XMM–Newton observations of high-luminosity radio-quiet quasi-stellar objects

K. L. Page 2 J. N. Reeves 1 P. T. O'Brien 2 M. J. L. Turner 2 D. M. Worrall 0 0 H.H. Wills Physics Laboratory, University of Bristol , Bristol BS8 1TL 1 Laboratory for High Energy Astrophysics , Code 662, NASA Goddard Space Flight Center , Greenbelt, MD 20771, USA 2 X-Ray and Observational Astronomy Group, Department of Physics and Astronomy, University of Leicester , Leicester LE1 7RH A B S T R A C T XMM-Newton observations of five high-luminosity radio-quiet quasi-stellar objects (QSOs; Q 01443938, UM 269, PG 1634+706, SBS 0909+532 and PG 1247+267) are presented. Spectral energy distributions were calculated from the XMM-Newton European Photon Imaging Camera (EPIC) and Optical Monitor (OM) data, with bolometric luminosities estimated in the range from 7 1045 to 2 1048 erg s1 for the sample, peaking in the ultraviolet. At least four of the QSOs show a similar soft excess, which can be well modelled by either one or two blackbody components, in addition to the hard X-ray power law. The temperatures of these blackbodies (100-500 eV) are too high to be direct thermal emission from the accretion disc, so Comptonization is suggested. Two populations of Comptonizing electrons, with different temperatures, are needed to model the broad-band spectrum. The hotter of these produces what is seen as the hard X-ray power law, while the cooler (0.25-0.5 keV) population models the spectral curvature at low energies. Only one of the QSOs shows evidence for an absorption component, while three of the five show neutral iron emission. Of these, PG 1247+267 seems to have a broad line (with an equivalent width of 250 eV), with a strong, associated reflection component ( R 2), measured out to 30 keV in the rest frame of the QSO. Finally, it is concluded that the X-ray continuum shape of active galactic nuclei remains essentially constant over a wide range of black hole mass and luminosity. 1 I N T R O D U C T I O N The X-ray continuum in Seyfert galaxies has been well studied; in these objects, a hard spectral component dominates the emission above 2 keV. A proposed origin for this component is in a hot corona above the accretion disc surface, in which optical/ultraviolet (UV) photons from the disc are Comptonized to X-ray energies. A fraction of the hard X-ray power-law continuum then illuminates the disc (and possibly a molecular torus); some of these photons are absorbed, forming an iron K-edge at >7 keV, while others are reprocessed into an Fe K line at 6.4 keV and a Compton reflection hump, caused by Compton down-scattering of the hard X-ray photons (Pounds et al. 1990). It has been found that, although these features are generally observed in Seyfert 1 galaxies (Nandra & Pounds 1994), they are less common in the spectra of quasi-stellar objects (QSOs; e.g. Reeves et al. 1997; Lawson & Turner 1997). In many cases this could be related to a lack of signal-to-noise, but it is important to determine whether such features are ubiquitous in QSOs and, hence, whether this emission mechanism is common over the full range of active galactic nuclei (AGN) luminosities. Observing the most luminous QSOs allows an investigation into objects where the accretion rate may be close to the Eddington limit and/or where the black hole mass may be large (i.e. 109 M ). Work with ASCA (e.g. Reeves & Turner 2000) found evidence for iron-line emission in a number of QSOs, but often originating from ionized material, rather than the cold emission found for Seyferts. If this were true in general, it could be explained by the more luminous AGN having a higher accretion rate, which causes the surface of the disc to become ionized. The QSOs in this paper form a small sample of high-luminosity, radio-quiet objects, with X-ray luminosities from 7 1044 to 3 1046 erg s1 (bolometric luminosities of 7 1045 to 2 1048 erg s1); the redshifts cover a range from 0.244 up to 2.038. Radio-quiet objects form the majority of luminous AGN (e.g. Kukula et al. 1998), but are less X-ray luminous than their radio-loud counterparts, for a given optical luminosity (Zamorani et al. 1981; Worrall et al. 1987; Wilkes et al. 1994). Because radio-quiet QSOs (RQQs) are fainter in the X-ray band, the high throughput of the XMMNewton X-ray telescopes makes the European Photon Imaging Camera (EPIC) instruments ideal for an X-ray investigation of distant RQQs. The aim of this study is to investigate the properties of the central engine in some of the most luminous QSOs (i.e. the extreme end of the accretion rate, black hole mass and/or luminosity parameter space). It is possible to do this with RQQs because the jet is thought not to contribute significantly to the X-ray emission, whereas, in radio-loud quasars, synchrotron or inverse Compton emission from a relativistic jet may dilute some of the spectral features (such as the iron line or soft excess) which are thought to originate from the accretion disc. 2 XMMNEWTON OBSERVATIONS Table 1 lists the dates and the instrumental set-up for each of the EPIC observations, while Table 2 gives the redshifts, Galactic absorbing column and radio measurements for each of the QSOs. A value of R L < 1 defines the AGN as being radio-quiet, where RL is given by the log of the ratio of the radio (5 GHz) to optical (B-band) fluxes (Wilkes & Elvis 1987; Kellerman et al. 1989). The pipeline-produced event-lists were filtered using XMMSELECT within version 5.4 of the SAS (Science Analysis Software); singleand double-pixel events (patterns 04) were used for the PN, while patterns 012 were chosen for the MOS instruments. Spectra were extracted within a small circular region, centred on the source, with a radius of between 25 and 40 arcsec, depending on the source brightness. (Smaller regions were used for the fainter sources, to minimize the contribution from the background.) Background spectra (within the same size, or larger, region) were produced from an offset position free of other sources. MOS 1 and MOS 2 spectra were subsequently co-added, after confirmation that the results were consistent. Source and background light curves were also extracted for each object. Time intervals of relatively high, flaring background were identified for both of the PG QSOs, and these periods were excluded from the following analysis. None of the source light curves showed variability over the duration of the observations after the removal of the background, however. After grouping the spectra to obtain a minimum of 20 counts per bin, version 11.1.0 of XSPEC was used to analyse the data. The most recent (time-dependent) response matrices (rmfs) were used when fitting the spectra, together with an ancillary response file (arf) generated by running ARFGEN within the SAS. The rmfs take into account the degradation of the instruments over the years since launch, and specifically model how the charge transfer inefficiency (CTI) has changed. Optical/UV magnitudes were obtained from the Optical Monitor (OM) where possib (...truncated)