observations, supporting the HNe
The weak s-process in massive stars seems to efficiently
produce only elements with a mass number4up to 90
(Rayet & Hashimoto 2000). The Y overabundance may re-
quire an enhanced flux of neutrons, which would also con-
tribute to the production of Ni. The Ba overabundance, how-
ever, is more difficult to understand. It is not clear whether
this same mechanism would result in the overproduction of
Ba. Moreover, a significant amount of Ba is expected to be
produced only by the main s-process in AGBs or by the r-
process, usually associated to massive stars. Since pollution
by AGBs is not possible (see below), the Ba overabundance
is likely a product of a massive star. For example, although
not expected, Mazzali, Lucy & Butler (1992) found Ba to be
overabundant by a factor of 5 in the spectrum of SN 1987A.
Although Ba might pose a problem for our scenario, we re-
call that theoretical predictions for the r-process elements
in HNe are not available. Therefore, whether this scenario is
able to explain the Ba abundance is still an open question.
The HNe scenario, at least qualitatively, is able to ex-
plain most features observed in HD 106038 within a single
peculiar event. More work, however, is still needed to show
whether the scenario still holds quantitatively. A detailed
comparison with nucleosynthetic predictions of theoretical
models is necessary to validate or not the HNe hypothesis.
3.2 Other scenarios
In this subsection we present some alternative scenarios for
the origin of the Be enhancement in HD 106038, which were
discarded for the reasons presented below.
A pollution by AGB stars, or any kind of evolved star,
can at once be discarded. Although these may explain the
s-process elements, and maybe Li, they do not explain the
Be overabundance. On the contrary, the material ejected by
an AGB or by a massive star, after successive mixing events,
would be depleted in Be.
The SBs scenario (Parizot 2000) would be able to re-
produce the observed Li/Be and Be/O ratios only with an
extreme model where particles are accelerated from pure
SNe ejecta. The SBs evolution models, however, predict that
most material inside of a SB comes from the ISM. Moreover,
the remaining chemical peculiarities are not typical of SNe
II ejecta and thus can not be explained within the same
Another possibility is the engulfing of a sub-stellar ob-
ject, a planet or planetesimals debris as in star J37 of the
open cluster NGC 6633 (Ashwell et al. 2005). This, however,
can be excluded with a robust quantitative argument. The
accreted material would be confined to the surface convec-
tive layer of the star. In a metal poor star this layer is much
shallower than in a solar metallicity star. With the equation
given by Murray et al. (2001) we estimate the surface con-
vective layer of HD 106038 to have ∼ 4.5 × 10−3M⊙. The
3The coordinate, in mass, separating the part of the star that is
ejected and the one that forms the remnant.
4The mass number of Y is A = 89 and of Ba is A ∼ 136.
A beryllium rich halo dwarf5
mass of Be in this layer is ∼ 7.7×10−13M⊙ while in a star
with normal abundance of Be it would be ∼ 4.8×10−14M⊙.
Assuming the accreted material to have a composition sim-
ilar to chondrites meteorites (Lodders 2003) a mass of Fe of
5.3×10−6M⊙would also be accreted with the required mass
of Be. However, the total mass of Fe in the convective layer
of the star is ∼ 3.3×10−7M⊙. In this scenario all, or almost
all, Fe in the convective layer of the star would come from
the accreted material. HD 106038 would then originally be
a population III metal free star; an extremely unlikely pos-
sibility. In addition, we note that the most metal poor star
found to host planets has [Fe/H] = −0.68 (Cochran et al.
A longer exposure to EPs would be the natural expla-
nation if HD 106038, for some reason, was younger than halo
stars of the same metallicity. Its Be abundance would be a
result of the accumulated action of EPs in a cloud where
star formation was, somehow, delayed. Its Be abundance
higher than the solar photospheric one could be a sign of
a solar or younger age, in agreement with the suggestion
of Pasquini et al. (2004) that Be abundances could be used
as a cosmic clock. The position of the star in the HR dia-
gram, although not favourable for a good age determination,
favours an older age and argues against this hypothesis.
If the star originated in or near the Galactic centre, the
enhanced star formation and supernovae events could pro-
vide an enhanced EPs flux. This flux might also originate
from a non-stellar source such as the central black hole. A
bulge origin for the star, however, seems unlikely. In particu-
lar the abundances of Ni, Y, and Ba are not compatible with
the ones of bulge stars (McWilliam & Rich 1994). Since it
requires another source for the Ni and s-process overabun-
dances, we also discard this hypothesis.
This work was developed during the visit of R.S. to ESO
made possible by a CAPES fellowship (1521/06-3) and sup-
port from the ESO DGDF.
Ashwell, J. F., Jeffries, R. D., Smalley, B., Deliyannis, C.
P., Steinhauer, A., King, J. R., 2005, MNRAS, 363, L81
Asplund, M., Lambert, D. L., Nissen, P. E., Primas, F.,
Smith, V. V., 2006, ApJ, 644, 229
Boesgaard, A. M., Deliyannis, C. P., King, J. R., Ryan, S.
G., Vogt, S. S., Beers, T. C., 1999, AJ, 117, 1549
Boesgaard, A. M., Novicki, M. C., 2006, ApJ, 641, 1122
Castelli, F., Kurucz, R. L., 2003, in Proceedings of the IAU
Symp. 210, A20
Cochran, W. D., Endl, M., Wittenmyer, R. A., Bean, J. L.,
2007, ApJ, 665, 1407
Coelho, P., Barbuy, B., Mel´ endez, J., Schiavon, R. P.,
Castilho, B. V., 2005, A&A, 443, 735
Dekker, H., D’Odorico, S., Kaufer, A., Delabre, B., Kot-
zlowski, H., 2000, SPIE, 4008, 534
Fabbian, D., Asplund, M., Carlsson, M., Kiselman, D.,
2006, A&A, 458, 899.
Fields, B. D., Daigne, F., Cass´ e, M., Vangioni-Flam, E.,
2002, ApJ, 581, 389
Iwamoto, K., Brachwitz, F., Nomoto, K., Kishimoto, N.,
Umeda, H., Hix, W. R., Thielemann, F. K., 1999, ApJS,
Iwamoto, K., Mazzali, P. A., Nomoto, K., Umeda, H., et
al., 1998, Nature, 395, 672
Iwamoto, K., Nomoto, K., Mazzali, P. A., et al., 2003, Lec-
ture Notes in Physics, 598, 243
King, J. R., 2001, AJ, 122, 3115
Lodders, K., 2003, ApJ, 591, 1220
Mazzali, P. A., Iwamoto, K., Nomoto, K., 2000, ApJ, 545,
Mazzali, P. A., Lucy, L. B., Butler, K., 1992, A&A, 258,
McWilliam, A., Rich, R. M., 1994, ApJS, 91, 748
Mel´ endez, J., Shchukina, N. G., Vasiljeva, I. E., Ram´ ırez,
I, 2006, ApJ, 642, 1082
Molaro, P., Bonifacio, P., Castelli, F., Pasquini, L., 1997,
A&A, 319, 593
Murray, N., Chaboyer, B., Arras, P., Hansen, B., Noyes, R.
W., 2001, ApJ, 555, 801
Nakamura, K., Shigeyama, T., 2004, ApJ, 610, 888
Nakamura, T., Umeda, H., Iwamoto, K., Nomoto, K.,
Hashimoto, M., Hix, W. R., Thielemann, F. K., 2001,
ApJ, 555, 880
Nissen, P. E., Akerman, C., Asplund, M., Fabbian, D., Ker-
ber, F., 2007, A&A, 469, 31
Nissen, P. E., Chen, Y. Q., Schuster, W. J., Zhao, G., 2000,
A&A, 353, 722
Nissen, P. E. & Schuster, W. J., 1997, A&A, 326, 751
Nomoto, K., Tominaga, N., Umeda, H., Kobayashi, C.,
Maeda, K., 2006, Nuclear Physics A, 777, 424
Parizot, E., 2000, A&A, 362, 786
Parizot, E., Drury, L., 2000, A&A, 356, L66
Pasquini, L., Bonifacio, P., Randich, S., Galli, D., Gratton,
R. G., 2004, A&A, 426, 651
Pasquini, L., Bonifacio, P., Randich, S., Galli, D., Gratton,
R. G., Wolff, B., 2007, A&A, 464, 601
Pasquini, L., Galli, D., Gratton, R. G., Bonifacio, P.,
Randich, S., Valle, G., 2005, A&A, 436, L57
Podsiadlowski, P., Nomoto, K., Maeda, K., Nakamura, T.,
Mazzali, P., Schmidt, B., 2002, ApJ, 567, 491
Primas, F., Duncan, D. K., Pinsonneault, M. H., Deliyan-
nis, C. P., Thorburn, J. A., 1997, ApJ, 480, 784
Rayet, M., Hashimoto, M., 2000, A&A, 354, 740
Richard, O., Michaud, G., Richer, J., 2002, ApJ, 580, 1100
Reeves, H., Fowler, W. A., Hoyle, F., 1970, Nature, 226,
Spite, F., Spite, M., 1982, A&A, 115, 357
Suzuki, T. K., Yoshii, Y., 2001, ApJ, 549, 303
Valle, G., Ferrini, F., Galli, D., Shore, S. N., 2002, ApJ,