Sr and Ba abundances: Comparing machine-learning with star-by-star analyses

Context. A new large sample of 895 s-process-rich candidates out 454 180 giant stars surveyed by LAMOST at a low spectral resolution ( R ~ 1800) has been reported Norfolk et al. (2019, MNRAS, 490, 2219; hereafter N19). Aims. This study is aimed confirming the s-process enrichment higher 86 000) offered HERMES- Mercator spectrograph for 15 brightest targets N19 sample, which consists 13 Sr-only and two Ba-only (designating with only Sr or Ba lines strengthened). Methods. Abundances were derived elements Li, C (including 12 C/ isotopic ratio), N, O, Na, Mg, Fe, Rb, Sr, Y, Zr, Nb, Ba, La, Ce, using TURBOSPECTRUM radiative transfer LTE code MARCS model atmospheres. Binarity tested comparing Gaia DR2 radial velocity (epoch 2015.5) HERMES obtained 1600–1800 days (about 4.5 yr) later. Results. Among programme stars, 4 show no overabundances ([X/Fe] < 0.2 dex), 8 mild (at least three heavy [ X ? Fe ] 0.8), 3 have strong [X/Fe] ? 0.8). classified as previous investigation, overabundances, are barium 1 star. The turn to be both stars. Especially noteworthy fact that these actually dwarf Two among clear evidence in support their binary character, expected objects produced through mass-transfer. results more surprising; namely, no-s there binaries four, whereas one eight diagnosed shows signature radial-velocity variations. Conclusions. Blending effects saturated considered very carefully when machine-learning techniques, especially applied low-resolution spectra. from about 60% (8/13) them can true 8% stars; this fraction likely close 100% (2/2). Therefore, we recommend limit needs an unpolluted mass-transfer (i.e., extrinsic) objects.