LAMOST J011939.222−012150.45: The most barium-enhanced CEMP-s turnoff star

Abstract

We have performed chemical abundance analyses for a newly discovered metal-poor turn-off star (T_eff = 6276 K, log g = 3.93, [Fe|$/$|H] = −2.93), LAMOST J011939.222−012150.45, based on high-resolution and high signal-to-noise ratio spectra in both optical and near-UV obtained by Subaru. Abundances have been derived for 20 elements, including 11 light elements such as C, N, Na, Mg, etc., and 9 neutron-capture elements from Sr to Pb. This object is a carbon-enhanced metal-poor star with a large carbon excess of [C|$/$|Fe] = +2.26. LAMOST J011939.222−012150.45 shows extreme enhancement in s-process elements, especially for Ba, La, and Pb ([Ba|$/$|Fe] = +3.16 ± 0.18, [La|$/$|Fe] = +2.29 ± 0.24, [Pb|$/$|Fe] = +3.38 ± 0.12). A very clear radial velocity variation has also been detected, providing evidence of the existence of a companion. Interestingly, even without any scaling, the observed abundance pattern from light to heavy neutron-capture elements agrees well with predictions of accretion from a companion asymptotic giant branch (AGB) star. Considering the evolutionary status of this object, its surface material is very likely to be completely accreted from its AGB companion and has been preserved until today.

1 Introduction

Very metal-poor (VMP; [Fe|$/$|H] < −2.0) stars and extremely metal-poor (EMP; [Fe|$/$|H] < −3.0) stars have been observed to trace the chemical composition of the early interstellar medium of the universe, which reflects the nucleosynthesis products of the first stars and their associated supernovae. Decades of studies on metal-poor stars have revealed that more than 10% show a significant excess of carbon, and the frequency of carbon-enhanced metal-poor (CEMP) stars increases at lower metallicities (Beers & Christlieb 2005; Carollo et al. 2012; Spite et al. 2013; Yong et al. 2013; Frebel & Norris 2015). Based on the abundance pattern of neutron-capture (n-capture) elements, CEMP stars are further classified into four sub-classes (Beers & Christlieb 2005): CEMP-no (with no enhancement in neutron-capture elements), CEMP-s (enhanced in s-process elements), CEMP-r/s (enhanced in both r- and s-process elements), and CEMP-r (enhanced in r-process elements). The binary frequency of CEMP-s (including CEMP-r/s) stars is ∼82% ± 10% (Starkenburg et al. 2014; Hansen et al. 2016a, 2016b; Jorissen et al. 2016), which supports the scenario that in these objects the excesses of s-process elements and of carbon can be attributed to mass transfer from the envelope of an asymptotic giant branch (AGB) companion (Suda et al. 2004; Lucatello et al. 2005; Aoki et al. 2008),

Meanwhile, the majority of CEMP stars studied so far are located on the giant branch. This is probably because giants are relatively brighter and show stronger CH-band absorption in their spectra, and are thus easier to identify when selecting CEMP (candidate) stars. However, the outer layers of most giants have already been affected by the mixing during the first dredge-up. By contrast, main-sequence turnoff stars basically preserve in their surface atmosphere the initial material from their birth place or the accreted material from the companion star. Hence, abundance analyses of CEMP turnoff stars are of great importance in studying the origin of CEMP stars.

In this letter we report on the discovery of a CEMP turn-off star that shows large excesses of heavy neutron-capture elements, and further discuss the constraints on s-process nucleosynthesis at low metallicity obtained from this unique object.

2 Observations and data reduction

LAMOST J011939.222−012150.45 was identified from a large sample of VMP stars (Li et al. 2018) selected from Data Release 3 of the LAMOST¹ (The Large Sky Area Multi-Object Fiber Spectroscopic Telescope, also known as the Guoshoujing telescope, GSJT) survey (Zhao et al. 2006, 2012; Cui et al. 2012; Luo et al. 2012). Readers may refer to Li et al. (2018) for more details about the VMP candidate selection and the LAMOST–Subaru project on VMP stars. High-resolution optical and near-UV spectra of this object were later obtained with the Subaru Telescope High Dispersion Spectrograph (HDS: Noguchi et al. 2002) in 2015 and 2017.

The high-resolution spectra of LAMOST J011939.222−012150.45 cover the wavelength range from 4000 Å to 6800 Å with a resolution of R = 45000, and from 3520 Å to 5260 Å with R = 60000. The signal-to-noise ratio (S|$/$|N) per pixel around 4250 Å is given in table 1. The weather conditions for all the runs were clear and the seeing size was at the |${0{^{\prime\prime}_{.}}6}$|–|${0{^{\prime\prime}_{.}}8}$| level. Data reduction was carried out with standard procedures using IRAF.² The uncertainty of the wavelength calibration is 1–2 mÅ.

Radial velocities were obtained using the IRAFfxcor procedure, where a synthetic high-resolution spectrum with low metallicity was employed as a template for cross-correlation. The radial velocities derived from high-resolution spectra are V_helio = −92 km s⁻¹ in 2015 November and V_helio = −55 km s⁻¹ in 2017 August. The velocities estimated from eight LAMOST spectra show a variation from −90 km s⁻¹ to −38 km s⁻¹ from 2013 October through 2014 January.

3 Stellar parameters and abundance analysis

The equivalent widths of atomic lines are measured by fitting Gaussian profiles to isolated atomic absorption lines which mainly come from the line list of Frebel and Norris (2013) and Aoki et al. (2013). The effective temperature T_eff of the program stars was estimated from the dereddened (V − K)₀, adopting the calibration from Alonso, Arribas, and Martinez-Roger (1996). We have not adopted the spectroscopic method to estimate the effective temperature, primarily because the number of Fe i lines is not sufficient to derive a reliable temperature. For the photometric data [V = 15.002, K_s = 13.775, E(B − V) = 0.0468], we have adopted the V magnitude from APASS (AAVSO Photometric All-Sky Survey: Henden et al. 2012) and the K magnitude converted from K_s from 2MASS (Two Micron All Sky Survey: Skrutskie et al. 2006) following the calibration described in equation (1c) of Ramírez and Meléndez (2004). We adopted the reddening correction E(B − V) from the dust map of Schlafly and Finkbeiner (2011). The derived T_eff from (V − K)₀ is 6276 ± 103 K. The surface gravity log g was estimated based on the corrected distance (1515.69 pc; Bailer-Jones et al. 2013) obtained from Gaia DR2 (Gaia Collaboration 2018; Lindegren et al. 2018), with an assumed mass of |$0.8\, M_{\odot }$| and M_{bol, |${\odot }$|} = 4.74. The micro-turbulent velocity ξ is determined by minimizing the trend between the derived abundances and measured equivalent widths of Fe i lines. The derived log g and ξ are 3.93 dex and 0.40 km s⁻¹, respectively.

To check the reliability of our measurement, we re-analyzed the spectrum of the CEMP star LP 706-7, which has been studied by Aoki et al. (2002a) and used as a comparison star here. The mean difference between the measured equivalent widths and those from the previous study is ∼0.7 mÅ with σ = 1.7 mÅ. Using our method, the derived parameters for LP 706-7 are T_eff = 6250 K, log g = 4.45, ξ = 1.78 km s⁻¹, and [Fe|$/$|H] = −2.55, which agree well with the results in Aoki et al. (2002a) (T_eff = 6250 K, log g = 4.5, ξ = 1.5 km s⁻¹, and [Fe|$/$|H] = −2.55). The derived elemental abundances for LP 706-7 are also in agreement with those reported by Aoki et al. (2002a), within measurement uncertainties.

For the abundance analysis, we have adopted the one-dimensional local thermodynamic equilibrium model atmospheres of the ATLAS9 NEWODF grid from Castelli and Kurucz (2003), and the updated version of the abundance analysis code MOOG (Sneden 1973, version 2017), which treats continuous scattering as a source function (Sobeck et al. 2011).

Abundances of carbon and nitrogen were determined from molecular bands of CH and CN using spectral synthesis fitting. The line data for CH and CN were taken from Masseron et al. (2014) and 〈kurucz.harvard.edu〉, respectively. The top panel of figure 1 shows the fitting of the CH band around 4310 Å. A carbon abundance of [C|$/$|Fe] = +2.26 was derived for this object, as shown by the solid line. The carbon isotope ratio ¹²C|$/$|¹³C was estimated by fitting the ¹³CH feature at 4230 Å. The nitrogen abundance of J011939.222−012150.45 was determined from the CN band around 4215 Å to be [N|$/$|Fe] = +2.56. The aluminum and silicon abundances were determined from the Al i 3961 Å, 4258 Å lines and the Si i 3905 Å line, respectively.

Among light neutron-capture elements, the abundances of Sr and Y were measured. The abundance of Sr was determined based on the equivalent widths of the Sr ii 4077 and 4215 Å lines. The Y abundance was estimated from Y ii lines at 3788, 3832, and 3950 Å by fitting synthetic spectra because of the blended feature of the 3832 Å line. Abundances of six heavy neutron-capture elements with 56 ≤ Z ≤ 72, including the second s-process peak, were measured. The effect of hyperfine splitting (HFS) was taken into account during the determination of Ba abundance, assuming the s-process isotope ratio (McWilliam 1998). The two resonance lines of Ba have been excluded from the analysis because the absorption lines are severely saturated, and the three subordinate lines of Ba ii 5853, 6141, and 6497 Å were adopted. The three La ii lines 3949, 3988, and 4086 Å were used to determine the La abundance adopting the HFS splitting data from Lawler, Bonvallet, and Sneden (2001). For the Eu and Yb measurements, the Eu ii 3819 Å line and the Yb ii 3694 and 3988 Å lines were used, adopting the isotopic splitting data from table 2 of Jonsell et al. (2006). The Pb abundance was estimated by spectrum synthesis around the Pb i 4057.8 Å line, using the line list for the Pb hyperfine and isotope splitting from Aoki et al. (2001). The Nd, Er, and Yb abundances were determined by spectrum synthesis fitting the Nd ii 4156 Å, Yb ii 3694 and 3988 Å, and Er ii 3692 and 3830 Å lines. Fittings of the Ba ii, Eu ii, and Pb i lines are also shown in figure 1.

The uncertainties in our abundance (σ_[X|$/$|Fe]) analysis mainly come from two aspects. The first is the observational error (σ_obs), which refers to the standard deviation of the measured abundances from different lines. When σ_obs is smaller than 0.15 dex, or there is only one line available, σ_obs is conservatively set to be 0.15. The other comes from the uncertainty of the atmospheric parameters, which is estimated by varying the adopted parameters one by one (T_eff by +100 K, log g by +0.3 dex, and ξ by +0.3 km s⁻¹). As the uncertainty of log g resulting from parallax error is ∼0.06 and that of ξ resulting from [Fe|$/$|H] is even smaller, relatively conservative errors which are typical values for EMP stars have been chosen. The final adopted elemental abundances and relevant uncertainties are listed in table 2.

4 Discussion and conclusions

In this letter we have determined the chemical abundances of 20 elements for LAMOST J011939.222−012150.45. The result shows that it is a CEMP turnoff star with large excesses of N, Na, Ba, La, Nd, Yb, and Pb. In order to investigate the three s-process peak elements, i.e., light-s (ls; Sr and Y), heavy-s (hs; Ba, La, and Nd), and Pb, we define [Ls|$/$|Fe] = 〈[Sr|$/$|Fe], [Y|$/$|Fe]〉 and [hs|$/$|Fe] = 〈[Ba|$/$|Fe], [La|$/$|Fe], [Nd|$/$|Fe]〉. This object shows extreme s-process enhancement, in particular for hs and Pb ([Ls|$/$|Fe] ≈1.08, [hs|$/$|Fe] ≈ 2.59, and [Pb|$/$|Fe] = 3.38).

The excess of Ba and Eu in LAMOST J011939.222−012150.45 is exceptionally large even among CEMP-s stars. Indeed, compared with CEMP stars in the literature taken from the SAGA database (Suda et al. 2008, 2011; Yamada et al. 2013), it is the most Ba-enhanced CEMP star known to date, as shown in figure 2. We first inspect the CEMP sub-class of this object based on the abundance ratios of Ba and Eu. Several studies have proposed various criteria for CEMP classification, among which we discuss the classification based on two popular criteria. The first one is defined in Masseron et al. (2010) and Hampel et al. (2016), as presented in the left panel of figure 2: stars with [Ba|$/$|Fe] > 1, [Ba|$/$|Eu] > 0, and [Eu|$/$|Fe] ≤ 1 are classified as CEMP-s; those with [Ba|$/$|Fe] > 1, [Ba|$/$|Eu] > 0, and [Eu|$/$|Fe] > 1 are CEMP-r/s stars; and those with [Eu|$/$|Fe] > 1 and [Ba|$/$|Eu] < 0 are CEMP-r stars. The remaining stars that do not show enhancement in heavy elements are categorized into CEMP-no stars. According to this classification, LAMOST J011939.222−012150.45 is a CEMP-r/s star. The other classification we discuss here was proposed by Beers and Christlieb (2005) and Frebel (2018). According to this criterion, LAMOST J011939.222−012150.45 is classified as a CEMP-s star as it has [Ba|$/$|Fe] > 1.0, [Ba|$/$|Eu] > +0.5, and [Ba|$/$|Pb] > −1.5, as shown in the right panel of figure 2.

We further investigate the observed Ba|$/$|Eu ratio by comparing with solar values. In figure 2, the dashed and dotted lines indicate the pure r-process and s-process ratios of Ba|$/$|Eu in solar system material, with [Ba|$/$|Eu]_{|${\odot}$|⁠,r} = −0.84 and [Ba|$/$|Eu]_{|${\odot }$|⁠,s} = 1.20, respectively (Goriely 1999; Masseron et al. 2006). The solid and the dash-dotted lines refer to [Ba|$/$|Eu] = 0 and [Ba|$/$|Eu] = 0.5, respectively. LAMOST J011939.222−012150.45, together with another two Ba-enhanced CEMP stars (LP625-44, [Ba|$/$|Fe] = 2.86, [Eu|$/$|Fe] = 1.76; HE 0336+0113, [Ba|$/$|Fe] = 2.63, [Eu|$/$|Fe] = 1.18), are very close to the dotted line that indicates almost pure s-process.

More recently, Hansen et al. (2019) have indicated that the abundance ratio of [Sr|$/$|Ba] could be an efficient tracer to classify CEMP stars. According to their table 6, LAMOST J011939.222−012150.45 and LP625-44 should be categorized as CEMP-r stars and HE 0336+0113 as a CEMP-r/s star, because they all have [Sr|$/$|Ba] < −0.5 (only stars with −0.5 < [Sr|$/$|Ba] < 0.75 are classified into CEMP-s). However, as mentioned above, the abundance pattern of heavy neutron-capture elements of these objects is consistent with the s-process abundance pattern. We suspect that for such an extreme case of s-process enhancement, the Sr/Ba ratio may not be a reliable indicator for separating CEMP-r, CEMP-s, and CEMP-r/s stars.

The radial velocity variations of LAMOST J011939.222−012150.45 described in section 2 indicate that it should belong to a binary system. Therefore, the enhancement of heavy elements, as well as carbon, is expected to come from an AGB companion, which is consistent with the general interpretation of the origin of the observed abundance pattern of CEMP-s stars. Since this object is a main-sequence turnoff star whose surface atmosphere preserves the material well from the accretion, it provides us a unique opportunity to examine the model predictions for yields of the progenitor AGB star from light to heavy elements.

Here we compare the observed abundances of LAMOST J011939.222−012150.45 with the yields of theoretical AGB models with different ¹³C pockets (a narrow mass zone enriched in ¹³C in the He/C-rich layers) provided by FRANEC calculations at [Fe|$/$|H] ≈ −2.6 (Bisterzo et al. 2010). Figure 3 compares the observed abundance pattern with the FRANEC model prediction for AGB stars with different initial masses and cases of ¹³C pockets.

The models are set with no pre-enrichment of r-process elements, and abundances of all heavy elements are considered to be produced through s-process. The Na and Mg abundances, [Na|$/$|Fe] = 1.93 and [Mg|$/$|Fe] = 0.94, suggest an initial mass of |${\sim } 1.4\, M_{\odot }$|⁠, which is in good agreement with the deductions of Hansen et al. (2019), i.e., CEMP-s stars are companions of low-mass metal-poor AGB stars (⁠|$\sim 1.5\, M_{\odot }$|⁠), while CEMP-r/s could be relevant to the contribution of more massive AGB stars (∼2–5|$\, M_{\odot }$|⁠). As can be seen from figure 3, the best-fit (solid) line is the theoretical AGB model in the case of ST|$/$|6, where ST|$/$|6 refers to dividing the ¹³C abundances in the pocket of the standard model (Arlandini et al. 1999) by factors of 6. It should be noted here that the observed pattern for most elements can be reproduced well by the model without any scaling. For other Ba-enhanced CEMP-s stars, e.g., LP625-44, the pattern prediction by models needs to be scaled to fit the observation (Aoki et al. 2006). Note that LP625-44 is already in the evolutionary stage close to the red giant branch. By contrast, the turnoff CEMP-s star LAMOST J011939.222−012150.45 would completely preserve what has been transferred from its AGB companion.

The estimated isotopic ratio ¹²C|$/$|¹³C and the abundance ratio of [C|$/$|N] are ¹²C|$/$|¹³C = |$10^{+40}_{-6}$| and [C|$/$|N] = −0.30 ± 0.10, respectively, which reflects the degree of internal mixing with material affected by the CNO cycle. The observed [C|$/$|N] ratio cannot be fitted by the theoretical AGB models, as shown in figure 3. This, together with the relatively low ¹²C|$/$|¹³C, indicates that the material accreted from the AGB companion has been affected by the CN cycle. This is a unique constraint on the process that has built up the surface abundances of the low-mass metal-poor AGB star obtained by observations of CEMP-s turnoff stars; even if excesses of N and ¹³C are found in CEMP-s red giants, we cannot distinguish the effect of the CN cycle in the object itself from that in the progenitor AGB star. Further modeling of AGB nucleosynthesis and evolution is needed to fully explain the abundance ratios found in LAMOST J011939.222−012150.45.

Being the most Ba-rich CEMP turnoff star, LAMOST J011939.222−012150.45 is a very interesting object that shows a quite peculiar abundance pattern and has preserved pure material transferred from an AGB companion. In order to constrain the accretion process, long-term spectroscopic observations will be needed to monitor the variation of its radial velocity, which will allow us to fully solve the orbital parameters of this system. Future large-scale stellar surveys (e.g., LAMOST-II, Subaru/PFS) will enable us to find a larger sample of such objects which would then allow more systematic investigation of their origins.

Acknowledgments

This work is based on data collected at the Subaru Telescope, which is operated by the National Astronomical Observatory of Japan. This work was supported by NSFC grants nos. 11890694, 11573032, and 11390371, JSPS KAKENHI grant numbers 16H02168, 16K05287, and 15HP7004, and the JSPS–CAS Joint Research Program. Guoshoujing Telescope (the Large Sky Area Multi-Object Fiber Spectroscopic Telescope, LAMOST) is a National Major Scientific Project built by the Chinese Academy of Sciences. Funding for the project has been provided by the National Development and Reform Commission. LAMOST is operated and managed by the National Astronomical Observatories, Chinese Academy of Sciences. TM is supported by Grant-in-Aid for JSPS Fellows (grant number 18J11326).

Footnotes

References