The stellar host in star-forming low-mass galaxies: Evidence for two classes

Abstract

Context. The morphological evolution of star-forming galaxies provides important clues to understand their physical properties, as well as the triggering and quenching mechanisms of star formation. Aims. We analyze the morphology of galaxies hosting star-forming events at low redshift (z <  0.36). We aim at connecting morphology and star-formation properties of low-mass galaxies (median stellar mass ∼108.5 M⊙) beyond the local Universe. Methods. We use a sample of mediumband selected star-forming galaxies from the GOODS-North field. Hα images for the sample are created combining both spectral energy distribution fits and HST data. Using them, we mask the star forming regions to obtain an unbiased two-dimensional model of the light distribution of the host galaxies. For this purpose we use PHI, a new Bayesian photometric decomposition code. We applied it independently to 7 HST bands, from the ultraviolet to the near-infrared, assuming a Sérsic surface brightness model. Results. Star-forming galaxy hosts show low Sérsic index (with median n ∼ 0.9), as well as small sizes (median Re ∼ 1.6 kpc), and negligible change of the parameters with wavelength (except for the axis ratio, which grows with wavelength in 46% of the sample). Using a clustering algorithm, we find two different classes of star-forming galaxies: A more compact, redder, and high-n (class A) and a more extended, bluer and lower-n one (class B). This separation holds across all seven bands analyzed. In addition, we find evidence that the first class is more spheroidal-like (according to the distribution of observed axis ratios). We compute the color gradients of the host galaxies finding that 48% of the objects where the analysis could be performed show negative gradients, and only in 5% they are positive. Conclusions. The host component of low-mass star-forming galaxies at z <  0.36 separates into two different classes, similar to what has been found for their higher mass counterparts. The results are consistent with an evolution from class B to class A. Several mechanisms from the literature, like minor and major mergers, and violent disk instability, can explain the physical process behind the likely transition between the classes.