The Galactic Archaeology with HERMES (GALAH) survey is a large-scale stellar spectroscopic survey of the Milky Way and designed to deliver chemical information complementary to a large number of stars covered by the Gaia mission. We present the GALAH second public data release (GALAH DR2) containing 342,682 stars. For these stars, the GALAH collaboration provides stellar parameters and abundances for up to 23 elements to the community. Here we present the target selection, observation, data reduction and detailed explanation of how the spectra were analysed to estimate stellar parameters and element abundances. For the stellar analysis, we have used a multi-step approach. We use the physics-driven spectrum synthesis of Spectroscopy Made Easy (SME) to derive stellar labels (T eff , log g, [Fe/H], [X/Fe], v mic , v sin i, A K S ) for a representative training set of stars. This information is then propagated to the whole survey with the data-driven method of The Cannon. Special care has been exercised in the spectral synthesis to only consider spectral lines that have reliable atomic input data and are little affected by blending lines. Departures from local thermodynamic equilibrium (LTE) are considered for several key elements, including Li, O, Na, Mg, Al, Si, and Fe, using 1D stellar atmosphere models. Validation tests including repeat observations, Gaia benchmark stars, open and globular clusters, and K2 asteroseismic targets lend confidence to our methods and results. Combining the GALAH DR2 catalogue with the kinematic information from Gaia will enable a wide range of Galactic Archaeology studies, with unprecedented detail, dimensionality, and scope.

The element abundance ratios of four low-mass stars with extremely low metallicities indicate that the gas out of which the stars formed was enriched in each case by at most a few, and potentially only one low-energy, supernova 1,2,3,4 . Such supernovae yield large quantities of light elements such as carbon but very little iron. The dominance of lowenergy supernovae is surprising, because it has been expected that the first stars were extremely massive, and that they disintegrated in pair-instability explosions that would rapidly enrich galaxies in iron 5 . What has remained unclear is the yield of iron from the first supernovae, because hitherto no star is unambiguously interpreted as encapsulating the yield of a single supernova. Here we report the optical spectrum of SMSS J031300.36-670839.3, which shows no evidence of iron (with an upper limit of 10 -7.1 times solar abundance). Based on a comparison of its abundance pattern with those of models, we conclude that the star was seeded with material from a single supernova with an original mass of ~60 M  (and that the supernova left behind a black hole). Taken together with the previously mentioned low-metallicity stars, we conclude that low-energy supernovae were

We investigate departures from local thermodynamic equilibrium (LTE) in the line formation of Fe for a number of well‐studied late‐type stars in different evolutionary stages. A new model of the Fe atom was constructed from the most up‐to‐date theoretical and experimental atomic data available so far. Non‐LTE (NLTE) line formation calculations for Fe were performed using 1D hydrostatic marcs and mafags‐os model atmospheres, as well as the spatial and temporal average stratifications from full 3D hydrodynamical simulations of stellar convection computed using the stagger code. It is shown that the Fe i/Fe ii ionization balance can be well established with the 1D and mean 3D models under NLTE including calibrated inelastic collisions with H i calculated from Drawin's formulae. Strong low‐excitation Fe i lines are very sensitive to the atmospheric structure; classical 1D models fail to provide consistent excitation balance, particularly so for cool metal‐poor stars. A better agreement between Fe i lines spanning a range of excitation potentials is obtained with the mean 3D models. Mean NLTE metallicities determined for the standard stars using the 1D and mean 3D models are fully consistent. Moreover, the NLTE spectroscopic effective temperatures and gravities from ionization balance agree with that determined by other methods, e.g. the infrared flux method and parallaxes, if one of the stellar parameters is constrained independently.

Neutral sodium is a minority species in the atmospheres of late-type stars, and line formation in local thermodynamic equilibrium (LTE) is often a poor assumption, in particular for strong lines. We present an extensive grid of non-LTE calculations for several Na I lines in cool stellar atmospheres, including metal-rich and metal-poor dwarfs and giants. For the first time, we constructed a Na model atom that incorporates accurate quantum mechanical calculations for collisional excitation and ionisation by electrons as well as collisional excitation and charge exchange reactions with neutral hydrogen. Similar to Li I, the new rates for hydrogen impact excitation do not affect the statistical equilibrium calculations, while charge exchange reactions have a small but non-negligible influence. The presented LTE and non-LTE curves-of-growth can be interpolated to obtain non-LTE abundances and abundance corrections for arbitrary stellar parameter combinations and line strengths. The typical corrections for weak lines are −0.1... −0.2 dex, whereas saturated lines may overestimate the abundance in LTE by more than 0.5 dex. The non-LTE Na abundances appear very robust with respect to uncertainties in the input collisional data.