The star formation rate (SFR) and black hole accretion rate (BHAR) functions are measured to be proportional to each other at z < ~3. This close correspondence between SF and BHA would naturally yield a BH mass-galaxy mass correlation, whereas a BH m
ass-bulge mass correlation is observed. To explore this apparent contradiction we study the SF in spheroid-dominated galaxies between z=1 and the present day. We use 903 galaxies from the COMBO-17 survey with M* >2x10^10M_sun, ultraviolet and infrared-derived SFRs from Spitzer and GALEX, and morphologies from GEMS HST/ACS imaging. Using stacking techniques, we find that <25% of all SF occurs in spheroid-dominated galaxies (Sersic index n>2.5), while the BHAR that we would expect if the global scalings held is three times higher. This rules out the simplest picture of co-evolution, in which SF and BHA trace each other at all times. These results could be explained if SF and BHA occur in the same events, but offset in time, for example at different stages of a merger event. However, one would then expect to see the corresponding star formation activity in early-stage mergers, in conflict with observations. We conclude that the major episodes of SF and BHA occur in different events, with the bulk of SF happening in isolated disks and most BHA occurring in major mergers. The apparent global co-evolution results from the regulation of the BH growth by the potential well of the galactic spheroid, which includes a major contribution from disrupted disk stars.
We analyze the infrared (IR) spectral energy distributions (SEDs) for 10micron < lambda(rest) < 100micron for ~600 galaxies at z~0.7 in the extended Chandra Deep Field South by stacking their Spitzer 24, 70 and 160micron images. We place interesting
constraints on the average IR SED shape in two bins: the brightest 25% of z~0.7 galaxies detected at 24micron, and the remaining 75% of individually-detected galaxies. Galaxies without individual detections at 24micron were not well-detected at 70micron and 160micron even through stacking. We find that the average IR SEDs of z~0.7 star-forming galaxies fall within the diversity of z~0 templates. While dust obscuration Lir/Luv seems to be only a function of star formation rate (SFR; ~ Lir+Luv), not of redshift, the dust temperature of star-forming galaxies (with SFR ~ 10 solar mass per year) at a given IR luminosity was lower at z~0.7 than today. We suggest an interpretation of this phenomenology in terms of dust geometry: intensely star-forming galaxies at z~0 are typically interacting, and host dense centrally-concentrated bursts of star formation and warm dust temperatures. At z~0.7, the bulk of intensely star-forming galaxies are relatively undisturbed spirals and irregulars, and we postulate that they have large amounts of widespread lower-density star formation, yielding lower dust temperatures for a given IR luminosity. We recommend what IR SEDs are most suitable for modeling intermediate redshift galaxies with different SFRs.
