Quasar Evolution

It has been well established for nearly two decades that the quasar population evolves on a cosmological timescale. (I use quasar and the more formal QSO interchangeably here, since the strength of radio emission is not important for these arguments.) Statistically, quasar evolution is described by changes in their luminosity function with redshift. At the powerful luminosities which can be traced to high redshifts, the space density increases with redshift out to about z = 2.2, reaching a peak several orders of magnitude greater than the local density that we can measure in our own neighborhood. At still higher redshifts, the density drops rapidly (the so-called QSO cutoff). Radio galaxies do something similar up to a point, but the relative ease of getting redshifts for large QSO samples in comparison has made QSOs the easier tracer with much more statistically secure results.

Establishing how the luminosity distribution of QSOs has changed with redshift does not give a unique physical picture for just what is evolving. The luminosity function is close to a power-law form, lacking any characteristic reference feature such as L* in the galaxy luminosity function, which engendered years of debate about whether the evolution was expressed in luminosity or space density. For such a featureless distribution, measures of the luminosity function itself could not distinguish changes that corresponded to vertical (space density) or horizontal (luminosity) displacements of the curve. Improved data sets, and ways of accounting for the inevitable selection biases at various redshifts, eventually showed changes in the form of the luminosity function which made it clear that the amount of density evolution depends on luminosity; or, more transparently, the shape of the luminosity function becomes flatter at earlier epochs for the most luminous objects. In more physical terms, the relative number of very luminous QSOs was larger at early times, above and beyond the entire population having a greater space density (in comoving coordinates, so that the space density as measured at that time would have been still greater by a factor (1 + z)3).

Evolution in the QSO luminosity function, while an adequate description of changes in the population with redshift, doesn't tell us the whole story. For AGN in the local Universe, there are hints that the duty cycle of powerful activity is much less than unity, as we might expect if special conditions are needed to feed the central engine and produce the radiating byproducts that make AGN so bright. At low luminosity, Seyfert galaxies comprise about 5% of the population of luminous galaxies. This might in principle mean that these 5% are special and are always Seyferts, or it might mean that all similar galaxies spend 5% of their time as Seyfert galaxies, with the central mass quiescent and invisible the rest of the time. There are indeed also physical reasons to suspect that quasars are "on" for only a fraction of cosmic time. There is a simple and powerful argument based on the luminosities of luminous QSOs, since we believe these to be fed by accretion of matter around a central black hole. Such accretion can convert up to half the rest mass of this matter into energy, though the efficiency could be much less depending on the details of accretion. The nearest really powerful quasar, 3C 273, has an overall energy output of about 1046 ergs/s. This would require one-third of a solar mass per year at the highest possible efficiency, meaning that keeping it so bright over a Hubble time would have required the consumption of over 5 billion solar masses' worth of stars and gas. We can still see the host galaxy around 3C 273, so that it hasn't been consumed (and this probably didn't happen for the even more powerful QSOs we can see at higher redshifts). More circumstantially, HST images of the galaxies around QSOs at moderate redshifts (z = 0.3) show a remarkable fraction interacting with very compact companion galaxies, a high enough fraction that these unusual kinds of interaction seem to have something to do with the occurrence of QSO activity. If we see a statistical connection between, say, powerful nuclear activity and some kind of galaxy interaction, the episode of activity cannot last much longer than the interaction does, or the correlation would be weakened. This suggests that, at least for this kind of trigger phenomenon, an episode of luminous QSO symptoms lasts for a time comparable with a galaxy interaction or merger, typically a few times 108 years.

Recently, Amy Barger et al. (2001) have used a census of X-ray sources in deep exposures taken by the Chandra X-ray Observatory to address this problem in a different way, tallying the total amount of X-rays from AGN at various flux levels. Since X-rays are an energetically important part of the total output of AGN, and insensitive to absorption by ordinary amounts of gas in galaxies, this should be a powerful and unbiased survey. The X-ray intensity can be linked to the accretion rate, using the kinds of efficiency arguments above. Taking the estimated present-day statistics of massive black holes in galaxies, they find that a typical AGN has spent a total of about 0.5 Gyr "on", a duty cycle of about 4%. While there may be wide variations from galaxy to galaxy about this characteristic value, and in how the 0.5-Gyr span was occupied in one or many episodes, this is interestingly close to the local demographics as to the fraction of galaxies that are "on" at a given time. The history of accretion into massive black holes may also be bounded from above, as an important contributor to the X-ray background. The integrated, redshifted X-ray emission from all AGN cannot exceed the mean X-ray intensity from "blank sky" at any energy. As imaging techniques allow resolution of ever-larger fractions of this background emission into individual sources, it has become clear that we must recognize an important contribution from objects that are heavily absorbed at low energies by surrounding gas, so that the mean background spectrum is "harder" than we would expect from nearby, well-observed AGN.

In this light, QSO evolution may have consisted of changes in the typical luminosity of an episode of activity, its duration, or both. The link between stellar and central masses (below) suggests that, if the growth of the two was linked, the central engine was being fed constantly if not necessarily continuously by accretion. The overall evolution of QSOs may then indicate that they were "on" a larger fraction of the time earlier in the Universe. In fact, when they were most numerous at epochs around z = 2.2, they accounted for a substantial fraction of all luminous galaxies.

0 0

Post a comment