Coevolution of black hole accretion and star formation in galaxies up to z = 3.5

Abstract

Aims. We study the coevolution between the black hole accretion rate (BHAR) and the star formation rate (SFR) in different phases of galaxy life: main-sequence star-forming galaxies, quiescent galaxies, and starburst galaxies at different cosmic epochs. Methods. We exploited the unique combination of depth and area in the COSMOS field and took advantage of the X-ray data from the Chandra COSMOS-Legacy survey and the extensive multiwavelength ancillary data presented in the COSMOS2015 catalog, including in particular the UVista Ultra-deep observations. These large datasets allowed us to perform an X-ray stacking analysis and combine it with detected sources in a broad redshift interval (0.1 <  z <  3.5) with unprecedented statistics for normal star-forming, quiescent, and starburst galaxies. The X-ray luminosity was used to predict the black holeAR, and a similar stacking analysis on far-infrared Herschel maps was used to measure the corresponding obscured SFR for statistical samples of sources in different redshifts and stellar mass bins. Results. We focus on the evolution of the average SFR-stellar mass (M*) relation and compare it with the BHAR-M* relation. This extends previous works that pointed toward the existence of almost linear correlations in both cases. We find that the ratio between BHAR and SFR does not evolve with redshift, although it depends on stellar mass. For the star-forming populations, this dependence on M* has a logarithmic slope of ∼0.6 and for the starburst sample, the slope is ∼0.4. These slopes are both at odds with quiescent sources, where the dependence remains constant (log(BHAR/SFR) ∼ −3.4). By studying the specific BHAR and specific SFR, we find signs of downsizing for M* and black hole mass (MBH) in galaxies in all evolutionary phases. The increase in black hole mass-doubling timescale was particularly fast for quiescents, whose super-massive black holes grew at very early times, while accretion in star-forming and starburst galaxies continued until more recent times. Conclusions. Our results support the idea that the same physical processes feed and sustain star formation and black hole accretion in star-forming galaxies while the starburst phase plays a lesser role in driving the growth of the supermassive black holes, especially at high redshift. Our integrated estimates of the M* − MBH relation at all redshifts are consistent with independent determinations of the local M* − MBH relation for samples of active galactic nuclei. This adds key evidence that the evolution in the BHAR/SFR is weak and its normalization is relatively lower than that of local dynamical M* − MBH relations.