On the Lx - L6 μm ratio as a diagnostic for Compton-thick AGN

As the mid-IR luminosity represents a good isotropic proxy of the AGN power, a low X-ray to mid-IR luminosity ratio is often claimed to be a reliable indicator of Compton-thick AGN. We assess the efficiency of the X-ray to mid-IR luminosity ratio diagnostic by examining the 12 μm IRAS AGN sample (intrinsic L_{2-10 keV} > 10⁴² erg s^-1) for which high signal-to-noise ratio XMM-Newton observations are now available. We find that the vast majority (ten out of eleven) of the AGN that were classified as Compton-thick on the basis of X-ray spectroscopy by Brightman & Nandra have a low L_X/L_{6 μm} luminosity ratio, i.e. lower than a few percent of the average AGN ratio, as is typical of reflection-dominated Compton-thick sources. At low L_X/L_{6 μm} ratios, we also find a comparable number of AGN, most of which are heavily absorbed but not Compton-thick. This implies that although most Compton-thick AGN have low L_X/L_{6 μm} ratios, at least in the local, Universe, the converse is not necessarily true. We then extend our analysis to higher redshifts. We perform the same analysis in the Chandra Deep Field South, for which excellent quality Chandra (4 Ms) and XMM-Newton (3 Ms) X-ray spectra are available. We derive accurate X-ray luminosities for Chandra sources using X-ray spectral fits, as well as 6 μm luminosities from spectral energy distribution fits. We find in total eight AGN (intrinsic L_{2-10 keV} > 10⁴² erg s^-1) with low L_X/L_{6 μm} ratios, after excluding one source where the 6 μm emission primarily comes from star-formation. One of these sources has been already found to host a Compton-thick nucleus, while for another one at a redshift of z = 1.22 we argue it is most likely Compton-thick on the basis of its combined Chandra and XMM-Newton spectrum. In agreement with the low redshift sample, we find a large number of non Compton-thick "contaminants" with low X-ray to mid-IR luminosity ratios. Our results suggest that a low L_X/L_{6 μm} ratio alone cannot help us to ascertain whether a Compton-thick AGN is present, albeit the majority of low L_X/L_{6 μm} AGN are heavily obscured. More interestingly, the two most reliable Compton-thick AGN in the high redshift Universe have high L_X/L_{6 μm} ratios, showing that this method cannot provide complete Compton-thick AGN samples.