In order to fully understand galaxy formation we need to know when in the cosmic history are supermassive black holes (SMBHs) growing more intensively, in what type of galaxies this growth is happening and what fraction of these sources are invisible at most wavelengths due to obscuration. Active Galactic Nuclei (AGN) population synthesis models that can explain the spectral shape and intensity of the cosmic X-ray background (CXRB) indicate that most of the SMBH growth occurs in moderate-luminosity (Lx~1044 erg/s) sources (Seyfert-type AGN), at z~0.5-1 and in heavily obscured but Compton-thin, NH~10^23 cm^-2, systems.
However, this is not the complete history, as a large fraction of black hole growth does not emit significantly in X-rays either due to obscuration, intrinsic low luminosities or large distances. The last frontier in our census of SMBH growth are the so-called Compton Thick AGN, those with NH>10^24 cm^-2 in the line of sight. In this talk I will review the latest effort in the search for this heavily obscured AGN population taking advantage of hard X-ray surveys using INTEGRAL, Swift/BAT and NuSTAR. Current estimates, after correcting for observational biases, suggest that Compton-thick accretion can account for ~20-30% of the total SMBH growth, at least in the local Universe.
I will conclude by showing the first results from a NuSTAR survey aimed to unveil heavily obscured SMBH accretion in nearby Ultra-Luminous Infrared Galaxies (ULIRGs), which seems to suggests that there is a strong connection between SMBH growth and major galaxy mergers. Furthermore, these results provide support for an evolutionary scenario in which gas-rich major galaxy merger trigger first a heavily obscured, Compton-thick, quasar, which then, thanks to the the strong radiation pressure gets rid of most of the surrounding gas to reveal a (classical) unobscured AGN. We then find that AGN activity triggered by major galaxies is responsible for ~60% of the total black hole growth.