The formation history of the Galactic bulge

The distributions of the stellar metallicities of K giant stars in several fields of the Galactic bulge, taken from the literature and probing projected Galactocentric distances of similar to500 pc to similar to3 kpc, are compared with a simple model of star formation and chemical evolution. Our model assumes a Schmidt law of star formation and is described by only a few parameters that control the infall and outflow of gas and the star formation efficiency. Exploring a large volume of parameter space, we find that very short infall time-scales are needed (less than or similar to0.5 Gyr), with durations of infall and star formation greater than 1 Gyr being ruled out at the 90 per cent confidence level. The metallicity distributions are compatible with an important amount of gas and metals being ejected in outflows, although a detailed quantification of the ejected gas fraction is strongly dependent on a precise determination of the absolute stellar metallicities. We find a systematic difference between the samples of Ibata & Gilmore, at projected distances of 1-3 kpc, and the sample in Baade's window (Sadler et al.). This could be caused either by a true metallicity gradient in the bulge or by a systematic offset in the calibration of [Fe/H] between these two samples. This offset does not play an important role in the estimate of infall and formation time-scales, which are mostly dependent on the width of the distributions. The recent bulge data from Zoccali et al. are also analysed, and the subsample with subsolar metallicities still rules out infall time-scales greater than or similar to1 Gyr at the 90 per cent confidence level. Hence, the short time-scales we derive based on the observed distribution of metallicities are robust and should be taken as stringent constraints on bulge formation models.