Titanium abundances in late-type stars, II. Grid of departure coefficients and application to a sample of $70\,000$ stars

Abstract

Rapidly growing datasets from stellar spectroscopic surveys are providing unprecedented opportunities to analyse the chemical evolution history of our Galaxy. However, spectral analysis requires accurate modelling of synthetic stellar spectra for late-type stars, for which the assumption of local thermodynamic equilibrium (LTE) has been shown to be insufficient in many cases. Errors associated with LTE can be particularly large for Ti I, which is susceptible to over-ionisation, particularly in metal-poor stars. The aims of this work are to study and quantify the 1D non-LTE effects on titanium abundances across the Hertzsprung-Russell diagram for a large sample of stars. A large grid of departure coefficients, $\beta_\nu$, were computed on standard MARCS model atmospheres. The grid extends from 3000K to 8000K in T$_{\mathrm{eff}}$, -0.5 to +5.5 dex in log(g), and -5.0 to +1.0 in [Fe/H], with non-LTE effects in this grid reaching up to 0.4 dex. This was used to compute abundance corrections that were subsequently applied to the LTE abundances of over 70,000 stars selected from the GALAH survey and additional metal-poor dwarfs. The non-LTE effects grow towards lower [Fe/H], lower log(g), and higher T$_{\mathrm{eff}}$, with a minimum and maximum $\Delta$A(Ti I) of 0.02 and 0.19 in the GALAH sample. For metal-poor giants, the non-LTE modelling reduces the average ionisation imbalance from -0.11 dex to -0.01 dex at [Fe/H] = -1.7, and the enhancement in titanium abundances from Ti I lines results in a [Ti/Fe] versus [Fe/H] trend that more closely resembles the behaviour of Ti II at low metallicities. Non-LTE effects on titanium abundances are significant. Neglecting them may alter our understanding of Galactic chemical evolution. We have made our grid of departure coefficients publicly available, with the caveat that the Ti abundances of metal-poor dwarfs need further study in 3D non-LTE.