Next Contents Previous

2.5. Bolometric Corrections

Given an intrinsic dispersion in the SED of quasars, any flux-limited sample selected in a given spectral band is biased towards high ratios between the flux in the selection band and the bolometric emission. This effect must be carefully taken into account in the construction of an AGN SED.

A well investigated example of the relevance of this selection effect is the average value of the optical-to-X-ray flux ratio <alphaOX> obtained in different quasar samples. E94 estimate <alphaOX> = -1.35 for their X-ray selected sample. On the other hand, Laor et al. (1997) find <alphaOX> = -1.55 for local, optically selected PG quasars. The difference, Delta alphaOX = 0.2, is a factor of ~ 3.3 in the flux ratio. It is possible to use the distribution of observed alphaOX in the two samples to estimate the effect of the selection bias. Elvis et al. (2002) showed that after this correction, the values estimated from the two samples match, with <alphaOX = -1.43. These corrections are important when estimating the accretion luminosity of the universe and comparing this with the mass spectrum of local black holes (Fabian & Iwasawa 1999; Elvis, Risaliti, & Zamorani 2002).

The analogous correction has not been computed so far for the IR emission of PG quasars. We do this here. In Figure 6a, we plot the logarithmic ratio alphaIR of the IR (3 - 1000 µm) to > 1 µm (bolometric) emission for the z < 0.4 PG quasars observed with ISO (Haas et al. 2003). The emission from 2 to 100 µm has been estimated from the B-band magnitude and the bolometric correction in E94. Approximating the distribution in Figure 6a with a Gaussian with a mean observed ratio <alphaIR>obs = -0.56 and sigma = 0.3, one obtains that the average alphaIR, corrected for the observational bias, is <alphaIR> = <alphaIR>obs + sigma2 / 2 = -0.51. The corresponding fraction of the bolometric luminosity emitted in the IR is 31%.

Figure 6

Figure 6. (a) Distribution of logarithmic IR-to-bolometric ratio for local (z < 0.4) PG quasars observed with ISO (Haas et al. 2003). (b) Same, for objects with luminosity LIR < 1012 ergs s-1 (empty histogram) and LIR > 1012 ergs s-1 (shaded histogram).

A summary of the average contribution of several spectral bands to the bolometric emission of local quasars is shown in Table 2. In this compilation, we made use of the data discussed above on PG quasars, as well as HST data of optically selected quasars (Telfer et al. 2002). These data do not show any spectral dependence with redshift in the optical/UV and therefore are assumed to be representative of local quasars. The 1 - 3 µm continuum, which is not covered in any of the works discussed above, has been taken from E94.

Table 2. Bolometric corrections for local quasars

Band Range (lambda) Range (nu) Range (energy) F / FTOT

Radio 3m-0.1mm 108 - 3 × 1011 Hz 4 × 10-7 - 1.2 × 10-3 eV 0%
Submillimeter 1000-150 µm 3-20 × 1011 Hz 1.2-8.3 × 10-3 eV 0.2%
far-IR 150-40 µm 2-7.5 × 1012 Hz 8.3-31 × 10-3 eV 4.9%
mid-IR 40-10 µm 7.5-30 × 1012 Hz 3.1-12 × 10-2 eV 13.9%
mid-IR 10-3 µm 3-10 × 1013 Hz 0.12-0.41 eV 11.9%
near-IR 3-1 µm 1-3 × 1013 Hz 0.41-1.25 eV 7.0%
Opt 1 µm-3000 Å 3-10 × 1014 Hz 1.25-4.16 eV 12.2%
UV 3000-1200 Å 1-2.5 × 1015 Hz 4.16-10.4 eV 16.5%
EUVa 1200-12.5 Å 2.5-240 × 1015 Hz 10.4 eV-1.0 keV 29.1%
X-ray 12.5-0.125 Å 2.4-240 × 1017 Hz 1-100 keV 4.2%
         

Table notes - a Based on high redshift quasars (Telfer et al. 2002).

Next Contents Previous