Galaxy morphologies provide important clues about their formation mechanism. In the nearby universe, a clear relation exists between a galaxy's morphology and its star formation status: galaxies with no or little star formation (quiescent galaxies) are primarily ellipticals while star-forming galaxies are disks. It suggests the processes that lead to morphological transformation may also lead to shutting down (quenching) star formation. Thus understanding the origin and evolution of this relation is essential to study the drivers of the evolution of quiescent and star-forming galaxies, a central question in galaxy formation and evolution. Both observational and theoretical results now suggest that the critical epoch for massive galaxy evolution and mass assembly is at 1 < z < 3. Thus, a complete census of a well-defined massive galaxy sample at z ~ 2 is crucial to understanding the origin of this relation, as well as the formation and evolution of massive galaxies.
It turns out that quiescent galaxies at z ~ 2 are remarkably different from their nearby counterparts. Despite the well-known fact that they are much more compact than present-day galaxies of the same mass, recent studies show that roughly half of them show significant disk components. So if quiescent and star-forming galaxies can be no longer be separated based on whether they have disks, what is the major difference between their morphologies?
In a recent paper from the CANDELS collaboration, we explored the relation between galaxy morphology and star-formation for a massive galaxy sample at z ~ 2 selected in the GOODS-South field. These galaxies were shown to form a bimodal distribution in their mid-infrared colors, which are sensitive to star formation rates, and thus were classified into quiescent and star-forming galaxies.
We then performed both visual classification and quantitative measurements of their morphologies based on new high-resolution H-band imaging from WFC3/HST. We used two parameters to quantify their morphologies: the Gini coefficient and M20. Gini measures the relative distribution of the galaxy pixel flux values and M20 represents the second-order moment of the brightest 20% pixels, respectively. In most cases, Gini is correlated with the concentration index and increases with the fraction of light in a compact component while M20 is very sensitive to merger signatures. These quantities have previously been shown to be very effective at separating different morphological types for galaxies at low to intermediate redshifts.
We show that the quiescent and star-forming galaxies can be well separated by the Gini coefficient: all of the quiescent galaxies have higher Gini values while nearly all the star-forming galaxies have lower ones. Visual classifications derived consistent results: the most prominent difference between the two populations is whether they have a significant bulge, which leads to high Gini values.
This indicates that the quenching process for star formation must lead to or be accompanied by the increase of galaxy concentration, or the formation of a prominent bulge. This process, however, may not necessarily destroy the disks: we confirm the existence of a significant population of quiescent galaxies with prominent bulges yet also with significant disks. We also show that most massive star-forming galaxies are regular disks with no clear evidence of merger/interactions. Together, these pose challenges to the merging scenario as the main mode of massive galaxy formation.