Highly obscured active galactic nuclei (AGN) are common in nearby galaxies, but are difficult to observe beyond the local Universe, where they are expected to significantly contribute to the black hole accretion rate density. Furthermore, Compton-thi
ck (CT) absorbers (NH>10^24 cm^-2) suppress even the hard X-ray (2-10 keV) AGN nuclear emission, and therefore the column density distribution above 10^24 cm^-2 is largely unknown. We present the identification and multi-wavelength properties of a heavily obscured (NH>~10^25 cm^-2), intrinsically luminous (L(2-10keV)>10^44 erg s^-1) AGN at z=0.353 in the COSMOS field. Several independent indicators, such as the shape of the X-ray spectrum, the decomposition of the spectral energy distribution and X-ray/[NeV] and X-ray/6{mu}m luminosity ratios, agree on the fact that the nuclear emission must be suppressed by a 10^25 cm^-2 column density. The host galaxy properties show that this highly obscured AGN is hosted in a massive star-forming galaxy, showing a barred morphology, which is known to correlate with the presence of CT absorbers. Finally, asymmetric and blueshifted components in several optical high-ionization emission lines indicate the presence of a galactic outflow, possibly driven by the intense AGN activity (L(Bol)/L(Edd) = 0.3-0.5). Such highly obscured, highly accreting AGN are intrinsically very rare at low redshift, whereas they are expected to be much more common at the peak of the star formation and BH accretion history, at z~2-3. We demonstrate that a fully multi-wavelength approach can recover a sizable sample of such peculiar sources in large and deep surveys such as COSMOS.
We study the relation of AGN accretion, star formation rate (SFR), and stellar mass (M$_*$) using a sample of $approx$ 8600 star-forming galaxies up to z=2.5 selected with textit{Herschel} imaging in the GOODS and COSMOS fields. For each of them we d
erive SFR and M$_*$, both corrected, when necessary, for emission from an active galactic nucleus (AGN), through the decomposition of their spectral energy distributions (SEDs). About 10 per cent of the sample are detected individually in textit{Chandra} observations of the fields. For the rest of the sample we stack the X-ray maps to get average X-ray properties. After subtracting the X-ray luminosity expected from star formation and correcting for nuclear obscuration, we derive the average AGN accretion rate for both detected sources and stacks, as a function of M$_{*}$, SFR and redshift. The average accretion rate correlates with SFR and with M$_*$. The dependence on SFR becomes progressively more significant at z$>$0.8. This may suggest that SFR is the original driver of these correlations. We find that average AGN accretion and star formation increase in a similar fashion with offset from the star-forming main-sequence. Our interpretation is that accretion onto the central black hole and star formation broadly trace each other, irrespective of whether the galaxy is evolving steadily on the main-sequence or bursting.
We study a sample of Herschel-PACS selected galaxies within the GOODS-South and the COSMOS fields in the framework of the PACS Evolutionary Probe (PEP) project. Starting from the rich multi-wavelength photometric data-sets available in both fields, w
e perform a broad-band Spectral Energy Distribution (SED) decomposition to disentangle the possible active galactic nucleus (AGN) contribution from that related to the host galaxy. We find that 37 per cent of the Herschel-selected sample shows signatures of nuclear activity at the 99 per cent confidence level. The probability to reveal AGN activity increases for bright ($L_{rm 1-1000} > 10^{11} rm L_{odot}$) star-forming galaxies at $z>0.3$, becoming about 80 per cent for the brightest ($L_{rm 1-1000} > 10^{12} rm L_{odot}$) infrared (IR) galaxies at $z geq 1$. Finally, we reconstruct the AGN bolometric luminosity function and the super-massive black hole growth rate across cosmic time up to $z sim 3$ from a Far-Infrared (FIR) perspective. This work shows general agreement with most of the panchromatic estimates from the literature, with the global black hole growth peaking at $z sim 2$ and reproducing the observed local black hole mass density with consistent values of the radiative efficiency $epsilon_{rm rad}$ ($sim$0.07).
