2000A&A...364..517R

We analyze a morphologically-selected complete sample of 52 late-type (spiral and irregular) galaxies in the Hubble Deep Field North with total K-magnitudes brighter than K=20.47 and typical redshifts z∼0.5 to 1.4. This sample exploits in particular the ultimate imaging quality achieved by HST in this field, allowing us to clearly disentangle the early- from late-type galaxy morphologies, based on accurate profiles of the surface brightness distributions. Our purpose was to investigate systematic differences between the two classes, as for colours, redshift distributions and ages of the dominant stellar populations. Our analysis makes also use of an exhaustive set of modellistic spectra accounting for a variety of physical and geometrical situations for the stellar populations, the dusty Interstellar Medium (ISM), and relative assemblies. The high photometric quality and wide spectral coverage allow to estimate accurate photometric redshifts for 16 objects lacking a spectroscopic measurement, and allow a careful evaluation of all systematics of the selection [e.g. that due to the surface-brightness limit]. This sample appears to miss significantly galaxies above z=1.4 (in a similar way as an early-type galaxy sample previously studied by us), a fact which may be explained as a global decline of the underlying mass function for galaxies at these high redshifts. Differences between early- and late-types are apparent - particularly in the colour distributions and the evolutionary star-formation (SF) rates per unit volume -, although the complication in spectro-photometric modelling introduced by dust-extinction in the gas-rich systems prevents us to reach conclusive results on the single sources (only future long-wavelength IR observations will allow to break the age/extinction degeneracy). However, we find that an integrated quantity like the comoving star-formation rate density as a function of redshift Ψ(z) is much less affected by these uncertainties: by combining this with the previously studied early-type galaxy sample, we find a shallower dependence of Ψ(z) on z between z=0.2 and z=1.5 than found by Lilly et al. (1995ApJ...455..108L). Our present results, based on a careful modelling of the UV-optical-NIR SED of a complete galaxy sample - exploiting the observed time-dependent baryonic mass function in stars as a constraint and attempting a first-order correction for dust extinction - support a revision of the Lilly-Madau plot at low-redshifts for both UV- and K-band selected samples, as suggested by independent authors (Cowie et al., 1999AJ....118..603C).