The bolometric luminosity of type 2 AGN from extinction-corrected [OIII]: no evidence for Eddington-limited sources

There have been recent claims that a significant fraction of type 2 AGN accrete close or even above the Eddington limit. In type 2 AGN the bolometric luminosity (L_b) is generally inferred from the [OIII] emission line luminosity (L_OIII). The key issue, in order to estimate the bolometric luminosity in these AGN, is therefore to know the bolometric correction to be applied to L_OIII. A complication arises from the fact that the observed L_OIII is affected by extinction, likely due to dust within the narrow line region. The extinction-corrected [OIII] luminosity (L^c_OIII) is a better estimator of the nuclear luminosity than L_OIII. However, so far only the bolometric correction to be applied to the uncorrected L_OIII has been evaluated. This paper is devoted to estimate the bolometric correction C_OIII=L_b/L^c_OIII in order to derive the Eddington ratios for the type 2 AGN in a sample of SDSS objects. We have collected from the literature 61 sources with reliable estimate of both L^c_OIII and X-ray luminosities (L_X). To estimate C_OIII, we combined the observed correlation between L^c_OIII and L_X with the X-ray bolometric correction. We found, contrary to previous studies, a linear correlation between L^c_OIII and L_X. We estimated C_OIII using the luminosity-dependent X-ray bolometric correction of Marconi et al. (2004), and we found a mean value of C_OIII in the luminosity ranges log L_OIII=38-40, 40-42, and 42-44 of 87, 142 and 454 respectively. We used it to calculate the Eddington ratio distribution of type 2 SDSS AGN at 0.3<z<0.4 and we found that these sources are not accreting near their Eddington limit, contrary to previous claims.