A few decades ago, our understanding of the chemical evolution of the Milky Way relied on two observational facts. First, the chemical composition of old low-mass stars in the halo closely resembled that of the Sun, scaled to their low metallicity. Second, in disk stars - including the Sun - the abundances of elements were identical (within an error of about 50%). These "facts" found a simple explanation within the monolithic collapse scenario for galaxy formation, where disk galaxies form through the collapse of a large gas cloud. In this framework, stellar nucleosynthesis of heavy elements or metals occurred before the halo collapsed to form the disk, i.e during the first 0.1—0.3 billion years of the early Galactic evolution. Since then, the average composition of the interstellare medium (ISM) was thought to remain basically unchanged.

Advances in instrumentation and spectrum analysis methods in the past 20 years made it possible to compute stellar abundances accurately with an error of only about 10%. This revealed interesting regularities in the abundance variations between stars of different stellar populations. For instance, regularities for alpha-elements (e.g. O, Mg, Ca) or those of the iron peak (22 < Z < 28, i.e. Ti, V, Cr, Mn, Fe, Co, Ni) indicate that the chemical and dynamical evolution of the Galaxy is, in fact, very complex. Various substructures of the Milky Way (disk, bulge, halo, globular clusters) formed on different timescales and by different mechanisms, including mergers with satellite galaxies as well as accretion of gas.

Curiously enough, even today the most advanced Galactic chemical evolution models fail to explain the observed abundance distributions of different iron peak elements in Galactic metal-poor stars simultaneously. Spectroscopic studies report decreasing [Cr/Fe] and increasing [Ti/Fe] ratios with decreasing metallicity, while the [Co/Fe] ratio remains solar down to the lowest metallicities. (Square brackets denote logarithmic abundance ratios of two elements in a star relative to their ratio in the Sun.) These trends are inconsistent with the stellar nucleosynthesis theory, which predicts that stable odd-Z nuclei (V, Mn, Co) must be suppressed relative to their stable even-Z neighbours (Cr, Fe, Ni) in a low-metallicity environment.

Although most studies attribute this problem to deficiencies of the Galactic chemical evolution models, we became concerned about the accuracy of the spectroscopically determined abundances in stars. In fact, all previous abundance estimates for iron peak elements contain a systematic error, since they did not consider the true kinetic equilibrium, also known as non-local thermodynamic equilibrium (NLTE), of an element throughout a stellar atmosphere. NLTE calculations take into account the interaction of atoms with the radiation field explicitly by solving radiative transfer and statistical equilibrium equations for each of the atomic energy levels and ionisation stages.

We constructed complete atomic models for several iron peak elements and performed, for the first time, calculation of NLTE spectral line formation for their neutral and singly-ionized atoms using models of stellar atmospheres. The synthetic spectra were compared to observed spectra of Galactic metal-poor field stars and of the globular cluster Omega Centauri to derive element abundances.

Unlike earlier studies, we find that Galactic metal-poor stars are rich in the odd-Z element Cobalt, but have nearly solar proportions of the even-Z element Chromium and the odd-Z Manganese relative to iron. The Titanium abundances in Galactic metal-poor stars mimic the well-known trend of alpha-process elements (Mg, Ca) with [Fe/H]. Now, the trend of [Cr/Fe] with [Fe/H] can be reproduced by Galactic chemical evolution models without the need to invoke additional assumptions, such as peculiar conditions in the ISM. However, the models are still fully inadequate to represent the halo trends of [Mn/Fe], [Co/Fe], and [Ti/Fe]. These abundance ratios are largely insensitive to the majority of parameters in models except for stellar nucleosynthesis, which is expressed through theoretically computed element yields from stars of various masses and metallicities. Iron group elements are produced in explosive burning of silicon, which occurs in massive stars, exploding as Type II supernovae, and in exploding white dwarfs in binary systems (SN Ia). Thus, our results are useful to constrain models of supernovae and the properties of their progenitors.

An interesting pattern of Manganese abundances was obtained for giants of the globular cluster Omega Centauri. The [Mn/Fe] values in the metal-poor populations of Omega Centauri ([Fe/H] ~ -1.5 ... -1.8) overlap those of Milky Way halo stars. However, unlike in Galactic disk stars, [Mn/Fe] declines in two more metal-rich stars in the red giant branch of Omega Centauri. These results suggest that low-metallicity supernovae of either Type II or Type Ia dominated the enrichment of the more metal-rich stars in this globular cluster.

The potential of using abundances of iron group elements in old metal-poor stars is very large. They are useful not only as a diagnostic tool for physical conditions in stellar atmospheres, but also for understanding nucleosynthesis in supernovae, chemical enrichment of the galactic ISM, and, hence, chemical evolution of our Galaxy and galaxies in general.

Maria Bergemann

Publications

M. Bergemann and G. Cescutti, 2010, submitted to A & A
Bergemann et al., 2010, MNRAS 401, 1334-1346
K. Cunha, V. V. Smith, M. Bergemann, N. Suntzeff and D. Lambert, 2010, submitted to ApJ