We investigate the co-evolution of black-hole-accretion-rate (BHAR) and star-formation-rate (SFR) in $1.5<z<2.5$ galaxies displaying a greater diversity of star-forming properties compared to previous studies. We combine X-ray stacking and far-IR pho
tometry of stellar mass-limited samples of normal star-forming, starburst and quiescent/quenched galaxies in the COSMOS field. We corroborate the existence of a strong correlation between BHAR (i.e. the X-ray luminosity, L_X), and stellar mass (M*) for normal star-forming galaxies, although find a steeper relation than previously reported. We find that starbursts show a factor of 3 enhancement in BHAR compared to normal SF galaxies (against a factor of 6 excess in SFR), while quiescents show a deficit of a factor 5.5 at a given mass. One possible interpretation of this is that the starburst phase does not coincide with cosmologically relevant BH growth, or that starburst-inducing mergers are more efficient at boosting SFR than BHAR. Contrary to studies based on smaller samples, we find the BHAR/SFR ratio of main sequence (MS) galaxies is not mass invariant, but scales weakly as M*^(0.43pm0.09}, implying faster BH growth in more massive galaxies at $zsim2$. Furthermore, BHAR/SFR during the starburst is a factor of 2 lower than in MS galaxies, at odds with the predictions of hydrodynamical simulations of merger galaxies that foresee a sudden enhancement of L_X/SFR during the merger. Finally, we estimate that the bulk of the accretion density of the Universe at $zsim2$ is associated with normal star-forming systems, with only 6(+/-1)% and 11(+/-1)% associated with starburst and quiescent galaxies, respectively.
We compare various star formation rate (SFR) indicators for star-forming galaxies at $1.4<z<2.5$ in the COSMOS field. The main focus is on the SFRs from the far-IR (PACS-Herschel data) with those from the ultraviolet, for galaxies selected according
to the BzK criterion. FIR-selected samples lead to a vastly different slope of the SFR-stellar mass ($M_*$) relation, compared to that of the dominant main sequence population as measured from the UV, since the FIR selection picks predominantly only a minority of outliers. However, there is overall agreement between the main sequences derived with the two SFR indicators, when stacking on the PACS maps the BzK-selected galaxies. The resulting logarithmic slope of the SFR-{$M_*$} relation is $sim0.8-0.9$, in agreement with that derived from the dust-corrected UV-luminosity. Exploiting deeper 24$mu$m-Spitzer data we have characterized a sub-sample of galaxies with reddening and SFRs poorly constrained, as they are very faint in the $B$ band. The combination of Herschel with Spitzer data have allowed us to largely break the age/reddening degeneracy for these intriguing sources, by distinguishing whether a galaxy is very red in B-z because of being heavily dust reddened, or whether because star formation has been (or is being) quenched. Finally, we have compared our SFR(UV) to the SFRs derived by stacking the radio data and to those derived from the H$alpha$ luminosity of a sample of star-forming galaxies at $1.4<z<1.7$. The two sets of SFRs are broadly consistent as they are with the SFRs derived from the UV and by stacking the corresponding PACS data in various mass bins.
Two main modes of star formation are know to control the growth of galaxies: a relatively steady one in disk-like galaxies, defining a tight star formation rate (SFR)-stellar mass sequence, and a starburst mode in outliers to such a sequence which is
generally interpreted as driven by merging. Such starburst galaxies are rare but have much higher SFRs, and it is of interest to establish the relative importance of these two modes. PACS/Herschel observations over the whole COSMOS and GOODS-South fields, in conjunction with previous optical/near-IR data, have allowed us to accurately quantify for the first time the relative contribution of the two modes to the global SFR density in the redshift interval 1.5<z<2.5, i.e., at the cosmic peak of the star formation activity. The logarithmic distributions of galaxy SFRs at fixed stellar mass are well described by Gaussians, with starburst galaxies representing only a relatively minor deviation that becomes apparent for SFRs more than 4 times higher than on the main sequence. Such starburst galaxies represent only 2% of mass-selected star forming galaxies and account for only 10% of the cosmic SFR density at z~2. Only when limited to SFR>1000M(sun)/yr, off-sequence sources significantly contribute to the SFR density (46+/-20%). We conclude that merger-driven starbursts play a relatively minor role for the formation of stars in galaxies, whereas they may represent a critical phase towards the quenching of star formation and morphological transformation in galaxies.
