BD+44°493: A NINTH MAGNITUDE MESSENGER FROM THE EARLY UNIVERSE; CARBON ENHANCED AND BERYLLIUM POOR^*

Chemical abundance analyses of metal-poor stars based on high-resolution spectroscopy play an important role in our understanding of the chemical evolution of the early universe (e.g., Beers & Christlieb 2005) because their atmospheres preserve (in most cases) the chemical composition of the gas from which they formed. In particular, the nucleosynthesis signatures from the first generations of (likely massive) stars are believed to be recorded by lower-mass stars at the lowest observed metallicity. However, the number of extremely metal-poor stars having [Fe/H] <−3.5 known to date is very small (∼15), and the chemical-enrichment processes that were in operation at such low metallicity are still unclear.

Surveys for metal-poor stars over the past few decades have shown that a large fraction of metal-poor stars exhibit enhancements of carbon. Such stars are often referred to as carbon-enhanced metal-poor (CEMP) stars. In particular, all three ultra metal-poor ([Fe/H] <−4) stars discovered so far exhibit strong C excesses. Several sources that could produce a significant amount of carbon in the early universe have been suggested to account for these observations. For example, mass transfer from an asymptotic giant branch (AGB) companion is proposed to explain the so-called "CEMP-s" stars, which have excesses of s-process elements as well as carbon. However, some CEMP stars, in particular for [Fe/H] <−2.6 (Aoki et al. 2007), are not rich in s-process elements ("CEMP-no"). Among the handful of CEMP stars with [Fe/H] <−3.5 studied to date, none exhibits a clear excess of the s-process elements. One must therefore consider other scenarios where the C excess at extremely low metallicity may have been produced by previous generations of stars, e.g., mass loss from rapidly rotating massive stars (Meynet et al. 2006) or the ejecta of so-called "faint" supernovae (Umeda & Nomoto 2005). In order to distinguish between these, and other possible scenarios, measured abundances of N and O, as well as for the heavy neutron-capture elements, are required.

In this Letter, we present a one-dimensional LTE abundance analysis of the extremely metal-poor ([Fe/H]=−3.7) CEMP star BD+44°493. Anthony-Twarog & Twarog (1994) reported the metallicity of this object to be [Fe/H]=−2.71, based on uvby photometry, while Carney et al. (2003) pointed out that a synthetic spectrum with lower metallicity produced a better fit to the spectrum they used to measure its radial velocity. Our analysis reveals the presence of strong CH bands, which might lead to the overestimate of this star's iron abundance, and that the lower metallicity speculation by Carney et al. (2003) is indeed correct. BD+44°493, with apparent magnitude V = 9.1, is quite remarkable in that it has the lowest iron abundance of any star with V < 12 found to date, and is by far the brightest extremely low metallicity star among all objects that have reported [Fe/H] based on high-resolution spectroscopy (according to the SAGA database; Suda et al. 2008).

Thanks to its brightness, our high signal-to-noise ratio (S/N) UV spectrum also allows inspection of the strength of the Be ii lines at 3130 Å for BD+44°493, permitting measurement of a meaningful upper limit for beryllium at the lowest metallicity yet achieved. In the past decade, observations have revealed a clear linear correlation between stellar Be abundances and metallicity (e.g., Boesgaard et al. 1999). This result was at first surprising, because standard cosmic-ray spallation processes, in which high-energy protons and α-particles accelerated by supernovae impact interstellar CNO, predicted a quadratic correlation ("secondary" process). The observations indicated the need for a "primary" process, in which accelerated CNO nuclei impinge on interstellar protons and α-particles (e.g., Yoshii et al. 1997). Such a primary process is expected to be especially important in the early universe, due to the lack of large amounts of interstellar CNO. The mechanism to accelerate CNO nuclei is still uncertain, but superbubbles (e.g., Parizot & Drury 1999) and type Ic supernovae (e.g., Nakamura & Shigeyama 2004) have been suggested as possibilities. Our analysis of Be for BD+44°493 provides new insight for the origin of Be in the early universe, as well as a constraint on models that predict a low-metallicity plateau of Be abundances, including the Inhomogeneous Big Bang Nucleosynthesis (IBBN) scenario (Orito et al. 1997).

High-resolution spectroscopy of BD+44°493 was carried out on 2008 October 5 with HDS (Noguchi et al. 2002) on the Subaru Telescope for three grating settings, covering 3100–9350 Å with a resolving power of R ∼ 90, 000 (no on-chip binning). The exposure time for the UV observation is 120 minutes, while those for the other settings used are 15 minutes. In order to obtain a spectrum over 3900–3950 Å, which fell in the CCD gap in the October run, an additional exposure of 5 minutes was made with the same instrument on 2008 November 15, with a resolving power of R ∼ 60, 000 and 2 × 2 on-chip binning.

Data reduction was performed with the IRAF⁵ echelle package. The S/N per pixel achieved was ∼100/1 at 3100 Å, ∼400/1 at 4500 Å, and ∼250/1 at 5000 Å. The spectrum covering the range 3400–6800 Å, where the S/N is the highest and many atomic lines exist, is used for the abundance analysis for most elements. We use the UV spectrum (<3400Å) to measure the Be lines and the molecular features of NH and OH. Equivalent widths are measured by fitting Gaussian profiles to isolated atomic lines. For blended lines and molecular features, the spectrum synthesis technique is employed.

We adopt the effective temperature T_eff of 5510 K determined by Carney et al. (2003), who used the color-T_eff relations derived by Alonso et al. (1999) based on the infrared flux method. Although they adopted a higher metallicity of [Fe/H] =−2.7 in the calculation, the results hardly change even if a lower metallicity is assumed. The surface gravity is obtained from the LTE ionization equilibrium between Fe i and Fe ii, and the microturbulent velocity is determined from 107 Fe i lines by demanding that no trend of Fe abundances with equivalent widths is found. For this analysis, as well as for the following measurements of elemental abundances, a one-dimensional LTE abundance analysis was performed using the grid of ATLAS9 NEWODF model atmospheres (Kurucz 1993; Castelli & Kurucz 2003). The results are log g = 3.7 and v_t = 1.3 km s^-1; the derived metallicity is [Fe/H] =−3.68. The non-LTE (NLTE) effect on Fe i is estimated to be 0.1–0.3 dex by previous studies (e.g., Asplund 2005). NLTE effects on determinations of abundances and stellar parameters will be discussed separately (H. Ito et al. 2009, in preparation). For elements for which no line is detected, we estimate upper limits based on the 3σ error of equivalent width measurement, employing the formula given in Norris et al. (2001). The derived abundances and uncertainties are listed in Table 1.

Table 1. Abundance Results and Uncertainties

X	Ion	Nalines	log (X)	[X/Fe]b	Random error	Total error
Li	1	Syn	1.04	...	0.03	0.09
Be	2	Syn	<−2.0	...	⋅⋅⋅	⋅⋅⋅
C	CH	Syn	6.02	+1.31	0.10	0.23
N	NH	Syn	4.42	+0.32	0.25	0.32
O	OH	Syn	6.57	+1.59	0.15	0.26
Na	1	2	2.76	+0.27	0.09	0.13
Mg	1	7	4.37	+0.52	0.04	0.10
Al	1	3961 Å	2.12	−0.57	0.13	0.16
Si	1	3905 Å	4.24	+0.41	0.13	0.19
Ca	1	9	2.90	+0.27	0.02	0.06
Sc	2	6	−0.20	+0.43	0.04	0.12
Ti	1	5	1.54	+0.32	0.02	0.11
Ti	2	13	1.53	+0.31	0.01	0.11
V	1	4379 Å	<−1.00	<−0.42	⋅⋅⋅	⋅⋅⋅
Cr	1	6	1.52	−0.44	0.04	0.11
Mn	1	2	0.59	−1.22	0.09	0.14
Fe	1	107	3.77	−3.68	0.01	0.11
Fe	2	7	3.77	−3.68	0.03	0.11
Co	1	6	1.72	+0.48	0.02	0.10
Ni	1	5	2.59	+0.04	0.03	0.11
Zn	1	4722 Å	<1.29	<0.37	⋅⋅⋅	⋅⋅⋅
Sr	2	2	−1.02	−0.26	0.09	0.15
Y	2	3774 Å	−1.81	−0.34	0.13	0.17
Zr	2	4149 Å	<−1.31	<−0.22	⋅⋅⋅	⋅⋅⋅
Ba	2	2	−2.10	−0.59	0.09	0.15
Eu	2	3819 Å	<−3.03	<+0.13	⋅⋅⋅	⋅⋅⋅
Pb	1	4057 Å	<−0.27	<+1.41	⋅⋅⋅	⋅⋅⋅

Notes. ^a"Syn" indicates the use of spectrum synthesis. In the case that the number of lines used is one, as well as for upper limits, the wavelength of the line is given. ^bSolar abundances were taken from Asplund et al. (2005). For iron [Fe/H] is given.

Download table as:  ASCIITypeset image

We measure C abundances from the CH G band at 4312 Å (Figure 1) as well as the feature at 4324 Å. The CH line list of Aoki et al. (2002) is used. The two C abundances differ by about 0.1 dex. We take the simple average to obtain our final C abundance, [C/Fe] =+1.3, which indicates that BD+44°493 is a carbon-enhanced object.

Zoom In Zoom Out Reset image size

The N abundance is determined from the NH band feature at 3360 Å. The synthetic spectra are calculated using the NH line list of Kurucz (1993) (Figure 2). Aoki et al. (2006) found that the gf-values of the Kurucz (1993) NH line list needed to be corrected to reproduce the solar N abundance of Asplund et al. (2005). The same correction is applied to our analysis, with the result that N is not overabundant in BD+44°493 ([N/Fe] =+0.3).

Zoom In Zoom Out Reset image size

We measure O abundances via spectrum synthesis to match the OH features at 3123.95, 3127.69, 3128.29, 3139.17, and 3140.73 Å. All values agree with each other within 0.2 dex; we adopt their straight average as our final O abundance. The result is [O/Fe] =+1.6, indicating that BD+44°493 has a strong excess of O. However, the O i triplet at 7775 Å is not detected in our spectrum, resulting in an upper limit from this feature to be [O/Fe] <+1.41, which is lower than the O abundance derived from the OH lines. Such a problem was also found for the carbon-enhanced hyper metal-poor star HE 1327−2326 by Frebel et al. (2006), and a significant three-dimensional effect on the OH features was suggested. It should be noted that the C/O ratio, which is important for the discussion of the origin of the carbon excess for this star, is expected to be comparatively robust, because both C and O abundances are determined from molecular features.

The Li abundance is derived from the Li i doublet at 6708 Å using the spectrum synthesis approach. The derived abundance, A(Li)=1.04, agrees well with the values found in other metal-poor subgiants of similar effective temperature (see Figure 5 in Frebel et al. 2007).

The Be ii resonance lines at 3130 Å are not detected. Using the line list based on Kurucz (1993) and Boesgaard et al. (1999), we synthesize spectra for different Be abundances and compare them with the observed spectrum (Figure 3). The upper limit estimated by this approach is A(Be) < −2.0. The upper limit estimated from the nondetection of 3131 Å, employing the formula mentioned above, is A(Be) < −2.18. We adopted an upper limit of A(Be) < −2.0, which is the lowest Be abundance limit so far for metal-poor dwarfs or subgiants that have normal Li abundances.

Zoom In Zoom Out Reset image size

The abundances of the α-elements Mg, Si, Ca, and Ti are enhanced by ∼0.3–0.5 dex with respect to Fe, as found in most metal-poor stars of the halo.

No excess of neutron-capture elements is found. Note that the measurements (or upper limits) for Sr, Zr, and Y are all sub-solar. The Ba abundance ([Ba/Fe] =−0.59) is derived from the Ba ii resonance lines at 4554 Å and 4934 Å, including the effect of hyperfine splitting using the line list of McWilliam (1998) (assuming the isotope ratios of the r-process component of solar-system material). We only have an upper limit available for Eu, which is quite low as well ([Eu/Fe] <+0.13).

Random errors arising from the measurements of atomic lines are estimated to be σN^−1/2, where σ is the standard deviation of the individual derived abundances and N is the number of lines used in the analysis. When the number of lines is smaller than five, we use the σ of Fe i instead. For molecular features, abundance variations from different features and continuum placement are taken into account to estimate random errors. In order to examine how stellar parameters affect abundances, we performed an abundance analysis after changing each stellar parameter by its uncertainty (ΔT_eff = 100 K,  Δlog g = 0.3 dex,  Δv_t = 0.3 km s^-1). The root-sum-square of all the error sources is adopted as the total error of log (X).

From its overabundance of C ([C/Fe] > + 1) and low Ba abundance ([Ba/Fe] <+0.5), BD+44°493 is classified as a "CEMP-no" star, according to the nomenclature of Beers & Christlieb (2005). The abundance pattern of BD+44°493 is very similar to that of HE 1300+0157, which is a CEMP-no subgiant with [Fe/H]=−3.9 (Frebel et al. 2007), whose [C/Fe] and [O/Fe] are comparable with those of BD+44°493. The brightness of BD+44°493 enables us to measure the abundances of N, Sr, Y, and Ba, which were not determined for HE 1300+0157. We can also set stronger upper limits for Zn, Pb, and so on. We now discuss the origin of the C excess in BD+44°493 and implications of the very low Be upper limit.

4.1. The Origin of the Carbon Excess

4.2. Implications of the Low Beryllium Abundance

Figure 4 shows the Be abundance upper limit for BD+44°493 in the A(Be) versus [Fe/H] plane, along with the results of previous studies. Our result is consistent with a linear trend between Be and Fe. Although Be is a fragile element, previous measurements for subgiants having T_eff ∼ 5500 K showed no depletion. It is of significance that the Be abundance keeps decreasing and exhibits no plateau, at least to the level of A(Be) < −2.0. Previous measurements of Be for stars with [Fe/H] <−2.5 permitted the possibility of a Be plateau around A(Be) ∼ −1.3 (Primas et al. 2000a), which would now be clearly called into question.

Zoom In Zoom Out Reset image size

Models that have been invoked to account for a Be plateau include IBBN (Orito et al. 1997; Suzuki et al. 1999), interstellar accretion (Yoshii et al. 1995), pre-Galactic cosmic rays generated by massive stars (Kusakabe 2008), and a particular superbubble model (Vangioni-Flam et al. 1998). Some of them predict a plateau as high as A(Be) = −1.5 ∼ −2.0. The Be upper limit of BD+44°493 provides a meaningful constraint on these models, as it appears that any Be plateau, if it exists at all, must occur at A(Be) < −2.0.

Our analysis is the first attempt to measure a Be abundance for a CEMP star. Since Be is produced via the spallation of CNO nuclei, their abundances, especially O abundances, have been expected to correlate with Be abundances. However, our low Be upper limit shows that the high C and O abundances in BD+44°493 are irrelevant to its Be abundance. Moreover, the origin of the C excess in CEMP-no stars like BD+44°493, which we have argued is most likely due to faint supernovae, is unlikely to be a significant source of high-energy CNO nuclei that participate in the primary process associated with Be production.

W.A. is supported by a Grant-in-Aid for Science Research from JSPS (grant 18104003). T.C.B. acknowledges partial funding of this work from grants PHY 02-16783 and PHY 08-22648: Physics Frontier Center/Joint Institute for Nuclear Astrophysics (JINA), awarded by the U.S. National Science Foundation.