Astron. Astrophys. 347, 532-549 (1999)

1. Introduction

The Initial Mass Function (IMF) of starburst clusters has become critical for our understanding of the Universe at high redshifts. Starbursts are a natural consequence of tidal interactions and merging (e.g. Mihos & Hernquist 1994, and references therein) and there are good reasons to believe that these processes were more frequent in the past (Barnes 1998): first, recent work suggests that the number of pairs of galaxies was larger in the past than it is today (Abraham 1999); second, most of the high z galaxies, as revealed for example by the Hubble Deep Field (HDF) can be classified as Irr/Pec/Merger (Glazebrook et al. 1995; Abraham et al. 1996). In addition, the signatures of starbursts have been actually observed at very large redshifts (Pettini et al. 1998). Thus, violent star formation and starburst clusters have gained a new status: a large fraction of the old stellar population of present day galaxies appears to have formed in starbursts. 30 Doradus, as the nearest example of a starburst cluster and the only one whose stellar population can be spatially resolved from the ground, is a natural laboratory for our studies of the IMF of these objects.

Beyond its importance in helping us decode the information contained in the light we receive from distant objects, the IMF is interesting in its own right. In what follows it will be useful to recognize, somewhat arbitrarily, three mass ranges: (1) the low mass end defined by ; (2) the intermediate mass range defined by ; and the high mass end defined by . It has recently being proposed that the stellar IMF is the direct result of a random sampling of a fractal molecular cloud system (Elmegreen 1997). In this view the form of the IMF has its origin in the turbulent processes that give molecular clouds their fractal structure, so the intermediate and high mass range IMF is expected to be a universal power law function, , with an exponent close to the Salpeter value, (Salpeter 1955).

Notwithstanding its attractiveness, the idea of universality is contradicted by some observations of clusters in the Large Magellanic Clouds (LMC) and the Milky Way. These observations reveal a mass function subject to important regional variations in the intermediate to high mass range (Parker et al. 1992; Walborn & Parker 1992), variations that have been used to support the idea of propagating or contagious star formation. However, a proper control over systematic effects is a pre-requisite to properly analyze this type of observations. In this paper, the third in a series of papers studying the IMF and star-formation history of the 30 Doradus superassociation, we develop a method that reduces the magnitude of systematic effects thus permitting the determination of stellar physical parameters in an unbiased way. In particular, we show that a hitherto neglected systematic effect might be responsible for some of the claims of regional variations of the IMF mentioned above.

In Paper I we used Daophot II (Stetson 1997; Davis 1994) to analyze a set of UBV frames of 30 Doradus obtained in sub-arcsecond seeing and photometric conditions. The overall completeness limit, defined as the magnitude at which the probability of detection in all three filters equals 50%, was found to be V=19.2. These observations are used, together with the spectroscopy presented in Paper II (Bosch et al. 1999), to determine the reddening law of the region. In this paper we combine the photometry and reddening determination from Paper I with the spectroscopy from Paper II to "read" the star-formation history of the region, and to determine its IMF for almost the full intermediate to high mass range. This will give us the keys to interpret other regions of star formation. Reddening is shown to play a crucial role, as the culprit of a hitherto scarcely described systematic effect, possibly responsible for some of the variations that have been found in the IMF slopes of various systems.

© European Southern Observatory (ESO) 1999

Online publication: June 30, 1999
helpdesk.link@springer.de