arXiv:1212.2957v1 [astro-ph.HE] 12 Dec 2012 iin rebl,M 20771 MD Greenbelt, vision, ne eigo,MA Lexington, enue, tt nvriy nvriyPr,P 60,USA 16802, PA Park, University University, State tt nvriy 2 ae a,Uiest akP 16802 PA Park University Lab, Davey 525 University, State ak MD Park, in nohrwvbns hsls etr simportant is feature last observa- This with possible wavebands. is other than in sight tions of absorbing line of the amounts in greater bands, material through other in penetrate subject present can not dilution and are host-galaxy heavy is X-rays the them to observations. understanding X-ray to by key wave-provided invaluable ultraviolet an and including but optical lengths, spectrum, infrared, AGN entire sub-mm, from the radio, emission across at Strong observed been holes 1984). has black Rees supermassive luminous onto e.g. their accretion (‘SMBHs’, and to Universe, due is the output in sources energy erful * 5 4 3 2 1 pow- most the among are (AGN) nuclei galactic Active rpittpstuigL using ApJ typeset Preprint in publication for Accepted -A RPRISO H OTENGLCI A ORE NTHE IN SOURCES CAP GALACTIC NORTHERN THE OF PROPERTIES X-RAY [email protected] AAGdadSaeFih etr srpyisSineDi Science Astrophysics Center, Flight Space NASA/Goddard topei n niomna eerh 3 atelAv- Hartwell 131 Research, Environmental and Atmospheric nttt o rvtto n h oms h Pennsylvania The Cosmos, the and Gravitation for Institute eateto srnm srpyis h Pennsylvania The Astrophysics, & Astronomy of Department eateto srnm,Uiest fMrln,College Maryland, of University Astronomy, of Department a apefrawd ait fftr tde nAGN. ut on the studies emphasize future We of variety catalog. headings: wide Subject 9-month a the for in sample seen Cap than fainter times o h ape hc erdcste11 e -a background log( X-ray keV 2–10 keV The 1–10 sample. the AGN reproduces BAT which month sample, the for motn nalrefato forsml sn joint using sample our of C fraction hidden. large as a identified not in previously important objects seven including sample, niaiglgtbnigo xrml ope bopin ihenergy High absorption. complex ( band extremely BAT or the bending light indicating wihmyhv emtial hc ou reaitdscattering emaciated or torus thick geometrically a have may (which te e uniisfrtectlg eueacnitn approac consistent hig a abs use at luminosity, We AGN of catalog. already of from distributions the gathered observations are for present field’ that quantities We key ‘deep data other the multi-wavelength Hole). to of Lockman analog range COSMOS, low-redshift wide a a as to due study rae bobn oundniydsrbto.Tefato fob of fraction The distribution. density column absorbing broader a httepoesrsosbefrpouigtesf xesi not is excess soft so the the N where producing (log another for sources and responsible Compton-thick well-detected is process excess the soft that a de which K- in previous one iron reinforce We for ratio. effect (Iwasawa-Taniguchi) signal-to-noise sufficient with jects h apeis sample the h -ot A aao ( catalog BAT 9-month the nteNrhr aatcCpo h 8mnhSitBrtAetTeles Alert Burst Swift 58-month with the AGN of 100 Cap of Galactic consisting Northern the in epeettepoete fio ie,sf xessadinzda ionized and excesses soft lines, iron of properties the present We epeetadtie -a pcrlaayi facmlt apeo sample complete a of analysis spectral X-ray detailed a present We ajnV Vasudevan V. Ranjan a n .Baumgartner H. Wayne 1. INTRODUCTION A T ∼ E XMM-Newton tl mltajv 08/22/09 v. emulateapj style X > E 0,ad4–6 ftesml xiis‘ope’041 e spectra keV 0.4–10 ‘complex’ exhibits sample the of 43–56% and 60%, 0kV u r eni oesucs epeetteaeae1– average the present We sources. some in seen are but 200keV) 1,* h z ila .Brandt N. William , wf/R,AC n wf/A.W rb odee esit th redshifts deeper to probe We Swift/BAT. and ASCA Swift/XRT, , i H 5 > b hmsT Shimizu T. Thomas , 0 = > 24 . 50 4 oprdto compared 043 . cetdfrpbiaini ApJ in publication for Accepted ◦ 5 norsml is sample our in 15) hssyae a xeln oeta o ute dedicated further for potential excellent has area sky This . α ie.W loietf w itntppltoso sources; of populations distinct two identify also We lines. ABSTRACT CATALOG N - )-log( 2,3 1 ihr .Mushotzky F. Richard , h oad .Schneider P. Donald. , S hnotcloe eg,Msozy20) hs AGN column those Hydrogen 2004), neutral density (with Mushotzky obscuration (e.g., heavy with ones optical than absorbing the whole. of about as range conclusions population draws a AGN one complete across before as possible densities have as AGN column AGN to overall of essential survey the therefore to a of is thought it proportion are population; AGN significant absorbed a because constitute studies AGN in ihreege,adeetal h .–0kV adex- band as keV, such observatories 0.4–10 by successively the plored at eventually fluxes and larger energies, the Progressively higher depress absorption satellites. of imaging amounts X-ray typical of o hsproe n spouiga l-k uvyof survey all-sky the an band, keV producing 14–195 is the in and AGN purpose, al. et this Barthelmy for (BAT, the Telescope on Alert 2005) Burst at The Sensitivity sus. AGN. highly-absorbed > in levels sorption dra n et steBTisrmn otne osre the survey to continues instrument increas- BAT at the detections as by depth augmented ing is survey The AGN. z i ltipiscmltns ont fluxes to down completeness implies plot ) hl -a uvy fANaemr penetrating more are AGN of surveys X-ray While 0kVi eurdt banamr opeeANcen- AGN complete more a obtain to required is keV 10 0 = XMM-Newton eoe nucett dniyadcntanab- constrain and identify to insufficient becomes , ∼ . 3frte9mnhctlg,aduncover and catalog), 9-month the for 03 % efidta hde/uidAGN’, ‘hidden/buried that find We 9%. N H twr nalAN h rcinof fraction The AGN. all in work at ofi e n rhvlXrydata X-ray archival and new fit to h > emntoso h -a Baldwin X-ray the of terminations Swift tecs sudtce,suggesting undetected, is excess ft lp sfudfrtebihe 9- brighter the for found as slope 10 ein)constitute regions) crd(o N (log scured mtnrflcini on obe to found is reflection ompton BTfis( fits +BAT 23 lt fti otenGalactic Northern this of ility sresfrtesbe fob- of subset the for bsorbers e esit eg CDFN/S, (e.g. redshifts her vial,adw rps it propose we and available, − aelt a rvneteeyuseful extremely proven has satellite u-ff eeal i outside lie generally cut-offs rigclm est,and density, column orbing adXryslce AGN selected X-ray hard f 24 oe(wf/A)catalog, (Swift/BAT) cope tl a aloto h purview the of out fall can still ) 2,3 1 iaM Winter M. Lisa , onNousek John , 8MNHSWIFT-BAT 58-MONTH H h R XMM-Newton 0kVspectrum keV 10 ≥ i . 2 = 2 bet in objects 22) Swift ∼ 4 four of 14% . 7 BTctlgof catalog /BAT ± 2 0 4 . , 75), ∼ an and 4 Chan- 2 sky. Source lists and sample properties have been pre- the importance of each of these goals below. sented for the catalog after the first 9 months of survey- Understanding the true distribution of column den- ing (Tueller et al. 2008), 22 months (Tueller et al. 2010), sities in the AGN population has been a ques- 36 months (Ajello et al. 2009), 58 months7, 60 months tion of particular interest in X-ray studies of AGN. (Ajello et al. 2012), and the 70 month catalog source list Studies of the X-ray background (e.g., Gilli et al. is now in preparation. 2007, Brandt & Hasinger 2005, Treister & Urry 2005, Much work has been done on the 9-month BAT cat- Worsley et al. 2005, Gandhi & Fabian 2003) suggest that alog (consisting of 153 sources), such as studies of their the majority of accretion in the Universe must be ob- X-ray properties (Winter et al. 2009, W09 hereafter), op- scured. If we wish to determine the true, intrinsic AGN tical properties (Winter et al. 2010), host-galaxy prop- power output, a knowledge of the amount of absorbing erties (Koss et al. 2011a) and construction the nuclear material is needed. We aim to determine the true dis- AGN spectral energy distributions (Vasudevan et al. tributions of X-ray absorption and emission properties 2009, 2010). The X-ray absorption properties of the using a complete and representative subset of the least- AGN in the 36-month catalog have been presented in biased sample of local AGNs. We use the best-quality Burlon et al. (2011) (199 sources). This work draws X-ray spectral data available: therefore XMM-Newton is from the published 58-month catalog which contains 1092 employed preferentially if archival data are present (sup- sources, of which ∼ 720 are AGN candidates (i.e., have a plemented by 13 new XMM-Newton observations taken counterpart identified as a galaxy, AGN, Seyfert, blazar, specifically for this study); other sources of X-ray data or BL Lac, but confirmed to not have a counterpart are detailed in §2. We determine the column density that is a Galactic black hole binary/neutron star/white and nature of the X-ray absorption. In W09, complex dwarf/pulsar). As the catalog becomes increasingly sen- X-ray absorption was found to be common (≈ 55%) in sitive, the data present a great opportunity to improve the 9-month catalog BAT AGNs; we produce new esti- and refine the conclusions drawn from the previous ver- mates of the covering fraction of such complex absorption sions of the catalog, in particular the detailed X-ray anal- where it is present. Quality measurements of the level ysis of W09. and nature of X-ray absorption are important for de- The numbers of AGN in the BAT catalog make it pro- riving reliable absorption-corrected luminosities over the hibitive to perform pointed observations and analysis for broadest possible X-ray bandpass; the luminosities pre- the X-ray properties for all ∼ 720 AGN. The fraction sented here will therefore be useful in constructing low- of AGN with good (XMM-Newton quality, at & 4000 redshift luminosity functions used to assess the amount counts) 0.4–10 keV data over the whole sky is much of accretion and black-hole growth in the local universe. smaller for the 58-month sources than it is for the 9- Our absorption measurements can also serve as a use- < month sources, requiring a more targeted approach. The ful z ∼ 0.1 “anchor” for assessments of the dependence various studies on properties of BAT AGN mentioned of X-ray absorption upon luminosity, Eddington ratio above have concentrated on subsamples from the BAT (L/LEdd), and redshift. The 0.4–10 keV spectra are catalog, including a recent, very thorough X-ray spec- fit jointly with Swift/BAT spectra from 14–195 keV to tral analysis of 48 Seyfert 1–1.5 AGN in (Winter et al. constrain absorption reliably in heavily obscured AGNs > 23 −2 2012). However, the sources in that paper were selected (NH ∼ 10 cm ), including those with Compton-thick 24 −2 based on optical type and specifically to understand the X-ray absorption (NH > 1.4 × 10 cm ). prevalence of ionized absorbers, and therefore the aver- Renewing the W09 analysis of detailed spectral fea- age properties of the sample cannot be directly compared tures will provide further constraints on the physical pro- with the complete sample analysed in W09 as they miss cesses at work in AGN. In W09, the authors search for heavily obscured sources by construction. Our overall iron K-α lines at or near 6.4 keV, soft excesses (spec- goal here is to update the analysis of the complete 9- tral excesses below ∼ 1.5 keV) and signatures of ion- month catalog from W09 and the subsequent analysis ized absorbers or winds (edges in the spectrum). With of the 36-month catalog in Burlon et al. (2011) with a a knowledge of the abundance of such features and any representative, unbiased subsample from the 58-month correlations present (such as the X-ray Baldwin effect catalog. We therefore concentrate on a more manage- linking iron line equivalent width with 2–10 keV intrin- able sample in the Northern Galactic Cap (b> 50◦, 4830 sic luminosity (e.g, Iwasawa & Taniguchi 1993), we can deg2 or 1.47 steradians, 11% of the sky); the sample we better understand the importance of processes such as focus on in this paper has almost exactly the same num- X-ray reflection (e.g., Ross & Fabian 2005) or complex ber of objects as W09’s uniform sample (102 objects), so absorption (e.g., Done & Nayakshin 2007). we can manageably perform our analysis to a comparable The 14–195 keV BAT data have improved in quality level of detail as done in W09. significantly since the W09 study. Only four energy chan- The aims of our study are threefold: firstly, to ob- nels were available in the publicly available BAT spectra tain the absorbing column density for this complete sam- for the 9-month catalog, whereas the 8-channel spectra ple; secondly, to provide a detailed spectral analysis of a for the 58-month catalog sources can be downloaded eas- ‘uniform sample’ akin to the one identified in W09; and ily from the link provided in Footnote 1. Knowledge of thirdly, to fit the higher quality 8-channel BAT spectra the spectrum across the entire 0.4–200 keV energy range alongside 0.4–10 keV data when they provide extra in- will better constrain the absorbing column density in formation on the processes at work in these AGN, not heavily obscured objects (as done by Burlon et al. 2011) attempted previously in the W09 analysis. We discuss and, taking our cue from Winter et al. (2012), we also use the latest BAT spectra to constrain X-ray reflection in a 7 http://swift.gsfc.nasa.gov/docs/swift/results/bs58mon/ subset of our sources with XMM-Newton data (extend- 3 ing the analysis to the absorbed sources in our sample). and workflows perform a wide range of useful functions One problem that has plagued such analyses in the past on data from different X-ray detectors. All data (re- is that the BAT spectra are averaged spectra from 58 gardless of detector used) are fit with the same suite of months of monitoring, whereas the 0.4–10 keV spectra models, and chi-squared comparisons are automatically are snapshots taken for periods of typically ∼ 2 − 20 ks. performed on all of the models fit to determine the best- We therefore do not have simultaneous data across the fit model for a particular source. If the peculiarities of a whole X-ray band, a significant concern since the X-ray particular instrument require a different fitting approach, emission from AGN varies rapidly throughout the 0.4– this can be readily extended due to the modular, object- 200 keV bandpass (Beckmann et al. 2007), and the cross- oriented way in which the workflows have been designed. normalization between the 0.4–10 keV data and the BAT The fitting is semi-automated; i.e., models are initially fit data for a given observation is therefore not known. We to the data and can be inspected and modified if needed, present a strategy in this paper by which the BAT spec- then re-fit; but preliminary ‘first-look’ fits of large num- tra can be renormalized to be quasi-simultaneous with bers of spectra for many sources can also be done in the 0.4–10 keV observations, using the publicly available the background without user interaction. The combined BAT light curves. properties of all sources can be readily gathered together We also present an analysis of ‘hidden’ or ‘buried’ AGN in minutes, and the full range of properties, plots, corre- as done in W09 in order to unearth examples of obscured lations presented in this paper can be readily extended AGNs with a small fraction of scattered nuclear X-rays to other subsamples in the BAT catalog or indeed other (e.g., Ueda et al. 2007). These sources have small levels catalogs. These tools can potentially be used to produce of scattered X-ray continuum. Such objects constitute a complete, consistent X-ray analysis of the entire BAT 24% of the 9-month BAT sources in W09; they may have catalog in the longer term. obscuration subtending most of the sky as seen by the In §2.1 we outline the sample and data sources used. In X-ray source, or emaciated scattering regions. §3 we describe the data processing and fitting of the X- A higher fraction of good-quality 0.4–10 keV coverage ray data in detail. In §4 we present the results from our is available for our chosen sky region (augmented by an analysis, including the absorbing column density and lu- XMM-Newton proposal by PI Brandt to complete the minosity distributions along with analyses of features in 22-month catalog XMM-Newton coverage) than for the the 0.4–10 keV data. In §5 we discuss the utility of joint whole catalog. This region of the sky has high value 0.4–10 keV fits with the 14–195 keV BAT data. In §6 for AGN researchers because it also has extensive multi- we present the average stacked spectrum for the entire band imaging and spectroscopic coverage in the Sloan sample and the log(N)-log(S) diagram to estimate the Digital Sky Survey (SDSS, York et al. 2000) and other sample completeness. In §7 we compare results for the surveys including WISE (IR, Wright et al. 2010), 2MASS smaller 22-month catalog source list (with 90% XMM- (IR8), GALEX (UV9), and FIRST (radio, Becker et al. Newton coverage) with our full results for the 58-month 1995), providing a ready opportunity to extend this catalog (49% XMM-Newton coverage), to 1) to identify work to construct broad-band spectral energy distribu- whether the higher quality data available for the 22- tions and to understand the host-galaxy properties for month subset allows more robust determination of the a complete sample. Using the Northern Galactic Cap sample properties, and 2) quantify any differences be- also ensures low Galactic absorption and more reliable tween the properties of this deeper sample and earlier source identification, due to less potential for confusion versions of the catalog. Finally, in §8 we summarize our with Galactic sources. Our complete subsample is there- findings and present the conclusions. We employ the cos- −1 −1 fore optimally selected for its potential for future multi- mology H0 = 71 kms Mpc , ΩM =0.27 (assuming a wavelength studies. flat Universe, i.e., ΩM +ΩΛ = 1) throughout. We present this subsample as a low-redshift ana- log to more distant samples, such as the Chandra 2. THE SAMPLE AND THE DATA Deep Fields (CDFS-N/CDFS-S, e.g. Falocco et al. 2012, 2.1. Sample Definition Lehmer et al. 2012), the COSMOS field (Brusa et al. 2007) and the Lockman Hole (Rovilos et al. 2011). The We employ the 58-month BAT catalog source list as multi-wavelength work done on the deep fields has proved presented online (Footnote 1) as the definitive source very illuminating for studies of AGN, and our chosen sky list for the whole, all-sky catalog. The catalog contains region has excellent potential for similar wide-ranging 1092 entries of which we determine 720 to be AGN candi- multi-wavelength work, as additional observational cam- dates, with a mean redshift of hzi =0.16 (0.045 excluding paigns are directed at this region of the sky. The low blazars and BL Lacs) providing an excellent local AGN redshifts offer the distinct advantage of better quality survey for our purposes. The signal-to-noise ratio (S/N) data for most objects; we outline in this work how we is above 4.8 for all objects in the catalog list available; can use this sky region to understand AGN accretion Tueller et al. (2010) identify that such a threshold yields comprehensively in the local universe with an unbiased 1.54 spurious detections in the entire 22-month catalog sample, offering the potential to link our findings to those (i.e., there should be 1.54 BAT detections due to random from the deep fields at higher redshifts. fluctuations with S/N > 4.8), and this number should One of the key features of this study is the systematic hold good for the 58-month catalog also, yielding that and uniform way in which we have analysed all of our the number of a false detections in our region of the sky X-ray spectra. The methods implemented in our scripts is less than 0.5, and is therefore negligible. We generate a potential list of targets by filtering the 8 http://www.ipac.caltech.edu/2mass/ objects in the 58-month catalog to select only those 9 http://www.galex.caltech.edu/about/overview.html with Galactic latitude b > 50◦. This criterion pro- 4 duces a list of 143 targets in the desired region of the sky, out of which 31 are blazars, BL Lacs, flat spec- trum radio quasars (as identified from NED and the gn BAT catalog counterpart types) and are excluded from further analysis. We also exclude 3 galaxy clusters (Coma Cluster, ABELL 1914 and ABELL 2029). Of g( the remaining 109 sources, 3 lack counterpart identifi- cations at the time of writing (SWIFT J1138.9+2529, go SWIFT J1158.9+4234, SWIFT J1445.6+2702), and we leave them out of the current study in the absence of 3 a reasonable AGN/Galaxy counterpart identification in 1XPEHU another waveband. The remaining 106 sources con- s sist of Seyferts (Type 1 and 2 and intermediate types), quasars/AGN, LINERs, and the more general category n of galaxies, including 2 ‘X-ray Bright Optically Normal Galaxies’ (XBONGs). ;00 It is essential to have good-quality X-ray spectral data ( to determine the absorbing column densities accurately $6&$ for these AGN, and such data in the soft X-ray regime o o g ( c n t s ) (0.4–2 keV) are particularly important if we wish to iden- tify any signatures of ionized absorption such as O VII ORJrFRXQWV0 and OVIII edges at 0.73 and 0.87 keV, or ‘soft ex- cess’ components which often peak at ∼0.1–0.4 keV. At Fig. 1.— Histogram of total counts (0.4–10 keV) per observation for the 100 objects in our study. XMM-Newton (49 objects) clearly present, a large fraction of the BAT AGN in the most shows far superior counts statistics compared to Swift-XRT (46 recent 58-month catalog do not have good-quality XMM- objects) and ASCA (6 objects). Newton data available in the archives. To try to obtain the true underlying NH distribution for local AGN from (LBQS 1344+0233, VCC 1759, NGC 3718, Mrk 653 and a relatively complete sample, we have therefore selected 2MASXi J1313489+365358). This constitutes a final list an area of the sky for which obtaining XMM-Newton fol- of 100 objects for which we have obtained data from low up to complete the sample is most economical (i.e., XMM-Newton, XRT, or ASCA. The data sources used requires the fewest number of new observations to pro- for each object are provided in Table 1. duce a complete sample). We have already discussed the advantages and detailed properties of the chosen North- 3. SPECTRAL ANALYSIS ern Galactic Cap region (Galactic latitude b> 50◦, solid 2 We present the distribution of total counts per obser- angle 4830 deg ) in the previous section. A successful vation (in the 0.4–10 keV band) obtained using differ- XMM-Newton proposal by PI Brandt has extended the ent observatories in Fig. 1. We can see that the XMM- XMM-Newton coverage of this region with 13 new ob- Newton data clearly have far better statistics than XRT servations. As mentioned above, confusion with Galactic or ASCA, but some of the XMM-Newton observations sources is minimized at this high Galactic latitude, and at the lower-counts end of the distribution may also not the Galactic neutral hydrogen absorbing column density 20 −2 be suitable for a comprehensive analysis of spectral fea- is also low (0.61–4.08 ×10 cm ). tures. We therefore only assess the significance of iron 2.2. Sources of X-ray Spectral Data K-α lines, soft excesses and warm absorber signatures for objects with > 4600 counts in the observation used We use the XMM-Newton X-ray Science Archive (summing over all detectors in the observatory, e.g. pn (XSA) to download all the available XMM-Newton data counts + mos1 counts + mos2 counts for XMM-Newton for the 106 sources, and find that 49 of these objects have observations). This threshold allows, for example, the XMM-Newton spectra (including the new observations detection of a 100 eV equivalent width iron line over a from the XMM-Newton proposal). We therefore provide Γ=1.8 continuum at the 3σ level. For all remaining ob- as complete an analysis as possible for the 49 AGN candi- jects, we perform more basic fits to determine fundamen- dates with XMM-Newton data, and turn to the Swift-X- tal properties such as luminosity and intrinsic absorbing ray Telescope (XRT) and ASCA archives to provide ba- column density. sic coverage for the remaining sources. The Swift/XRT is of particular utility here since the XRT has been used 3.1. Data Reduction specifically to target BAT sources, and the BAT team routinely obtains XRT exposures to gather counterpart 3.1.1. XMM-Newton data information on BAT detections. We prefer Swift-XRT The XMM-Newton data were downloaded from the over Chandra in this study since Chandra observations XMM-Newton Science Archive (XSA) and were reduced are not available as extensively for BAT sources as XRT according to the standard guidelines in the XMM-Newton observations are. Additionally, for sources with the typ- User’s Manual10, using the XMM-Newton Science Anal- ical X-ray fluxes seen in this sample, Chandra obser- ysis Software (sas) version 9.0.0. The tasks epchain and vations would usually exhibit pile-up. We downloaded emchain were used to reduce the data from the pn and XRT data for 45 of the remaining sources, and ASCA MOS instruments, respectively. Initially a circular source data for a further six sources. We find that five sources do not have any data available at the time of writing 10 http://heasarc.nasa.gov/docs/xmm/sas/USG/ 5 region of radius 36′′ was used to extract a source spec- used to provide the X-ray data, observation dates, counts trum, checking for nearby sources in the extraction re- in the observation and optical types. The optical types gion and reducing the source region size to exclude them have been gathered from a variety of sources, primarily if needed. Background regions were either chosen to be the NASA/IPAC Extragalactic Database (NED14) and circles near the source or annuli which exclude the cen- visual inspection of multiwavelength images and spectra; tral source. Additionally, the background light curves as a result these types are very heterogeneous. We do not (between 10–12 keV) were inspected for flaring, and a use these types further in our analysis but provide them comparison of source and background light curves in the for completeness, and urge interested readers to check same energy ranges was used to determine the portions of the types before using them in multi-wavelength work. the observation in which the background was sufficiently low compared to the source; the subsequent spectra were 3.2. Spectral Fitting generated from the usable portions of each observation. We consistently fit a suite of models to all the 0.4– The sas tool epatplot was used to determine whether 10 keV spectral data available to determine the best- pile-up was present in the observations; if strong pile-up fitting model in each case, using python and tcl script- was found, we followed the recommended approach of ing with the xspec package (Arnaud 1996). For some using annular source regions to excise the piled-up core XRT observations, or for XMM-Newton observations of the source, and re-calculated spectra and light curves with few counts or with large portions of the observa- until the strong pile-up was removed. Response matrices tion excluded due to flaring, we impose a lower energy and auxiliary files were generated using the tools rmf- limit greater than 0.4 keV. gen and arfgen, and the final spectra were grouped By default, we fit only the 0.4–10 keV data in order to with a minimum of 20 counts per bin using the grppha concentrate on the accurate determination of soft X-ray tool. feature parameters, but where the counts in the soft band (< 10keV) are insufficient to obtain a good constraint 3.1.2. XRT data on NH or to analyse features (we adopt the threshold of We downloaded the XRT data from the High-Energy 4600 counts for this purpose), we include the BAT data Astrophysics Science Archive (HEASARC)11 . Pipeline- in the fit, extending beyond the W09 analysis. Inclusion processed ‘level 2’ FITS files were readily available from of the BAT data introduces its own complications, par- HEASARC, ready for further processing. The pipeline- ticularly due to the variability in the BAT band over the processed event files from the XRT detector were pro- 58 months during which the spectra were constructed; cessed using the XSELECT package, as directed in the we return to these issues in §3.3 and §5. For all ob- Swift-XRT user guide12. Source regions of 50 ′′ were jects, we perform a comparison between a detailed fit used, with larger accompanying background regions (av- in the 0.4–10 keV band and a broader, ‘continuum-only’ erage radius 150 ′′), and care was taken to exclude other fit to 0.4–200 keV before arriving on a best-fit model. sources in source and background regions. Background For objects with sufficient counts where the BAT data light curves were determined from the event files and are not included in our final best-fit model, we perform inspected for flaring, but this was not found to be a a check to ensure that inclusion of the BAT data does problem in any of our observations. Source and back- not modify the spectral properties recovered from 0.4– ground spectra were extracted, and the source spectra 10 keV data alone; surprisingly we find that the addition were grouped with a minimum of 20 counts per bin by de- of the higher energy BAT data do not significantly alter fault, or a lower limit of 10 counts per bin for observations the best-fit parameters found from analysing the lower- with few counts. The following objects had their spectra energy XMM-Newton results alone. In summary, we find grouped to 10 counts per bin: 2E 1139.7+1040, 2MASX the BAT data are only required to constrain the contin- J13105723+0837387, 2MASX J13462846+1922432, B2 uum and absorption in very high column density sources, 1204+34, CGCG 291-028, MCG -01-30-041, MCG +05- where there are insufficient counts in the 0.4–10 keV band 28-032, NGC 4939, NGC 5106 and Ark 347. Some ob- to constrain the soft X-ray spectral features. jects have too few counts to construct a spectrum (NGC All models include Galactic absorption by default (de- 4180, NGC 4500, MCG -01-33-063, CGCG 102-048 and termined using the nh tool from the ftools suite of util- 2MASX J13542913+1328068); for these we present ba- ities, Blackburn 1995); this is by design uniformly low for Gal 20 −2 sic luminosity estimates and other quantities (including this sample (0.61 < NH < 4.08 × 10 cm ). We fol- upper limits where appropriate) in Appendix B. low W09 by classifying AGN spectra into two broad cat- egories: those with ‘simple’ absorbed power-law spectra 3.1.3. ASCA data (with or without features such as an iron line at 6.4 keV, ASCA spectra (pre-reduced) were downloaded from ionized absorption edges, or a soft excess below ∼ 2 keV); the Tartarus archive13 for the sources 3C 303.0, Was and ‘complex’ spectra for which partially covering ab- 49b, Mrk 202, NGC 4941, Mrk 477 and NGC 4619, for sorbers or double power-law models must be invoked to which Swift-XRT or XMM-Newton data were not found provide a statistically acceptable fit to the spectrum. at the time of writing. The various model sub-types are presented in Table 2. The details of all observations used are presented in We differ slightly from W09’s analysis by not employing Table 1, including their sky positions, the instruments a model combination that includes both partial covering and a black-body component, as used for a handful of 11 http://heasarc.gsfc.nasa.gov/cgi- their sources; we use the term ‘soft excess’ in this work bin/W3Browse/w3browse.pl to refer uniformly to an excess above a clear power-law 12 http://heasarc.nasa.gov/docs/swift/analysis/ 13 http://tartarus.gsfc.nasa.gov/ 14 http://ned.ipac.caltech.edu/ 6 in which the spectrum shows low neutral absorption, and when these features produce a reduction in chi-squared, do not use the term to refer to the soft features often seen but are not significant according to the ∆χ2 > 4.0 crite- in heavily absorbed spectra, as these are probably due to rion. different physical processes. We systematically and self- We also present a plot of χ2 against the number of de- consistently fit all of the model combinations using our grees of freedom (n d.o.f) for all the fits in our sample semi-automated system which initially fits the data with in Fig. 2 to illustrate the quality of the fits, as done a model, presents the findings to the user allowing any by Mateos et al. (2010) (Fig. 2 of their paper). The necessary adjustments or re-seeding/freezing of parame- solid line represents 1:1 correspondence between χ2 and ters, and re-fits the data with the refined model. In this n d.o.f, and the dashed lines represent the extremal val- study, we also introduce the category of an ‘intermediate’ ues of χ2 above or below which we would expect less model type, where either two models fit the data equally than a 1% probability of obtaining such a χ2 if the well (e.g., χ2/d.o.f < 1.0 in both cases) or visual inspec- model is correct. We color code the objects by spec- tion of a fitted spectrum revealed that, whilst a partial tral type, with red triangles showing ‘simple’ spectral covering model may fit the data better in a formal sense, types and green squares showing ‘complex’ ones. At low the spectrum did not show obvious signatures of strong n d.o.f (hence low counts), almost all fits lie within these complexity. For this class of objects, we present the re- limits, but at higher n d.o.f, we do see some worse fits sults for both models in all subsequent tables and figures (the maximal reduced chi-sqared for the whole sample is and indicate them clearly. Thirteen such intermediate χ2/n d.o.f ≈ 4.0), with a majority of these objects ex- spectral-type objects in our sample highlight the need hibiting ‘complex’ spectral types. A comparison of our for better quality data (for cases where both simple or plot with Fig. 2 of Mateos et al. (2010) shows that our complex absorption models over-fit poorer quality data), sample extends to higher n d.o.f, due to a number of or for further investigation of the spectra (in the cases datasets with greater counts than those in their study. where good-quality data reveals an indeterminate spec- Our plot also shows that a majority of the objects with tral shape). poor fits have complex spectral types. In the high counts Additionally, we emphasize that the complement of regime, we encounter spectra of very high quality (with models used here encompasses the most common charac- high counts statistics) exhibiting strong spectral com- terisations of AGN X-ray spectra, and do not represent plexity that cannot be modeled by the suite of model an exhaustive list of physical scenarios. We do not at- combinations used here, including notably NGC 4151 tempt to model all of the more intricate features that and Mrk 766. Another notable outlier, II SZ 010, whilst may be present, and as a result we do see some poor fits having a ‘simple’ spectral type, exhibits a poor fit due 4 (null hypothesis probabilities < 5 × 10 ), primarily due to the inclusion of the BAT data in the fit, and very low to not modelling complex iron lines and line emission at flux in the BAT band at the time of the XRT observation soft energies in complex-absorption sources. These cases (see §3.3). We find that removing the BAT data brings are briefly discussed in Appendix C. χ2/n d.o.f much closer to 1.0 without altering the key We fit the models in Table 2 to all spectra (omit- results (photon index, column density) for this object. ting model combinations with features such as soft ex- We present the rest-frame 2–10 keV observed lumi- cesses, iron lines, or edges for XRT or ASCA data due nosity (not corrected for absorption) against redshift in to lower signal-to-noise ratio or lower sensitivity below Fig. 3, to characterize the redshift distribution of the 1 keV), and determine the best-fitting model by order- sample. All sources are located at redshifts z < 0.2, with ing the model fits based on their reduced chi-squared an average of hzi =0.043, slightly higher than the aver- values. Although we only analyze the properties of age of 0.03 obtained in both W09 for the 9-month BAT soft excesses, lines and edges for objects with > 4600 catalog and in Burlon et al. (2011) for the 36-month cat- counts, we fit models including these features to all alog. XMM-Newton datasets and later exclude objects below the counts threshold when determining the prevalence and sample-wide properties of such features. We also es- 3.3. On the issue of simultaneity across the entire timate the significance of components such as lines, edges 0.4–200 keV band and soft excesses by requiring a reduction in chi-squared The BAT spectra have been gathered over the entire of 4.0 per degree of freedom for a feature to be deemed duration of the survey, and are therefore not in any ‘significant’ (corresponding to a ≈ 95% confidence detec- sense ‘simultaneous’ with the 0.4–10.0 keV data used tion of the feature). The basic fit results are presented in from XMM-Newton, Swift/XRT or ASCA. It is impor- Table 3, and the analysis of detailed features (iron lines, tant to use simultaneous observations wherever possible soft excesses, and ionized absorber edges) is presented in when combining multi-wavelength data due to the vari- Table 4. Fig 4 shows some example spectral fits, show- able nature of AGNs. This is even more pertinent at ing both the raw spectrum with the ratio of the data high energies, where the short-time scale variability is to the model (counts s−1 keV−1 against energy in keV in reflective of rapidly changing accretion processes occur- the upper panel, ratio of data-to-model against energy in ring close to the inner regions of the accretion flow. keV in the lower panel) and a ν Fν spectrum (unfolded Whilst we cannot obtain a BAT spectrum simultane- through the response, keV2 ×(Photons s−1 cm−2 keV−1)) ous with the 0.4–10 keV observation due to insufficient for 3 objects. Table 8 shows basic results for objects with counts in the BAT instrument in such short (1 − 100ks) very few counts, including upper limiting luminosities. In time intervals, we can attempt to account for this effect Table 4, we show upper limiting equivalent widths, soft- in some measure using the BAT light curves. These are excess strengths and warm-absorber edge optical depths available for each BAT catalog source, spanning the en- tire 58 months of the survey. The variability displayed 7

to-noise ratio to paramaterize properly, and a full treat- ment of this effect will be presented in Shimizu et al. (in 1*&Sfdd d2f prep.). Their preliminary analysis of hardness ratios for the brightest ∼ 30 BAT sources reveals minimal spectral 0UNS variability across 14–195 keV, but this analysis is only 1*&Sf2d possible on the brightest sources on variability timescales d222 greater than 30 days. Ideally we prefer truly simulta- 0UNSfo neous data (such as will be obtained with co-ordinated ,,S6=S2d2 NuSTAR and 0.4–10 keV campaigns, or ASTROSAT )

. alongside the 0.4–10 keV data to be able to interpret the Ȥ d22 broad-band spectral shape fully. Many of our soft (0.4–10 keV) X-ray observations have been taken within the timeframe of the BAT survey. Where possible, the BAT light curve is used to estimate d2 the variation in the overall normalization of the BAT spectrum, by considering the BAT flux ratio relative to the full 58-month average, at the date of observation of the soft X-ray data.

d f We then re-normalize the BAT spectrum accordingly, d2 d22 d222 d2 Q whenever the soft X-ray data have been taken within the SG7R7I span of the BAT survey. We show an example in Fig. 5. The key improvement in this approach is produced when Fig. 2.— Plot of χ2 against the number of degrees of freedom fitting the XMM-Newton/XRT/ASCA spectra jointly (n d.o.f) for the best-fit models for each source. Red triangles repre- sent objects with ‘simple’ model spectral types, and green squares with the BAT data within xspec: without such renor- represent ‘complex’ ones (as defined in Table 2). The solid line malization, we would have to allow the normalizations of indicates χ2 = n d.o.f, and the dashed lines indicate the limiting the XMM-Newton/XRT/ASCA and BAT data to ‘float’ χ2 for a given n d.o.f above or below which we expect less than a 2 with respect to each other because of this uncertainty 1% probability of seeing such extremal values of χ if the model is in the absolute normalization of the BAT spectrum due correct (see Mateos et al. 2010). to non-simultaneity with the 0.4–10 keV data. However, by re-normalizing, we can lock the normalizations of the BAT and 0.4–10 keV (XMM-Newton/XRT/ASCA) com- ft ponents together, removing a degree of freedom from the fit and providing more stringent constraints on the pa- rameters obtained from model fits. This is particularly ff useful when performing fits to determine reflection pa- rameters, which we will return to in §5.2. The utility of this renormalisation is illustrated in Fig. 6, where we plot the 2–10 keV flux against the V fi REV BAT (14–195 keV) flux, color-coding the observations based on the measured column density from spectral fit- h4dRNH9 ting. We overplot lines showing the expected ratio of fh − − ORJe/ F2 10keV/F14 195keV for different fiducial absorption lev- els and intrinsic photon indices, to indicate the predicted locus of objects in this plot depending on absorption fd and spectral slope. In the left panel we see that be- fore re-normalization, the fluxes cluster tightly close to the BAT flux limit for our sample, irrespective of ab- fR sorption and do not lie in the regions expected for their dRíi R3Rd R3d measured column density. After re-normalization (right 5HGVKLIWg] panel of Fig. 6), the objects overwhelmingly shift into the expected regions for the three fiducial column den- sity ranges shown. (obs) Fig. 3.— Observed 2–10 keV luminosity L2−10keV (not cor- There is one outlier in the far left of the right rected for absorption) against redshift for the 100 sources in panel of Fig. 6, the low-absorption source 2MASX our sample. J12055599+4959561 for which the re-normalization does not appear to work well. When re-normalization is ap- for the BAT AGN in these light curves indicates that the plied, this object lies far from the expected position in true 14–195 keV spectrum at a particular epoch within F2−10keV − F14−195keV. This behaviour is contrary to the survey may look significantly different to the final expectations, since we would expect that re-normalizing averaged 58-month spectrum. Such variation will be the BAT data to be contemporaneous with the 2–10 keV composed of two components: variation in the overall data would improve the congruence between the two lu- normalization (i.e., flux) and spectral shape. We aim minositites. This object also displays an unusual ratio to account for the former effect in this work. Variabil- of L14−195keV/L2−10keV (Fig. 7). Inspection of the joint ity in spectral shape requires particularly good signal- XRT+BAT spectrum reveals a renormalized BAT spec- 8

NGC5899 NGC5899

0.1 0.01

0.05 −1 keV −1 0.02 ) −1

0.01 −3 keV 10 −1 s

5×10−3 −2 normalized counts s

2×10−3 (Photons cm 2

1.4 −4 keV 10

1.2 ratio

1

1 2 5 1 2 5 Energy (keV) Energy (keV) NGC5290 NGC5290

0.01 −1

keV −3 −1 10 ) −1 10−4 keV −1 s −2 10−5 0.01 0.1 normalized counts s

10−6 (Photons cm 2 4 keV

3 −3 10 ratio 2

1 1 10 100 1 10 100 Energy (keV) Energy (keV) 3C303.0 3C303.0

0.01 −3 −1 10 keV −1

−3

10 ) −1 −4 10 keV −1

10−4 s −2 normalized counts s −5 10 10−5 (Photons cm

3 2 −6 keV

2 10

1 ratio

0 −7 10 −1 1 2 5 1 2 5 Energy (keV) Energy (keV) Fig. 4.— Some example spectra from our sample, with the left panel plots showing the spectrum and ratio plot, and the right panel plots showing the νFν plots. The objects are, from top-to-bottom, NGC 5899 (XMM-Newton data), NGC 5290 (XRT+BAT data) and 3C 303.0 (ASCA data).

trum that lies below the XRT spectrum in flux. If we hlog(L14−195keV)i = 43.7, σL14−195keV = 0.8 from the 9- use the raw BAT spectrum without re-normalization, the month catalog (using W09’s results), and for consis- BAT and XRT spectra link continuously in a νFν plot, tency, we exclude any objects with jets analyzed in W09 with a hard, simple power-law spectrum with Γ = 1.5 in calculating this average. We also present plots of intrinsic and negligible NH . Inspection of the BAT light the BAT-X-ray colors/hardness ratios as done in W09 curve shows that there is a pronounced dip at the time for easy comparison of our present, deeper sample to the XRT data were taken, and that the source is very the 9-month catalog results. In Fig. 7 we see the soft faint in the BAT band. This might imply that the re- color F0.5−2keV/F2−10keV plotted against the hard color normalization occasionally fails for very faint sources, F14−195keV/F2−10keV. The range of colors spanned in the but for all other objects the re-normalization produces 58-month catalog appears larger on both axes than that results consistent with the lines of constant Γ and NH in seen in W09. In the same region of parameter space Fig. 6. spanned by the 9-month catalog, we see the same di- vision into regimes occupied by high, intermediate, and 4. RESULTS low absorption sources, but there are three sources with extreme values (outside the range of the plot): these are 4.1. Average Sample Properties and Distributions 2MASX J12055599+4959561 (for which logNH < 22, The average BAT luminosity for our sample is F14−195keV/F2−10keV < 0.1, and F0.5−2keV/F2−10keV ≈ hlog(L14−195keV)i = 43.5, σL14−195keV = 1.1. This 0.4), 2MASX J13105723+0837387 and CGCG 291−028 result is similar to the average BAT luminosity of (log NH > 23, F14−195keV/F2−10keV ≈ 10, and 9

i ic −5 F0.5−2keV/F2−10keV < 10 ). For CGCG 291−028 and 2MASX J13105723+0837387, the model fits have very little 0.5–2 keV flux, but this is likely due to their poor XRT data quality, since these objects required the inclu- fîciía sion of BAT data to obtain a fit at all. For 2MASX J12055599+4959561, the highly unusual ratio of BAT flux to 2–10 keV flux indicates a renormalized BAT spec- ž trum that lies below the XRT spectrum in flux, due to a i dip in the BAT light curve at that date, as discussed in 5$7( §3.3. W09 presented the diagnostic plot intrinsic intrinsic L2−10keV/L14−195keV versus L2−10keV + L14−195keV, ífîciía showing the regions of the plot populated by objects of different photon indices Γ; we reproduce this plot in Fig. 8. We note the unusual location of 2MASX J12055599+4959561 again as in Fig. 7, and additionally íi ic i ci ei ai ri fi li gi highlight NGC 5683, identified as a Seyfert 1. The 0RQWKTVLQFHTVWDUWTRITOLJKWFXUYHs%$7TVXUYH\ measured photon index for NGC 5683 is 2.15, but this source does not have a re-normalized BAT spectrum. As Unfolded Spectrum a result, the ratio of the measured BAT luminosity to the 2–10 keV luminosity places it in the region expected for extremely hard sources with Γ < 1.0. We now present the distributions of key quantities in- 0.01

) cluding the absorbing column density (Fig. 10), photon −1 index Γ (Figs. 16,17,18) and 2–10 keV intrinsic luminos- keV −1 s

−2 ity (Figs. 19,20). −3

10 4.2. Radio loudness (Photons cm 2

keV We present the radio loudness values for our int sample (defined as νLν (ν = 5 GHz)/L2−10keV, Terashima & Wilson 2003) plotted against the intrinsic 2–10 keV luminosity in Fig. 9, to provide an indication of −4

10 1 10 100 the radio properties of the sample. The radio luminosi- Energy (keV) ranjan 1−May−2011 21:43 ties are taken from the FIRST survey at 1.4 GHz, and Unfolded Spectrum we convert the fluxes to 5 GHz using a standard spectral index of α = 0.7 typical for synchrotron emission (for −α flux density fν ∝ ν ; see e.g., Mel´endez et al. 2010 for a discussion of radio spectral indices). Where no match 0.01 is found within 5 ′′ for a given source, we assume a flux ) −1 limit of 0.75 mJy for the survey and use it to calculate an keV −1 upper limiting radio luminosity. The large angular size s −2 of the FIRST beam is likely to introduce significant con- tamination from the host galaxy, but our main purpose −3

10 here is to catch significant outliers where the nuclear ra- (Photons cm 2 dio emission is heavily boosted by a jet. Such objects keV will easily stand out from the rest of the distribution. All of our sources have radio-loudness values below −2, and all but two of our sources are below the values typi-

−4 cally seen for strongly-beamed sources such as BL Lacs or 10 1 10 100 Energy (keV) flat-spectrum radio quasars (Terashima & Wilson 2003). ranjan 1−May−2011 21:44 The two objects with the highest radio-loudness param- Fig. 5.— Illustration of the renormalization of the BAT eters are Mrk 463 and 3C 303.0. The object 3C 303.0 spectrum using the light curve for Mrk 1310. The top panel is expected to be radio loud based on its inclusion in shows the BAT light curve, with a red bar indicating the the 3C catalog. Further investigation of Mrk 463 reveals date of observation of the 0.4–10 keV data. The 0.4–10 keV data are from XRT in this case. The middle panel shows that it contains two nuclei at very close separation (3.8 the 58-month averaged BAT spectrum used as-is along with kpc, Bianchi et al. 2008), with the eastern source Mrk XRT data, whereas the bottom panel shows the renormalized 463E dominating the X-ray emission by a factor of ∼ 4. BAT spectrum (thereby made ‘quasi-simultaneous’ with the The brighter Mrk 463E nucleus was previously found to XRT data). Prior to re-normalization, the spectrum looks have a Seyfert 2-type optical spectrum, but a fuller con- like a continuous unabsorbed power-law across 0.4–200 keV, sideration of the radio morphology reveals that it is a but after re-normalization, the spectrum has the appearance ‘hidden’ Seyfert 1 nucleus (Kukula et al. 1999). Both nu- of strong reflection or complex absorption. clei show moderate-to-weak radio emission (Drake et al. 2003). Our XMM-Newton and Swift/BAT analysis of 10

īg,g050 īg,g-5V īAeA(.( īAeAg.) ORJ31+>g9g00 ORJ9bu+AkA((

00g9gORJ31+>g9g0k ((AkAORJ9bu+AkA(N

ORJ31+>gfg0k ORJ9bu+AVA(N

5 5 5 5 > + .0 í-- =( ígg -F 5 go 5 gFP AFP

.- 5 =g 5 55 5 5 555 5 5 5 55 5 5 55 5555 5 5 55555 g3HUJgV 555 A9HUJAV 5 5 5 5 5 5 5 5 5 5 55 5 5 5 0.-FgNH9 5 55 5 (=goANH9 5 55 5 ) 5 5 i 5 5 5 5 5 5 5 55 5 55 5 5 5 5 55 5 5 55 55 5 5 5 5 5 5 -Fí-0 goíg( 5 5 5 5 (vc6;Awg(o)))RR6HR)R)

5 5 -Fí-k -Fí-0 -Fí-- -Fí-F -Fím goígN goíg( goígg goígo goíR .- .0 =g =( )-e.-mVgNH9g3HUJgV gFP 1gIURPgFDWDORJ> igH=gR)ANH9A9HUJAV AFP ΓAIURPAILWΓAXVLQJAUHQRUPAZKHUHADYDLODEOH+

Fig. 6.— Observed 2–10 keV flux (F2−10keV) against BAT flux (F14−195keV ). The red triangles represent sources with NH < 22 −2 23 −2 10 cm , while the green circles represent sources with NH > 10 cm . Blue squares indicate the intermediate column sources. Sources for which renormalization of the BAT data was done are indicated using an ‘R’. The left panel shows the comparison if the BAT flux is calculated from the average spectra presented in the 58-month catalog, and the right panel shows the results obtained if we re-normalize the BAT spectrum as detailed in §3.2 . The vertical line indicates the flux limit of our sample based on the average fluxes reported in the 58-month catalog.

Γ9l  =0r6;C-Γ=4lllLLJ7LlLloΓ Γ

 S 49l

Γ70ΓLlCNH9 īCVC=94 +/ 4 9NH9

) īCVCΓ9gl  ALQWULQVLFS í49l 9.NH9 ) =0Γ4CNH9

ORJA/ íΓ í  īCVCΓ94 1*cClo(. íΓ9l

í í=    7Γ 7= 7. 77 7l 7o ALQWULQVLFS 0Γ )9.NH9)9NH9 ORJA/=0Γ4CNH9 CJC/Γ70ΓLlCNH9SCAORJC/+HUJCV S

◦ intrinsic Fig. 7.— The 58-month BAT AGNs with b > 50 with Fig. 8.— Right: ratio of L210keV /L14−195keV vs. the to- soft and hard fluxes plotted on the color-color plot initially tal luminosity in the 2–10 keV and 14–195 keV bands. The presented in Winter et al. (2008), and later in W09. The dotted lines show values of constant Γ between both bands 22 −2 ( red triangles represent sources with NH < 10 cm , while at constant ratios of L intrinsic)210keV /L14−195keV , following 23 −2 the circles represent sources with NH > 10 cm . Squares the presentation in W09. The different absorption levels are indicate the intermediate column sources. distinguished using the same key as in Fig. 7

4.3. Column density NH this source therefore is likely to include the combined As the BAT survey increases its exposure, we expect it emission from both sources (as discussed in Bianchi et al. to uncover a more accurate reflection of the true absorp- 2008). We discuss these two cases in Appendix A, but tion distribution for the AGN population, and to see dif- a more detailed study of the X-ray properties of these ferences from the earlier 9-month catalog analysis. If we objects is needed. compare the absorption distribution seen here (Fig. 10) 11

are very close to the threshold for being Compton-thick, í= and the errors on their column densities could push them over the threshold. This result suggests that ∼ 9% of our

a 0UNkgνl

LQW sample is Compton-thick. However, at such high column

l7klelue densities, basic photoelectric absorption is not sufficient

=LseNH9 íl to model the level of absorption present and more so- phisticated absorption models must be used to calculate the column density for such objects. This exercise is be- íg yond the scope of this paper, but we discuss alternate R(k*+]aknk/ Ȟ measures of Compton-thickness in §5.1. The histograms in Fig. 10 show a clear bifurcation be- í( tween spectral types in terms of their absorption. Simple spectra overwhelmingly fit objects with low absorption, and complex spectra are generally required for objects with high absorption. However, the intermediate class íν of objects identified in this study introduces some un- certainty in the distributions, since for these objects the 5DGLRk/RXGQHVVk2kORJRȞ/ different model types yield different estimates of NH. We í5 therefore show the ‘worst-case’ scenarios in Fig. 10, as- ge gs g= gl gg g( LQW suming that the intermediate objects are all simple or /=LsekNH9 complex in the left and right panels, respectively. As- suming that these objects take the NH values of their Fig. 9.— Radio-loudness (νLν (ν = 5GHz)/L2−10keV ) complex model fits, we see a more pronounced peak in against intrinsic 2–10 keV luminosity. Downward pointing high-absorption sources. We also note that a substantial arrows show upper limits where FIRST radio detections were fraction of sources have negligible intrinsic absorption not available (assuming a flux limit of 0.75 mJy to calculate the luminosities). (log NH < 20). The relationship between absorption and intrinsic 2– 10 keV luminosity is given in Fig. 12. This plot shows with that seen in the 9-month catalog (W09, see Fig. 25 a broad absorption distribution at all luminosity levels. of this paper), we indeed see that our distribution shows a The thirteen sources with ambiguous spectral types can tail at higher column densities than that seen previously, have highly uncertain column densities (indicated by the and the average absorbing columns from our distribu- blue shaded boxes). We plot the absorbed fractions (us- tion are hlog NH i = 20.80, σ =1.18 (simple, 51 objects) ing thresholds of log NH = 22 and 23 to defined ‘ab- and hlog NH i = 23.55, σ = 0.71 (complex, 44 objects), sorbed’) as a function of 2–10 keV luminosity in Fig. 13, assuming the ‘simple’ model type for any objects with using 10 objects per bin to determine the absorbed frac- dual best-fits. We assume all objects with log(NH) < 20 tion. We do not see as strong a decrease in the ab- to have a lower-limiting absorption of log(NH) = 20 for sorbed fraction with luminosity as previously reported a consistent comparison with W09. If we assume dual in Fig. 13 of Burlon et al. (2011) and the earlier INTE- objects are by default complex, we find slightly differ- GRAL AGN survey (Beckmann et al. 2009), although a ent distributions: hlog NH i = 20.67, σ = 1.12 (simple, similar trend is present. Our work reveals a more ho- 38 objects) and hlog NH i = 23.27, σ = 0.95 (complex, mogenous distribution of absorption throughout lumi- 57 objects). We contrast these results with those from nosity space, and notably three heavily absorbed sources W09, who find hlog NH i = 20.58, σ = 0.74 (simple, 46 (log NH > 23.5) are found at the highest luminosities (3C objects) and hlog NH i = 23.03, σ = 0.71 (complex, 56 234, 2E 1139.7+1040 and 2MASX J11491868-0416512). objects), verifying a tail of higher-absorption objects in The Burlon et al. (2011) and (Beckmann et al. 2009) our sample. If we split the objects based on the spec- samples contain more sources at high luminosities than tral complexity exhibited, we find that the percentage of our sample, and it is above luminosities of 1044 ergs−1 ‘complex’ objects is in the range 43–56% (of the whole where the decrement in the absorbed fraction is more sample, 100 objects), with the range again due to the pronounced. It is therefore possible that we do not have presence of sources with ambiguous spectral types. How- enough sources at high luminosities to see the decrement. ever, inspection of Fig. 11 shows that the absorption dis- We see from Fig. 13 how the distribution of absorbed tributions reported in our sample do not vary appreciably sources depends on the threshold absorption used; we depending on BAT flux. Interestingly, in the region of quantify this effect in Fig. 14 by employing three thresh- flux space already covered by the 9-month catalog, we olds of log(NH) = (22, 22.5, 23), and plotting the fraction find an increase in the average absorbing column density of objects above these thresholds against the threshold towards the highest BAT fluxes. itself. We present a comparison with the 9-month cat- We also identify the proportion of Compton-thick alog results. The shaded area in the figure again shows 24 −2 (NH > 1.4 × 10 cm ) sources in our sample. In the uncertainty due to the non-unique model fits for 13 contrast to W09 who find no Compton-thick ob- sources. It is clear that the absorbed fractions are uni- jects by this criterion, we find eight Compton-thick formly higher at all thresholds in our work than in W09, sources in our sample: these are NGC 4102, 2MASX and that the slower fall-off with absorption threshold J10523297+1036205, 2MASX J11491868-0416512, B2 indicates that a substantial proportion of our absorbed 1204+34, MCG -01-30-041, MRK 1310, PG 1138+222 sources are heavily absorbed (log NH > 23), about 10% and UGC 05881; additionally, NGC 4941 and NGC 5106 more than in the 9-month catalog. Inspection of a plot 12

 

 

 

  1 1  

 

 

             

ORJH1+) ORJH1+)

Fig. 10.— Histograms of absorbing column density log(NH). The blue shaded portions represent simple spectrum objects, whereas the semi-transparent grey shaded portions represent complex spectrum objects (semi-transparent to show the underlying blue histogram when present). In the left panel, any objects with ambiguous spectral classifications (both a simple and a complex spectral model describe the data well) have been assumed to be simple spectrum objects; in the right panel those same objects are assumed to possess complex spectra.

of absorbing column density against redshift (which we omit for brevity) indicates no evolution of the absorption s( distribution with redshift over the redshift range probed )OX[ VSDFH H[SORUHG )OX[ VSDFH H[SORUHG E\ dHuPRQWK FDWDORJXH E\ )uPRQWK FDWDORJXH by this survey. Therefore, the deeper flux limit must be responsible for picking up more asborbed objects. At log(NH) = 23.5, ∼ 5% of the intrinsic flux in the BAT sr band is absorbed, so we would expect that the deepening of the survey would produce this kind of increase in the proportion of highly-absorbed objects. c + ss 4.4. Photon Index ORJh1 We restrict photon indices to lie between 1.5 < Γ < 2.2 in our fits for observational and physical reasons: observationally, the AGN in the XMM-Newton Bright sF Serendipitous Survey (Corral et al. 2011) indicate 3σ limits on their absorption-corrected photon indices con- sistent with this restriction; physically, photon indices much lower (harder) than 1.5 indicate unphysical Comp- xH ton y parameters in standard inverse-Compton scatter- x xF xFF xFFF uxs us ing scenarios for modelling the X-ray power-law emission )xrux)d NH9 hxF HUJ9V9FP c hIURP SXEOLVKHG FDWDORJXHc from the corona (e.g., Zdziarski et al. 1990). Fig. 15 is a −2 plot of Γ against the intrinsic X-ray luminosity L − , Fig. 11.— Absorbing column density log(N /cm ) against 2 10keV H and the distribution of photon indices as histograms is BAT flux (F14−195 keV); black points show the results for the 58-month catalog b > 50◦ sources and the small green presented in Figs. 16 and 17. triangles show the results from the 9-month catalog (from Earlier studies have investigated whether photon in- W09). The blue (thin) and brown (thick) error bars and dex is correlated with luminosity or Eddington ratio; points show the results binned by F 14−195keV from two differ- larger photon indices (softer X-ray spectra) are seen to ent approaches; 1) using a constant number of objects in each accompany higher accretion rates with more prominent bin, or 2) using a constant interval in F14−195keV for each bin. accretion disc components in Galactic Black Hole can- −2 The mean NH/cm in each bin is calculated, and the error didates, constraining the physics of different accretion bars are the standard deviation to show the degree of spread states (for a review see e.g., Remillard & McClintock (not the error on the mean). The distribution of absorption 2006). This parallel has also been explored with AGN probed in this region of the sky appears to be independent of (e.g., K¨ording et al. 2006, Shemmer et al. 2008). Fig. 15 BAT flux, with a wide range of absorbing columns seen at all shows that there is no observed correlation between pho- flux levels. We note the different distribution found in W09 and discuss this in the text. ton index and 2–10 keV luminosity, although the more physically relevant correlation is with Eddington ratio, 13

ec scsi5aa

cp c ec scib5ab ORJA1+bVti >+eg(@ i(bo5big co ORJA1+bVtt ]

nc a 6 cb rFP +

ca a .

cc

a n

cs $EVRUEHGgIUDFWLRQ

,QWULQVLF DEVRUSWLRQ ORJ91 DEVRUSWLRQ ,QWULQVLF ci a t

s(

bi bs bc ba bb bo a LQW nc nt ni nn ORJ9/cnsi NH9 ] dc ORJg/tdcaNH9gAHUJgV b

−2 Fig. 12.— Absorbing column density log(NH/cm ) Fig. 13.— Absorbed fractions against intrinsic 2–10 keV against intrinsic 2–10 keV luminosity L2−10keV (absorption- luminosity (10 objects per bin). Filled circles connected by corrected). The black circles represent objects for which a thin solid lines show the fraction of sources with log(NH) > unique best-fit model was determined. For objects where a 22, whereas empty squares connected by thick lines show the unique best-fit model could not be determined and two ‘best- fraction of sources with log(NH) > 23. The solid grey and blue fit’ models were identified, blue shaded areas represent the hatched shading reveal the uncertainty in these fractions due range in log(NH) and L2−10keV spanned by those two mod- to the thirteen sources with ambiguous spectral types (and els (with the model fit results themselves signified by blue hence two estimates for their log NH). The absorbed fraction crosses), and the blue oval points represent the central, aver- in the highest luminosity bin (indicated by the red square) is age values for that object. The dashed horizontal line shows more uncertain since it contains only four objects. the conventional threshold between ‘absorbed’ and ‘unab- sorbed’ objects (log NH = 22), and the dot-dashed line shows the threshold for Compton-thick objects (log NH = 24.15). 19 −2 All objects for which NH was below 10 cm are shown at log(NH)=19; in these sources absorption by neutral gas has negligible effect on the X-ray spectrum. fi g + which we do not explore here as we lack black hole mass estimates for the entire sample. The limits imposed on Γ could prevent us from seeing any correlation, if present; ei however, if the range 1.5 < Γ < 2.2 is an appropriate Pb)PRQWKmFDWDORJmlWKLVmZRUNg physical constraint to place on Γ, then this should not gm3mWKUHVKROGmORJl1 be an issue. In any case, our study reinforces the re- + sult in W09 that the photon index does not appear to Pi be correlated with 2–10 keV luminosity, at least in our j)PRQWKmFDWDORJml:ijg complete sample of objects, although W09 suggest that they may be correlated when comparing multiple obser- vations of an individual object. We also note that for the high-absorption sources, the photon index is more ki prone to uncertainty due to complex absorption, and that the photon indices in such sources are more likely to serve simply as an indication of the general spectral 3HUFHQWDJHmRImREMHFWVmZLWKmORJl1 ri shape. If we therefore restrict our view to unabsorbed, ww ww5w ww5k ww5e ww5b wr log(NH) < 22, sources in our search for a correlation 7KUHVKROGmORJl1+g between Γ and L2−10keV, we still do not observe any ob- vious correlation, but note that there is a larger range Fig. 14.— Absorbed fraction vs. threshold log(NH), for of Γ values at high luminosities, whereas low L2−10 keV the 9-month catalog (W09, green triangles and dashed line) luminosities seem to accompany lower values of Γ. We and for the 58-month catalog Northern Galactic Cap sources defer further investigation of this topic to a dedicated (solid lines and shaded area, this work). The shaded area study on multi-epoch observations of BAT sources. shows the uncertainty in the absorbed fractions in this work We find that in high-absorption objects, the photon due to the sources with ambiguous spectral types and non- index hits our hard limits imposed on Γ more frequently. unique best-fitting column densities. In objects with log(NH) > 22, we see a larger fraction of 14

4.5. 2–10 keV intrinsic luminosity The average 2–10 keV luminosity for the sample −1 is hlog(L2−10keV/ergs )i = 43.00, with σlogL = sos 0.91 − 0.92, similar to the distribution seen in W09 9month −1 (hlog(L2−10keV/ergs )i = 43.02, with σlogL = 0.87), but we notice differences when we split the sample based on the absorption level or spectral complexity. In- s specting the histograms in Figs 19 and 20 shows that complex/high-absorption sources appear to have a wider distribution in luminosity than simple/low-absorption ī sources, notably showing more of a spread to low lu- ro minosities. Indeed, Fig. 12 demonstrates this broader distribution of absorbed sources. The simple/complex bifurcation again closely corresponds to the low/high ab- ro sorption split. Simple/low absorption objects have an av- erage luminosity of log(L2−10keV) = 43.02 − 43.11, with σlogL = 0.87 − 0.98 (the ranges take into account the uncertainty in spectral type for some sources). For com- ro3 3/ 3r 3s 3) 33 35 plex/absorbed sources, we find a slightly lower average lr log(L2−10keV)=42.88 − 42.97, σlogL =0.89 − 0.96. /slr/.NH9. ORJ./gHUJ.V k

Fig. 15.— Photon index Γ against intrinsic (absorption- 4.6. Hidden/Buried Sources corrected) 2–10 keV luminosity L2−10keV . The grey shaded This class of sources was identified as a poten- areas delineate the hard limits imposed on Γ in the fit. Red tially important component of the X-ray background triangles depict low absorption (log NH < 22) objects, blue in Ueda et al. (2007), and the proportion of such AGN circles depict high absorption objects (log NH > 22) and green squares represent objects with intermediate absorptions in the BAT catalog has been discussed in Winter et al. between these two limits. (2008) and W09; the latter find that ≈ 24% of the 9- month BAT AGN are hidden. The criteria employed to define such hidden AGN are that the model is com- objects with pegged extremal (1.5 or 2.2) spectral indices plex (best fit by a partial-covering model) with a cov- (20/60, 33%) compared to low-absorption log(NH) < 22 ering fraction f ≥ 0.97 and a ratio of soft (0.5–2 keV) objects (7/40, 18%), indicating that the spectral com- to hard (2–10 keV) flux ≤ 0.04. A partial covering frac- plexity due to higher absorption ultimately needs a more tion greater than 0.97 implies a scattering fraction below sophisticated modelling approach. 3% and is suggestive of a geometrically thick torus or an We present histograms of the photon index in Figs. 16, emaciated scattering region. We identify a total of 13-14 17 and 18. Low-absorption sources appear to peak at Γ ≈ hidden/buried sources in our sample using these criteria. (lowabs) Of these, two were previously identified as hidden sources 1.8 (hΓlowabsi = 1.81, σΓ = 0.21), whereas high- absorption sources show a wider spread in Γ and often hit in W09 (CGCG 041-020 and SWIFT J1309.2+1139 - (highabs) NGC 4992), one narrowly missed identification as a hid- the hard limits imposed (hΓ i = 1.84, σ = highabs Γ den source in W09 (Mrk 417, using the same data), three 0.26). Complex sources appear to have a wide range of (complex) were analysed in W09 but did not pass the criteria for photon indices (hΓcomplexi =1.87, σΓ =0.25), but being deemed hidden (NGC 4138, NGC 4388 and NGC hit the Γ = 2.2 boundary more often than simple model 4395), and the remainder are newly identifed hidden ob- (simple) sources (hΓsimplei = 1.79, σΓ = 0.22) which, as ex- jects (KUG 1208+386, Mrk 198, NGC 5899, NGC 4258, pected, show a peak at around Γ ≈ 1.8 (since they over- NGC 4686, Mrk 268, NGC 4939 and MCG +05-28-032). whelmingly overlap with low-absorption sources). Due For NGC 4138, we again use the same data as analyzed to the large error bars on some values of photon indices, in W09, but the best-fit adopted by W09 is taken from we present histograms of the values of Γ with absolute the Cappi et al. (2006) study. Cappi et al. fit NGC 4138 errors less than 0.05 in Fig. 18, and find that these trends with a power-law plus soft-excess model, whereas our are borne out even when restricting ourselves to the well- systematic model comparisons reveal that a partial cov- determined values of photon index. High-absorption ob- ering model is clearly preferred over a soft-excess model jects with well-determined Γ values show a distribution for this particular observation. skewed toward slightly higher values of Γ (hΓhighabsi = We point out that in this work, a uniform model-fit- (highabs) comparison approach has been adopted for all objects, 1.90, σΓ =0.28) than found above, as do complex (complex) unlike W09, where some model fits were taken from the objects (hΓcomplexi =1.90, σΓ =0.28), but simple literature; for consistency, we perform all our analysis on spectrum or low-absorption objects with well-determined the results from our own fits. For NGC 4388, we have photon indices do not show significant differences in their used XMM-Newton data in our study, whereas ASCA distributions, when comparing with the whole sample of data were used previously in W09 where it narrowly es- simple/low-absorption objects. The uncertainties in the caped classification as ‘hidden’; we prefer our analysis fits for the thirteen ambiguous spectral-type objects do of the better-quality XMM-Newton data for defining the not affect these statistics by more than 0.02 for any of hidden status of this object. The ‘hidden’ classification the quantities presented. of NGC 4395 can be called into question since it exhibits 15

 

 

 

1  1 

 

 

            ī ī

Fig. 16.— Histograms of 2–10 keV photon index, Γ. The grey shaded portions represent high-absorption objects (log NH > 22), whereas the blue shaded portions represent low-absorption objects (log NH < 22). In the left panel, in order to split our objects into the two absorption categories, any objects with ambiguous spectral classifications and two ‘best-fit’ NH estimates (both a simple and a complex spectral model describe the data well) have been assumed to have the smaller of the two NH estimates. In the right panel, the maximal NH is assumed for objects with ambiguous spectral classification.

 

 

 

1  1 

 

 

            ī ī

Fig. 17.— Histograms of 2–10 keV photon index, Γ. The grey shaded portions represent complex spectrum objects, whereas the blue shaded portions represent simple spectrum objects. In the left panel, in order to split our objects into the two spectral-type categories, any objects with ambiguous spectral classifications (both a simple and a complex spectral model describe the data well) have been assumed to be simple spectrum objects; in the right panel those same objects are assumed to possess complex spectra. rapid variability (e.g., Vaughan et al. 2005), which may well determined and has a large error due to poor quality argue for a reflection interpretation instead of the com- data, so it may not be hidden. plex absorption seen here; we present its reflection prop- Our finding of 13–14 hidden sources indicates a per- erties in §5.2 where we are only able to produce an upper centage of hidden sources of ≈ 14%, lower than the 24% limit on the reflection parameter (R< 0.71). Further de- fraction found in W09. However, if we assume Poissonian tailed study of this source is needed. MCG +05-28-032 errors on the counts used to calculate these percentages, has only XRT data and is of an intermediate model type; the proportions of hidden objects in the two samples are close inspection reveals that the covering fraction is not consistent to within 2σ. We emphasize that we have 16

6 6

681 (UURUVoRQoJDPPDo o414 681 (UURUVoRQoJDPPDo o414

64 64

1 1 1 1

 

81 81

4 4 61 61 61 8 818 61 61 61 8 818 ī ī

Fig. 18.— Histograms of 2–10 keV photon index, Γ, with objects with large errors on Γ removed. The grey shaded portions represent complex spectrum objects, whereas the blue shaded portions represent simple spectrum objects. In the left panel, the grey shading represents high absorption objects and blue shading represents low absorption (dividing line between these 23 −2 designations is NH = 10 cm as before). The right panel shows the same values split on model type, such that grey represents complex spectrum objects and blue represents simple spectrum objects. For objects with dual spectral classifications, we assume simple spectral types and the smaller of the two NH estimates.

 

 

1  1 

 

             

ORJ-/4NH9k ORJ-/4NH9k

Fig. 19.— Histograms of intrinsic 2–10 keV luminosity log(L2−10keV ). The grey shaded portions represent complex spectrum objects, whereas the blue shaded portions represent simple spectrum objects. In the left panel, any objects with ambiguous spectral classifications (both a simple and a complex spectral model describe the data well) have been assumed to be simple spectrum objects; in the right panel those same objects are assumed to possess complex spectra. adopted a uniform strategy for fitting models to all of age soft-to-hard flux ratio is F0.5−2keV/F2−10keV = 0.02 our data, whereas in W09 many fits were gathered from with σ =0.01, consistent with the W09 results for hidden the literature. This difference may partly explain the objects. differing proportion of hidden sources identified. The identification of seven new hidden objects (three with 4.7. Detailed Features newly obtained XMM-Newton data) is nevertheless inter- We present the fraction of objects for which soft ex- esting. The average absorption for these hidden sources is log(N ) = 23.42 with σ = 0.44, and their aver- cesses, iron lines, and warm absorber edges are well de- H logNH tected, for the subset of objects with at least 4600 counts 17

 

 

1  1 

 

             

ORJ-/4NH9k ORJ-/4NH9k

Fig. 20.— Histograms of intrinsic 2–10 keV luminosity log(L2−10keV ). The grey shaded portions represent high-absorption objects (log NH > 22), whereas the blue shaded portions represent low-absorption objects (log NH < 22). In the left panel, any objects with ambiguous spectral classifications and two ‘best-fit’ NH estimates (both a simple and a complex spectral model describe the data well) have been assumed to have the smaller of the two NH estimates. In the right panel, the maximal NH is assumed for objects with ambiguous spectral classification. in the observation (39 objects). The data sets are all from sented; the alternative Buckley-James algorithm yields XMM-Newton, and the properties of these features are very similar results with slightly steeper slopes. How- detailed in Table 4 below. We are careful to calculate up- ever, the definition of the equivalent width requires a per limiting parameters for these components in all cases good knowledge of the intrinsic continuum over which where fitting does not reveal a significant improvement the line is detected; significant or complex absorption to the fit; this approach allows a more complete analysis in many sources may make it difficult to recover the of properties later. ‘true’ iron line equivalent width in those cases. There- fore, we also check the presence of an anti-correlation for 4.7.1. Iron K-α lines the 24 sources with log(NH) < 22. We find a slightly steeper relation, log(EW/keV) = (5.41 ± 3.14) − (0.14 ± We find that 79% of the objects with > 4600 counts exhibit iron lines, which is very close to the 81% found 0.07) × log(L2−10keV) with a Spearman’s Rank coeffi- cient of −0.356 (Spearman probability 0.088). These re- in W09. For the remaining sources, we include a zgauss sults are consistent with the Page et al. (2004) finding model fixed at an energy of 6.4 keV with a width of σ = 0.1 keV and fit the normalization. We obtain and later works by Jiang et al. (2006) and Bianchi et al. (2007) for radio-quiet AGN. the upper limiting equivalent width by setting the nor- We inspect the residuals of all of the XMM-Newton malization to its upper limiting value as determined in xspec, and determining the equivalent width using the fits for hints of broad iron lines that can indicate the presence of strong-gravitational processes at work in the eqwidth command. inner part of the accretion flow near the black hole. A From Fig. 21 (left panel), we see that there are sug- gestions of an anti-correlation between the iron line systematic analysis of such lines in our sample is not presented here, but we identify four sources with XMM- equivalent width and 2–10 keV absorption-corrected Newton data that display hints of complex iron lines by luminosity, known as the ‘X-ray Baldwin Effect’ (Iwasawa & Taniguchi 1993). However, the presence of inspection of residuals, but have too few counts (< 4600) to fit an iron line successfully (Mrk 417, UGC 06527, many upper limiting equivalent widths complicates this NGC 4686 and SWIFT J1309.2+1139), two sources that picture. Since the upper limiting equivalent widths oc- cupy the same range as those for well-detected iron lines, definitely exhibit structure in the lines beyond what is shown here using our simple zgauss fits (Mrk 766 and we use the asurv (Astronomy Survival Analysis) pack- NGC 4051), and five sources that show possible signs of age for censored data (from the STATCODES suite of 15 broad lines by inspection of residuals (NGC 5252, NGC utilities; Feigelson & Nelson 1985 ) to determine the 5273, UM 614, NGC 4151, NGC 4579). Many of the more correlation parameters. We obtain a very shallow anti- well-known sources have been analysed in more detail correlation of log(EW/keV) = (3.678 ± 2.406) − (0.104 ± (e.g., Nandra et al. 2007, Cappi et al. 2006), and in other 0.056) × log(L2−10keV) with a Spearman’s Rank coeffi- works on individual sources. cient of −0.275 (Spearman probability 0.078). We use the E-M algorithm whenever results from asurv are pre- 4.7.2. Soft excesses 15 http://astrostatistics.psu.edu/statcodes/sc censor.html 18

L H

glL glH H . (: /31k o NH9 o /31k (: . H NH9 o /31k (: . H ) )

glgL glgH ig iL i2 i- ii in H2 H8 ig iH ii in it ic LQW 1L LQW 1i ORJ//21Lg NH9 oHUJ V k ORJ/1+ oFP k

Fig. 21.— Left panel: iron line equivalent width from the PN instrument (for objects with > 4600 counts in all detectors) vs. L2−10keV (absorption-corrected). Downward pointing arrows with horizontal lines show upper limiting iron line equivalent widths wherever a source did not have a statistically significant iron line (using the zgauss model). Triangles (red) show log(NH) < 22 sources, circles (blue) show log(NH) > 23 sources, and squares (green) show the remaining sources with intermediate absorbing columns. Right panel: iron line equivalent width (PN) against absorbing column density; the symbols and color-coding are the 19 −2 19 −2 same as for the left panel. Absorbing columns below 10 cm are shown at NH = 10 cm . In both figures, dashed lines or ellipses connect two different measurements for an individual source when a unique best-fit model was not found. The black thin solid line shows a best-fit (including upper limits using asurv) using all objects; the red dashed line shows the fit to only unabsorbed (log NH < 22) objects. The grey dotted line shows the anti-correlation found in W09, using binned data.

Many X-ray spectra of low-absorption AGN reveal an as edges and iron lines. This approach is an extension of excess at energies below ∼1 keV. A number of possible the concept presented in W09 where the power present in origins have been suggested for this soft excess: blurred the soft excess was compared to that seen in the power- reflection (e.g., Ross & Fabian 2005), complex absorp- law component, but here we adopt the fractional measure tion (e.g., Sobolewska & Done 2007) or an extension of since in some reprocessing models, we expect some frac- thermal emission from the accretion disc, possibly more tion of the coronal power-law emission to be responsible visible in low black hole mass AGN (e.g. Narrow Line for the soft excess. We also extend the previous analy- Seyfert 1 AGN). A definitive physical picture has not sis by producing upper-limiting soft-excess strengths for yet emerged to explain the soft excess in all AGN; we those objects where soft excesses are not detected, and therefore employ a simple redshifted black body compo- show the soft-excess strength against power-law lumi- nent (e.g., Crummy et al. 2006) as a phenomenological nosity in Fig. 22. The upper limits are calculated by description of the feature, and we do not attempt to fit including a black body component with a temperature the more complex models described above to the soft ex- fixed at the canonical soft-excess temperature 0.1 keV cess. An inventory of the observed properties of the soft (e.g., as found in the sample of Crummy et al. 2006 or excesses in our sample will allow future investigations Gierli´nski & Done 2004), finding the upper limiting nor- into their physical origins. We find that 31–33% of the malization, thereby determining the upper limiting black objects with > 4600 counts in their spectra have soft ex- body luminosity. Amongst the objects with detected soft cesses (with the range due to ambiguous spectral types excesses, we see a small decrease in soft-excess strength for a few sources), compared to 41% in W09 (consis- with higher power-law luminosities (in constrast to the tent within Poisson errors). All of the objects for which simple proportionality seen in W09 between soft excess soft excesses are detected have intrinsic absorbing col- luminosity and power-law luminosity), but there are too umn densities log(NH) < 21.22, so we restrict ourselves few points at low luminosities to be able to constrain explicitly to soft excesses seen above ‘unabsorbed, simple the slope of any such trend. This result may suggest power-law’ type spectra as mentioned in §3.2. that a straightforward reprocessing scenario, where the We define the ‘soft-excess strength’ Ssoftex as the ra- power in the power-law component (due to the corona) is tio of the luminosity in the black body component only somehow recycled in the soft excess, is not favored. How- (setting the power-law normalization to zero, integrating ever, this hypothesis would be better tested on a larger the luminosity from 0.4 to 3 keV) to the luminosity in sample of objects. the power-law component only (setting the black body The remainder of the objects (for which upper lim- normalization to zero), measured between 1.5 and 6 keV its have been calculated) occupy a completely different (PL) part of parameter space than those with detected soft (Ssoftex = LBB/L1.5−6keV ); these energy ranges were se- lected so we could be confident of avoiding features such excesses. The objects with upper-limiting soft-excess 19

)P )P n n 3/ ) 3/ ) // // ,, ll

PB) PB) IWLH[FHVVfVWUHQJWKfr/ R PBP) PBP) 6 6RIWLH[FHVVfVWUHQJWKfr/

)Pí0 )Pí0 )P−P )P−) )P−1 )P−0 )P−− )P−3 )B− )B. )Bd 1 1B1 L) 3RZHUfODZfOXPLQfr/3/=/)B3L.fNH9(fHUJfV n 3KRWRQf,QGH[fīfIURPfEHVWLILWWLQJf;LUD\fSRZHULODZ

Fig. 22.— Soft-excess strength against power-law luminos- Fig. 23.— Soft-excess strength against photon index Γ. The ity (L1.5−6keV). The soft excess is modelled as a black body, soft-excess strength is parameterized as described in Fig. 22. and its strength is parameterized as the ratio of the luminos- Downward pointing arrows show upper limiting soft-excess ity in the black body component only to the luminosity in strengths wherever a source did not have a statistically sig- the power law between 1.5 and 6 keV (determined from the nificant soft excess (using the zbbody model). PN instrument). Downward pointing arrows show upper lim- iting soft-excess strengths wherever a source did not have a statistically significant soft excess (using the zbbody model). Gierli´nski & Done 2004), we find a narrow distribution of temperatures (hkTsoftexi = 0.113, σkT = 0.032) and can identify no discernible correlation with either soft-excess strengths span a range of luminosities and lie well be- strength or photon index. This result lends credence to the idea that the soft excess is not part of the direct, low the anti-correlation between Ssoftex and L1.5−6 keV for the objects with detected soft excesses. This dis- thermal accretion disk emission since the temperature at joint distribution suggests that there are two classes of which it is seen is uniformly and tightly clustered around objects: those which show measurable, statistically sig- ∼ 0.1keV. nificant soft excesses, in which there is a weak anti- correlation between soft-excess strength and luminosity, 4.7.3. Ionized absorber edges and those without any detectable soft excess, in which We present the edge depths due to ionized absorbers the stringent upper limits on the soft-excess strength dic- in Table 4. Winter et al. (2011) noted that X-ray obser- tate that there can be little scope for any kind of correla- vations do not offer the same sensitivity for picking up tion or anticorrelation in those sources. This dichotomy warm absorbers that, for example, UV spectroscopy does suggests that the physical process responsible for the soft (such as COS spectra used in Winter et al. 2011). How- excess occurs in some AGN only, and that a soft excess ever, the detection rates and properties here can at least is not an intrinsic part of all AGN spectra. We return serve as an indicator of what is detected in X-rays using to this issue in a companion paper on the relation be- an unbiased sample and can be compared with UV stud- tween reflection/absorption and soft-excess strength in ies. We find that 18% of our high-counts subsample show our sample (see §5.2). warm absorber signatures in the form of an OVII edge at We also want to ensure that the stronger soft excesses 0.73 keV. Only 8% exhibit a significant additional warm are not biased to being found in sources with harder spec- absorber edge at 0.87 keV. However, this is measured as tra (Γ < 2.0): a harder spectrum leaves more scope for a fraction of all sources with > 4600 counts; if we in- soft features to be seen as an excess. We see from Fig. 23 stead only consider the 21–23 unabsorbed (logNH < 22) that this is not the case; for the two objects with the sources within this high-counts subset, we find that the largest soft-excess strengths (NGC 4051 and Mrk 766, fraction of such sources with at least one well-detected which also happen to show pronounced spectral variabil- ionized absorber edge rises to ∼ 32% (two sources with ity) we do indeed see that they have hard (Γ = 1.5) spec- ambiguous NH values introduce an uncertainty of ±1% tra, but for all other objects, the opposite trend seems to to this figure). The study of Seyfert 1-1.5 BAT-selected be evident, i.e., the strength of the soft excess increases AGN presented in Winter et al. (2012) reveals that 53% with Γ (as the spectrum gets softer), in line with the of their 48 sources exhibit a detectable ionized absorber trend previously found in Brandt et al. (1997). edge using XMM-Newton and Suzaku X-ray data; the We lastly investigate whether there is any relation pioneering work of Reynolds (1997) finds similarly that between the temperature of the soft excess and the 50% of their sources exhibit such edges in ASCA spec- soft-excess strength or Γ. In line with what has tra, and the work of Crenshaw et al. (1999) using UV been found previously (e.g., W09, Crummy et al. 2006, spectroscopy from HST reveals that 47% of their sources 20

(

) n 3/ / / %%

up(

PB) 6RIWLH[FHVVfVWUHQJWKfr/ fH[FHVVfEODFNfERG\fWHPSHUDWXUHflNH9a

6RIW upu(

PBP) PBP) PB)) (p0 (p1 (ph V VpV Vp0 6RIWfH[FHVVfEODFNfERG\fWHPSHUDWXUHfrNH9n 3KRWRQf,QGH[fī

Fig. 24.— Left: soft-excess strength against soft excess temperature (using the black body model as a parameterization for the soft excess). The soft excess temperatures are clustered at ∼ 0.1 keV, as found in previous works. Right: soft excess black body temperature against photon index Γ. show evidence for such absorbers. The latter two studies of particular note is Mrk 50, a very bright unabsorbed are not, however, from an unbiased sample, so we empha- Seyfert 1 galaxy with a soft excess but no measurable size the utility of Winter et al. (2012) and this work on Iron line; this object would benefit from further detailed BAT-selected AGN. Considering the small sample sizes study. involved in all the above studies, our finding that ∼ 32% of unabsorbed sources have measurable ionized absorp- 5. INCLUDING THE BAT DATA tion is broadly consistent with previous findings. The sources that do exhibit well-detected warm ab- 5.1. Compton-thick sources sorbers are clustered around luminosities of L2−10keV = Using our simple ztbabs model to model absorption, 43.8 −1 10 ergs , with only two sources having luminosites we identify eight Compton-thick sources (NH > 1.4 × 44 −1 above 10 ergs . The upper limiting optical depths of 1024cm−2): NGC 4102, 2MASX J10523297+1036205, the OVII edge for the remaining sources show increas- 2MASX J11491868-0416512, B2 1204+34, MCG -01-30- ingly stringent upper limits that indicate less scope for 041, MRK 1310, PG 1138+222 and UGC 05881; NGC ionized absorption at higher luminosities. This is con- 4941 and NGC 5106 may be Compton thick within the sistent with a general picture where absorption of any errors on their fitted column densities. This constitutes kind (neutral or ionized) is less prevalent at high intrin- ∼ 9% of our sample; this is a notable change from W09 sic luminosities (e.g., Winter et al. 2012 and §4.3 of this (where no Compton thick sources are detected by this work). definition) and Burlon et al. (2011) where 4.6% of their sample (the 36-month BAT catalog AGN) are Comp- 4.8. New XMM-Newton observations ton thick. However, we have not taken into account the Our sample of 100 objects in this study includes 13 effect of Compton scattering for our heavily absorbed objects with new XMM-Newton observations, gathered sources; more sophisticated models such as plcabs or specifically for improving the coverage of the BAT cat- MyTorus (Murphy & Yaqoob 2009) or that presented alog at b > 50◦. We highlight some of the interesting by Brightman & Nandra (2011) are required to model features of these 13 objects here. Two of these XMM- these effects. Burlon et al. (2011) examine the effect of Newton datasets have been studied in more detail in using MyTorus for a fraction of their sources and also other works already: Mrk 817 (Winter et al. 2011, a use the built-in xspec model cabs to take Compton scat- multi-wavelength study including UV spectroscopy from tering into account. MyTorus and similar models are the Cosmic Origins Spectrograph - COS along with sufficiently complex to warrant a separate study since HST and IUE archival data, looking for outflows and they have many more parameters than simple absorp- broad-band variability in this source) and NGC 3758 tion models, and determining these parameters individ- (also known as Mrk 739, Koss et al. 2011b, identifying a ually for each object is beyond the scope of this study. faint counterpart Mrk 739W using high spatial resolution However, a number of other measures have also been Chandra data thereby identifying a dual AGN in this sys- used to identify Compton-thick sources, such as a high tem). The remaining new XMM-Newton datasets reveal Fe-Kα line equivalent width, a flat photon index (here diverse properties for these 13 objects (including three we take this to mean Γ pegs at the minimal value of 1.5) hidden AGN, KUG 1208+386, Mrk 198, NGC 5899 and or an unusually high reflection fraction (R> 1) (see W09 one Compton-thick candidate NGC 4102). One object and Table 5 for reflection values). Using these alternative 21 metrics, W09 note that the proportion of Compton-thick tenuation’ factors for the BAT flux for different column AGN in the 9-month catalog could increase to 6%, al- densities (ratios for the observed-to-intrinsic BAT fluxes though they do not use BAT data to constrain absorption calculated using the MyTorus model). We attempt to in the highly absorbed sources. We employ these metrics recover the observed absorption distribution for different with our sample and include our full 0.4–200 keV band flux limits corresponding to the 9-month, 58-month, and fits and find that a few more sources may be Compton- future, deeper surveys. However, we discover that the thick: Mrk 766, NGC 4051, UM 614, Mrk 744, NGC precise degree of improvement expected with deepening 3227, CGCG 041-020. We caution that we can only exposure is difficult to predict and relies on a realistic use the Fe-Kα equivalent width metric on the objects source input BAT flux distribution. We defer publishing with XMM-Newton data since XRT data are not of suffi- of this simulation to a future study. cient quality to analyse these line properties, and in this study we restrict ourselves to reflection fits to objects 5.2. Reflection with XMM-Newton data. The object Mrk 766 exhibits The hard X-ray BAT data provide an opportunity to all three of these alternative signatures despite having a constrain the reflection properties of the sample, since low column density (log NH < 21). Mrk 766 is known the Compton reflection hump peaks in the BAT band. to be highly variable and has been extensively studied We present the reflection properties for the subset of ob- in the literature (e.g., Turner et al. 2007, where the pro- jects with XMM-Newton data. We fit a pexrav model to nounced variation in spectral shape between epochs is the combined XMM-Newton and BAT data, again renor- shown); this conclusion also holds for NGC 4051 which malizing the BAT spectrum wherever possible (and link- exhibits both a broad Iron line and a high reflection ing BAT and PN normalizations). In our pexrav fits, we fraction, alongside well-studied variability favouring the allow the reflection fraction R, the photon index Γ, the reflection scenario (Ponti et al. 2006). There are four folding energy Efold and the normalization to vary, fix sources with both high reflection and Γ = 1.5 (UM 614, the redshift of the source and freeze all other parameters Mrk 744, NGC 3227 and CGCG 041-020), the last three (abundance of elements heavier than Helium relative to of which have absorptions at or above log(NH)=23; they solar abundances, iron abundance and cosine of the in- may also be good Compton-thick candidates. The ob- clination angle) at their default values. We ignore any ject B2 1204+34, which only has XRT data, also exhibits data below 1.5 keV to avoid soft excesses or edges and Γ=1.5 reinforcing its status as Compton-thick. Based ignore data between 5.5 and 7.5 keV to avoid the iron on these considerations, we may add 2–6 sources to the line and edge which are not modelled by pexrav. We existing list of ∼ 9 with measurably Compton-thick col- include either a simple absorption component or a par- umn densities, to yield a Compton-thick fraction of 11- tial covering absorber, depending on whether the basic 15%. However, what is ultimately required is the use of fit from Table 3 indicates that the 0.4–10 keV spectral models that fully include Compton scattering and all the shape is simple or complex, respectively. For complex possible types of reflection, coupled with an understand- objects, we seed the absorbing column density with the ing of the geometry in each source, to determine the true value of NH obtained from our 0.4–10 keV fits. We do Compton-thick fraction. not impose any restriction on Γ as done before in the In Fig. 25, we show how the log(NH) distributions vary power-law fits, since we wish to use Γ here purely to con- between the 9-month (W09) and 58-month (b > 50) strain the spectral shape, and can then more easily probe catalog subsamples. We present these distributions as any correlations between the different reflection parame- a fraction of the total number of objects in each sam- ters. For objects where the error calculation on R, Efold ple for easy comparison. As before, we allow for the or Γ fails, we perform a detailed contour-plot using the uncertainty in NH for some sources by assuming any steppar command in xspec to better constrain the pa- sources with ambiguous spectral types have the ‘sim- rameters. We present the results in Table 5 and show ple’ absorption fit in the left panel and the ‘complex’ plots of the three key variables in the reflection scenario, (partial covering) absorption fit in the right panel. The R, Γ and Efold in Figs. 26, 27 and 28. distributions for the earlier catalog and the 58-month We find that reflection can be constrained in a large catalog are consistent within errors (calculated accord- fraction of our sources, and that for the majority of the ing to the Gehrels 1986 Poisson approximation) for all sample, some level of reflection is required (R> 0). The columns up to log(NH) ≈ 23.6; the bin centered on average value of the reflection parameter for the sample log(NH) ≈ 24 shows a twofold increase in objects in is hRi = 2.7 ± 0.75, indicating that reflection is impor- the 58-month catalog, and a discrepancy remains even tant in this unbiased sample of AGN, in both absorbed when errors are taken into account. Our 58-month cat- and unabsorbed objects. This large average value indi- alog does show some objects at even higher columns, cates that strong reflection involving light bending may but the numbers are small. In general, these results be common (R > 1) or that highly complex absorp- show how the increased sensitivity of the 58-month cat- tion could be important. In our dataset, there are no alog allows detection of intrinsically bright objects with ‘strong’ reflection (R> 4) sources with sufficient signal- log(NH) & 23.6. As discussed in Burlon et al. (2011), the to-noise ratio (when considering the combined XMM- increased proportion of highly absorbed sources as the Newton and BAT data) to distinguish these two pos- BAT catalog deepens in exposure suggests that the BAT sibilities. However, the soft excess has been linked to hard X-ray survey is still missing a significant fraction reflection (e.g., Ross & Fabian 2005), and in a compan- of log(NH) > 24 sources. We constructed simple simu- ion study we aim to explore how a comparison between lations to estimate the fraction of such missed sources the soft-excess strength and the reflection parameter may at higher columns, by assuming a given underlying (in- allow determination of which of the two scenarios, reflec- trinsic) BAT flux and NH distribution and using the ‘at- tion or absorption, is favored in individual objects. In- 22

W9r W9r

cupPRQWK-19-*DO9-&DS cupPRQWK-19-*DO9-&DS 1pPRQWK-.:W1G 1pPRQWK-.:W1G

W9F W9F

W9) W9) )UDFWLRQ-RI-SRSXODWLRQ )UDFWLRQ-RI-SRSXODWLRQ W90 W90

W W )W )0 )) )F )r )c )W )0 )) )F )r )c

ORJ.1+G ORJ.1+G

Fig. 25.— Comparison of log(NH) distributions between the 58-month catalog (red diagonal cross-hatching shading, solid error bars) and the 9-month catalog (solid green shading, dotted error bars). The left panel shows the 58-month column density distribution assuming that any ambiguous spectral-type objects have the ‘simpler’ model, and the right shows the results assuming that such objects are better represented by their ‘complex’ (partial-covering absorption) model fit. spection of Fig. 26 reveals that absorbed objects (blue shape (via the photon index Γ). filled circles) exhibit similar ranges in reflection param- We also briefly contrast our results with those eter as less absorbed sources. The average fold energy of Winter et al. (2012), who fit a reflection model is well above the maximum energy of BAT, indicating to a sample of Seyfert 1-1.5 AGN using a slightly that we cannot reliably detect high-energy cut-offs us- different approach (fitting a model combination ing BAT data alone. There is significant uncertainty tbabs(ztbabs(cutoffpl + zbbody + zgauss + in determining Efold, since when the fit yields a value pexrav)) in xspec) to simultaneously model the soft outside the BAT band, we typically see very large error excess, iron line, power-law, absorption and reflection, bars on the fit value. In Table 5, we have shown objects with edges added where needed). For the 8 objects with Efold > 5000 keV (arbitrarily chosen) as lower lim- that are common between our sample and theirs, they its, using their negative error bar to determine the lower employ Suzaku data alongside BAT data and fit them limit in Efold. For some objects with fold energies below with the above model combination. All of the common 5000 keV, the positive error bar on Efold is difficult to objects were therefore observed at different epochs in constrain due to the poor sampling in the BAT band. Winter et al. (2012) than presented here. We note that For those objects, we do not plot a positive error bar some objects exhibit very similar reflection fractions in but note that the fold energy is difficult to determine in the two studies (e.g., NGC 4151, NGC 4593, NGC 5548, those cases. The average photon index from pexrav for Mrk 841) but some objects display large discrepancies the whole sample is 1.80 with a standard deviation of (e.g., NGC 4051, Mrk 766). This could be in part be 0.32; the 1σ range in photon indices falls within within attributed to re-normalization employed here providing the limits we previously imposed when performing our new constraints on the fit, but also due to intrinsic 0.4–10 keV fits. There is no evidence of any correla- changes in the spectrum between observations. NGC tions between any of the three parameters R, Efold and 4051 and Mrk 766 in particular are known for their Γ (see Figs. 27, 28). Given the limited signal-to-noise strong variability in both spectral shape and intensity ratio and bandpass of the BAT data, one finds that one (see §5.1 above), which is capable of producing such cannot separately constrain the three components of the pronounced spectral changes. pexrav model using BAT data alone, and that degen- Another interesting comparison can be made with the eracies exist between these three parameters. However, Ricci et al. (2011) analysis of INTEGRAL AGN detected inclusion of the 0.4–10 keV data constrains the fit better, in the 15–1000 keV band by the observatory’s soft γ- and the BAT re-normalization removes one free param- ray imager instrument. Their work differs from ours in eter and provides better constraints on fit parameters in that they use only hard X-ray spectra (> 10 keV) to many cases. Overall, our reflection fits show that re- calculate reflection parameters, whereas we restrict our- flection is present in some form in the majority of our selves to the subset of objects with XMM-Newton data sample. The fold energy is generally outside the BAT below 10 keV to provide an additional constraint on re- range and we only observe a few definitive high-energy flection parameters. Additionally,the BAT survey is flux- cut-offs in our BAT data. Our determinations of R and limited and taken across the whole sky, unlike the INTE- Efold are given more credence by using the 0.4–10 keV GRAL survey which consists of pointed observations of to help lock down the overall broad-band X-ray spectral sources of interest. The 165 Seyfert galaxies in the INTE- 23

GRAL sample show a very similar average photon index of Γ ≈ 1.8 and an unobservable average hard X-ray cut- E... off (EC & 200 keV). They also show a range of reflection values, and in particular find that moderately-obscured Seyfert 2 nuclei (with 23 < log NH < 24) have a higher E.. reflection component than other classes of AGN, with +4.5 hRi =2.2−1.1 for this class of sources. We do not notice E. any such trend in our sample and find a broad distribu- tion of reflection amplitudes at every log NH level, but our XMM-Newton+BAT sample is significantly smaller E than theirs (49 AGN vs. 165 AGN) and we therefore may

not have sufficient statistics to make a detailed compari- OHFWLRQfSDUDPWHUf5 I son. We defer more detailed discussion of this issue to a H .−E companion paper on the stacked spectrum of the North- 5 ern Galactic Cap BAT AGN (Vasudevan et al. 2013 in prep). .−.E Both our results and those of Ricci et al. (2011) may be subject to selection bias when using hard X-ray selec- E.í/ tion, since the effect of reflection is to boost the flux in the E. E.. E... E.k

BAT band. Using a simple pexrav model in xspec, we (IROGf3fNH9 determine the BAT flux boost factor for different reflec- 6 tion strengths, assuming Γ = 1.8 and Efold = 10 keV. If we divide the observed BAT fluxes of these sources by Fig. 26.— Reflection parameter R against folding energy the boost factors appropriate for each source to predict Efold from our pexrav fits. Red triangles and small up- per/lower limit arrows depict low absorption (log NH < 22) what flux they would have without reflection, we find objects, blue circles and large upper/lower limit arrows depict that 14% of the XMM-Newton-BAT sample would drop high absorption objects (log NH > 22) and green squares and below the flux limit of this survey (not including sources intermediate-sized upper/lower-limit arrows represent objects with upper-limiting values of R), and all of these sources with intermediate absorptions between these two limits. The have R> 1. This is an important consideration when at- dark red square shows the mean R and Efold for the whole tempting to determine average reflection properties using sample; the ‘formal’ mean Efold lies well outside the BAT hard X-ray selected surveys. band. We plot any objects with R < 0.01 as an upper limit at R = 0.01, and for objects Efold above 5000 keV, we treat these fold energies as unobservable and compute a lower limit 6. AVERAGE SPECTRUM AND LOG(N)-LOG(S) on Efold using the negative error; these limits are indicated 6.1. Stacked Spectrum by the gray shaded areas. We present the stacked/summed spectrum for all sources (excluding the five sources with insufficient counts to construct their spectra) in Fig. 29, for com- finding from W09 that the slope of the XRB below 15 keV parison with the corresponding average spectrum gen- found from HEAO-1 (Γ ≈ 1.4) can be reproduced by the erated from the 9-month catalog in W09. While W09 summed emission from BAT AGN. We confirm this mea- compiled their fits from a variety of different sources in surement here with the deeper 58-month catalog, with a the literature in addition to their own fits, here we have greater fraction of obscured AGN (compared to the 9- month catalog), but using a sample that covers 11% of performed a uniform analysis of all of the X-ray spectra ◦ for the objects in our sample. We are therefore able to the sky instead of 74% of the sky ( |b| > 15 ) from W09. simply sum the model fits to all of the spectra to ob- In a companion paper, we extend this stacking analysis tain the stacked spectrum, and avoid the need to employ to include the BAT data above 10 keV and comment on the complex method outlined in W09. We also distin- the relevance of our results to X-ray background studies guish, as done in W09, between the contributions due (Vasudevan et al. 2013 in prep). to simple power-law sources and complex (those with a Although we do not have data of sufficient quality to partial-covering spectral shape) sources. Due to the pres- detect soft excesses and iron lines in our XRT or ASCA ence of 13 intermediate-spectral type objects, we produce sources (∼ 50% of the sample), we find that based solely two versions of the summed spectrum: one assuming all stacking the objects with XMM-Newton detections, we of the intermediate objects are best fit by the simple see both a clear soft excess and an iron line in the power-law, and one where all such objects are assumed summed spectrum for the simple power-law sources (thin to have a complex spectrum. The relative importance of solid line in both panels of Fig. 29). The iron line is also the two source types is clearly seen in Fig. 29, but the notably prominent in the complex spectrum sources (thin two versions of the summed spectrum at energies above dashed line). It would be highly desirable to complete the 0.2 keV show no appreciable differences. Although AGN XMM-Newton-quality coverage for the entire sample to emission is thought to be responsible for the X-ray back- acquire better statistics on the prominence of this fea- ground (XRB), the XRB is only well-known for energies ture. This discussion highlights the utility of completing E > 0.6 keV (McCammon et al. 2002, Markevitch et al. the XMM-Newton coverage of this region of the sky, or 2003, Soltan 2007, Gupta & Galeazzi 2009). If we fit a obtaining other high-quality data to understand the de- power-law to the 1–10 keV region, we find an average tailed features in AGN spectra in the Northern Galactic power-law slope of Γ = 1.37 − 1.38, in line with the key Cap. 24

Γ...

va5 Γ..

Γ. vaaa

Γ oroNH9 IROG (

.−Γ 5HIOHFWLRQfSDUDPWHUf5 vaa

.−.Γ

Γ.íx va .−) ΓΓ−)((−) x ax2 v vx2 ) )x2 . īfpSH[UDYa īo1SH[UDY0

Fig. 27.— Reflection parameter R against photon index Γ Fig. 28.— Folding energy Efold against photon index Γ from from our pexrav fits. Red triangles and small upper/lower our pexrav fits. Red triangles and small upper/lower limit limit arrows depict low absorption (log NH < 22) objects, arrows depict low absorption (log NH < 22) objects, blue cir- blue circles and large upper/lower limit arrows depict high cles and large upper/lower limit arrows depict high absorption absorption objects (log NH > 23) and green squares and objects (log NH > 23) and green squares and intermediate- intermediate-sized upper/lower-limit arrows represent objects sized upper/lower-limit arrows represent objects with inter- with intermediate absorptions between these two limits. The mediate absorptions between these two limits. The dark red dark red square shows the mean R and Γ for the whole sam- square shows the mean Efold and Γ for the whole sample. For ple.. We plot any objects with R < 0.01 as upper limits at objects for which the value of Efold is above 5000 keV, we R = 0.01 (emphasized by the gray shaded area). treat them as having an unobservable fold energy outside the BAT band and use the negative error bar to compute a lower limit on Efold. The 5000 keV limit is indicated using the gray 6.2. log(N)-log(S) diagram shaded area. In order to estimate the completeness of our sample in the 2–10 keV band, we present the 2–10 keV log(N)- log(S)≈−11.6. log(S) plot for the sample in Fig. 30. We adopt all of the We can also contrast the properties of the source in the same conventions as in W09; N is the number of sources ‘complete’ and ‘incomplete’ fractions of the sample from above a threshold 2–10 flux S. The slope of the log(N)- the earlier and later catalogs, in their column density dis- log(S) relation is expected to be −1.5 for a uniform dis- tribution and spectral shapes as done by W09. For clar- tribution of objects throughout a Euclidean volume (i.e., ity, we state that ‘complete’ refers to objects with fluxes for low redshifts) with a broad luminosity distribution, log(S)> −11.0 when discussing the 9-month catalog, but and we find our results consistent with this down to a refers to objects with log(S)> −11.6 when discussing the flux of log(S)≈−11.6 (by inspection; we do not attempt 58-month catalog. In order to compare absorption distri- to fit these points due to the complications involved in butions, we redefine any lower-limiting log(NH)=20 val- fitting points with correlated errors). This result con- ues from W09 to log(NH)=19 as done in W09. In the trasts with a completeness limit of log(S)≈ −11.0 from 9-month catalog, the objects above completeness show the 9-month catalog, indicating that we have a complete both low absorption (hlog NH i = 20.9 ± 0.2) and a wide census of objects up to fluxes ∼ 4 times fainter than the range of absorption (19.0 < log NH < 23.5), whereas 9-month catalog completeness limit. This log(S) com- objects above completeness in our latest sample exhibit pleteness limit is consistent with the sensitivity expected slightly higher absorption levels (hlog NH i = 21.4−21.6± from the all-sky survey. 0.2) and a greater range (19.0 < log NH < 24.5), includ- Our Northern Galactic Cap sample covers 11% of the ing four Compton-thick candidates. Below completeness, sky, in contrast to 74% coverage of the sky in the com- where we expect to be missing a substantial fraction of plete sample of W09. In order to more easily compare objects, we find that the incomplete 50% of the 9-month the results from the 9-month and 58-month catalogs, we sample has high absorption (hlog NH i = 22.5 ± 0.2) and therefore scale the results obtained in W09 down to the a wide range (19.0 < log NH < 24.1) including two expected numbers of objects for 11% of the sky, and almost-Compton-thick candidates, NGC 1365 and NGC overplot the scaled 9-month catalog log(N)-log(S) rela- 612, whereas our latest sample shows even higher absorp- tion using empty diamonds in Fig. 30, along with the tions below completeness (hlog NH i = 22.9 − 23.2 ± 0.3) best fit above log(S)≈ −11.0 from W09. We find that and a wider range (19.0 < log NH < 25.3) including the the objects above the W09 completeness limit constitute remaining four Compton-thick candidates in our sam- half of the sample in the 9-month catalog, whereas 65% ple. There is larger uncertainty in the average log(NH) of our objects lie above our latest completeness limit of for objects below completeness since there are more ob- 25

d d

VRXUFHVuZLWKuDPELJXRXVuVSHFWUDOuW\SHVuDVVXPHGuWRuEHutVLPSOHt VRXUFHVuZLWKuDPELJXRXVuVSHFWUDOuW\SHVuDVVXPHGuWRuEHutFRPSOH[t au au p' ídl) p' ídl) uNH9 uNH9 p' p' uV uV x p( í' p í'

í'l) í'l) uîuSKRWRQVuFP 9 H N

í( h íx u [a X

í(l) h)O íxl) ORJh)OX[auhNH9uîuSKRWRQVuFP ORJ

íF í( í' ídl) d dl) ' í' ídl) d dl) ' ORJh(QHUJ\yNH9a ORJh(QHUJ\yNH9a

Fig. 29.— Summed spectrum for the whole sample. Left panel: the summed spectrum assuming a simple spectral type for any intermediate spectral-type objects. Right panel: the summed spectrum assuming a complex spectral type for such intermediate spectral-type objects. We stack the model spectra for all objects between 0.4–10 keV. The thick black solid line shows the summed overall spectrum in both panels, and the thin solid black and thin dashed black lines show the contributions from simple and complex spectrum-type sources, respectively. The red straight line in both panels shows a fit to the 1–10 keV data (yielding a best-fit photon index of Γ ≈ 1.37 − 1.38, matching that from the X-ray background).

sample, we have a wider range of absorption levels than seen in W09 and a marked increase in the number of ) Compton-thick candidates (subject to further investiga- SrePRQWK uePRQWK-(VFDOHG-WR-ffh-RI-VN\s tion of their spectra with more physical models). The 6ORSH-efdS-(IRU-XQLIRUP-VDPSOHs broad-band spectral shape, as parameterized by the ratio VFDOHG-:%u-EHVW-ILW-IRU-ORJ(6sgeff y FBAT/F2−10keV, also shows a marked variation. In the 9-month catalog, W09 found indications that the missing sources in the sample were likely to be heavily absorbed k since this flux ratio changed from an average 3.0 ± 0.3 above the completeness flux limit to 16.9 ± 2.3 in the incomplete part of the sample. This effect is less pro- nounced in our latest sample: there is a change from ORJ-(1-g-6s f 5.3 ± 0.7 to 12.4 ± 1.9 for the average value of this flux ratio, indicating that our sample is more homogenous than the previous 9-month catalog. % Below completness, W09 estimate ∼ 3000 missing objects at log(S)≈ −12, which corresponds to ∼ 400 sources if scaled to the 11% of sky covered by our North- íf ern Galactic Cap sample. At the same flux level, we ífkdS ífk íffdS íff íf%dS íf% íudS íu are missing ∼ 300 sources. Based upon the discussion ORJ-6-(kef%-NH9s-(HUJ-Vef-FPeks of flux ratios above, it is likely that the missing sources are more heavily absorbed than those above the com- Fig. 30.— Plot of log(N >S) against log(S), for 2–10 keV pleteness limit, but since our sample is more homoge- −1 −2 flux S in erg s cm , where log(N>S) is the logarithm of nous than the 9-month catalog, it is difficult to predict the number of sources with flux greater than S. The thin line the properties of the missing sources. shows a slope of −1.5 expected for a uniform distribution. Our sample shows a slope consistent with this down to fluxes of log(S)=−11.6. 7. RESULTS FROM THE NORTHERN GALACTIC CAP SOURCES IN THE 22-MONTH BAT CATALOG jects with ambiguous spectral types in this regime. Since Throughout this paper, all of our results pertain to the objects below completeness are by definition fainter, we 58-month BAT catalog objects in the Northern Galactic expect them to have fewer counts in their observations, Cap. We choose this source list since the 58-month cat- and therefore require longer exposures before the spectral alog is the deepest edition of the catalog for which the shape can be properly constrained. source list and BAT spectra are publicly available. Us- In both the ‘complete’ and ‘incomplete’ regions of our ing the deepest version of the catalog provides the largest 26 possible sample, and therefore allows a more robust de- We present a detailed X-ray spectral analysis of the termination of the true column density distribution and non-beamed AGN in the Northern Galactic Cap of the luminosity distribution. Ideally this requires 0.4–10 keV 58-month BAT catalog, consisting of 100 AGN with data of sufficient quality, but high-quality XMM-Newton b > 50◦. This field has excellent potential for further data is only available for 49% of the objects in our 58- investigations due to a wide range of multi-wavelength month catalog subsample. However, the 13 new XMM- data that is already available, and we propose the field Newton observations in this sky region from the proposal as a low-redshift analog to the ‘deep field’ observations by PI Brandt were taken to complete XMM-Newton cov- at higher redshifts (e.g. CDFN/S, Lockman Hole). We erage for the 22-month catalog (Tueller et al. 2010) in present distributions of the redshift, luminosity, absorb- this sky region, and the observations that were success- ing column density and other key quantities for the cat- fully obtained provide XMM-Newton coverage for 90% of alog. We use a consistent approach to fit all data, using the b> 50◦ 22-month BAT catalog sources. We therefore our semi-automated X-ray spectral fitting workflow, use- present some key comparisons between results for the 9- ful for fitting suites of models to large samples of AGN, month catalog (W09’s uniform sample, 102 objects, 74% producing consistent comparisons between models and of the sky), the 22-month catalog (39 objects, 11% of determining the significance of various spectral compo- the sky), and the 58-month catalog results (100 objects, nents. In summary, we find that: 11% of the sky) in Tables 6 and 7 below, to illustrate the utility of the deepening sample, and to identify if any • We probe to deeper redshifts with this representa- insight can be gained using a complete subsample with tive subsample of 100 objects from the 58-month higher-quality data. We contrast the flux limits, com- catalog (hzi = 0.043 compared to 0.03 from W09 pleteness limits, percentage of sources with ambiguous and 0.03 from Burlon et al. 2011) spectral types, and percentages of objects with different levels/types of absorption in Table 6; we do the same for • The average X-ray luminosity found here the luminosity distributions, frequency of spectral fea- (hlog L2−10keV i = 43.0) is identical to that tures, and percentage of ‘hidden/buried’ sources in Ta- seen in W09 (hlog L2−10keV i = 43.0), but with ble 7. A few sources in the 22-month catalog list do not more pronounced tails in the distribution at low occur in the 58-month catalog (due to intrinsic variabil- and high luminosities. The average 14–195 keV ity of those sources); for our 22-month catalog analysis BAT luminosity is hlog LBAT i = 43.5, compared we only use those sources that survive in the 58-month to 43.7 from W09. catalog. • We uncover a broader absorbing column density The 22-month catalog flux limit is 1.8 times fainter distribution. The obscured fraction (log N ≥ 22) than the 9-month catalog flux limit, and the 58-month H is ∼ 60%, an increase from the measurements for catalog flux limit is 5 times fainter than that for the 9- the 36-month catalog (Burlon et al. 2011) and the month. While the 58-month catalog completeness limit 9-month catalog (W09). The obscured fraction (from the logN − log S relation) is estimated to be 4 broadly overlaps with complex spectrum (partial times fainter than the 9-month catalog, the 22-month covering-type) sources, which constitute 43-56% of catalog is only complete to fluxes 1.8 times fainter than the sample. the 9-month catalog. Despite the better XMM-Newton coverage for the 22-month catalog subsample, we find • Thirteen objects have ambiguous spectral types in that there is still a large degree of uncertainty in the that a unique best-fit model could not be found, percentages of sources at different absorption levels and and both simple or complex absorption can de- with different absorption types (simple/complex) due to scribe the observed spectral shape. These are typ- four ambiguous spectral-type sources in the 22-month ically objects where the existing (generally XRT) catalog list. The fraction of complex absorption sources data does not have sufficient counts below 1 keV to appears to change little between the catalogs, within the distinguish between the two models. For these ob- uncertainties. However, there does seem to be a trend jects, we have presented the results for both models towards uncovering a higher fraction of absorbed sources, in all subsequent analyses, and take into account in particular Compton-thick objects, as the catalog gets the resulting uncertainty in model fit parameters. deeper. The most significant uncertainty stemming from The luminosity distributions in all three catalogs show these ambiguous sources is the absorbing column a trend for unabsorbed sources to have marginally density: two different model fits sometimes pro- brighter luminosities, and percentages of sources with duce extremely different log(NH) values. This be- iron K-α lines or soft excess appears relatively stable be- haviour highlights the need to obtain better quality tween the different catalogs. We do not consider warm data for these sources. absorbers here, as this requires a more considered ap- proach; Winter et al. (2012) outline such an approach us- • We present the properties of iron lines, soft excesses ing XMM-Newton and Suzaku data. In summary, while and ionized absorbers detected in the 39 objects the 22-month catalog objects in this sky region have with > 4600 counts in their spectra. Iron lines are better data quality on average, the smaller number of detected in 79% of the XMM-Newton objects, simi- objects in that subset make the percentages presented lar to 81% found in W09. Soft excesses are detected above more prone to uncertainty. at a frequency of 31–33% compared to W09’s 41%, and ionized absorption edges are detected in 18% of our sample. We present upper limiting iron line 8. DISCUSSION AND CONCLUSIONS equivalent widths for sources where the feature was 27

not detected as a significant addition to the spec- curves) to the epoch during which the 0.4–10 keV trum, and confirm the X-ray Baldwin effect to the data were taken removes a degree of freedom from degree identified by e.g., Page et al. (2004) in our the fit, further constraining the fit for some objects. sample. • We present the summed spectrum for the sam- • We introduce the concept of the soft-excess ple. The slope of the summed spectrum between strength, defined as the luminosity in the soft ex- 1–10 keV reproduces the X-ray background slope, cess (modelled as a black body) divided by the as found in W09, but we find this with a deeper power-law luminosity between the ‘clean range’ sample that exhibits a different absorption distri- 1.5–6 keV. In W09, a linear relation was found bution. Iron lines and soft excesses appear to be between the soft excess power and power-law lu- significant in the whole sample, but higher quality minosity. We find a deviation from this descrip- data are needed for about half the sample (that tion, suggesting that the soft excess is not sim- does not yet have XMM-Newton coverage) to un- ply powered by the power-law coronal emission, as derstand the frequency of these components prop- the fraction of power seen in the soft excess drops erly. slightly at higher power-law luminosities. We also find a disjoint distribution of sources: one set in • The 2–10 keV log(N)-log(S) plot for the North- which a soft excess is well-detected and for which ern Galactic Cap reveals completeness down to the soft excess fraction is > 0.1, and another fam- log(S)≈ −11.6, a factor of ∼ 4 fainter than the 9- ily of sources where the soft excess is not detected month catalog. A larger proportion of our sources at all (indicated by stringent upper limits in the lie above the completeness limit than in W09, and soft-excess strength). This result suggests that the whilst the missing sources are expected to be more process responsible for producing the soft excess is heavily absorbed and possibly Compton-thick due not ubiquitous in AGN. to their FBAT/F2−10keV ratio, the absorption prop- erties of our 58-month BAT catalog subsample • The fraction of unabsorbed (logNH < 22) sources seem to be more homogenous both above and below with ionized absorber edges is ∼ 32%, lower than completeness than found in W09 for the 9-month the 53% found for Seyfert 1–1.5 BAT AGN in catalog. Winter et al. (2012). However, given that our study and previous studies estimating the preva- • We consider the properties of the complete subsam- lence of ionized absorbers have small sample sizes, ple of 39 objects drawn from the 22-month BAT our result is broadly consistent with previous re- catalog source list in this sky region, for which the sults. XMM-Newton coverage is 90% (compared to 49% for our full 58-month sample), to identify whether • The fraction of Compton-thick sources (log N > H the improved data quality allows more robust de- 24.15) in our sample is ∼ 9%, using a simple termination of the subsample properties. However, absorption model (ztbabs). Other measures of we do not find significantly more stringent con- Compton thickness, such as high Fe-K line equiv- straints on fractions of complex spectra, luminos- alent width, high reflection R or a flat photon ity distributions, or the proportions of sources with index (pegging at 1.5), were also investigated, iron lines and soft excesses. This is in part due to which may add 2–6 Compton thick candidates, the smaller sample size of the 22-month subset and increasing the proportion of such sources to 11– four sources in the 22-month catalog with ambigu- 15%. The true Compton-thick fraction involves es- ous spectral types. There is some evidence, how- timating the number missed due to the attenua- ever, for an increasing fraction absorbed (including tion of Compton-thick AGN spectra in the BAT Compton-thick sources) with survey depth. band. Burlon et al. (2011) suggest that the true Compton-thick fraction could be 20%. In summary, we have analysed an unbiased sample that • We identify seven new ‘hidden’ sources, unidenti- is representative of local AGN activity spanning a wide fied in W09, three by using newly obtained XMM- range of parameters (absorption, luminosity), with de- Newton data. The fraction of such sources in our tailed information on the X-ray properties of the sam- sample is 13–14%, lower than the proportion found ple. There is substantial scope to build a suite of multi- in W09. wavelength studies on this sample, to further our un- derstanding of AGN accretion at z < 0.2. We defer • Reflection is found to be important in a large frac- some of the following issues to later studies, for brevity. tion of our sample, with the average value of the The complete sample assembled here will provide a use- reflection fraction found to be hRi = 2.7 ± 0.75, ful base for understanding long-term AGN variability, by suggestive of strong light bending or highly com- analysing multi-epoch observations of these AGN; many plex absorption. The average fold energy of the such observations exist in the archives, and such a study sample is well outside the BAT band, but we do would complement other such variability studies on AGN observe a well-defined high-energy cut-off in some samples (e.g., Paolillo et al. 2004). It would also be in- sources. The use of BAT data in conjunction with teresting to examine low-energy X-ray emission (below XMM-Newton data allows reflection parameters to ∼1 keV) in buried AGNs for potential signatures of hot- be better constrained, and our technique for re- gas emission due to nuclear starburst or broader galac- normalizing the BAT data (using the BAT light tic activity. We again remind the reader of the existing 28 multi-band coverage and therefore potential for gener- ting package with which most of the plots in this pa- ating broad-band SEDs and accretion rates, which we per were generated, and Mike Koss for his comments on identify as a key priority for future work. We are cur- the paper. We thank Jack Tueller, Craig Markwardt rently working on companion studies related to under- and Neil Gehrels for their work on the proposal to ex- standing reflection in AGN; including a study of corre- tend the XMM-Newton coverage of this region, and all lations between soft X-ray features (soft excesses, iron of the BAT team for their work on the BAT catalog. We lines) and the Compton reflection hump. Our previous also thank the anonymous referee for useful comments studies using the BAT catalog (Mushotzky et al. 2008, that improved the paper. This work makes use of data W09, Vasudevan et al. 2009, 2010) all reveal the com- from XMM-Newton, an ESA science mission with instru- plexities of determining accurate black-hole mass esti- ments and contributions directly funded by ESA Mem- mates for this sample; an object-specific approach is re- ber States and NASA. This research has made use of quired to determine black-hole masses in a diverse AGN the NASA/IPAC Extragalactic Database (NED) which sample such as this, making this another priority for the is operated by the Jet Propulsion Laboratory, California future. Institute of Technology, under contract with the National Aeronautics and Space Administration (NASA). 9. ACKNOWLEDGEMENTS RVV would like to thank Jeremy Sanders for design- ing and providing extensive support for the Veusz plot- APPENDIX A. APPENDIX: MODERATELY RADIO LOUD SOURCES MRK 463 AND 3C 303.0 The objects Mrk 463 and 3C 303.0 were identified as having the highest radio loudness parameters in the sample (RL & −3). It is not straightforward to account for how this will affect the X-ray spectral fitting, since the radio emission could be due to a combination of star formation (e.g., Condon et al. 2002) and jet emission. In the case of Mrk 463, this is a known dual AGN system (§4.2) the radio images available on NED for this object 16 reveal a non-standard morphology for the radio source which cannot be straightforwardly classified as one of the traditional Fanaroff-Riley classes (FR-I/FR-II). There is little evidence for a significant component of X-ray jet emission in radio-loud non-blazar AGN (Sambruna et al. 1999), and indeed the X-ray analysis of Mrk 463 in Bianchi et al. (2008) does not consider any X-ray component in the Chandra or XMM-Newton data. For 3C 303.0, we inspect the archival Chandra images and find that while the 2–10 keV emission has a negligible contribution from extended jet emission, the 0.5–2 keV emission may be contaminated by jet emission at a level of 17% of the nuclear emission. This is likely to influence our spectral fit for this object. However, inspection of the spectrum reveals a smooth power law from 0.5–10 keV. If the intrinsic nuclear spectrum is more heavily absorbed, the radio emission due to jets at soft energies may give the illusion of less absorption. However, determining the precise degree of jet contamination is complex, and since these considerations only affect two of our objects, they will not influence the absorption distributions and other results significantly.

B. APPENDIX: SOURCES WITH LOW COUNTS The five sources in Table 8 had XRT data but lacked sufficient counts to construct a spectrum, and due to their poor data quality we have not included them when calculating sample-wide properties presented in this paper. For completeness, we present a basic analysis of these datasets here. We examined the counts in the 0.5–2 keV and 2– 10 keV in the source and background regions for these objects. When sources were detected at a ≥ 2σ level above background, we calculate basic fluxes and luminosities (in 0.5–2 and 2–10 keV bands) using WebPIMMS assuming a Galactic-absorbed powerlaw with a photon index of 1.9; where the sources were not detected at greater than 2σ, we present 95% confidence upper limits on fluxes and luminosities. These upper limits were calculated using the Gehrels (1986) prescription. These sources would clearly benefit from longer, higher signal-to-noise ratio observations to identify whether there is any sustained X-ray emission. We attempt to recover an estimate for the intrinsic column density in these sources, since the very faint X-ray fluxes may be indicative of heavily absorbed, potentially Compton-thick objects, when considered alongside the available BAT spectra. However, we find that this approach produces uncertain results with large uncertainties on the inferred log(NH), unless we are able to renormalize the BAT spectrum and thereby have a better estimate of the absolute ratio between the 2–10 keV and 14–195 keV fluxes (only possible for three of these objects). These results are presented in the final column of Table 8 for completeness. We do not find any convincing hints of Compton-thick candidates amongst these five objects, but better data are needed.

C. APPENDIX: POOR FITS From the 95 with spectra with sufficient counts to attempt fitting, we identify 19 which have null-hypothesis prob- abilities of their best-fitting models less than 5 × 104, indicating that some key features of these spectra have not been modeled fully. In this work, we do not attempt to account for all of the detail in each spectrum, but try to account for the key properties of the spectrum such as the photon index, intrinsic absorption, and the basic set of features discussed in §4.7 and restrict ourselves to these features. AGN spectra often exhibit much more complexity

16 http://ned.ipac.caltech.edu/ 29 than the model combinations used here, and we examine if this complexity can account for the poor fits in the 19 sources with low null-hypothesis probabilities. On inspection, these objects can be split into three categories: 1) those with few counts, 2) partial covering sources with un-modelled residuals at low energies, such as bumps or other types of structure below 1 keV, and 3) sources with broad iron lines with complex structure not modelled by the zgauss model, which were already discussed above. The soft features in the second category may be due to features in the host galaxies of these AGN (such as emission lines from photo-ionized/hot gas), but are not addressed further here.

REFERENCES Ajello, M., Alexander, D. M., Greiner, J., et al. 2012, ApJ, 749, 21 Mushotzky, R. 2004, in Astrophysics and Space Science Library, Ajello, M., Costamante, L., Sambruna, R. M., et al. 2009, ApJ, Vol. 308, Supermassive Black Holes in the Distant Universe, ed. 699, 603 A. J. Barger, 53–+ Arnaud, K. A. 1996, in Astronomical Society of the Pacific Mushotzky, R. F., Winter, L. M., McIntosh, D. H., & Tueller, J. Conference Series, Vol. 101, Astronomical Data Analysis 2008, ArXiv e-prints, 0807.4695 Software and Systems V, ed. G. H. Jacoby & J. Barnes, 17 Nandra, K., O’Neill, P. M., George, I. M., & Reeves, J. N. 2007, Barthelmy, S. D., Barbier, L. M., Cummings, J. R., et al. 2005, MNRAS, 382, 194 Space Sci. Rev., 120, 143 Page, K. L., O’Brien, P. T., Reeves, J. N., & Turner, M. J. L. Becker, R. H., White, R. L., & Helfand, D. J. 1995, ApJ, 450, 559 2004, MNRAS, 347, 316 Beckmann, V., Barthelmy, S. D., Courvoisier, T. J.-L., et al. Paolillo, M., Schreier, E. J., Giacconi, R., Koekemoer, A. M., & 2007, A&A, 475, 827 Grogin, N. A. 2004, ApJ, 611, 93 Beckmann, V., Soldi, S., Ricci, C., et al. 2009, A&A, 505, 417 Ponti, G., Miniutti, G., Cappi, M., et al. 2006, MNRAS, 368, 903 Bianchi, S., Chiaberge, M., Piconcelli, E., Guainazzi, M., & Matt, Rees, M. J. 1984, ARA&A, 22, 471 G. 2008, MNRAS, 386, 105 Remillard, R. A., & McClintock, J. E. 2006, ARA&A, 44, 49 Bianchi, S., Guainazzi, M., Matt, G., & Fonseca Bonilla, N. 2007, Reynolds, C. S. 1997, MNRAS, 286, 513 A&A, 467, L19 Ricci, C., Walter, R., Courvoisier, T. J.-L., & Paltani, S. 2011, Blackburn, J. K. 1995, in Astronomical Society of the Pacific A&A, 532, A102 Conference Series, Vol. 77, Astronomical Data Analysis Ross, R. R., & Fabian, A. C. 2005, MNRAS, 358, 211 Software and Systems IV, ed. R. A. Shaw, H. E. Payne, & Rovilos, E., Fotopoulou, S., Salvato, M., et al. 2011, A&A, 529, J. J. E. Hayes, 367 A135 Brandt, W. N., & Hasinger, G. 2005, ARA&A, 43, 827 Sambruna, R. M., Eracleous, M., & Mushotzky, R. F. 1999, ApJ, Brandt, W. N., Mathur, S., & Elvis, M. 1997, MNRAS, 285, L25 526, 60 Brightman, M., & Nandra, K. 2011, MNRAS, 413, 1206 Shemmer, O., Brandt, W. N., Netzer, H., Maiolino, R., & Kaspi, Brusa, M., Zamorani, G., Comastri, A., et al. 2007, ApJS, 172, S. 2008, ApJ, 682, 81 353 Sobolewska, M. A., & Done, C. 2007, MNRAS, 374, 150 Burlon, D., Ajello, M., Greiner, J., et al. 2011, ApJ, 728, 58 So ltan, A. M. 2007, A&A, 475, 837 Cappi, M., Panessa, F., Bassani, L., et al. 2006, A&A, 446, 459 Terashima, Y., & Wilson, A. S. 2003, ApJ, 583, 145 Condon, J. J., Cotton, W. D., & Broderick, J. J. 2002, AJ, 124, Treister, E., & Urry, C. M. 2005, ApJ, 630, 115 675 Tueller, J., Mushotzky, R. F., Barthelmy, S., et al. 2008, ApJ, Corral, A., Della Ceca, R., Caccianiga, A., et al. 2011, A&A, 530, 681, 113 A42 Tueller, J., Baumgartner, W. H., Markwardt, C. B., et al. 2010, Crenshaw, D. M., Kraemer, S. B., Boggess, A., et al. 1999, ApJ, ApJS, 186, 378 516, 750 Turner, T. J., Miller, L., Reeves, J. N., & Kraemer, S. B. 2007, Crummy, J., Fabian, A. C., Gallo, L., & Ross, R. R. 2006, A&A, 475, 121 MNRAS, 365, 1067 Ueda, Y., Eguchi, S., Terashima, Y., et al. 2007, ApJ, 664, L79 Done, C., & Nayakshin, S. 2007, MNRAS, 377, L59 Vasudevan, R. V., Fabian, A. C., Gandhi, P., Winter, L. M., & Drake, C. L., McGregor, P. J., Dopita, M. A., & van Breugel, Mushotzky, R. F. 2010, MNRAS, 402, 1081 W. J. M. 2003, AJ, 126, 2237 Vasudevan, R. V., Mushotzky, R. F., Winter, L. M., & Fabian, Falocco, S., Carrera, F. J., Corral, A., et al. 2012, A&A, 538, A83 A. C. 2009, MNRAS, 399, 1553 Feigelson, E. D., & Nelson, P. I. 1985, ApJ, 293, 192 Vaughan, S., Iwasawa, K., Fabian, A. C., & Hayashida, K. 2005, Gandhi, P., & Fabian, A. C. 2003, MNRAS, 339, 1095 MNRAS, 356, 524 Gehrels, N. 1986, ApJ, 303, 336 Winter, L. M., Lewis, K. T., Koss, M., et al. 2010, ApJ, 710, 503 Gierli´nski, M., & Done, C. 2004, MNRAS, 349, L7 Winter, L. M., Mushotzky, R. F., Reynolds, C. S., & Tueller, J. Gilli, R., Comastri, A., & Hasinger, G. 2007, A&A, 463, 79 2009, ApJ, 690, 1322 Gupta, A., & Galeazzi, M. 2009, ApJ, 702, 270 Winter, L. M., Mushotzky, R. F., Tueller, J., & Markwardt, C. Iwasawa, K., & Taniguchi, Y. 1993, ApJ, 413, L15 2008, ApJ, 674, 686 Jiang, P., Wang, J. X., & Wang, T. G. 2006, ApJ, 644, 725 Winter, L. M., Veilleux, S., McKernan, B., & Kallman, T. R. K¨ording, E. G., Jester, S., & Fender, R. 2006, MNRAS, 372, 1366 2012, ApJ, 745, 107 Koss, M., Mushotzky, R., Veilleux, S., et al. 2011a, ApJ, 739, 57 Winter, L. M., Danforth, C., Vasudevan, R., et al. 2011, ApJ, Koss, M., Mushotzky, R., Treister, E., et al. 2011b, ApJ, 735, L42 728, 28 Kukula, M. J., Ghosh, T., Pedlar, A., & Schilizzi, R. T. 1999, Worsley, M. A., Fabian, A. C., Bauer, F. E., et al. 2005, MNRAS, ApJ, 518, 117 357, 1281 Lehmer, B. D., Xue, Y. Q., Brandt, W. N., et al. 2012, ApJ, 752, Wright, E. L., Eisenhardt, P. R. M., Mainzer, A. K., et al. 2010, 46 AJ, 140, 1868 Markevitch, M., Bautz, M. W., Biller, B., et al. 2003, ApJ, 583, 70 York, D. G., Adelman, J., Anderson, Jr., J. E., et al. 2000, AJ, Mateos, S., Carrera, F. J., Page, M. J., et al. 2010, A&A, 510, A35 120, 1579 McCammon, D., Almy, R., Apodaca, E., et al. 2002, ApJ, 576, Zdziarski, A. A., Ghisellini, G., George, I. M., et al. 1990, ApJ, 188 363, L1 Mel´endez, M., Kraemer, S. B., & Schmitt, H. R. 2010, MNRAS, 406, 493 Murphy, K. D., & Yaqoob, T. 2009, MNRAS, 397, 1549 30 ) are presented. observations where the XRT XMM-Newton or 8.73 (5.10) 112.78 (56.21) 14.85 (4.84) 13.69 (5.04) 20.94 (10.42) 17.52 (6.63) 11.90 (5.17) 8.86 (5.38) 5.52 (4.97) 4.37 (4.97) 15.16 (6.40) 9.08 (5.58) 8.93 (4.93) 17.81 (11.05) 16.84 (8.59) 8.09 (7.32) 17.51 (7.12) 11.46 (8.16) 20.85 (12.12) 17.19 (8.13) 15.75 (7.21) 15.47 (5.93) 15.99 (9.36) 11.06 (6.38) 23.79 (10.20) 11.83 (5.99) 13.69 (5.64) 16.89 (9.83) 21.18 (9.61) 13.23 (6.11) 37.62 (24.17) 29.15 (11.73) 9.99 (5.79) 12.37 (5.30) 28.55 (14.07) 16.91 (5.04) 21.94 (12.32) 30.67 (14.92) 533.09 (275.00) 21.63 (12.63) 14.44 (6.04) 11.94 (5.18) 31.39 (14.07) 23.92 (11.71) BAT flux (SNR) 33.63 (15.22) 20.49 (8.70) 13.64 (5.28) 15.40 (7.96) 8.07 (4.82) 21.42 (14.64) for ASCA mos2 + mos1 Optical Sy1/Sy2 Sy1.5 Sy1 galaxy Sy2 Sy1 Sy1.9 Sy1 QSO/BLAGN Sy2 Sy1.5 Sy2 Sy1 Sy1 Sy1.8 Sy2 Sy1 Sy1 Sy1.5 Sy1 Sy1 Sy1 galaxy Sy1.5 LINER Sy1 Sy1.8 Sy1/LINER Sy2 Sy1 Sy1.5 Sy2 Sy1.2 BLAGN LINER Sy2 Sy2 Sy1.9 Sy1.5 Sy1 AGN Sy1 Sy1 Sy1.9/LINER , along with the signal-to-noise ratio of the BAT Type Sy2 Sy1.8 Sy1.9 Sy2/binary AGN Sy1 Sy1.5 2 + − or the identified soft X-ray (0.4–10 keV) counterpart pn cm 1 − i.e. Usable % of obs. 84 99 – – – – 100 – – – 100 – 97 – 100 95 – 100 – – – 66 – – – – – 94 100 – 86 – 100 – 41 – 92 100 100 71 – – 100 100 75 42 88 – – 91 erg s 12 − 9 1 2 2 2 1 5 9 7 9* 8 2* 9 9* 4 9 8 9 0 0 8 1 7 3 5 5 6 0 5 ...... 8 8 2 1 6 1 9 9 0 8 0 0 4 7 6 2 6 7 0 3 6 ...... Obs. time (ks) 39 40 9 8 16 8 11 11 15 43 8 11 12 8 11 9 19 8 8 5 13 12 51 10 17 7 31 26 15 33 15 3 4 13 12 4 14 9 4 9 29 1 2 6 6 39 8 39 20 95 datasets. Source Counts 9214 53612 2085 133 217 1788 190 21479 1049 910 46 31401 94 42913 3076 25011 976 781 30020 3521 10239 2693 25 17251 435 1357 139 634 44 37289 7098 1789 147191 141 57522 15499 204 2085 110 15537 7674 247586 3691 7 534 1693 13865 4519 454079 10415 rt of the observation; this was not done for Obs. date 2006-04-24 2000-11-28 2010-10-29 2010-10-31 2008-07-03 2006-06-15 2008-07-13 2002-05-23 2009-07-12 2010-08-30 2010-08-30 2009-12-20 2008-11-07 2002-05-14 2009-02-15 2009-05-29 2004-05-02 2010-07-23 2009-06-14 2005-12-14 2004-05-24 2008-10-26 2008-07-23 2009-05-23 2009-02-17 2010-07-02 2009-10-20 2009-11-09 2009-08-04 2001-05-09 2006-06-26 2005-12-25 2002-11-22 2006-07-04 2002-05-30 2009-12-26 2010-04-26 2009-10-30 2008-10-20 2009-11-07 2001-11-26 2000-12-21 2009-06-14 2007-11-30 2010-11-21 1995-05-22 2004-06-09 09-11-1999 2005-05-23 2001-05-06 XMM-Newton TABLE 1 Observations Obs. ID 0405340101 0101040301 00040954001 00040955004 00037314002 0312191501 00037131004 0103861801 00039831001 00040957001 00040957001 0601780201 00031290001 0109080801 00037369002 0601780301 0200430501 00038055002 0601780401 00035265001 0204650301 00037371001 00036986001 0601780501 00038346001 00040962001 00090176001 00038057001 00037373002 0090020101 0312191701 00035361001 0157560101 00035599002 0109080101 0601780601 00040963001 0601780701 00037315003 0601780801 0112551201 0112310101 0601780901 00036654002 00040964003 73036000 0204650201 77076000 0304030101 0059140101 sources. BAT fluxes are provided in units of 10 ons are indicated using bold type. The positions quoted are f Instrument XMM XMM XRT XRT XRT XMM XRT XMM XRT XRT XRT XMM XRT XMM XRT XMM XMM XRT XMM XRT XMM XRT XRT XMM XRT XRT XRT XRT XRT XMM XMM XRT XMM XRT XMM XMM XRT XMM XRT XMM XMM XMM XMM XRT XRT ASCA XMM ASCA XMM XMM ion at energies 0.4–10 keV (in all detectors used for fitting, mn therefore only applies to b 52.708 55.446 55.932 61.326 62.148 62.143 57.525 59.481 65.101 53.357 53.320 61.057 57.569 64.494 58.546 69.111 60.320 64.978 72.080 55.580 73.704 73.238 66.650 72.821 56.162 66.498 75.999 55.192 54.678 60.064 66.407 56.896 70.085 77.289 79.613 79.034 65.525 63.072 78.396 68.413 71.404 75.064 75.920 67.940 56.128 81.484 68.468 57.931 82.271 68.842 XMM-Newton l 200.208 216.992 234.761 182.222 208.222 214.722 237.757 239.366 181.567 145.639 145.481 245.603 149.169 164.675 147.029 230.898 147.136 251.453 226.837 139.255 191.584 227.137 255.411 174.169 138.147 260.274 198.919 275.039 277.014 138.172 270.116 279.540 148.882 242.830 205.968 188.499 140.753 138.079 174.707 142.750 147.305 155.074 157.646 276.792 286.006 194.394 279.183 131.140 190.681 138.319 Dec 28.785 19.865 11.089 38.181 25.932 22.964 10.606 11.047 37.490 58.856 58.946 9.586 54.389 44.226 54.383 19.156 52.949 10.296 21.596 59.198 31.909 21.939 10.394 36.886 58.977 9.041 29.642 -4.280 -5.208 55.453 6.806 -3.678 44.531 20.316 27.902 31.177 49.999 52.711 33.879 47.059 43.685 39.406 38.336 7.038 -5.506 29.529 7.191 58.660 29.813 47.304 observations, corrections for flaring required excising pa RA 150.457 155.878 160.860 161.203 161.679 162.379 163.137 165.258 165.918 166.407 166.496 168.457 168.833 169.277 171.401 171.818 173.170 173.205 174.123 174.786 174.928 175.317 175.570 176.125 176.387 176.980 177.190 177.327 178.160 179.484 180.242 180.310 180.791 181.124 181.176 181.180 181.483 181.596 181.887 182.309 182.374 182.637 182.686 183.262 183.477 183.574 184.292 184.479 184.611 184.740 XMM-Newton Redshift 0.1849 0.0039 0.0476 0.0259 0.0206 0.0328 0.0878 0.0356 0.0739 0.1930 0.0477 0.0292 0.0703 0.1438 0.0211 0.1055 0.0274 0.0437 0.0299 0.0601 0.0089 0.0632 0.1505 0.0381 0.0099 0.0688 0.0230 0.0845 0.0188 0.0035 0.0360 0.0194 0.0023 0.0224 0.1653 0.0250 0.0631 0.0028 0.0791 0.0242 0.0030 0.0033 0.0228 0.0070 0.0660 0.0640 0.0080 0.0210 0.0129 0.0015 . — Table of observations used for each object. New observati Note NGC 3227 SDSS J104326.47+110524.2 UGC 05881 Mrk 728 FBQS J110340.2+372925 IC 2637 MCG +09-19-015 PG 1114+445 ARP 151 UGC 06527 IC 2921 SBS 1136+594 Mrk 744 KUG 1141+371 MCG+10-17-061 2MASX J11491868-0416512 NGC 3998 CGCG 041-020 MRK 1310 NGC 4051 Ark 347 PG 1202+281 UGC 7064 B2 1204+34 Mrk 198 NGC 4138 NGC 4151 KUG 1208+386 NGC 4180 Mrk 202 AGN 3C 234 MCG +06-24-008 Mrk 417 2MASX J11053754+5851206 CGCG 291-028 1RXS J1127+1909 NGC 3758 2E 1139.7+1040 2MASX J11475508+0902284 MCG +05-28-032 MCG -01-30-041 NGC 4102 2MASX J12135456-0530193 Was 49b 2MASX J10523297+1036205 PG 1138+222 2MASX J12055599+4959561 NGC 4235 Mrk 766 NGC 4258 whole observations were used, and the ‘Usable % of obs.’ colu detection (in brackets). For Asterisks indicate that pile-up was corrected for these to the BAT source. The total number of counts in the source reg 31 8.69 (5.78) 14.83 (5.31) 25.43 (8.38) 10.94 (5.19) 16.79 (10.92) 13.97 (5.51) 20.01 (7.62) 19.14 (8.05) 10.71 (6.54) 9.70 (5.00) 7.31 (4.91) 19.30 (9.15) 13.61 (5.71) 15.79 (6.53) 23.43 (10.56) 16.77 (7.70) BAT flux (SNR) 23.95 (10.18) 275.78 (110.73) 26.08 (14.28) 14.84 (6.17) 10.41 (4.93) 88.68 (33.45) 6.64 (5.59) 27.88 (13.12) 10.32 (4.92) 17.93 (9.00) 34.56 (18.53) 19.41 (7.41) 55.64 (21.63) 14.55 (5.71) 6.95 (5.27) 17.24 (7.46) 111.13 (42.40) 18.97 (10.14) 14.15 (5.41) 16.18 (6.77) 20.26 (10.35) 11.61 (5.22) 242.76 (95.00) 73.57 (32.55) 19.42 (10.55) 12.29 (6.23) 17.66 (7.16) 13.39 (7.99) 27.39 (15.65) 13.44 (7.71) 8.12 (4.90) 36.08 (14.55) 18.81 (7.70) 20.95 (9.67) starburst galaxy? Sy1? Sy2 Sy2 Sy2 AGN Sy1.9 Sy2 galaxy galaxy galaxy Sy2 Sy1.9 XBONG Sy1 Sy2 Optical Type Sy1 Sy2 Sy1.9 Sy2 LINER Sy1 Sy1 XBONG galaxy Sy1.5 Sy1 Sy2 XBONG Sy1 Sy1.9 Sy2 Sy1.9 Sy2/gal. pair Sy1.9 Sy1 Sy1.5 Sy1/dual AGN Sy1.9 Sy1.5 Sy1.5 Sy1 Sy1 Sy1 Sy1.5 Sy1 Sy1.5/FR-II Sy1 Sy1 Sy2 – 69 – – – – – – – – – – – – – – Usable % of obs. 100 78 97 100 100 80 – 62 – – 100 – 98 – 100 100 78 76 100 34 100 90 100 97 55 – 100 – 75 – – 100 – 100 9* 8 9 5 7 1 4 3 9* 1 4 9 9 3 5 1 8 8 6 8 5 6 4 2* 4 2 1 9 ...... 5 4 6 8 6 8 8 4 9 3 6 3 4 2 1 8 2 0 7 2 8 4 ...... Obs. time (ks) 11 18 15 4 25 23 28 33 30 6 2 5 11 17 6 16 8 4 11 2 6 17 67 28 17 0 5 5 19 3 7 8 26 21 95 21 5 6 17 6 10 14 4 21 18 4 10 5 9 22 Source Counts 64116 3629 11194 24 95204 79300 278937 3179 1350 1136 10 679 64141 567 65 2365 50 1855 34064 152 73 13934 84375 6135 25885 3 589 629 3132 39 135 11416 4022 122599 1875711 59341 232 1873 22461 320 175 61638 163 871 3131 1613 128688 764 472 14459 Obs. date 2009-07-09 2002-07-07 2002-05-31 2010-04-25 2005-07-09 2003-06-12 2000-07-02 1997-06-03 2008-06-28 2003-07-14 2010-12-08 2008-05-09 2006-06-23 1996-07-19 2008-03-07 2006-06-27 2010-08-20 2009-02-25 2002-12-18 2007-09-20 2009-11-23 2010-01-30 2003-07-18 2008-07-20 2002-06-14 2008-06-25 2008-09-05 2010-04-04 2010-01-31 2010-06-01 2010-08-26 2002-12-10 2001-12-22 2004-07-11 2001-07-09 2002-05-27 2009-11-18 2007-01-10 2000-07-28 2010-09-10 2010-11-12 2009-12-13 2008-04-27 1995-12-04 1995-05-22 2008-02-01 2001-01-13 2008-12-23 2007-12-19 2011-02-13 Obs. ID 0601781001 0110930301 0112521901 00040965001 0301450201 0112840101 0109970101 75081000 0554500101 0149170701 00041773002 00037318001 0312192001 74040000 00031153005 0312192101 00041174001 00037378001 0094360501 00037093003 00038063001 0601781201 0152940101 0554500701 0112551701 00037319001 00038067001 00090327001 0601781301 00040970001 00040971001 0072340701 0094401201 0201830201 0089960301 0094740201 00090180002 00035307003 0102040501 00040977001 00038070002 0601781401 00037321001 73028000 73008000 00037322001 0112910201 00037323001 00036622001 0651850501 Instrument XMM XMM XMM XRT XMM XMM XMM ASCA XMM XMM XRT XRT XMM ASCA XRT XMM XRT XRT XMM XRT XRT XMM XMM XMM XMM XRT XRT XRT XMM XRT XRT XMM XMM XMM XMM XMM XRT XRT XMM XRT XRT XMM XRT ASCA ASCA XRT XMM XRT XRT XMM b 64.647 74.335 81.534 58.962 81.739 74.355 57.403 81.799 62.581 79.220 54.720 78.950 63.237 57.174 52.405 73.960 70.932 51.384 79.448 72.069 70.554 63.640 64.803 78.630 76.246 75.726 71.714 75.190 61.305 69.912 74.047 72.322 72.747 53.809 70.495 62.886 68.973 66.861 55.125 57.383 60.615 53.478 61.795 56.819 57.501 62.817 54.631 50.264 50.211 57.216 ed). l 286.395 279.123 162.095 128.094 269.447 290.398 297.483 136.975 124.432 314.092 306.739 317.528 118.811 308.806 308.096 318.766 317.846 311.463 98.060 108.983 325.355 328.569 331.299 52.466 74.348 3.747 89.276 4.296 334.600 353.206 67.887 77.329 5.821 339.150 31.960 88.564 31.299 19.700 349.218 355.897 86.979 100.299 11.223 93.037 90.528 37.352 11.209 2.754 3.137 69.398 Dec 2.679 12.662 33.546 57.965 20.158 11.818 -5.344 35.063 54.534 16.537 -8.086 16.407 53.791 -5.552 -10.338 11.633 8.627 -11.128 36.594 44.407 8.980 2.999 4.543 30.378 35.654 19.567 41.713 19.379 2.079 13.467 35.350 38.575 18.371 -3.208 25.137 47.790 24.615 19.831 1.285 5.459 48.662 58.794 14.256 53.504 52.027 25.910 10.437 3.708 3.863 42.050 RA 185.851 186.445 186.455 187.842 188.016 189.432 189.914 190.436 191.669 195.024 195.080 195.746 196.000 196.055 196.059 197.274 197.738 198.274 198.365 198.823 200.246 203.951 204.567 205.297 205.535 206.065 206.330 206.618 207.470 208.621 208.892 208.973 209.011 213.313 214.499 215.376 216.095 216.854 217.278 218.469 218.719 219.093 219.799 220.159 220.761 223.283 226.006 226.485 226.682 228.763 Redshift 0.0234 0.0084 0.0011 0.0104 0.0630 0.0051 0.0090 0.0231 0.0167 0.0800 0.0263 0.0672 0.0299 0.0037 0.0104 0.0251 0.0527 0.0343 0.0029 0.0366 0.0319 0.0218 0.0230 0.0399 0.0035 0.0269 0.0086 0.0840 0.0327 0.0635 0.1016 0.0501 0.0504 0.0062 0.0172 0.0723 0.0169 0.1105 0.0865 0.0249 0.0362 0.0315 0.0714 0.0377 0.1412 0.0490 0.0364 0.0361 0.0377 0.0085 AGN Mrk 50 NGC 4388 NGC 4395 NGC 4500 NGC 4579 NGC 4593 NGC 4619 NGC 4686 2MASX J13000533+1632151 MRK 0783 SWIFT J1303.9+5345 NGC 4941 NGC 4939 2MASX J13105723+0837387 NGC 5033 UGC 08327 NED02 NGC 5252 Mrk 268 NGC 5273 CGCG 102-048 2MASX J13542913+1328068 Mrk 463 NGC 5506 NGC 5548 H 1419+480 NGC 5610 Mrk 1383 NGC 5674 Mrk 817 2MASX J14391186+1415215 3C 303.0 2MASX J14530794+2554327 MRK 1392 2MASX J15064412+0351444 Ark 374 MCG -01-33-063 SWIFT J1309.2+1139 II SZ 010 NGC 5106 NGC 5231 NGC 5290 2MASX J13462846+1922432 UM 614 2MASX J13553383+3520573 Mrk 464 Mrk 813 NGC 5683 Mrk 477 Mrk 841 NGC 5899 Table 1: Table of observations used for each object (continu 32 orber line .87 keV (default energies) α line and soft excess line bsorption α ption α line and soft excess α line α line at (default) 6.4 keV α line at (default) 6.4 keV α Description Power-law with Galactic absorptionAbsorbed only power-law with GalacticAs and for intrinsic S2, (neutral) withAs a a for Fe S2, K- withAs a for soft S2, excess withAs modelled an for as edge S2, a at with blackAbsorbed 0.73 both body power-law keV a with (default) soft warm toAbsorbed excess absorber model power-law and edge a with Fe and warm warm K- Absorbed Fe abs absorber K- power-law edge, with Fe two K- Absorbed warm power-law absorber with edges two atAbsorbed warm 0.73 power-law absorber with and edges two 0 and warm a absorber Fe edges, K- FePartially-covered K- absorbed power-law with Galactic absor As for C1, including a Fe K- TABLE 2 Model combinations used model string xspec tbabs(powerlaw) tbabs(ztbabs(powerlaw)) tbabs(ztbabs(powerlaw+zgauss)) tbabs(ztbabs(powerlaw+zbbody)) tbabs(ztbabs(zedge(powerlaw))) tbabs(ztbabs((powerlaw+zbbody+zgauss))) tbabs(ztbabs(zedge(powerlaw+zgauss))) tbabs(ztbabs(zedge(powerlaw+zgauss+zbbody))) tbabs(ztbabs(zedge(zedge(powerlaw)))) tbabs(ztbabs(zedge(zedge(powerlaw+zgauss)))) tbabs(ztbabs(zedge(zedge(powerlaw+zgauss+zbbody)))) tbabs(zpcfabs(powerlaw)) tbabs(zpcfabs(powerlaw+zgauss)) Model identifier Simple power-law models S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 Complex models (partial covering) C1 C2 33 (3) ities, 2 − 2 were . 2 10keV cm − < 20 int 2 Γ /L < 5 . 5GHz L 331 403 798 961 930 342 191 091 010 121 554 530 392 030 982 851 215 ...... 5 5 5 4 4 4 5 5 5 5 5 5 5 5 4 5 5 = 917 895 671 245 277 872 442 998 569 567 600 923 089 590 714 810 132 550 550 101 799 011 619 394 022 ...... − − − − − − − − − − − − − − − − − 3 4 4 5 5 4 4 3 5 5 4 3 5 4 5 4 5 4 4 5 3 5 5 4 5 RL (10) < − < − − − < − − − < < < − < < < − − − − < − < < − − < < < − − − − < − − − − − − < ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) † † † † † † † † † † † † † † † † † † † † † † † † † † ovided in brackets: errors below 195keV 5 GHz fluxes or upper limits from − 14 L (9) 44.9 (44.6 42.6 (42.7) 43.9 (43.8) 43.3 (43.3) 43.3 (43.3) 43.3 (43.6 43.9 (43.9 44.5 (45.0 43.5 (43.5) 44.1 (44.2 44.8 (44.5) 44.8 (44.5) 43.4 (43.5) 43.5 (43.2 44.0 (43.9 44.0 (43.9 44.7 (44.7) 43.3 (43.4 43.3 (43.4 44.7 (44.6 43.2 (43.2) 43.9 (43.8) 43.4 (43.4 44.3 (43.8 44.3 (43.8 42.6 (42.6) 44.2 (44.5 45.0 (44.7 43.7 (43.2 43.7 (43.2 42.5 (42.1 44.1 (44.1) 43.5 (43.6 43.5 (43.6 44.3 (44.7 43.0 (43.3 41.7 (41.7) 43.8 (44.0 43.1 (43.5 41.7 (41.7) 43.5 (43.5 44.9 (44.8) 47 61 46 44 . . . . 01 00 02 10 81 47 07 56 03 15 18 12 68 05 01 09 06 01 04 11 05 06 05 03 16 08 03 15 08 16 16 31 54 03 04 21 27 02 01 02 01 10 90 33 48 07 62 02 14 22 11 47 04 74 70 01 09 05 01 06 21 05 09 04 02 15 42 09 03 20 08 17 17 30 42 03 04 20 23 02 ozen in the fit; limits of 1 ...... 0 0 0 0 0 0 43 0 0 0 0 0 0 0 0 43 43 0 0 0 0 0 0 0 0 0 0 0 44 0 0 0 0 0 0 0 0 0 0 0 0 0 +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − 10keV 86 64 58 43 20 85 47 51 38 03 64 27 27 28 75 61 46 06 96 95 09 90 08 01 75 69 13 33 44 85 78 03 65 87 87 56 08 49 36 29 78 39 − ...... int 2 40 L (8) 44 41 43 43 42 43 43 44 43 43 44 44 43 42 43 43 44 42 42 44 42 43 43 43 43 42 44 44 42 42 42 43 42 42 44 43 41 43 43 42 44 tral hydrogen column density in units of 10 2keV − 5 . int 0 40.77 L (7) 44.70 41.18 43.26 42.81 42.46 43.44 43.36 44.08 42.80 43.37 44.33 44.01 43.34 42.53 43.51 43.26 43.66 42.75 42.72 43.70 42.96 43.08 42.96 43.80 43.65 41.73 44.28 44.35 42.64 42.53 42.05 43.45 42.48 42.48 44.50 42.99 41.32 42.96 43.20 42.44 44.28 , (7,8) Absorption-corrected 0.5–2 keV and 2–10 keV luminos 2 − 10keV cm − 1 2 − F (6) 12.54 84.25 51.99 46.06 34.11 33.48 16.35 19.25 37.28 33.94 16.65 18.76 6.63 29.92 19.37 17.69 22.33 92.65 91.73 46.75 5.45 26.66 51.62 63.58 57.38 66.74 49.93 14.47 19.42 18.44 35.19 40.09 51.53 51.54 35.92 14.61 121.10 51.59 61.82 61.74 20.04 33.52 ergs 13 2keV − − 5 . 0 F 47.69 (5) 0.94 3.83 29.18 9.04 0.18 22.63 0.51 6.89 21.34 16.84 10.35 10.22 0.00 16.97 9.90 0.25 5.52 51.10 51.39 15.67 0.39 11.38 33.91 49.69 50.85 3.04 39.89 2.95 9.92 10.04 8.06 18.60 1.17 1.16 18.03 1.37 75.99 0.33 38.85 0.85 24.14 (if partial covering model used, the covering fraction is pr 12 08 04 04 08 43 29 02 01 09 31 04 63 02 31 06 03 21 04 06 03 05 04 03 12 51 51 12 26 02 07 58 15 15 04 01 05 08 25 36 02 01 10 46 38 15 04 40 03 01 07 11 04 03 04 02 06 04 03 07 12 12 13 20 03 04 07 15 ...... 2 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 1 − ∗ +0 − +0 − +0 − +0 +0 − +0 − ∗ +0 − +0 − ∗ ∗ +0 − − +0 +0 − +0 +0 +0 − +0 − +0 − +0 − +0 − +0 − − +0 − +0 − +0 − +0 − +0 +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 +0 − +0 − ), (10) Logarithm of radio loudness parameter derived using 69 71 69 50 75 78 53 88 20 50 83 50 50 05 87 63 77 19 20 95 80 78 50 20 09 96 17 05 50 02 95 80 75 13 51 51 01 96 84 50 96 58 TABLE 3 cm ...... 1 Γ 1 1 1 1 1 1 1 1 (4) 2 1 1 1 1 2 1 1 1 2 2 1 1 1 1 2 2 1 2 2 1 2 1 1 1 2 1 1 2 1 1 1 1 1 − 19 Basic fit results ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ergs 48 12 35 10 00 02 10 09 09 15 15 08 07 07 07 03 21 08 07 10 00 01 06 03 07 07 84 10 05 07 04 03 ...... / 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − 99) 79 99) 36 97) 91 24 00 99 79 98) 92) 92 89 61 16 89 79 85 90 99) 96 ) orly constrained), (4) Photon index Γ (* denotes that Γ was fr ...... ∗ 10keV (0 (0 (0 (0 (0 (0 (0 (1 (0 (0 (0 (0 (0 (0 (0 (0 (0 (0 (0 (0 (0 (0 64 − . 2 32 03 26 05 02 22 02 58 33 28 16 18 02 01 14 18 29 30 99 24 18 21 23 28 20 17 02 15 19 03 55 31 02 22 02 77 37 20 18 24 02 01 15 23 19 17 59 15 14 43 18 25 52 12 02 16 freedom and null hypothesis probability), (2) Galactic neu ...... L 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 ( 95 86 57 23 (0 39 90 46 +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − ...... (covering fraction) 82 05 81 02 81 96 39 08 41 26 87 73 85 05 39 59 52 87 16 91 71 70 23 16 37 10 23 66 log 20 20 20 25 23 19 20 as defined in Terashima & Wilson (2003)...... H 20 < < < 24 – N (3) 23 23 23 22 24 24 24 – 22 23 – < < – 21 < 21 21 22 < 22 24 23 23 – 22 – 22 22 24 24 20 23 24 – 23 keV and 2–10 keV bands in units of 10 RL Gal 47 64 61 08 09 77 76 99 26 26 51 88 29 02 61 61 23 78 78 77 08 40 21 00 94 94 03 02 72 65 65 31 97 56 56 18 16 01 18 50 15 30 H ...... 2 1 0 1 1 1 N (2) 1 1 1 1 2 1 2 2 0 0 2 0 0 1 1 1 3 2 0 0 2 2 3 1 1 1 1 1 1 2 2 1 1 2 1 2 )) null hyp. ( P f, . o . d / 2 χ ), (9) BAT luminosity quoted as 1 − : ‘–’ indicates an insignificant column density below 10 Model ( (1) C2 (658.26/361, 0.000) C2 (1988.13/1775, 0.000) S2+BAT (80.53/89, 0.728) C1+BAT (4.72/7, 0.694) S2+BAT (6.04/8, 0.643) C1+BAT (7.81/13, 0.856) C1+BAT (129.03/81, 0.001) C1+BAT (10.65/12, 0.559) S3 (616.39/633, 0.674) S2+BAT (52.36/49, 0.345) C1+BAT (17.93/39, 0.998) S2+BAT (25.06/40, 0.969) S2+BAT (8.81/7, 0.266) S6 (804.05/786, 0.320) C1+BAT (3.28/6, 0.773) S2+BAT (3.63/7, 0.821) S10 (1254.56/1133, 0.007) C1+BAT (107.26/117, 0.729) S2+BAT (108.53/118, 0.722) S7 (781.23/796, 0.639) C1 (84.84/35, 0.000) C1+BAT (52.41/38, 0.060) S4 (767.93/756, 0.374) C1+BAT (134.38/124, 0.247) S2+BAT (151.40/125, 0.054) C2 (517.06/538, 0.734) C1+BAT (116.81/99, 0.107) C1+BAT (10.23/7, 0.176) C2 (736.77/689, 0.101) S3 (748.57/690, 0.060) C1+BAT (30.58/23, 0.133) S5 (80.10/52, 0.007) C1+BAT (14.55/15, 0.485) S2+BAT (14.54/16, 0.558) C1+BAT (34.15/30, 0.275) C1+BAT (5.28/6, 0.509) S2 (640.90/675, 0.823) C1 (334.61/315, 0.214) C1+BAT (71.21/74, 0.570) S8 (2277.83/1202, 0.000) C1+BAT (16.63/16, 0.410) S6 (1044.22/903, 0.001) H ergs N / 10keV − 2 L ( log are not shown, and * denotes that the covering fraction was po 3 . — (1) Best-fitting model (chi-squared/number of degrees of − 10 Note × AGN 3C 234 NGC 3227 SDSS J104326.47+110524.2 MCG +06-24-008 MCG +06-24-008 UGC 05881 Mrk 417 2MASX J10523297+1036205 Mrk 728 FBQS J110340.2+372925 2MASX J11053754+5851206 2MASX J11053754+5851206 CGCG 291-028 IC 2637 MCG +09-19-015 MCG +09-19-015 PG 1114+445 ARP 151 ARP 151 1RXS J1127+1909 UGC 06527 IC 2921 NGC 3758 SBS 1136+594 SBS 1136+594 Mrk 744 PG 1138+222 2E 1139.7+1040 KUG 1141+371 KUG 1141+371 MCG+10-17-061 2MASX J11475508+0902284 MCG +05-28-032 MCG +05-28-032 2MASX J11491868-0416512 MCG -01-30-041 NGC 3998 CGCG 041-020 MRK 1310 NGC 4051 Ark 347 PG 1202+281 5 the FIRST survey and 2–10 keV intrinsic luminosities, with imposed on the fit on physical grounds), (5,6) Fluxes in 0.5–2 Intrinsic column density quoted as 34 10keV − int 2 /L 5GHz L 648 226 162 778 576 342 510 851 ...... 5 5 5 5 5 4 5 5 = 540 597 789 362 735 189 668 911 489 487 960 622 829 573 705 773 809 572 572 972 023 015 996 948 833 572 394 401 732 910 860 704 991 089 807 ...... − − − − − − − − 4 4 3 4 4 4 4 4 4 4 3 5 4 5 5 3 5 4 4 4 4 4 3 5 5 4 5 4 4 3 4 4 3 5 4 RL (10) − − − − − < − − − < − − < − − < − − − − − − − − − − − − − − < − < < − − − − − − − − < ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) † † † † † † † † † † † † † † † † † † † † † 195keV − 14 L (9) 43.3 (43.1 44.1 (41.6 41.7 (41.6 44.4 (44.4 43.5 (43.6 41.8 (41.8) 43.1 (43.1) 43.4 (43.3 44.1 (44.1) 44.2 (44.3) 42.7 (42.7) 42.7 (42.7) 42.9 (42.9) 42.9 (42.6 41.1 (41.0) 43.5 (43.4) 43.6 (43.6) 40.8 (40.9) 44.2 (43.7 41.8 (41.8) 43.2 (43.1) 42.9 (43.0) 42.9 (43.0) 43.2 (42.7 44.4 (44.3) 44.4 (44.3) 44.3 (44.4 43.9 (44.0 43.9 (44.0 41.8 (41.8) 42.8 (42.9 43.9 (44.0 43.9 (43.8) 43.6 (43.1 41.1 (41.2) 43.7 (43.7 43.5 (44.0 43.3 (43.5 44.1 (44.2) 43.8 (44.0 41.6 (41.5) 42.5 (42.8 44.3 (44.1 54 . 06 45 02 40 05 09 00 07 06 22 06 04 22 00 03 02 04 03 02 02 01 21 13 10 16 13 08 05 03 24 17 05 60 03 03 27 07 01 05 01 06 12 06 24 03 68 06 09 00 07 20 15 06 04 82 00 03 02 04 02 02 02 01 08 07 10 12 08 05 06 03 23 20 05 39 03 03 27 58 07 01 05 01 04 11 ...... 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 43 0 0 0 0 0 0 +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − 10keV 67 92 69 74 98 21 28 75 43 45 60 61 48 38 82 19 68 32 45 30 82 14 15 44 53 51 68 62 49 47 12 24 26 11 91 53 54 88 13 48 21 05 68 − ...... int 2 L (8) 42 42 41 43 42 41 42 42 43 43 41 41 42 42 40 43 42 40 43 41 42 42 42 42 43 43 43 43 43 41 42 43 43 43 40 43 43 42 43 43 41 42 43 2keV − 5 . int 0 L (7) 42.51 42.62 41.75 43.35 42.66 40.89 42.33 42.50 43.34 43.05 41.28 41.25 42.32 42.16 40.84 43.14 42.50 39.90 43.51 41.20 42.71 41.83 41.83 42.38 43.14 43.12 43.28 43.46 43.28 41.30 41.72 42.85 43.19 43.17 40.63 43.48 43.33 42.60 42.72 43.33 40.79 41.66 43.52 10keV − 2 F (6) 22.97 8.96 5.41 11.66 51.25 53.98 409.58 29.53 26.25 14.58 28.52 28.93 23.00 66.57 65.31 126.46 76.57 59.40 30.06 36.41 377.74 11.85 11.93 8.47 20.43 19.78 45.81 162.47 155.66 12.75 15.44 22.19 9.12 47.83 44.60 62.03 25.21 61.69 94.79 24.65 59.75 67.64 28.48 2keV − 5 . 0 F (5) 1.24 3.29 2.76 4.59 0.65 0.51 29.49 0.71 18.64 1.50 9.19 9.19 14.03 34.29 2.43 109.79 2.50 0.77 31.55 25.25 256.45 4.60 4.62 0.23 1.49 1.38 14.32 92.89 92.85 2.13 0.36 0.10 0.00 50.40 22.92 0.13 6.21 3.30 3.50 0.56 13.14 8.37 17.04 18 02 08 00 07 64 02 04 06 00 04 05 04 08 02 02 02 02 01 61 61 05 02 02 23 02 13 78 05 04 24 03 09 00 02 53 37 04 05 04 12 05 04 10 00 02 02 02 02 02 01 20 15 05 17 11 07 02 02 23 13 03 03 14 42 05 00 04 03 ...... 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 1 0 0 +0 − +0 − +0 − ∗ − +0 +0 − +0 − − +0 +0 − +0 − − +0 − +0 +0 − +0 − +0 − +0 +0 − +0 − +0 − +0 +0 − +0 − +0 − +0 − +0 − +0 +0 +0 +0 − +0 − +0 − +0 +0 +0 − +0 − +0 − +0 − +0 +0 − +0 96 14 50 84 88 64 20 50 61 61 20 72 50 62 57 87 50 14 95 82 50 01 83 61 61 00 50 50 50 85 71 84 50 50 99 20 70 02 78 68 50 89 50 ...... Γ 1 2 1 1 (4) 1 1 2 1 1 1 2 1 1 1 1 1 1 2 1 1 1 2 1 1 1 2 1 1 1 1 1 1 1 1 1 2 1 2 1 1 1 1 1 ) ) ) ) ) ) ) ) ) ) ) ) 15 00 05 33 06 01 05 79 02 07 01 13 21 01 04 62 07 01 08 07 03 03 01 07 ...... 0 0 0 0 0 0 0 0 0 0 0 0 +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − 97) 70 99) 97 87 33 33 00) 99) 95 99) 99) 82 98) 99) 98) 93 99) 95 97 98 93 99) 96) ...... (0 (0 (0 (0 (0 (0 (0 (1 (0 (0 (0 (0 (0 (0 (0 (0 (0 (0 (0 (0 (0 (0 (0 (0 00 06 19 17 09 08 01 04 20 30 06 17 02 02 02 03 02 12 05 01 02 02 10 09 65 72 03 07 24 17 13 39 02 01 00 07 26 15 08 56 01 04 18 74 08 17 02 02 02 04 02 10 04 01 02 02 08 08 95 32 03 10 19 12 16 17 02 01 ...... 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 01 07 26 52 08 +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − . . . . . (covering fraction) 23 41 63 16 05 00 51 35 00 12 16 59 47 22 80 45 26 81 01 44 57 08 01 89 51 16 66 12 81 00 13 57 72 72 21 21 21 20 20 ...... H 23 < 22 23 23 – 22 < N (3) 23 < 24 25 23 23 23 23 21 21 23 20 23 – 23 23 – 20 20 21 21 23 22 21 23 – 24 23 24 < < 24 22 22 23 Gal 30 54 98 18 48 92 94 84 35 98 68 36 66 25 84 77 45 45 44 78 60 58 58 85 75 97 89 39 39 35 98 89 69 69 17 30 93 65 06 71 88 14 37 H ...... 2 3 1 2 1 0 0 1 N (2) 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 1 2 2 1 1 1 1 1 1 1 1 2 3 1 2 1 1 1 2 1 )) null hyp. ( P f, . o . d / 2 χ Model ( (1) C2 (876.84/648, 0.000) S5 (3.35/5, 0.646) C1+BAT (143.35/80, 0.000) C1+BAT (8.50/13, 0.810) C2 (649.78/654, 0.539) C2 (408.63/332, 0.003) C2 (11118.00/2780, 0.000) C1+BAT (287.72/230, 0.006) S2+BAT (32.90/27, 0.200) C1 (120.87/139, 0.864) C2 (502.43/514, 0.634) S3 (511.43/515, 0.536) C1 (207.46/164, 0.012) S6 (3841.54/2358, 0.000) C1 (803.72/450, 0.000) S4 (902.32/879, 0.285) C1+BAT (346.01/217, 0.000) C2 (503.48/476, 0.185) S8 (1124.99/1015, 0.009) S3 (1568.92/1153, 0.000) S6 (1803.06/1562, 0.000) C1 (158.50/162, 0.563) S2 (158.56/163, 0.584) C1+BAT (119.36/61, 0.000) C1+BAT (67.97/84, 0.898) S2+BAT (76.32/85, 0.738) S2+BAT (42.52/34, 0.150) C1 (1069.08/1048, 0.318) S4 (1090.34/1045, 0.161) C1+BAT (57.40/64, 0.707) C1+BAT (16.78/9, 0.052) C1+BAT (227.36/106, 0.000) S2+BAT (14.22/7, 0.047) S2+BAT (316.86/76, 0.000) S3 (837.13/834, 0.463) S2+BAT (22.85/17, 0.154) C1+BAT (10.83/10, 0.371) C2 (607.80/582, 0.222) C2 (2827.17/2270, 0.000) C2 (505.01/339, 0.000) S10 (1054.39/841, 0.000) S2+BAT (30.61/31, 0.486) S2+BAT (35.21/35, 0.458) AGN UGC 7064 2MASX J12055599+4959561 NGC 4102 B2 1204+34 Mrk 198 NGC 4138 NGC 4151 KUG 1208+386 2MASX J12135456-0530193 Was 49b NGC 4235 NGC 4235 Mrk 202 Mrk 766 NGC 4258 Mrk 50 NGC 4388 NGC 4395 Ark 374 NGC 4579 NGC 4593 NGC 4619 NGC 4619 NGC 4686 2MASX J13000533+1632151 2MASX J13000533+1632151 MRK 0783 SWIFT J1303.9+5345 SWIFT J1303.9+5345 NGC 4941 NGC 4939 SWIFT J1309.2+1139 2MASX J13105723+0837387 II SZ 010 NGC 5033 UGC 08327 NED02 NGC 5106 NGC 5231 NGC 5252 Mrk 268 NGC 5273 NGC 5290 2MASX J13462846+1922432 Table 3: Basic fit results (continued) 35 10keV − int 2 /L 5GHz L 471 212 682 811 301 298 ...... 4 5 5 5 5 5 = 761 911 158 554 118 242 316 153 186 481 975 948 939 768 969 560 700 ...... − − − − − − 4 4 4 2 4 5 5 4 5 5 4 4 4 3 2 4 4 RL (10) − − − − − − − − − − − − < − < − − < < − < < − ) ) ) ) ) ) ) † † † † † † † 195keV − 14 L (9) 43.6 (43.3 nan (44.3) 44.1 (44.1) 43.8 (43.8) 43.3 (43.4) 43.7 (43.6) 44.4 (44.4) 43.1 (43.2 44.6 (44.2 44.5 (44.5) 43.3 (43.3) 43.3 (43.3) 43.6 (43.6) 43.8 (43.5 44.3 (44.3 43.6 (43.7) 44.6 (44.6) 44.1 (43.7 44.0 (44.0) 43.8 (43.4 43.7 (–) 43.7 (–) 42.5 (42.5) 07 35 18 05 02 01 03 22 05 05 19 29 25 03 24 51 14 06 02 29 49 30 06 03 46 37 03 02 00 02 22 05 05 26 37 28 02 17 09 12 06 02 29 48 24 06 ...... 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − 10keV 53 91 39 15 83 40 95 55 18 17 04 01 24 50 59 25 03 66 58 19 10 09 20 − ...... int 2 L (8) 42 43 43 43 42 43 43 42 44 44 43 43 42 43 43 43 44 43 43 43 43 43 42 2keV − 5 . int 0 L (7) 42.10 43.97 43.14 43.20 42.55 43.17 43.75 42.24 44.11 44.22 43.07 43.01 42.28 43.56 43.25 43.01 43.79 43.73 43.65 43.13 42.71 42.70 42.02 10keV − 2 F (6) 14.18 20.89 25.80 5.99 668.44 391.76 71.63 40.74 49.43 80.88 48.52 48.37 5.10 143.66 29.06 16.26 21.20 79.43 127.78 32.62 36.52 36.37 52.86 2keV − 5 . 0 F (5) 4.11 0.46 11.98 0.97 28.21 207.31 40.14 0.37 39.27 83.27 4.15 0.59 2.19 157.81 1.68 1.53 9.15 51.05 139.24 17.94 4.06 3.37 0.37 41 00 05 02 02 02 03 02 01 11 05 25 38 31 57 74 10 05 13 04 00 06 02 03 02 08 03 01 12 05 06 11 05 15 24 13 13 67 37 04 ...... 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 − +0 − +0 − +0 − +0 − +0 − +0 +0 − − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − − +0 − +0 +0 +0 − 20 71 99 13 50 72 20 69 83 63 06 14 09 15 09 57 74 73 20 02 50 50 84 ...... Γ 2 1 1 2 (4) 1 1 2 1 1 1 2 2 2 2 2 1 1 1 2 2 1 1 1 ) ) ) ) ) ) ) ) 31 22 01 25 07 28 05 07 11 28 01 16 05 12 03 05 ...... 0 0 0 0 0 0 0 0 +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − 89 49 96 57 99) 95 68 95 59 99) ...... (0 (0 (0 (0 (0 (0 (0 (0 (0 (0 16 14 21 19 43 03 35 11 01 12 48 17 38 22 14 01 13 10 21 54 34 03 36 11 01 19 42 17 22 17 20 01 ...... 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 38 01 12 +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − . . . (covering fraction) 89 94 50 33 11 98 82 63 05 56 80 23 02 56 23 24 23 20 20 ...... H < 22 22 22 22 N (3) 21 23 23 22 – < – – 23 22 < 23 21 22 – 23 22 23 Gal 23 90 48 42 73 79 42 03 08 55 64 59 60 48 87 15 05 71 26 22 80 73 80 H ...... 1 1 2 1 3 N (2) 1 1 2 4 1 1 2 2 2 2 1 1 1 3 2 3 3 1 )) null hyp. ( P f, . o . d / 2 χ Model ( (1) S3 (135.25/157, 0.895) S2+BAT (11.33/8, 0.183) C1 (397.20/390, 0.390) C2 (419.98/162, 0.000) C2 (2029.06/1850, 0.002) S11 (3263.75/2420, 0.000) S7 (1143.32/1051, 0.024) S2+BAT (21.63/14, 0.087) S2+BAT (75.65/76, 0.490) S4 (427.24/371, 0.023) C1+BAT (7.92/17, 0.968) S2+BAT (8.11/18, 0.977) C1+BAT (6.97/10, 0.729) S6 (587.32/562, 0.222) S2+BAT (20.57/15, 0.151) C1 (82.02/87, 0.631) S2 (156.02/148, 0.310) C1+BAT (79.55/72, 0.254) S6 (1053.49/941, 0.006) C1+BAT (31.35/35, 0.645) C1 (14.31/17, 0.645) S2 (14.38/18, 0.704) C2 (645.03/605, 0.126) AGN UM 614 2MASX J13553383+3520573 Mrk 464 Mrk 463 NGC 5506 NGC 5548 H 1419+480 NGC 5610 Mrk 813 Mrk 1383 NGC 5674 NGC 5674 NGC 5683 Mrk 817 2MASX J14391186+1415215 Mrk 477 3C 303.0 2MASX J14530794+2554327 Mrk 841 MRK 1392 2MASX J15064412+0351444 2MASX J15064412+0351444 NGC 5899 Table 3: Basic fit results (continued) 36 -law ignature) component used zbbody counts in the fit spectra (9) 4600 line, (3) Energy of > α ents are not included in the best-fit model. (1) 87 87 87 [OVIII] . . . E – – 0 – – – – – – – – – – – – – – – – 0 0 – – – – – – – – – – – – – – – – – – – – – bsorber signature) near 0.87 keV, (9) Energy of [OVIII] (8) 105 103 006 100 108 009 cts with ...... 0 0 0 +0 − +0 − +0 − 534 769 103 [OVIII] . . . putative iron K- τ – – – – 0 – – – – – – – – – – – – – – – – – – – – – – – – – – – – 0 – – 0 – – – – – ) divided by the 1.5–6 keV luminosity in the underlying power (7) BB , (6) Optical depth of a putative [OVII] edge (warm absorber s 1 L − 73 73 73 73 73 73 73 [OVII] ...... E – – – – 0 0 – – – – – – 0 – – – – – – – – – – – 0 – – – – – – – – 0 – – 0 0 – – – – erg s 44 127 069 109 047 020 017 009 139 071 118 052 022 029 009 (6) ...... 0 0 0 0 0 0 0 +0 − +0 − +0 − +0 − +0 − +0 − +0 − 017 043 032 100 000 019 080 048 032 000 045 048 000 059 027 008 component ( ...... 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 131 794 103 017 439 221 480 [OVII] ...... τ – – < < 2 0 < – – < < – 0 < – – – – – < < – < – 0 < < – < < – – – 1 – – 0 0 < < < – zbbody 002 009 000 036 000 003 004 001 005 000 061 009 006 002 014 000 046 000 003 007 002 005 000 058 020 009 ...... TABLE 4 0 0 0 0 0 0 0 0 0 0 0 0 0 (5) +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − 005 014 000 356 010 027 061 015 026 016 399 095 118 BB ...... L – – – 0 – – 0 – – – – – 0 0 – – – – – – 0 – 0 – 0 – 0 – 0 – – – – – – – 0 – 0 0 0 – 039 140 023 053 026 022 026 027 025 030 002 058 027 041 137 025 062 027 023 025 029 027 031 002 061 028 (4) ...... 0 0 0 0 0 0 0 0 0 0 0 0 0 f the corresponding component, and therefore, those compon +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − 011 007 012 021 023 011 007 006 007 003 ...... 0 0 0 0 0 0 0 0 0 0 110 167 913 184 640 212 251 289 111 084 304 338 357 softex ...... S – – < 0 < < 0 – – < < – 0 0 – – – – – < 0 – 0 – 0 < 0 – 0 < – – – < – – 0 < 0 0 0 – component luminosity in units of 10 the luminosity in the s quoted, (8) Optical depth of a putative [OVIII] edge (warm a 026 017 010 016 020 015 009 011 (3) ...... 0 0 0 0 +0 − ∗ ∗ ∗ ∗ ∗ ∗ +0 − ∗ +0 − ∗ ∗ +0 − lines, soft excesses and warm absorber signatures) for obje zbbody line, with * denoting a frozen value, (2) Equivalent width of 203 082 122 100 088 108 112 088 155 100 116 100 100 softex α ...... E – – – 0 – – 0 – – – – – 0 0 – – – – – – 0 – 0 – 0 – 0 – 0 – – – – – – – 0 – 0 0 0 – α (2) 010 055 020 089 111 196 103 230 034 014 071 056 066 106 109 041 052 040 076 014 006 069 184 212 034 060 158 035 045 033 084 043 033 028 070 036 139 082 200 132 186 036 008 067 074 084 104 079 033 044 052 164 012 007 079 160 209 035 052 146 044 037 029 209 048 046 ...... 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 ), (5) The FeK +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − 142 058 117 061 091 037 068 122 196 ...... 0 0 0 0 0 0 0 0 0 117 230 201 256 141 076 135 148 394 621 324 432 138 075 345 303 275 721 443 537 093 114 317 120 235 176 686 140 101 320 190 562 156 ...... 6keV EQW 0 0 0 0 0 0 < 0 0 0 < < 0 0 0 0 0 0 0 0 0 < < 0 0 1 0 < < 0 0 0 0 0 < 0 0 0 < 0 0 0 − 5 . 1 36 08 04 01 05 26 00 05 05 43 08 03 03 11 03 01 06 06 08 01 04 09 01 06 04 00 07 06 21 08 03 03 11 05 01 05 ...... (1) /L 00 00 00 6 0 0 0 0 0 0 0 0 6 0 0 0 0 0 0 0 . . . ∗ +0 − ∗ ∗ +0 − ∗ ∗ ∗ ∗ ∗ +0 0 ∗ ∗ ∗ +0 − ∗ ∗ +0 0 +0 − +0 − +0 0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − +0 − BB 40 40 36 40 40 40 48 40 40 30 90 40 40 41 41 42 42 30 40 43 66 40 43 40 40 38 42 42 41 40 40 30 35 FeK ...... L E 6 6 6 6 6 6 – 6 6 6 – – 6 6 6 6 6 6 6 6 6 – – 6 6 6 6 – – 6 6 6 6 6 – 6 6 6 – 6 6 6 ), i.e., ( Model C2 C2 S3 S6 S10 S7 S4 C2 C2 S3 S2 C1 S8 S6 C2 C2 C2 C2 C2 S3 S6 C1 S4 C2 S8 S3 S6 C1 S4 S3 C2 C2 C2 S10 C1 C2 S11 S7 S4 S6 S6 C2 component used to model iron K- 6keV − 5 . 1 L Fit results - detailed features (iron K- zgauss . — All upper limits correspond to no significant detections o Note AGN 3C 234 NGC 3227 Mrk 728 IC 2637 PG 1114+445 1RXS J1127+1909 NGC 3758 Mrk 744 KUG 1141+371 KUG 1141+371 NGC 3998 CGCG 041-020 NGC 4051 PG 1202+281 UGC 7064 Mrk 198 NGC 4138 NGC 4151 NGC 4235 NGC 4235 Mrk 766 NGC 4258 Mrk 50 NGC 4395 Ark 374 NGC 4579 NGC 4593 SWIFT J1303.9+5345 SWIFT J1303.9+5345 NGC 5033 NGC 5231 NGC 5252 Mrk 268 NGC 5273 Mrk 464 NGC 5506 NGC 5548 H 1419+480 Mrk 1383 Mrk 817 Mrk 841 NGC 5899 continuum ( near 0.73 keV,edge, (7) frozen Energy unless of errors [OVII] quoted. edge, frozen unless error to model a soft excess, (4) Strength of soft excess, defined as Energy of 37

AGN Partial covering? (1) BAT renormed? (2) R (3) Efold (4) Γpexrav (5) ∆Γ (6) +1594 +0.14 3C 234 Y Y < 0.58 138 2.03 −0.17 −85 −0.13 +3.10 +0.05 NGC 3227 Y – 12.86 > 636 2.08 0.58 −3.14 −0.09 +13 +0.08 Mrk 417 Y Y < 0.45 38 0.75 −1.31 −17 −0.31 +3.10 +0.38 Mrk 728 – – 0.07 616 1.70 −0.07 −0.07 −570 −0.09 +2.38 +0.24 IC 2637 – Y 1.09 > 156 1.79 0.10 −0.91 −0.09 +5.81 +0.60 PG 1114+445 – – 0.95 69 1.60 0.10 −0.95 −47 −0.22 +1.71 +0.12 1RXS J1127+1909 – Y 1.07 259 1.60 0.10 −0.99 −168 −0.09 +411 +0.72 UGC 06527 Y – < 16.61 75 2.12 −0.08 −62 −0.61 +2.34 +0.13 NGC 3758 – Y 2.87 277 2.06 0.10 −1.55 −192 −0.10 +3.33 +0.27 Mrk 744 Y – 1.62 > 122 1.71 0.21 −1.62 −0.34 +0.14 KUG 1141+371 – Y < 0.30 263 1.79 0.05 −212 −0.18 +0.12 NGC 3998 – – < 0.98 622 1.83 −0.01 −539 −0.05 +4.77 +0.08 CGCG 041-020 Y Y 15.43 > 116 2.37 0.87 −7.73 −0.30 +8.39 +0.11 NGC 4051 – – 32.67 > 381 2.49 0.96 −6.84 −0.12 +3.57 +0.17 PG 1202+281 – – 4.61 556 2.08 0.19 −2.26 −492 −0.16 +3.56 +0.29 UGC 7064 Y Y 2.03 > 326 1.68 −0.21 −1.71 −0.23 +0.09 NGC 4102 Y Y < 0.37 > 264 2.13 −0.07 −0.07 +0.65 +57 +0.32 Mrk 198 Y Y 0.51 91 1.62 0.01 −0.37 −32 −0.07 +2.85 +0.34 NGC 4138 Y – 0.05 148 1.51 −0.10 −0.05 −73 −0.20 +4 +0.03 NGC 4151 Y – < 0.01 79 1.31 −0.89 −4 −0.03 +51 +0.34 KUG 1208+386 Y Y < 0.83 44 1.30 −0.42 −16 −0.07 +2.96 +188 +0.20 NGC 4235 – – 0.06 89 1.49 −0.08 −0.06 −35 −0.07 +3.22 +102 +0.19 NGC 4235 Y – 0.95 64 1.53 −0.10 −0.95 −30 −0.13 +3.57 +7 +0.09 Mrk 766 – Y 8.12 21 1.56 0.06 −2.38 −6 −0.08 +2.33 +0.06 NGC 4258 Y – 0.43 > 284 1.82 −0.32 −0.43 −0.10 +0.96 +0.02 Mrk 50 – – 4.79 > 334 2.18 0.23 −0.91 −0.02 +0.21 +0.03 NGC 4388 Y – 0.22 > 2096 1.81 −0.25 −0.22 −0.05 +27 +0.11 NGC 4395 Y – < 0.71 45 1.18 −0.32 −10 −0.12 +2.62 +0.13 Ark 374 – Y 1.12 542 2.10 −0.04 −0.85 −478 −0.08 +4.28 +0.14 NGC 4579 – – 3.70 > 61 2.02 0.01 −2.44 −0.10 +0.82 +0.08 NGC 4593 – – 0.90 > 517 1.89 0.06 −0.65 −0.05 +0.13 NGC 4686 Y Y < 0.37 > 230 1.86 0.10 −0.08 +0.15 2MASX J13000533+1632151 – – < 416.96 > 104 1.67 0.17 −0.33 +0.35 2MASX J13000533+1632151 Y – < 0.00 > 116 1.70 0.20 −0.35 +0.40 +350 +0.05 SWIFT J1303.9+5345 – Y 0.25 197 1.73 0.02 −0.25 −82 −0.04 +0.07 SWIFT J1303.9+5345 Y Y < 0.00 275 1.81 −0.04 −150 −0.07 +6.92 +0.36 SWIFT J1309.2+1139 Y Y 3.81 301 1.53 −0.38 −1.85 −185 −0.26 +3.70 +0.15 NGC 5033 – – 3.85 > 33 1.90 0.20 −2.49 −0.15 +0.48 +0.05 NGC 5231 Y Y 0.45 > 195 1.90 0.22 −0.45 −0.15 +58 +0.09 NGC 5252 Y – < 0.49 111 1.38 −0.12 −18 −0.05 +3.54 +0.12 Mrk 268 Y Y 7.42 > 379 1.84 −0.04 −4.02 −0.15 +0.17 NGC 5273 – – < 2.62 279 1.37 −0.13 −227 −0.05 +3.85 +0.11 UM 614 – Y 7.66 > 323 1.63 0.13 −2.18 −0.19 +8.16 +0.34 Mrk 464 Y – 3.52 170 1.80 0.08 −3.52 −131 −0.26 +31.68 +0.44 Mrk 463 Y – 7.52 198 1.80 −0.40 −7.29 −153 −0.55 +0.70 +107 +0.02 NGC 5506 Y – 1.24 166 1.85 0.16 −0.24 −30 −0.10 +0.24 +827 +0.02 NGC 5548 – – 0.99 415 1.73 0.03 −0.23 −178 −0.02 +1.38 +0.08 H 1419+480 – – 0.87 > 152 1.84 0.01 −0.87 −0.10 +4.56 +0.24 Mrk 1383 – – 2.23 > 134 2.20 0.14 −2.23 −0.20 +0.04 Mrk 817 – Y < 0.10 > 150 2.37 0.28 −0.09 +2.21 +0.10 Mrk 841 – – 4.12 > 597 2.26 0.13 −1.63 −0.09 +3.21 +0.22 NGC 5899 Y – 2.22 > 210 2.05 0.20 −2.00 −0.18

TABLE 5 Reflection fits

Note.— Reflection fit results for objects with XMM-Newton data, fit in conjunction with BAT data. We fit an absorbed pexrav model to all the objects. We inspect the best-fit model from basic fits to 0.4–10 keV data (Table 3) to select whether partial covering absorption or standard absorption should be used, indicated in column (1). (2) - if the XMM-Newton spectrum was taken during the time of assembly of the BAT catalog, the BAT spectrum was renormalized as detailed in §3.3; if this was done, this is indicated by a ’Y’ in this column. (3) - the pexrav reflection fraction R. If below 0.01, we only present the upper limit and assume that the reflection fraction is zero. (4) - The pexrav fold energy in keV. All values above 5000 keV are presented as 5000 keV (well outside the BAT band) along with the lower limit determined from the negative error bar. (5) - the photon index Γpexrav from pexrav, with no limits imposed. (6) - the difference between the pexrav Γ and that from the 0.4–10 keV basic fits presented in Table 3. 38 , (4) - 1 − s ) (9) 2 σ , o any of the − i H log N h 43–56% (23.27–23.55,0.71–0.95) Complex abs. ( 55% (23.03,0.71) 54–64% (23.28–23.4, 0.57–0.68) n units of erg cm 23, (7) - percentage of Compton-thick > ) H ) (8) σ , i H log N h 5.1), (8) - percentage of simple absorption sources, with § ity and standard deviation. Ranges in these values are due Simple abs. ( 45% (20.58,0.74) 36–46% (20.47–20.56, 0.86–0.90) 38–50% (20.67–20.80, 1.12–1.18) rption properties 15) . 24 ounts to construct a spectrum, so these are not classified int > 6%) 18%) 15%) H < < < log(N (C-thick) (7) 0% ( 5% ( 9% ( 22, (6) - percentage of sources with log(N 23) > > ) H H ton-thickness metrics discussed in , (3) - Completeness limit, given as log(S) for 2–10 keV flux S i 1 − s TABLE 6 log(N (6) 33% 49–54% 41–45% 2 − 22) > H in ergcm log(N (5) 55% 59-64 % 57–61% age of complex absorption sources, with average column dens ercentage of sources with log(N † Ambiguous sources (4) 0% 10% 13% Comparing BAT catalogs: flux limits, completeness and abso - An additional 5% of our 58-month sources do not have enough c † 15 (upper limits are based on consideration of the other Comp . Completeness limit (3) -11.0 -11.25 -11.6 24 > ) H Flux limit (2) -10.70 -10.96 -11.40 . — (1) - Catalog, (2) - Logarithm of BAT flux limit (14–195 keV) Note Catalog (1) 9-month 22-month 58-month average column density andto sources standard with deviation ambiguous (9) spectral - types. percent sources, with log(N categories shown here. percentage of sources with ambiguous spectral types, (5) - p 39 22 in < H ron lines (for s with logN values in this table, and ranges in th soft excesses (for 22-month and 58-month percentages, eatures Hidden/buried % (7) 24% 28% 13–14% , (5) - percentage of sources with significant detections of i 1 − Soft excess % (6) 41% 32–36% 31–33% , (3) - logarithm of intrinsic 2–10 keV luminosity for source 1 − α Fe K- % (5) 81% 75% 79% 22 in units of erg s ried sources. Standard deviations are quoted for all log(L) > TABLE 7 H σ , i 22) (4) s with logN > 10keV − V luminosity in units of erg s H 2 4600 counts in their spectra), (6) - percentage of sources wi log L > h (log N 42.67 (0.78) 42.60–42.65 (0.93–0.95) 42.91–42.97 (0.89) es due to ambiguous spectral types. σ Comparing BAT catalogs: luminosity and prevalence of X-ray f , i 22) (3) < 10keV − H 2 log L h (log N 43.42 (0.79) 42.80–42.84 (0.90-0.95) 43.02–43.07 (0.96-0.98) σ , i 4600 counts in their spectra), (7) - percentage of hidden/bu > 10keV − 2 log L , (4) - logarithm of intrinsic 2–10 keV luminosity for source h (all) (2) 43.01 (0.87) 42.70 (0.93) 43.00 (0.91-0.92) 1 − . — (1) - Catalog, (2) - logarithm of average intrinsic 2–10 ke Note Catalog (1) 9-month 22-month 58-month units of erg s 22-month and 58-month percentages, we only use sources with we only use sources with the averages or standard deviations represent uncertainti 40 . * These 2 detections. assuming a − σ 2 cm < 1 ) − † † † † (est) H 1 9 9 2 4 . . . . . ergs 22 22 22 23 23 13 WebPIMMS BAT renormalisation was − . (2),(3),(4),(5) Fluxes in log(N (10) ∼ ∼ ∼ ∼ ∼ 2 †† − cm 20 195keV − 14 L (9) 42.2 42.3 43.2 43.5 44.0 nits of 10 ) esent 95 per cent upper limits (see e.g., Gehrels , with upper limits provided for 1 10keV − − 95 obs 2 . L 68 66 71 82 39 9. All fluxes are in units of 10 . . . . . log( (8) < 40 41 41 42 ) 2keV − m the observed 2–10 keV count rate using 5 . obs 0 45 65 74 . . . L 45 83 40 40 41 . . log( (7) 39 39 < < < † . 2keV atic errors dominate for these objects with few counts 599 − . 5 . 0 46 05 24 96 extrap 0 . . . . F (6) < 1 2 2 4 ∗ V, corrected for absorption, in units of erg s TABLE 8 843 10keV . − 0 03 01 19 06 int 2 . . . . F (5) < 2 3 3 7 counts to construct a spectrum. (1) Galactic absorption in u detection of the source in the source region; otherwise we pr 842 XRT observations with few counts 10keV . σ − 0 02 01 18 05 2 . . . . F (4) < 2 3 3 7 is a 2 ∗ assuming Galactic absorption and a power-law with index Γ = 1 2keV 183 275 585 − on the column density estimates. . . . 5 . 0 0 0 267 292 int 0 . . F (3) 0 0 < < < This is the expected observed 0.5–2 keV flux extrapolated fro † WebPIMMS 2keV 168 261 555 − . . . 5 . 0 0 0 256 285 0 . . F (2) 0 0 < < < . (10) Estimated column density; we omit errors since system 1 Gal H − N (1) 1.390 0.830 2.800 1.680 1.730 9. (7), (8) Logarithm of luminosities in the 0.5–2 keV 2–10 ke . . — Fluxes and luminosities for XRT observations with too few Note AGN NGC 4180 NGC 4500 MCG -01-33-063 CGCG 102-048 2MASX J13542913+1328068 (9) log of BAT luminosity in erg s power-law index of Γ = 1 not possible for these sources, giving greater uncertainty 1986). Fluxes determined from count-rates using fluxes have been corrected for Galactic absorption. 0.5–2 keV and 2–10 keV bands. Full fluxes are provided if there