The (anti)correlation of the sub-mm and X-ray background sources

The connection between the sub-mm and the hard X-ray backgrounds is studied by comparing data at 2–10 keV and at 850μm for a sample of 34 sources at fluxes (or limiting fluxes) which resolve most of the background in the two bands. These data were obtained with new SCUBA observations and by correlating data sets available from the literature. None of the 11 hard X-ray (2–10 keV) sources has a counterpart at 850μm, with the exception of a faint Chandra source, which is a candidate type 2 QSO at high redshift. These data indicate that 2–10 keV sources brighter than 10erg scm, which make at least 75% of the background in this band, contribute for less than 7% to the submillimetric background. Out of the 24 SCUBA sources 23 are undetected by Chandra. These data indicate that most of these SCUBA sources must be powered either by starburst activity, or by an AGN which is obscured by a column NH > 10 cm, with a reflection efficiency in the hard X-rays significantly lower than 1% in most cases. Following the original suggestion of Setti & Woltjer (1989), the hard X-ray background (2–100 keV) is commonly explained with the superposed emission of a large population of highly obscured, type 2 AGNs. However, obscured AGNs have also been proposed as the powerhouse of a fraction of the SCUBA sources which make most of the background at 850μm, thus providing a link between the two spectral windows. In particular, some models predict that obscured AGNs could contribute a substantial fraction (20–50%) of the submillimeter background (Almaini et al. 1999). With the aim of investigating the relation between the sub-mm and the 2–10 keV hard X-ray backgrounds, we started an observing program with SCUBA of a subsample of the sources detected by BeppoSAX in the HELLAS survey. This survey resolved about 30% of the 5–10 keV background into discrete sources at a flux limit of ∼ 5 × 10erg scm. None of the four hard X-ray sources observed so far was detected at 850μm at a 2σ limiting flux of about 3 mJy. In order to put these results into perspective, we follow Fabian et al. (2000) and compute a submillimeter–to–X-ray index αSX (Fν ∝ ν ), using the observer-frame flux densities at 850μm and 5 keV. In Fig. 1, which gives αSX as a function of redshift, our results are represented by filled squares. We have also cross-correlated the deep hard X-ray survey of the SSA13 field (F2−10keV > 2.5× 10erg scm), carried out by Mushotzky et al. (2000), with the deep (smallarea) and shallow (wide-area) SCUBA maps of the same field obtained by Barger et 1 On behalf of a collaboration including R. Maiolino, M. Salvati, D. Axon, A. Cimatti, F. Fiore, R. Gilli, F. La Franca, A. Marconi, G. Matt, G. Risaliti, and C. Vignali. al. (1999) at 850μm. Only one out of the three Chandra sources in the deep SCUBA map is detected at 850μm, which is indicated with a filled circle in Fig. 1. The optical, X-ray and sub-mm properties of this object suggest that this is a candidate type 2 QSO, heavily absorbed and hosted in a young galaxy at redshift between 1.9 and 2.7. The other two Chandra sources in this area, undetected at 850μm, are marked as upper limits with filled circles in Fig. 1. Eight out of the nine SCUBA sources in the SSA13 field are not detected by Chandra; since their redshift is not known the spread of the lower limits on αSX is indicated with two horizontal shaded areas in Fig. 1. The open squares in Fig. 1 indicate the limits on αSX for the sources in the lensing clusters A1835 and A2390 obtained by Fabian et al. (2000) by cross-correlating Chandra observations and SCUBA maps. Finally, the open circles are the limits on αSX for the sources in the Hubble Deep Field North obtained by Hornschemeier et al. (2000), or derived by us, with a deep Chandra observation (F2−8keV > 9× 10 erg scm). In Fig. 1 we also show the values of αSX as a function of redshift for various classes of objects. NGC4593 and the two PG QSOs are assumed to be representative of type 1 Seyferts and of radio quiet QSOs. Four QSOs at high-z detected by SCUBA are shown with stars. Circinus is shown as representative of heavily obscured Sy2s (NH = 5×10cm), but we also show the expected αSX in the case that the absorbing column density were an order of magnitude lower than observed (namely NH = 5× 10 cm). NGC1068 is the case of a completely Compton thick AGN (NH > 10 cm) whose hard X-ray flux is reflection dominated up to 100 keV. As a starburst template we show the case of Arp220. The thick upper horizontal segment in Fig. 1 gives the αSX of the cosmic background, under the assumption that most of the flux in both spectral windows comes from redshifts between 1 and 2 (in analogy with the soft X-ray background). The location of the αSX of the various classes of sources with respect to this value constrains their contribution to the sub-mm and to the hard X-ray background. In particular, the lower thick horizontal segment gives the αSX of sources which would contribute 100% of the 2-10 keV background and only 10% of the sub-mm background. Out of the 11 hard X-ray sources only one is detected at 850μm. The upper limits on αSX for the hard X-ray sources are much lower than the background requirements. In particular, we estimate that, under conservative assumptions, the 2–10 keV sources brighter than ∼ 10erg scm, which resolve at least 75% of the background in this band (Mushotzky et al. 2000), cannot contribute for more than 7% to the sub-mm background. A more detailed discussion is given in Severgnini et al. (2000). None of the 24 SCUBA sources, but the candidate type 2 QSO in the SSA13 field, is detected in the 2–10 keV band down to a limiting flux of F2−10keV ∼ 1 − 2 × 10erg scm. As shown in Fig. 1, their lower limits on αSX occupy the upper part of the plot and are presumably akin to the starburst template given by Arp220. Most of the SCUBA sources with known redshift are even inconsistent with the reflectiondominated template given by the nucleus of NGC1068. These sources are probably dominated by starburst activity. A significant contribution from an obscured AGN might be present if the latter is completely Compton thick (NH > 10 cm) and if the reflection efficiency is significantly lower than estimated for NGC 1068 (∼1%).