Finding needles in a haystack
Our understanding of the formation paths of supermassive black holes is still very sketchy. In 1982, Andrzej Soltan showed that the summed emission from all observed quasars yields a remarkably accurate estimate of the total mass of present-day black holes. His argument was based on the expected conversion efficiency of the rest mass energy of matter in an accretion disk falling into a black hole at the centre of a quasar - these distant objects are thus believed to signpost the main sites of black hole growth across the Universe.
Unfortunately, quasars are not ideal objects to study the mechanisms by which black holes grow. The emission from the central nucleus is more luminous than the underlying host galaxy by many orders of magnitude, making detailed studies of the host system extremely difficult. For this reason, studies to constrain possible triggering mechanisms for black hole growth focus on so called Type II active galactic nuclei (AGN). In these systems, the radiation from the accretion disk is believed to be blocked by a very dense layer of gas and dust (the so-called torus, see Figure 1). Large spectroscopic galaxy surveys such as the Sloan Digital Sky Survey have yielded samples of hundreds of thousands of nearby Type II AGN, which are selected according to their optical emission line ratios. At higher redshifts, Type II AGN are commonly selected at X-ray wavelengths.
So far, studies of the host galaxies of these systems appear to rule out theoretical scenarios in which black hole growth occurs when two or more galaxies merge together. Simulations of the gravitational interactions and gas dynamics of two merging galaxies show that tidal torques during the merger cause the gas to shock, lose energy and flow towards the centre of the merger remnant. Energetic processes that act on the gas very close to the black hole are, however, very difficult to simulate in a reliable way.
In recent work at MPA, a new technique combined data from several observing programmes using the Wide-field Infrared Survey Explorer satellite, the Very Large Array (VLA) FIRST Survey (Radio Images of the Sky at Twenty-Centimeters) and the Sloan Digital Sky Survey. This combination of data permits a more reliable selection of a large sample of active galaxies where there is strong hot dust emission from a central torus. The radio data turned out to be a critical element of the selection technique, because there are large number of galaxies where the hot dust extends over a large area and is probably not being heated by the black hole. This had not been accounted for in previous work.
Follow-up work demonstrated that the new selection, which includes only 1.6% of the sources in previous AGN samples, yields galaxies with properties that are very different. More than 80% of the galaxies in the new sample turn out to be merging or interacting systems. For many of them, their stellar spectra show strong bursts of star formation in their central regions. The emission lines indicate that the gas is highly ionized, with the main source of ionization likely being an accreting black hole rather than the young stars in the nucleus. The radio emission is usually compact and centrally located and is too luminous to be explained by the observed young stars in the nucleus. With 1300 galaxies the new sample is large enough to show conclusively that these AGN currently signpost the bulk of black hole formation in the most massive galaxies in the local Universe - with the "normal" AGN population dominant in lower mass galaxies.
The challenge now is to go back in time to younger galaxies, i.e. to extend this study to higher redshifts. Pulling similar samples out of surveys of galaxies at higher redshifts, however, will need comparable data sets in different wavelength bands. Then we can investigate how the accreting black holes in these younger systems are influencing the gas in and around their host galaxies.