A detailed non-LTE analysis of LB-1: Revised parameters and surface abundances
LB-1 has recently been proposed to be a binary system at 4 kpc consisting of a B star of 8 Msol and a massive stellar black hole of 70 Msol. This finding challenges our current theories of massive star evolution and formation of BHs at solar metallicity. Our objective is to derive the effective temperature, surface gravity and chemical composition of the B-type component in order to determine its nature and evolutionary status and, indirectly, to constrain the mass of the BH. We use the non-LTE stellar atmosphere code FASTWIND to analyse new and archival high resolution data. We determine (Teff, logg) values of (14000$\pm500$ K, 3.50$\pm0.15$ dex) that, combined with the Gaia parallax, implies a spectroscopic mass, from logg, of $3.2^{+2.1}_{-1.9}$ Msol and an evolutionary mass, assuming single star evolution, of $5.2^{+0.3}_{-0.6}$ Msol. We determine an upper limit of 8 km/s for the projected rotational velocity and derive the surface abundances, finding the star to have a silicon abundance below solar, to be significantly enhanced in nitrogen and iron, and depleted in carbon and magnesium. Complementary evidence derived from a photometric extinction analysis and Gaia yields similar results for Teff and logg and a consistent distance around 2~kpc. We propose that the B star is a slightly evolved main sequence star of 3-5 Msol with surface abundances reminiscent of diffusion in late B/A chemically peculiar stars with low rotational velocities. There is also evidence for CN-processed material in its atmosphere. These conclusions rely critically on the distance inferred from the Gaia parallax. The goodness of fit of the Gaia astrometry also favours a high-inclination orbit. If the orbit is edge-on and the B star has a mass of 3-5 Msol, the mass of the dark companion would be 4-5 Msol, which would be easier to explain with our current stellar evolutionary models.