Unveiling the nature of the unidentified gamma-ray sources IV: the Swift catalog of potential X-ray counterparts
A. Paggi, F. Massaro, R. D'Abrusco, H. A. Smith, N. Masetti, M. Giroletti, G. Tosti, S. Funk
vversion N ovember
14, 2018: fm Preprint typeset using L A TEX style emulateapj v. 5 / / UNVEILING THE NATURE OF THE UNIDENTIFIED GAMMA-RAY SOURCES IV:THE
SWIFT
CATALOG OF POTENTIAL X-RAY COUNTERPARTS
A. P aggi , F. M assaro , R. D’A brusco , H. A. S mith , N. M asetti , M. G iroletti , G. T osti , S. F unk version November 14, 2018: fm ABSTRACTA significant fraction ( ∼ Fermi
LAT (2FGL)catalog are still of unknown origin, being not yet associated with counterparts at lower energies. In order toinvestigate the nature of these enigmatic sources, we present here an extensive search of X-ray sources lyingin the positional uncertainty region of a selected sample of these Unidentified Gamma-ray Sources (UGSs)that makes use of all available observations performed by the
Swift
X-ray Telescope before March 31, 2013,available for 205 UGSs. To detect the fainter sources, we merged all the observations covering the
Fermi
LAT positional uncertainty region at 95% level of confidence of each UGSs. This yields a catalog of 357 X-raysources, finding candidate X-ray counterparts for ∼
70% of the selected sample. In particular, 25% of the UGSsfeature a single X-ray source within their positional uncertainty region while 45% have multiple X-ray sources.For each X-ray source we also looked in the corresponding
Swift
UVOT merged images for optical and ultravi-olet counterparts, also performing source photometry. We found ultraviolet-optical correspondences for ∼ γ -ray blazar candidates. In addition, comparing our results with previous analyses,we select 11 additional γ -ray blazar candidates. Subject headings:
X-rays: galaxies - gamma rays: observations - galaxies: active - radiation mechanisms:non-thermal - catalogs INTRODUCTION
One of the biggest challenges of modern γ -ray astronomyand one of the main scientific objectives of the ongoing Fermi mission is unraveling the nature of the Unidentified Gamma-ray Sources (UGSs) (e.g., Abdo et al. 2009; Atwood et al.2009).Since the Third EGRET catalog (3EG) (e.g., Hartman etal. 1999) the fraction of γ -ray sources without an assignedcounterpart at low energies has been significant ∼
30% (e.g.,Sowards-Emmerd, Romani, & Michelson 2003). This sit-uation was mostly unchanged in the revised EGRET cata-log (EGR; Casandjian & Grenier 2008), even though the im-proved background modeling applied in the EGR resulted infewer γ -ray detections (188 sources in total, in contrast to 271listed in 3EG); 87 out of 188 EGR entries remain unassoci-ated.The UGSs at low Galactic latitude ( | b | < ◦ ) are expectedto be associated with local objects lying in our Galaxy, such asmolecular clouds (as consequence of interaction with cosmic-rays), supernova remnants, massive stars, pulsars and pulsarwind nebulae, or X-ray binaries (see,e.g., Gehrels & Michel-son 1999; Casanova et al. 2010; Yan, Lazarian, & Schlick- Harvard - Smithsonian Astrophysical Observatory, 60 Garden Street,Cambridge, MA 02138, USA SLAC National Laboratory and Kavli Institute for Particle Astro-physics and Cosmology, 2575 Sand Hill Road, Menlo Park, CA 94025,USA INAF - Istituto di Astrofisica Spaziale e Fisica Cosmica di Bologna,via Gobetti 101, 40129, Bologna, Italy INAF Istituto di Radioastronomia, via Gobetti 101, 40129, Bologna,Italy Dipartimento di Fisica, Universit`a degli Studi di Perugia, 06123 Peru-gia, Italy http: // heasarc.gsfc.nasa.gov / W3Browse / cgro / egret3.html eiser 2012; Ackermann et al. 2013; Dermer & Powale 2013)although there are few rare cases of γ -ray blazars detectedthrough the Galactic plane (e.g. Fermi J0109 + Fermi discoveries, severalmillisecond pulsars have been found at high Galactic latitudes(Abdo et al. 2010a,b; Nolan et al. 2012).A large fraction of these UGSs could be blazars, the rarestclass of radio-loud active galactic nuclei, whose emissiondominates the gamma-ray sky (e.g., Mukherjee et al. 1997;Abdo et al. 2010c). Their observational properties are gener-ally interpreted in terms of a relativistic jet aligned within asmall angle to our line of sight (Blandford & Rees 1978).The blazar spectral energy distributions (SEDs) typicallyshow two peaks. The first one, lying in the range of radio -soft X-rays, is widely held to be due to synchrotron emissionby highly relativistic electrons within their jet. The secondone lies at hard X-ray or γ -ray energies, and is interpretedas inverse Compton upscattering by the same electrons of theseed photons provided by the synchrotron emission (Inoue &Takahara 1996; Finke, Dermer, B¨ottcher 2008) with the pos-sible addition of seed photons from outside the jets yieldingcontributions to the non-thermal radiations due to external in-verse Compton scattering (see Dermer & Schlickeiser 1993,2002; Dermer et al. 2009; Finke 2013) often dominating their γ -ray outputs (Ackermann et al. 2011).Blazars are also know X-ray sources since ROSAT
DXRBS(Perlman et al. 1998; Landt et al. 2001) and
Einstein
IPC(Elvis et al. 1992; Perlman, Schachter, & Stocke 1999) sur- a r X i v : . [ a s t r o - ph . H E ] A p r A. Paggi et al.veys (see also Perlman 2000). Since then, the X-ray prop-erties of blazars have been deeply investigated by many au-thors (see for example Giommi & Padovani 1994; Padovani& Giommi 1995; Massaro et al. 2011b; Massaro, Paggi, &Cavaliere 2011c). Massaro et al. (2008a) in particular stud-ied
Swift observations of a sample of low and intermediatepeaked BL Lacs, for which the X-ray emission is expectedto lie in the “valley” between the low and high energy spec-tral components, finding these sources to be bright in the X-ray with fluxes above ∼ − erg cm − s − . In addition wenote that ∼
75% of the γ -ray blazars listed in the Second LATAGN Catalog (2LAC, Ackermann et al. 2011) are also X-raysources with fluxes above ∼ − erg cm − s − .However, due to the incompleteness of the current radioand X-ray surveys used for the gamma-ray associations, it isnot always possible to identify a blazar-like counterpart to aUGS .Radio follow up observations of UGSs have been per-formed or are still in progress (e.g., Kovalev 2009a; Kovalevet al. 2009b; Mahony et al. 2010; Petrov et al. 2013). Mas-saro et al. (2013b) recently proposed a method for searching γ -ray blazar-like candidate counterparts of the UGSs basedon the combination of radio observations from WesterborkNorthern Sky Survey (WENSS; Rengelink et al. 1997), thoseof the NRAO Very Large Array Sky survey (NVSS; Condonet al. 1998) and the Very Large Array Faint Images of theRadio Sky at Twenty-Centimeters (FIRST, Becker, White, &Helfand 1995; White et al. 1997).In addition, a procedure to recognize blazar-like candidatecounterparts for UGSs on the basis of their infrared (IR) col-ors have been successfully implemented by D’Abrusco et al.(2012, 2013) and Massaro et al. (2012a, 2013a) making useof the Wide-Field Infrared Survey Explorer (WISE) all-skydata (Cutri et al. 2012a). WISE data also proven to be usefulto address the widely entertained field of mid-infrared AGNselection (Stern et al. 2005, 2012, see also Eckart et al. 2010;Park et al. 2010).Additional attempts have been recently developed to asso-ciate or to characterize the UGSs using pointed Swift observa-tions (e.g., Mirabal 2009; Mirabal & Halpern 2009; Kataokaet al. 2012), and / or with several statistical approaches (e.g.,Mirabal, Nieto, & Pardo 2010; Ackermann et al. 2012).Moreover, in the last two years the Chandra and
Suzaku
X-ray telescopes have been used to investigate the nature of theUGSs (e.g., Fujinaga et al. 2011; Maeda et al. 2011; Mu-rakami et al. 2011; Cheung et al. 2012; Mori et al. 2012).The characterization of X-ray emission from UGSs is ofparticular interest. All γ -ray sources associated in the sec-ond Fermi
LAT (2FGL) catalog have a clear radio counter-part (Nolan et al. 2012) leading to the so called radio- γ -rayconnection in the case of blazars (e.g., Ghirlanda et al. 2010;Ackermann et al. 2011; Massaro et al. 2013b). However thisis not the case for the X-ray sources. It is not clear at themoment if all γ -ray sources feature an X-ray counterpart andtherefore a systematic study of X-ray emission from UGS isuseful to investigate their nature.Motivated by these researches, we investigate the X-ray- γ connection presenting in this paper a catalog of X-ray sourceslying in the positional uncertainty region of all UGSs listed We note that, in the following, we will refer to a source lying into thepositional uncertainty region of a γ -ray source as “candidate counterpart”,while we will use the term “blazar candidate” for the γ -ray source togetherwith its unique blazar-like counterpart. nu m b e r UVOT-U110100 UVOT-W1110100 UVOT-M20 20 40 60 80Exposure time (ks)110100 UWOT-W2
Figure 1.
Histograms of total exposures of the merged observations dis-cussed in Section 2. in 2FGL without any γ -ray analysis flag, making use of allavailable observations performed by Swift
X-ray Telescope(XRT) up to March 31, 2013, and we investigate their multi-wavelength properties.For X-ray sources with a WISE counterpart we then ap-ply the Kernel Density Estimation (KDE) technique to com-pare their IR colors to those of known γ -ray blazars, selecting44 new blazar-like candidate counterparts and 6 γ -ray blazarscandidates as a result.The paper is organized as follows: Section 2 is devoted tothe UGS sample definition while Section 3 describes the maindata reduction procedure adopted for the Swift
XRT and
Swift
UVOT observations. The complete list of X-ray sources thatcould be potential counterpart of UGSs in the 2FGL catalogis presented in Section 4. In Section 5 we illustrate our selec-tion of new γ -ray blazar candidates. In Section 6 we compareour results with di ff erent, previous selections, and Section 7is dedicated to our conclusions. SAMPLE SELECTION
The initial sample considered in our analysis is constitutedby the 299 UGSs in the 2FGL catalog that do not present any γ -ray analysis flag (Nolan et al. 2012).Up to March 31, 2013, 205 of these sources feature at leastone X-ray observation in the Swift master catalog performedin photon counting (PC) mode, and covering the positionaluncertainty region at 95% level of confidence as reported inthe 2FGL. The final sample considered in this analysis istherefore constituted by the above selected 205 sources.The Swift observations have variable exposures, and to de-tect the fainter X-ray objects we merged all the observationscorresponding to each UGSs (see Section 3 for details on thereduction procedures), obtaining the total exposures shown inFigure 1. SWIFT
OBSERVATIONS AND DATA REDUCTION PROCEDURES
Swift has proven to be an excellent multi-frequency ob-servatory for blazar research, so far observing hundreds ofsources (e.g., Moretti et al. 2007, 2012; Dai, Bregman, & Analysis flags in 2FGL identify a number of conditions that can sheddoubt on a source, and they are described in detail in Table 3 of Nolan et al.(2012). http: // heasarc.gsfc.nasa.gov / W3Browse / all / swiftmastr.html nidentified Gamma-ray Sources IV 3Kochanek 2012) and yielding an extremely rich and uniquedatabase of multi-frequency (optical, UV, X-ray), simultane-ous blazar observations. Several papers on samples selectedwith di ff erent criteria have already been published, includ-ing: blazars detected at TeV energies (e.g., Massaro et al.2008b, 2011a,b; Massaro, Paggi, & Cavaliere 2011c), simul-taneous optical-to-X-ray observations of flaring TeV sources(e.g., Perri et al. 2007; Tramacere et al. 2007) as well as theinvestigation of low and high frequency peaked BL Lacs (e.g.,Maselli et al. 2010; Giommi et al. 2012). Swift has also beenused for UV-optical and X-ray follow-up observations of TeVflaring blazars (e.g., Aliu et al. 2011; Aleksi´c et al. 2012;H.E.S.S. Collaboration et al. 2013) and has also been usefulin obtaining photometric redshift constraints for many
Fermi -detected BL Lacs (Rau et al. 2012).Once
Fermi was launched, the
Swift
XRT Survey of
Fermi
Unassociated Sources was started to perform follow-up ob-servations of the UGSs in an attempt to find their potentialX-ray counterparts (PI A. Falcone). In the following sec-tions we analyze all the data collected between the beginningof the follow-up program until March 31, 2013, for the se-lected sample of UGSs described in Section 2.During these observations, Swift operated with all its in-struments in data taking mode. For our analysis, however,we consider only
Swift
XRT (Burrows et al. 2005) and
Swift
UVOT (Roming et al. 2005) data.
Swift XRT data reduction
The XRT data were processed using the XRTDAS software(Capalbi et al. 2005) developed at the ASI Science Data Cen-ter and included in the HEAsoft package (v. 6.13) distributedby HEASARC. For each observation of the sample, calibratedand cleaned PC mode event files were produced with the xrt - pipeline task (ver. 0.12.6), producing exposure maps for eachobservation. In addition to the screening criteria used bythe standard pipeline processing, we applied a further filterto screen background spikes that can occur when the anglebetween the pointing direction of the satellite and the brightEarth limb is low. In order to eliminate this so called brightEarth e ff ect, due to the scattered optical light that usually oc-curs towards the beginning or the end of each orbit, we usedthe procedure proposed by Puccetti et al. (2011) and D’Eliaet al. (2013). We monitored the count rate on the CCD borderand, through the xselect package, we excluded time intervalswhen the count rate in this region exceeded 40 counts / s; more-over, we selected only time intervals with CCD temperaturesless than − ◦ C (instead of the standard limit of − ◦ C) sincecontamination by dark current and hot pixels, which increasethe low energy background, is strongly temperature depen-dent (D’Elia et al. 2013).We then proceeded to merge cleaned event files obtainedwith this procedure using xselect , considering only obser-vations with telescope aim point falling in a circular regionof 12’ radius centered in the median of the individual aimpoints, in order to have a uniform exposure. The correspond-ing merged exposure maps were then generated by summingthe exposure maps of the individual observations with ximage (ver. 4.5.1).
Swift XRT source detection
To detect X-ray sources in the merged XRT images, wemade used of the ximage detection algorithm detect , which lo- http: // / unassociated / cates the point sources using a sliding-cell method. The aver-age background intensity is estimated in several small squareboxes uniformly located within the image. The position andintensity of each detected source are calculated in a box whosesize maximizes the signal-to-noise ratio. The net counts arecorrected for dead times and vignetting using the appropri-ate exposure maps, and for the fraction of source counts thatfall outside the box where the net counts are estimated, us-ing the PSF calibration. Count rate statistical and systematicuncertainties are added quadratically. The algorithm was setto work in bright mode, which is recommended for crowdedfields and fields containing bright sources, since it can recon-struct the centroids of very nearby sources.We also evaluated the net count rates for the detectedsources with the sosta algorithm that, besides the net countrates and the respective uncertainties, yields the statistical sig-nificance of each source. We note that the uncertainties inthe count rates returned by sosta are purely statistical - i.e.do not include systematic errors - and are in general smallerthan those given by detect . sosta also yields slightly di ff er-ent count rates from detect , which are in most cases moreaccurate, because detect uses a global background for the en-tire image, whereas sosta uses a local background. Thus wereport both values in our analysis.The catalog was then cleaned from spurious sources - usu-ally occurring at count rates higher than 0 . − - by vi-sual inspection of all the observations. Finally, we refinedthe source position and relative positional errors by the task xrtcentroid of the XRTDAS package, and considered onlysources falling in a circular region of radius equal to the semi-major axis of the ellipse corresponding to the positional un-certainty region of the Fermi source at 95% level of confi-dence and centered at the 2FGL position of the γ -ray source(consistently with Massaro et al. 2013a). The source des-ignation we adopt for a source with RA HH:MM:SS.s andDEC ± DD:MM:SS is SWXRTJHHMMSS.s ± DDMMSS, asper D’Elia et al. (2013). The results of the detection processare presented in Appendix A in Table 1.
Swift UVOT observations
We note that 203 out of the 205 UGSs that constitute oursample have been also observed in the optical and UV byUVOT. We then produced for each X-ray observation thecorresponding merged UVOT event files adopting standardprocedures . After checking the correct WCS alignment ofour images with USNO-B Catalog (Monet et al. 2003), wemerged them with fappend (part of FTOOLS package ver.6.13) and then merged the images with uvotimsum ; the sameprocedure was applied to produce merged exposure maps.For each X-ray source found with the procedure describedin 3.2, we looked in the corresponding UVOT images for UV-optical counterparts falling in the relative XRT positional er-ror. We performed source photometry using the uvotsource task using the appropriate exposure map. We adopted an aper-ture radius of 5”, independently of the image filter, and tookthe background region in the form of circle with typical radiusof 20” in a source-free region of the sky (e.g., Maselli et al.2013).As a comparison we also evaluated source photometry withthe uvotdetect task, which detects sources in UVOT imagesand extracts their count rates evaluating the background level.In general, we note that although the uvotsource task yields http: // / analysis / uvot / image.php A. Paggi et al. ρ ( a r c m i n - ) LAT unc. reg.XRT FOV Exposure time (s) n Figure 2. (Upper panel) Mean spatial density ρ of X-ray sources detectedinside the LAT positional uncertainty region (blue crosses) and in the whole Swift
XRT field of view (red crosses), as a function of the exposure time. Withcircles of the appropriate color we represent the average values of ρ in bins ofexposure time of 1 ks. (Lower panel) Ratio n of mean spatial density of X-raysources detected in the whole Swift
XRT field of view to mean spatial densityof X-ray sources detected inside the LAT positional uncertainty region, as afunction of the exposure time (black crosses).With black circles we representthe average values of n in bins of exposure time of 1 ks. more accurate results for extended sources, we expect to dealmostly with point-like sources. The results of the detectionprocess are presented in Table 2. Chance coincidence probability
Due to considerable size of the
Fermi
LAT positional un-certainty region (ranging from ∼ (cid:48) to ∼ (cid:48) with an averagesize ∼ (cid:48) ) several UGSs feature more than one X-ray sourcein their uncertainty region. For this reason, we performed foreach UGS listed in Table 1 simulations to evaluate the proba-bility of chance coincidence detections of X-ray sources.As a first step we evaluated the mean spatial density ρ ofX-ray sources detected in the whole Swift
XRT field of viewand inside the LAT positional uncertainty region. In the up-per panel of Figure 2 we present with red and blue crossesrespectively these two densities as a function of the exposuretime, while in the lower panel of the same figure we showwith black crosses the ratio n of these two densities. Despitethe spread, the average values of these quantities evaluated inbins of 1 ks (indicated with circles of the appropriate color)show that for exposure times higher than ∼
20 ks the twomean densities become comparable.The mean spatial densities, however, cannot be used toproperly evaluate the chance coincidence probability, sincethey do not take into account the spatial distribution of the X- ray sources, that is not uniform. In order to properly evaluatethe chance coincidence probability we adopted a method sim-ilar to that presented by D’Abrusco et al. (2013), that consistsin randomly shifting the searching region (in our case, theLAT positional uncertainty region) and evaluate how many X-ray sources fall into this shifted region. For each USG listedin Table 1 we generated 50 random regions of the same sizeof the relative LAT positional uncertainty region (and discon-nected from the latter) in order to cover the whole
Swift
XRTfield of view. We then counted how many of these random re-gions contain a number of X-ray sources equal or higher thanthe number of X-ray sources contained inside the LAT posi-tional uncertainty region, evaluating for each UGS the relativechance coincidence probability that, as shown in Figure 2, de-pends on the source exposure. We then evaluated the averagechance coincidence probability for all our UGS, that is ∼ ∼ (cid:46) THE X-RAY CATALOG OF CANDIDATE COUNTERPARTS FORTHE UNIDENTIFIED GAMMA-RAY SOURCES
Using the procedure described in 3.2, we obtained a catalogof 357 X-ray sources detected with a significance ≥ σ . Inparticular, we have 195 sources detected with a significance ≥ σ , 111 sources with a significance ≥ σ and 80 sourceswith a significance ≥ σ . We found X-ray sources consistentwith the locations of 143 UGSs, with 51 UGSs having a singleX-ray source and 92 UGSs having multiple X-ray sources intheir positional uncertainty region. The remaining 62 UGSs,although overlapping with XRT-PC observations, do not showany X-ray counterpart.In Figure 3 we show for each X-ray source of our catalogthe estimated X-ray flux evaluated with PIMMS × cm − . Fig-ure 3 clearly shows the flux limit for an X-ray source to bedetected with a specific exposure. Exposure time (s) -14 -13 -12 -11 X - r ay f l u x ( er g c m - s - ) Figure 3.
Total exposure for each source of our catalog compared with therespective observed X-ray flux evaluated with PIMMS software for a pow-erlaw spectra with spectral index 2 and an absorption column density of5 × cm − . We note that this model assumption induce an error of ∼ http: // heasarc.nasa.gov / docs / journal / pimms3.html nidentified Gamma-ray Sources IV 5We searched several major radio, IR, optical and UV cata-logs for possible counterparts within the positional errors ob-tained with xrtcentroid to obtain additional information onthe source nature.For the radio catalogs we considered NVSS (N; Condon etal. 1998), Sydney University Molonglo Sky Survey (SUMSS- S; Bock, Large, & Sadler 1999; Mauch et al. 2003), FIRST(F; Becker, White, & Helfand 1995) and WENSS (W; Ren-gelink et al. 1997) surveys. For the IR catalogs, we used theWISE (w; Wright et al. 2010) archival observations togetherwith the Two Micron All Sky Survey (2MASS - M; Skrut-skie et al. 2006) since each WISE source is already associ-ated with the closest 2MASS object by the default catalog(see Cutri & et al. 2012b, for more details), and the UKIRTInfrared Deep Sky Survey (UKIDSS - UK; Lawrence et al.2007) archival observations. For the UV catalog, we used theGalaxy Evolution Explorer (GALEX GR6 - g; Martin et al.2005) archival observations. In addition we searched for opti-cal counterparts, with possible spectra available, in the SloanDigital Sky Survey (SDSS dr9 - s; e.g. Pˆaris et al. 2012) andin the Six-degree-Field Galaxy Redshift Survey (6dFGS - 6;Jones et al. 2004, 2009). Finally, we searched for X-ray cor-respondences in the
Chandra
Source Catalog (CSC - C; e.g.Evans et al. 2010).As anticipated in Section 3.3, we cross- checked XRT-PCobservations with UVOT observations both in UV (u) and op-tical (o) filters. Then, we also considered the NASA Extra-galactic Database (NED) for other multifrequency informa-tion. Finally, we cross correlate our sample with the USNO-BCatalog (U; Monet et al. 2003) to identify the optical coun-terparts of our γ -ray blazar candidates; this is important toprepare and plan future follow up observations (see Table 3).In our catalog of 357 X-ray sources we find the followingcounterparts: 26 in the NVSS catalog, 6 in the SUMSS cata-log, 5 in the FIRST catalog, 2 in the WENSS catalog, 41 in theSDSS catalog (2 with spectral observations), 5 in the 6DFGScatalog, 194 in the USNO-B catalog, 44 in the GALEX cat-alog, 6 in the UKIDSS catalog, 197 in the WISE catalog (94with 2MASS counterpart) and 1 in the CSC catalog. The re-sults of this association procedure are presented in Table 1(column 10).Although a proper counterpart identification would requiremore sophisticated techniques (see for example Brand et al.2006), for the scope of this work we are simply presentinga list of counterparts associations only based on positionalmatch. We note that for the 197 X-ray sources for whichwe find WISE counterparts we only have one multiple match,while for the other catalogs considered here we have 7 mul-tiple matches for SDSS, 1 multiple match for GALEX, and1 multiple match for UKIDSS. When multiple counterpartswere found within the positional error we simply choose thecloser one.We add that we also checked Planck
PCCS (Planck Col-laboration et al. 2013), Catalina CRTS (Drake et al. 2009),ROSAT RASS (Voges et al. 1999), XMM-
Newton
XMM-MASTER (Arviset et al. 2002) and
Suzaku
SUZAMASTER catalogs, finding no correspondences. CANDIDATE γ -RAY-BLAZAR SELECTION Recently, D’Abrusco et al. (2013) proposed a classifica-tion method to identify γ -ray blazar candidates on the basis http: // ned.ipac.caltech.edu / http: // heasarc.gsfc.nasa.gov / W3Browse / all / suzamaster.html [4.6]−[12] (mag) [ . ] − [ . ] ( m ag ) ll ll l llllll l l llll l lll ll l lll ll ll ll ll l ll ll lll llll llll lll l l lll ll ll llll ll ll l ll llll ll ll ll llll llll l lll l ll ll l ll l lll l ll ll l ll ll l l ll ll lll lll lll ll ll lll lllll l ll WISE counterparts to XRT sources (RC)WISE counterparts to XRT sources (WRC)
Figure 4.
Projection of the three-dimensional WISE color space on the two-dimensional [3.4]-[4.6] [4.6]-[12] color-color plane for XRT-PC sources witha WISE counterpart. Black lines represent the two-dimensional densities ofWISE counterparts to know γ -ray blazars evaluated using the KDE tech-nique, with the outermost line indicating the 90% density contour normal-ized to the peak density. Grey circles represent XRT-PC sources without aradio counterpart (WRC), and red circles represent the XRT-PC sources witha radio counterpart (RC). Black dashed lines represent isodensity contours ofgeneric WISE sources (D’Abrusco et al. 2012; Massaro et al. 2012a). Theouter dashed line represent densities ∼ − times the peak density. of their positions in the three-dimensional WISE color space.As a matter of fact, blazars - whose emission is dominatedby beamed, non thermal emission - occupy a defined regionin such a space, well separated from that occupied by othersources in which thermal emission prevails (D’Abrusco et al.2012; Massaro et al. 2012a). This method, however can onlybe applied to WISE sources detected in all 4 WISE bands, i.e.,3.4, 4.6, 12 and 22 µ m.Since 414 out of 610 blazars used by D’Abrusco et al.(2013) are detected in X-rays, we here use the XRT detec-tion as additional information and consider the 148 sourcesin our catalog with WISE counterparts detected only in thefirst 3 WISE, bands; we present their projection on the two-dimensional [3.4]-[4.6] [4.6]-[12] color-color plane in Fig-ure 4. In order to select γ -ray blazar-like candidate counter-parts among these sources, we evaluate the two-dimensionaldensities of known γ -ray blazars using the KDE technique(see, e.g., Richards et al. 2004; D’Abrusco, Longo, & Walton2009; Laurino et al. 2011, and reference therein), and conser-vatively consider as γ -ray blazar-like candidate counterpartsthose sources with WISE colors compatible with the 90%KDE density contour normalized to the peak density. On thesame figure we indicatively show the isodensity contours ofgeneric WISE sources, clearly showing that γ -ray blazars arewell separated on this color-color plane from others sources(see also D’Abrusco et al. 2012; Massaro et al. 2012a).In this way we select 64 blazar-like candidate coun-terparts lying in the uncertainty region of 33 UGSs.In particular, among these 33 UGSs the sources2FGLJ0200.4-4105, 2FGLJ1033.5-5032 2FGLJ1328.5- A. Paggi et al. Figure 5. (left frame) Merged XRT-PC image (0.5-10 keV) of the UGS 2FGLJ0900.9 + µ m WISE image of the same region of right frame, indicating in red the name of the WISE counterparts to X-ray sources. + + γ -ray blazar candidates.We note that Massaro et al. (2013a) applied the classifi-cation method proposed by D’Abrusco et al. (2013) to thesame UGSs sample discussed here, selecting 75 blazar-likeWISE sources (see Sect. 6.1). Among these 75 sources28 have an X-ray counterpart in our catalog, and 26 out ofthese 28 - with the exceptions of SWXRTJ011619.2-615344and SWXRTJ174507.7 + γ -rayblazar-like candidate counterparts with the KDE techniqueproposed here. This is an excellent agreement, consider-ing that the method proposed by D’Abrusco et al. (2013)makes use of a three-dimensional modelization in the Prin-cipal Component space, while the KDE contours in Figure 4represent a two-dimensional source density in the color space(Massaro et al. 2012a). In addition, with the KDE techniquewe also select the source SWXRTJ060102.8 + + γ -ray blazar-like source by Massaro et al. (2013b) on the basisof its low-frequency radio properties (see Sect. 6.1). We so se-lect 37 new γ -ray blazar-like candidate counterparts, markedin Table 1 (column 10) with the “KDE” string, and presenttheir SEDs in Appendix B. COMPARISON WITH PREVIOUS ANALYSES
Gamma-ray blazar candidates
As anticipated in Sect. 5, we compare our results withthose of Massaro et al. (2013a), that applied the classificationmethod proposed by D’Abrusco et al. (2013) to the sameUGSs sample considered in this work, finding 75 blazar-likeWISE candidate counterparts in the
Fermi
LAT positionaluncertainty region of 61 UGSs. Among these UGSs, forthe 35 for which we have available XRT-PC observations we find no X-ray counterparts only for 5 of them. For theother 30 UGSs, Massaro et al. (2013a) find a total of 44blazar-like WISE candidate counterparts, and in our catalogwe find X-ray counterparts to 28 of the latter. These sourcesare marked in Table 1 (column 10) with the “WISE” string,and their SEDs are presented in Appendix B. In particular,among these 30 UGSs the sources 2FGLJ0116.6-6153,2FGLJ0227.7 + + + γ -ray blazarcandidates.We also compare our results with those of Massaro etal. (2013b), that investigate the low-frequency radio prop-erties of blazars and searched for sources with similar ra-dio properties combining the information derived from theWENSS and NVSS surveys, identifying 26 γ -ray blazar-like sources in the Fermi
LAT positional uncertainty re-gions of 21 UGSs. Among these 21 objects, we haveavailable XRT-PC observations for 17 UGSs, and we findno X-ray sources for 3 of them. For the remaining 18UGSs Massaro et al. (2013a) find a total of 20 γ -ray blazar-like sources, and in our catalog we find an X-ray counter-part to 1 of them - WN0557.5 + + + + + γ -ray blazar candidate.We stress that these three methods to identify γ -ray blazar-like sources - namely, the one proposed by D’Abrusco etal. (2013) based on three-dimensional WISE colors space,the one proposed by Massaro et al. (2013b) based on low-nidentified Gamma-ray Sources IV 7 -14 -13 -12 X-ray flux (erg cm -2 s -1 )00.10.20.30.40.50.6 no r m a li ze d c oun t s X-ray sourcesno X-ray sources
Figure 6.
X-ray fluxes reached by XRT-PC observation of the 62 UGSs thatshow no X-ray counterpart falling in the
Fermi
LAT positional uncertaintyregion (orange bars) compared with X-ray fluxes reached in the 143 UGSsthat show at least one X-ray candidate counterpart (black bars). The fluxlimit is estimated with the same spectral model considered in Sect. 4 (seeFigure 3). frequency radio properties, and the KDE technique appliedto the two-dimensional WISE colors space - do not neces-sarily select the same sources (see Tables 5 and 6), nor dothey necessarily select the brighter X-ray candidate counter-part of the UGS. As an example we show in the left frameof Figure 5 the merged XRT-PC image (0.5-10 keV) of theUGS 2FGLJ0900.9 + γ -ray blazar-like source SWXRTJ090121.8 + . ± .
10 10 − ph s − , selected on the ba-sis of the IR colors of its WISE counterpart. However, thebrighter X-ray source detected in the LAT positional uncer-tainty region is SWXRTJ090110.9 + . ± .
10 10 − ph s − is not selected as γ -ray blazar-likesource, as well as SWXRTJ090039.0 + . ± .
53 10 − ph s − , which is the only X-ray sourcein the LAT positional uncertainty region that shows a ra-dio counterpart within the XTR-PC positional error - namelyNVSSJ090038 + + + γ -ray blazar-like source with the KDE technique and has a count rate of1 . ± .
49 10 − ph s − . Sources without counterparts
As anticipated in Section 4, 62 UGSs of our sample (mostof them lying on the Galactic plane), although featuring XRT-PC observations, show no X-ray counterpart. The X-rayfluxes reached by XRT-PC observations of these sources arepresented in Figure 6 in comparison with the X-ray fluxesreached for UGSs that show X-ray candidate counterparts.The flux limit is estimated with the same spectral model con-sidered in Sect. 4 (see Figure 3). We see that the obser- vations of sources that show at least one X-ray candidatecounterpart reach lower fluxes ∼ − erg cm − s − with re-spect to observations of sources that show no X-ray coun-terparts, the latter reaching fluxes ∼ × − erg cm − s − .The two observations, however peak at the same X-ray fluxof ∼ − erg cm − s − . In particular we have 45 UGSsthat, despite a total exposure time > + + + + γ -ray blazar-like candidate counterpart in their positional un-certainty region, as reported by Massaro et al. (2013a) andMassaro et al. (2013b).Moreover, we have 35 UGSs that, in their Fermi
LATpositional uncertainty region, show X-ray candidate coun-terparts in XRT-PC observations, but without lower en-ergy counterparts in either UVOT observations or the cat-alogs we described in Section 4. To take into accountthe astrometric uncertainties of these catalogs, we searchedfor counterpart of these sources using a searching radiusequal to three times the positional error obtained with xrt - centroid , yielding 6 UGS - namely 2FGLJ0239.5 + + + Comparison with 1FGL catalog
We note that among the 299 UGSs analyzed, there are 66sources that were also unidentified according to the investi-gation performed in the first
Fermi γ -ray catalog (1FGL) buthave been classified as active galactic nuclei (AGNs) or aspulsars (PSRs) using two di ff erent statistical approaches: theClassification Tree and the Logistic regression analyses (seeAckermann et al. 2012, and references therein). In particu-lar, 38 out of the 66 show γ -ray properties similar to thoseof others γ -ray AGNs while 11 are potential PSRs with theremaining 17 of unknown origin.For the 49 UGSs classified on the basis of the above statis-tical methods, we performed a comparison with our results inparticular to check if the 2FGL sources having in their uncer-tainty region an X-ray source whose IR counterpart featuresblazar-like WISE colors according to the KDE technique il-lustrated in Sect. 5 were also classified as AGNs according tothe results of Ackermann et al. (2012). We found that 8 out of33 UGSs we associate with a γ -ray blazar-like source are alsoclassified as AGNs, all of them with a probability systemat-ically higher than 60%. There is only one case (i.e., 2FGL1328.5-4728) in which the statistical procedures assigned aPSR classification, with a low probability (i.e., 53%) whilethe KDE method identified the X-ray candidate counterpartof the Fermi source as a blazar-like object. SUMMARY AND CONCLUSIONS
In this work we present a catalog of X-ray sources lying inthe positional uncertainty regions of the 299 UGSs reported inthe 2FGL catalog without any γ -ray analysis flag. To this end,we made use of all available observations performed by Swift
XRT in PC mode up to March 31, 2013, that where avail-able for 205 UGSs. In order to detect the fainter sources, we A. Paggi et al.merged all the observations corresponding to each UGSs, andapplied to these merged observations di ff erent detection algo-rithms (i.e., ximage detect and sosta ). The source list wascleaned from spurious and extended sources by visual inspec-tion of all the observations, to yield a final catalog of 357X-ray sources. We searched several major radio, IR, opticaland UV surveys for any possible counterparts within the posi-tional error of our X-ray sources to obtain additional informa-tion on their nature, providing a comprehensive list of X-raysources with multi-wavelength properties.The main results of our analysis can be summarized as fol-lows: • We find X-ray candidate counterparts for ∼
70% of theUGSs investigated. In particular, we have ∼
25% UGSsfeaturing a single X-ray counterpart and ∼
45% UGSsfeaturing multiple X-ray candidate counterparts fallingin the positional uncertainty region at 95% level of con-fidence. • For each X-ray source we also looked in the corre-sponding UVOT merged images for UV-optical coun-terparts performing sources photometry, and findingUV-optical counterparts to ∼
71% of the X-ray sourcesin our catalog. • We find no X-ray counterparts for 62 UGSs in our sam-ple ( ∼ ≥ • Comparing our results with Massaro et al. (2013a) andMassaro et al. (2013b) we find X-ray candidate coun-terparts to 29 sources classified as γ -ray blazar-like. • Applying the KDE technique to IR colors of WISEcounterparts, we obtain an additional list of 37 γ -rayblazar-like sources for 33 UGSs (29 with a unique can-didate and 4 with a double candidate). In particular,10 out of these 33 2FGL sources have radio counter-parts, and for 4 UGSs out of 33 we add a di ff erent γ -rayblazar-like sources from those selected by Massaro etal. (2013a) and Massaro et al. (2013b). • Among the 51 UGSs that have a single X-ray coun-terpart, 17 have their X-ray counterpart selected as γ -ray blazar-like source with the three methods discussedabove, and are there considered as γ -ray blazar candi-dates. • The source 2FGL1328.5-4728, a γ -ray blazar candidateselected with the KDE technique, is classified as PSRby Ackermann et al. (2012).Even though blazars are expected to be bright in X-rays,the methods discussed here to find γ -ray blazar-like sourcesin UGSs uncertainty regions show that this is not always thecase.We note that 39 2FGL sources in our sample are in commonwith the analysis of 1FLG UGSs by (Takeuchi et al. 2013).Comparing our results with Ackermann et al. (2012) we notethat 38 2FGL sources in our sample are classified as AGN the1FGL catalog with high level of confidence, 11 2FGL sourcesin our sample are classified as PSR with low level of confi-dence, and 17 2FGL sources in our sample are unclassified.In particular, 8 2FGL sources with a γ -ray blazar-like sourceselected with the KDE technique are classified as AGN byAckermann et al. (2012). Ground-based, optical and near IR, spectroscopic followup observations will be planned for the Swift
XRT sourcesselected as γ -ray blazar-like candidate counterparts becausethey are crucial to confirm the nature of the selected sourcesand to obtain their redshift, as shown for the unidentified IN-TEGRAL and Swift sources (e.g., Masetti et al. 2012; Parisiet al. 2012, and references therein).We acknowledge useful comments and suggestions by ouranonymous referee. The authors gratefully acknowledge A.Falcone for the
Swift
XRT Survey of
Fermi
UnassociatedSources that produced most of the observations used in thiswork. F. Massaro is grateful to M. Ajello for his support.The work is supported by the NASA grants NNX12AO97G.R. D’Abrusco gratefully acknowledges the financial supportof the US Virtual Astronomical Observatory, which is spon-sored by the National Science Foundation and the NationalAeronautics and Space Administration. H. A. Smith acknowl-edges partial support from NASA / JPL grant RSA 1369566.The work by G. Tosti is supported by the ASI / INAF con-tract I / / /
0. TOPCAT (Taylor 2005) for the prepara-tion and manipulation of the tabular data and the images. TheWENSS project was a collaboration between the NetherlandsFoundation for Research in Astronomy and the Leiden Ob-servatory. We acknowledge the WENSS team consisted ofGer de Bruyn, Yuan Tang, Roeland Rengelink, George Mi-ley, Huub Rottgering, Malcolm Bremer, Martin Bremer, WimBrouw, Ernst Raimond and David Fullagar for the extensivework aimed at producing the WENSS catalog. Part of thiswork is based on archival data, software or on-line servicesprovided by the ASI Science Data Center. This research hasmade use of data obtained from the High Energy AstrophysicsScience Archive Research Center (HEASARC) provided byNASA’s Goddard Space Flight Center; the SIMBAD databaseoperated at CDS, Strasbourg, France; the NASA / IPAC Extra-galactic Database (NED) operated by the Jet Propulsion Labo-ratory, California Institute of Technology, under contract withthe National Aeronautics and Space Administration. Thisresearch has made use of software provided by the Chan-dra X-ray Center (CXC) in the application packages CIAO,ChIPS, and Sherpa. Part of this work is based on the NVSS(NRAO VLA Sky Survey); The National Radio AstronomyObservatory is operated by Associated Universities, Inc., un-der contract with the National Science Foundation. This pub-lication makes use of data products from the Two MicronAll Sky Survey, which is a joint project of the University ofMassachusetts and the Infrared Processing and Analysis Cen-ter / California Institute of Technology, funded by the NationalAeronautics and Space Administration and the National Sci-ence Foundation. This publication makes use of data productsfrom the Wide-field Infrared Survey Explorer, which is a jointproject of the University of California, Los Angeles, and theJet Propulsion Laboratory / California Institute of Technology,funded by the National Aeronautics and Space Administra-tion. Funding for the SDSS and SDSS-II has been provided bythe Alfred P. Sloan Foundation, the Participating Institutions,the National Science Foundation, the U.S. Department of En-ergy, the National Aeronautics and Space Administration, theJapanese Monbukagakusho, the Max Planck Society, and the http: // / mbt / topcat / nidentified Gamma-ray Sources IV 9Higher Education Funding Council for England. The SDSSWeb Site is http: // / . The SDSS is managed bythe Astrophysical Research Consortium for the ParticipatingInstitutions. The Participating Institutions are the AmericanMuseum of Natural History, Astrophysical Institute Potsdam,University of Basel, University of Cambridge, Case WesternReserve University, University of Chicago, Drexel University,Fermilab, the Institute for Advanced Study, the Japan Partic-ipation Group, Johns Hopkins University, the Joint Institutefor Nuclear Astrophysics, the Kavli Institute for Particle As-trophysics and Cosmology, the Korean Scientist Group, theChinese Academy of Sciences (LAMOST), Los Alamos Na-tional Laboratory, the Max-Planck-Institute for Astronomy(MPIA), the Max-Planck-Institute for Astrophysics (MPA),New Mexico State University, Ohio State University, Univer-sity of Pittsburgh, University of Portsmouth, Princeton Uni-versity, the United States Naval Observatory, and the Uni-versity of Washington. The United Kingdom Infrared Tele-scope is operated by the Joint Astronomy Centre on behalfof the Science and Technology Facilities Council of the U.K.The CSS survey is funded by the National Aeronautics andSpace Administration under Grant No. NNG05GF22G is-sued through the Science Mission Directorate Near-Earth Ob-jects Observations Program. The CRTS survey is supportedby the U.S. National Science Foundation under grants AST-0909182. REFERENCESAbdo A. A., et al., 2009, APh, 32, 193Abdo A. A., et al., 2010a, ApJ, 712, 1209Abdo A. A., et al., 2010b, ApJ, 712, 957Abdo A. A., et al., 2010c, ApJS, 188, 405Ackermann M., et al., 2011, ApJ, 743, 171Ackermann M., et al., 2012, ApJ, 753, 83Ackermann M., et al., 2012, ApJ, 753, 83Ackermann M., et al., 2013, Sci, 339, 807Arnaud K. A., 1996, ASPC, 101, 17Aleksi´c J., et al., 2012, A&A, 544, A142Aliu E., et al., 2011, ApJ, 742, 127Arviset C., Guainazzi M., Hernandez J., Dowson J., Osuna P., Venet A.,2002, astro, arXiv:astro-ph / // heasarc.nasa.gov / docs / swift / analysis / xrt swguide v1 2.pdfCardelli J. A., Clayton G. C., Mathis J. S., 1989, ApJ, 345, 245Casandjian J.-M., Grenier I. A., 2008, A&A, 489, 849Casanova S., et al., 2010, PASJ, 62, 769Cheung C. C., Donato D., Gehrels N., Sokolovsky K. V., Giroletti M., 2012,ApJ, 756, 33Condon J. J., Cotton W. D., Greisen E. W., Yin Q. F., Perley R. A., TaylorG. B., Broderick J. J., 1998, AJ, 115, 1693Cutri R. M., et al., 2012a, wise.rept, 1Cutri R. M., et al., 2012b, yCat, 2311, 0Dai X., Bregman J. N., Kochanek C. S., 2012, AAS, 219, Perri M., et al., 2007, A&A, 462, 889Petrov L., Mahony E. K., Edwards P. G., Sadler E. M., Schinzel F. K.,McConnell D., 2013, MNRAS, 432, 1294Puccetti S., et al., 2011, A&A, 528, A122Rau A., et al., 2012, A&A, 538, A26Reimer O., 2001, ASSL, 267, 17Rengelink R. B., Tang Y., de Bruyn A. G., Miley G. K., Bremer M. N.,Roettgering H. J. A., Bremer M. A. R., 1997, A&AS, 124, 259Richards G. T., et al., 2004, ApJS, 155, 257Roming P. W. A., et al., 2005, SSRv, 120, 95Skrutskie M. F., et al., 2006, AJ, 131, 1163Sowards-Emmerd D., Romani R. W., Michelson P. F., 2003, ApJ, 590, 109Stern D., et al., 2005, ApJ, 631, 163 Stern D., et al., 2012, ApJ, 753, 30Su M., Finkbeiner D. P., 2012, arXiv, arXiv:1207.7060Takeuchi, Y., Kataoka, J., Maeda, K., et al. 2013, arXiv:1307.5581Taylor M. B., 2005, ASPC, 347, 29Thompson D. J., 2008, RPPh, 71, 116901Tramacere A., et al., 2007, A&A, 467, 501Vandenbroucke J., et al., 2010, ApJ, 718, L166Voges W., et al., 1999, A&A, 349, 389White R. L., Becker R. H., Helfand D. J., Gregg M. D., 1997, ApJ, 475, 479Wright E. L., et al., 2010, AJ, 140, 1868Zechlin H.-S., Fernandes M. V., Els¨asser D., Horns D., 2012, A&A, 538,A93 nidentified Gamma-ray Sources IV 11
APPENDIX A. CATALOG TABLES
Here we present the catalog of X-ray sources with their main properties.In Table 1 we list all the X-ray sources found in XRT-PC observations in the positional uncertainty region of each UGS. Thecolumns contain the following information: (1) NAME XRT: source designation as described in Section 3 and corresponding2FGL UGS; (2) OTHER NAME: name of the counterpart found in the catalogs described in Section 4. If more than onecounterpart is found, the order we choose for the alternate name is the following: NVSS, FIRST, SUMSS, WENSS, WISE,SDSS, 6DFGS, NED; (3) RA: right ascension as given by xrtcentroid ; (4) DEC: declination as given by xrtcentroid ; (5) ERR:positional error in arcseconds as given by xrtcentroid ; (6) EXP: XRT-PC total exposure in seconds; (7) COUNTRATE: countrateand relative error as given by detect in 10 − ph s − ; (8) SIGN: signal to noise threshold above which the source is detected by detect ; (9) SOSTA: countrate and relative error as given by sosta in 10 − ph s − ; (10) SNR: signal to noise ratio as given by sosta ;(11) NOTES: results of the cross-matching with the catalogs discussed in Section 4 within the positional error reported in columnERR: NVSS = N, FIRST = F, SUMSS = S, WENSS = W, WISE = w, 2MASS = M, UKIDSS = UK, SDSS = s, 6 = = g,UVOT(optical filter) = o, UVOT(UV filter) = u, USNO-B = U, CSC = C; (12) CAND: γ -ray blazar-like sources according to Massaroet al. (2013a) (WISE), to Massaro et al. (2013b) (WENSS) and to the KDE technique as discussed in Section 6.1; (12) REDSHIFT:redshift for the source counterpart as reported by SDSS, 6DFGS or NED.In Table 2 we list, for each source in Table 1, the properties of the UV-optical counterpart found in merged UVOT observations.The columns contain the following information: (1) NAME XRT: source designation as described in Section 3 and corresponding2FGL UGS; (2) RA: right ascension of the UVOT counterpart; (3) DEC: declination UVOT counterpart; (4) SEP: angularseparation in arcseconds between the XRT-PC source and the UVOT counterpart; (5) E(B-V): galactic extinction value as derivedby the Infrared Science Archive (IRSA); (6) EXPV: exposure of the UVOT-V filter merged observation in seconds; (7) MAGV:UVOT-V filter magnitude (Vega system) and relative error as given by uvotsource (not corrected by galactic extinction). Upperlimits are indicated with 0.00 errors, while * indicate filter saturation; (8) MAGVS: UVOT-V filter magnitude (Vega system) andrelative error as given by uvotdetect (not corrected by galactic extinction). Upper limits are indicated with 0.00 errors, while *indicate filter saturation; (9) EXPB, (10) MAGB, (11) MAGBS: same as columns (6), (7) and (8) but for UVOT-B filter; (12)EXPU, (13) MAGU, (14) MAGUS: same as columns (6), (7) and (8) but for UVOT-U filter; (15) EXPW1, (16) MAGW1, (17)MAGW1S: same as columns (6), (7) and (8) but for UVOT-W1 filter; (18) EXPM2, (19) MAGM2, (20) MAGM2S: same ascolumns (6), (7) and (8) but for UVOT-M2 filter; (21) EXPW2, (22) MAGW2, (23) MAGW2S: same as columns (6), (7) and (8)but for UVOT-W2 filter.In Table 3 we list all the XRT-PC sources that features a USNO-B counterpart within the positional error and present themagnitudes of this counterpart. The columns contain the following information: (1) NAME XRT: source designation as describedin Section 3; (2) B1: first epoch blue magnitude; (3) B2: second epoch blue magnitude; (4) R1: first epoch red magnitude; (5)R2: second epoch red magnitude; (6) I: second epoch near-IR magnitude.In Table 4 we list all UGS that, although featuring XRT-PC observations, show no X-ray counterpart. The columns contain thefollowing information: (1) NAME 2FGL: UGS name as reported in the 2FGL, with boldface indicating those sources that have a γ -ray blazar-like candidate counterpart in their positional uncertainty region as reported by Massaro et al. (2013a) and Massaroet al. (2013b); (2) EXP: XRT-PC total exposure in seconds. http: // irsa.ipac.caltech.edu / applications / DUST / T a b l e S a m p l eca t a l ogo f X R T - P C d e t ec t e d s ou r ce s i n t h e po s iti on a l un ce r t a i n t y r e g i ono f eac h UG S a s r e po r t e d i n t h e F G L . C o l u m nd e s c r i p ti on i s g i v e n i n A pp e nd i x A . NA M E X R T O T H E R NA M E R AD E C E RR E X P C OUN T R A TE S I GN S O S T A S N R NO TE S C AND R E D S H I F T J J a r c s ec s − ph s − − ph s − F G L J . + S W X R T J . + W I S E J . + . : : . + : : . . . ( . ) . ( . ) . w , M , UK , s , g , o , u , U S W X R T J . + S D SS J . + . : : . + : : . . . ( . ) . ( . ) . UK , s , o , u S W X R T J . + F I R S T J . + : : . + : : . . . ( . ) . ( . ) . F , w , UK , s , o , u , UKD E F G L J . + S W X R T J . + W I S E J . + . : : . + : : . . . ( . ) . ( . ) . w , M , o , U W I S E S W X R T J . + : : . + : : . . . ( . ) . ( . ) . S W X R T J . + : : . + : : . . . ( . ) . ( . ) . nidentified Gamma-ray Sources IV 13 T a b l e S a m p l e o f UVO T c oun t e r p a r t s t o t h e X R T - P C d e t ec t e d s ou r ce s w it h t h e i r pho t o m e t r i c p r op e r ti e s . C o l u m nd e s c r i p ti on i s g i v e n i n A pp e nd i x A . NA M E X R T R AD E C S E P E ( B - V ) E X P V M AGV M AGV S E X P B M AG B M AG B S E X P U M AGU M AGU S E X P W M AG W M AG W S E X P M M AG M M AG M S E X P W M AG W M AG W S J J a r c s ec m a g s m a g m a g s m a g m a g s m a g m a g s m a g m a g s m a g m a g s m a g m a g 2 F G L J . + S W X R T J . + : : . + : : . . . ------ . ( . ) . ( . ) . ( . ) . ( . ) . ( . )---- S W X R T J . + : : . + : : . . . ------ . ( . )- . ( . )- . ( . ) . ( . )--- S W X R T J . + : : . + : : . . . ------ . ( . ) . ( . ) . ( . ) . ( . ) . ( . ) . ( . )--- F G L J . + S W X R T J . + : : . + : : . . . ------ . ( . ) . ( . )--------- S W X R T J . + : : . + : : . . . ------ . ( . )---------- S W X R T J . + : : . + : : . . . ------ . ( . )---------- Table 3
Sample of XRT-PC sources feautring a USNO-B counterpart within the positional error. Column description is given in Appendix A.NAME XRT B1 B2 R1 R2 ISWXRTJ003054.8 + + B. BLAZAR-LIKE SOURCES SPECTRAL ENERGY DISTRIBUTIONS
Here we present the XRT-PC counterparts of γ -ray blazar-like sources, with their SEDs.In Table 5 we list the 30 XRT-PC counterparts of γ -ray blazar-like sources according to the classification methods proposed byD’Abrusco et al. (2013) and Massaro et al. (2013b). In boldface we indicate sources with radio counterparts within the positionalerror listed in Table 1. Columns contain the following information: (1) NAME 2FGL: UGS name as reported in the 2FGL;(2) NAME XRT: source designation as described in Section 3; (3) ALT NAME: name of the WISE counterpart (as reported byWISE All-Sky data catalog, Cutri & et al. 2012b) or of the WENSS counterpart (as reported by WENSS catalog, Rengelink et al.1997) closer to the XRT-PC coordinates (as reported in Table 1); (4) CLASS: for γ -ray blazar-like sources selected by D’Abruscoet al. (2013), every source is assigned to class A, B, or C depending on the probability of the WISE source to be compatiblewith the model of the WISE Fermi
Blazar (WFB) locus: class A sources are considered the most probable candidate blazarsfor the high-energy source, while class B and class C sources are less compatible with the WFB locus but are still deemed ascandidate blazars. For γ -ray blazar-like sources selected by Massaro et al. (2013b), with A we indicate radio sources having − . ≤ α ≤ .
55 and with B those with 0 . ≤ α ≤ .
65, where α is the radio spectral index between 325 MHz and1.4 GHz; (4) TYPE: classification of the candidate blazar according to D’Abrusco et al. (2013) based on the compatibility of theWISE source with the regions of the WFB locus model. BZB and BZQ indicate the regions dominated by BL Lac objects andFSRQs sources respectively, while MIXED indicate the region where the population is mixed in terms of spectral classes.In Table 6 we list the 44 XRT-PC counterparts of γ -ray blazar-like sources according to thee KDE technique illustrated in Sect.5. In boldface we indicate sources with radio counterparts within the positional error listed in Table 1. Columns contain thefollowing information: (1) NAME 2FGL: UGS name as reported in the 2FGL; (2) NAME XRT: source designation as describedin Section 3; (3) WISE NAME: name of the WISE counterpart (as reported by WISE All-Sky data catalog).SEDs of the sources listed in Table 5 are presented in Figures 7 and 8 for sources that feature and do not feature a radiocounterpart within the XRT positional error, respectively. In the same way, SEDs of the sources listed in Table 6 are presented inFigures 9 and 10 for sources that feature and do not feature a radio counterpart within the XRT positional error, respectively. Foreach XRT-PC source we show the spectral points corresponding to the various counterparts we found in the XRT-PC positionalerror as reported in Table 1 (see Section 4). Circles represent detections, while down triangles represent upper limits, with thecolor code presented in the legends. For IR, optical and UV points we present both observed (empty symbols) and de-reddened(full symbols) fluxes, the latter obtained using the extinction law presented by Cardelli, Clayton, & Mathis (1989) and the galacticextinction value as derived by IRSA. When possible, XRT-PC spectra are obtained form events extracted with xrtproducts taskusing a 20 pixel radius circle centered on the coordinates reported in Table 1 and background estimated from a nearby source-free circular region of 20 pixel radius. When the source count rate is above 0.5 counts s − , the data are significantly a ff ectedby pileup in the inner part of the point-spread function (Moretti et al. 2005). To remove the pile-up contamination, we extractonly events contained in an annular region centered on the source (e.g., Perri et al. 2007). The inner radius of the region wasdetermined by comparing the observed profiles with the analytical model derived by Moretti et al. (2005) and typically has a 4or 5 pixels radius, while the outer radius is 20 pixels for each observation. Source spectra are binned to ensure a minimum of 20counts per bin in order to ensure the validity of χ statistics. We performed our spectral analysis with the S herpa modeling andfitting application (Freeman, Doe, & Siemiginowska 2001) include in the CIAO (Fruscione et al. 2006) 4.5 software package,and with the xspec software package, version 12.8.0 (Arnaud 1996) with identical results. For the spectral fitting we used a modelcomprising an absorption component fixed to the Galactic value (Kalberla et al. 2005) and a powerlaw, and we plot intrinsicfluxes (i.e., without Galactic photoelectric absorption). When the extracted counts are not enough to provide acceptable spectralfits we simply converted the count rates reported in Table 1 to 0.3-10 keV intrinsic fluxes with PIMMS 4.6b software, assuminga powerlaw spectra with spectral index 2 and an absorption component fixed to the Galactic value. In this case we report with afilled circle the flux corresponding to the countrate as obtained with detect and with an empty box the countrate as obtained with sosta . http: // cxc.harvard.edu / sherpa nidentified Gamma-ray Sources IV 15 Table 4
UGSs without XRT-PC counterparts in the
Fermi
LAT positional uncertainty region. In boldface we indicate those sources that have a γ -ray blazar-likecandidate counterpart in their uncertainty region as reported by Massaro et al. (2013a) and Massaro et al. (2013b).NAME 2FGL EXPs + + + + + + + + + + + + + + + + + + + + + + + + + + + + Table 5
XRT-PC counterparts to γ -ray blazar-like sources selected according to D’Abrusco et al. (2013) and Massaro et al. (2013b). In boldface we indicate sourceswith radio counterparts within the positional error listed in Table 1. Column description is given in Appendix B. NAME 2FGL NAME XRT ALT NAME CLASS TYPE2FGLJ0039.1 + + + SWXRTJ011619.2-615344
WISEJ011619.59-615343.5 C BZB2FGLJ0133.4-4408
SWXRTJ013306.3-441423
WISEJ013306.35-441421.3 C BZBSWXRTJ013321.5-441319 WISEJ013321.36-441319.4 C BZQ2FGLJ0143.6-5844 SWXRTJ014347.1-584551 WISEJ014347.39-584551.3 C BZB2FGLJ0227.7 + SWXRTJ022744.0 + WISEJ022744.35 + SWXRTJ040946.5-040002
WISEJ040946.57-040003.4 B BZB2FGLJ0414.9-0855 SWXRTJ041457.1-085654 WISEJ041457.01-085652.0 C MIXED2FGLJ0600.9 + SWXRTJ060102.8 + WN0557.5 + + + + + SWXRTJ072355.1 + WISEJ072354.83 + SWXRTJ074627.1-022551
WISEJ074627.03-022549.3 C BZB2FGLJ0756.3-6433 SWXRTJ075624.1-643031 WISEJ075624.60-643030.6 C BZB2FGLJ0838.8-2828 SWXRTJ083842.4-282831 WISEJ083842.77-282830.9 C MIXED2FGLJ0900.9 + + + + + + SWXRTJ125422.8-220414
WISEJ125422.47-220413.6 C BZB2FGLJ1347.0-2956 SWXRTJ134707.1-295844 WISEJ134706.89-295842.3 C BZB2FGLJ1614.8 + SWXRTJ161541.3 + WISEJ161541.22 + + + + + SWXRTJ174507.7 + WISEJ174507.82 + + + + SWXRTJ202155.7 + WISEJ202155.45 + nidentified Gamma-ray Sources IV 17 Figure 7.
Sample SEDs of γ -ray blazar-like sources listed in Table 5 that have a radio counterpart within their XRT positional error. Symbol description is givenin Appendix B. SWXRTJ011619.2−615344 n ( Hz ) n F n ( e r g c m - s - ) - - - - llllllllllllllll llll lll l llllllll SUMSSWISE2MASSUVOTGALEXXRTLAT
SWXRTJ013306.3−441423 n ( Hz ) n F n ( e r g c m - s - ) - - - - llllllllllllllllll llll lll l llllllll SUMSSWISE2MASSUVOTGALEXXRTLAT
Figure 8.
Sample SEDs of γ -ray blazar-like sources listed in Table 5 without a radio counterpart within their XRT positional error. Symbol description is givenin Appendix B. SWXRTJ003858.3+432947 n ( Hz ) n F n ( e r g c m - s - ) - - - - llllllllll llll ll l llllll WISE2MASSUVOTXRTLAT
SWXRTJ013321.5−441319 n ( Hz ) n F n ( e r g c m - s - ) - - - - llllllllllll llll ll l llllll WISEUVOTGALEXXRTLAT
Table 6
XRT-PC counterparts to γ -ray blazar-like sources selected with KDE technique. In boldface we indicate sources with radio counterparts within the positionalerror listed in Table 1. Column description is given in Appendix B. NAME 2FGL NAME XRT WISE NAME2FGLJ0031.0 + SWXRTJ003119.9 + WISEJ003119.70 + + + + + SWXRTJ022051.5 + WISEJ022051.24 + + SWXRTJ035309.5 + WISEJ035309.54 + SWXRTJ042025.5-374445
WISEJ042025.09-374445.02FGLJ0427.2-6705 SWXRTJ042646.3-665954 WISEJ042646.88-665955.82FGLJ0540.1-7554 SWXRTJ054112.1-760249 WISEJ054111.58-760246.12FGLJ0737.1-3235 SWXRTJ073739.2-323255 WISEJ073738.91-323256.22FGLJ0737.5-8246
SWXRTJ073706.3-824836
WISEJ073706.06-824840.22FGLJ0745.5 + + + + + + + + + SWXRTJ101321.4 + WISEJ101321.17 + SWXRTJ132840.4-472749
WISEJ132840.61-472749.22FGLJ1517.2 + + + + + + + + + + + + + + + + SWXRTJ211521.9 + WISEJ211522.00 + + + + SWXRTJ222830.4-163643
WISEJ222830.19-163642.82FGLJ2246.3 + SWXRTJ224604.9 + WISEJ224604.98 + nidentified Gamma-ray Sources IV 19 Figure 9.
Sample SEDs of γ -ray blazar-like sources listed in Table 6 that have a radio counterpart within their XRT positional error. Symbol description is givenin Appendix B. SWXRTJ003119.9+072452 n ( Hz ) n F n ( e r g c m - s - ) - - - - llllllllllllllllllllllllll llll lll l lllllllllll FIRSTWISEUKIDSSSDSSUVOTXRTLAT
SWXRTJ022051.5+250930 n ( Hz ) n F n ( e r g c m - s - ) - - - - llllll llll lll l l lllllll NVSSWISEUVOTGALEXXRTLAT
Figure 10.
Sample SEDs of γ -ray blazar-like sources listed in Table 6 without a radio counterpart within their XRT positional error. Symbol description is givenin Appendix B. SWXRTJ004800.6−634956 n ( Hz ) n F n ( e r g c m - s - ) - - - - ll llll ll l lllll WISEUVOTXRTLAT
SWXRTJ010414.0+132427 n ( Hz ) n F n ( e r g c m - s - ) - - - - lllllllllllll lll ll l llllllllllllllllllll lll ll l lllllll