Multi-wavelength observations of 3FGL J2039.6-5618: a candidate redback millisecond pulsar
D. Salvetti, R. P. Mignani, A. De Luca, C. Delvaux, C. Pallanca, A. Belfiore, M. Marelli, A. A. Breeveld, J. Greiner, W. Becker, D. Pizzoccaro
aa r X i v : . [ a s t r o - ph . H E ] O c t Multi-wavelength observations of 3FGL J2039.6 − D. Salvetti , R. P. Mignani , , A. De Luca , , C. Delvaux , C. Pallanca , A. Belfiore , M.Marelli , A. A. Breeveld , J. Greiner , W. Becker , D. Pizzocaro , INAF - Istituto di Astrofisica Spaziale e Fisica Cosmica Milano, via E. Bassini 15,20133, Milano, Italy Janusz Gil Institute of Astronomy, University of Zielona G´ora, Lubuska 2, 65-265,Zielona G´ora, Poland Istituto Nazionale di Fisica Nucleare, Sezione di Pavia, Via Bassi 6, I-27100 Pavia, Italy Max-Planck Institut f¨ur Extraterrestrische Physik, Giessenbachstrasse 1, 85741 Garchingbei M¨unchen, Germany Dipartimento di Fisica e Astronomia, Universit`a degli Studi di Bologna, Viale BertiPichat 6-2, I-40127, Bologna, Italy Mullard Space Science Laboratory, University College London, Holmbury St. Mary,Dorking, Surrey, RH5 6NT, UK Universit´a degli Studi dell’Insubria, Via Ravasi 2, 21100 Varese, Italy
ABSTRACT
We present multi-wavelength observations of the unassociated γ -ray source3FGL J2039.6 − Fermi
Large Area Telescope. The source γ -ray properties suggest that it is a pulsar, most likely a millisecond pulsar, forwhich neither radio nor γ -ray pulsations have been detected yet. We observed3FGL J2039.6 − XMM-Newton and discovered several candidate X-raycounterparts within/close to the γ -ray error box. The brightest of these X-raysources is variable with a period of 0.2245 ± X = 1 . ± .
09, and hydrogencolumn density N H < × cm − , which gives an unabsorbed 0.3–10 keV X-ray flux of 1 . × − erg cm − s − . Observations with the Gamma-Ray BurstOptical/Near-Infrared Detector (GROND) discovered an optical counterpart tothis X-ray source, with a time-averaged magnitude g ′ ∼ .
5. The counterpartfeatures a flux modulation with a period of 0.22748 ± − Subject headings:
X-ray pulsars;
Fermi pulsars;
1. Introduction
The launch of the
Fermi
Gamma-ray Space Telescope in June 2008 marked a new era in γ -ray astronomy, thanks to the unprecedented performance of its Large Area Telescope (LAT;Atwood et al. 2009). The recently released Third Fermi -LAT γ -ray source catalogue (3FGL;Acero et al. 2015) derived from the first 4 years of observations contains 3033 sources. About70% of these sources have been either directly identified, e.g. either from the detection of γ -ray pulsations (pulsars) or correlated γ -ray and optical/radio variability (Active GalacticNuclei, Novae, X-ray binaries), or associated with objects that are either known or potential γ -ray emitters. The remaining 30% of the 3FGL sources have not been associated with anyobject yet, hence they are referred to as unassociated , and their nature is unknown.Being pulsars the largest family of γ -ray sources identified in the Galaxy ( ∼
160 andcounting ), a significant fraction of the unassociated Fermi -LAT sources might be γ -raypulsars. Some of them might have no or extremely faint radio emission, and escaped detectionin all radio pulsar surveys so far. Indeed, many of such radio-quiet (RQ), or radio faint (RF), γ -ray pulsars have been discovered through blind periodicity searches in the γ -ray data (e.g.,Abdo et al. 2009) thanks to the use of novel search techniques (e.g., Atwood et al. 2006;Pletsch et al. 2013). About 45% of the γ -ray pulsars discovered by the Fermi -LAT aremilli-second pulsars (MSPs). Interestingly, the vast majority of these MSPs ( ∼ . M ⊙ . M C . . M ⊙ , whereas others have a usually non-degenerate companion (a latemain sequence star or a brown dwarf) which is ablated by irradiation from the pulsar wind. https://confluence.slac.stanford.edu/display/GLAMCOG/Public+List+of+LAT-Detected+Gamma-Ray+Pulsars M C . . M ⊙ almost fully ablated by the pulsar wind, and the redback (RB) MSPs, wherethe companion is only partially ablated and has an higher mass of M C ∼ . . M ⊙ (Roberts2013).The use of automatic classification codes (e.g., Ackermann et al. 2012; Lee et al. 2012;Mirabal et al. 2012) based on the γ -ray characteristics is crucial to single out pulsar candi-dates among the many unassociated Fermi -LAT sources and optimise a systematic searchfor new γ -ray pulsars. Since γ -ray pulsars are also identified in the optical and X rays (e.g.,Abdo et al. 2013), multi-wavelength follow-ups of unidentified Fermi -LAT sources are stillkey to confirm the proposed pulsar classifications, though. In particular, optical observationsare an important aid in the search for binary MSPs, for which blind periodicity searches in γ rays must account for the unknown orbital parameters, requiring a massive use of super-computing power facilities (Pletsch & Clark 2014). Indeed, optical observations yielded theidentification of the two Fermi -LAT sources 2FGL J2339.7 − − γ -ray pulsations (Ray et al. in prepara-tion; Pletsch et al. 2012; Ray et al. 2013) through the discovery of orbital modulations in theflux of their companion stars (Romani & Shaw 2011; Kong et al. 2012; Romani et al. 2012;Kataoka et al. 2012). In a similar way, new binary MSP candidates have been identified forthe two unassociated Fermi -LAT sources 2FGL J1653.6 − − Fermi -LAT sources, we studied 3FGL J2039.6 − γ -ray source(detection significance ∼ σ ) that was listed in both the First (Abdo et al. 2010) and Second(Nolan et al. 2012) Fermi -LAT γ -ray source catalogues (a.k.a. 1FGL J2039.4 − − − Swift /X-Ray Telescope(XRT) during snapshot observations (1 and 3.6 ks exposure times) but no candidate X-raycounterpart was detected within the 2FGL γ -ray source error circle (Takeuchi et al. 2013).In radio, no potential counterpart was found in the Sydney University Molonglo Sky Survey(SUMMS) source catalogue (Mauch et al. 2003) and in dedicated observations of unassociated2FGL sources with the Australia Telescope Compact Array (Petrov et al. 2013) and theParkes radio telescope (Camilo et al. 2015). At very high energies, 3FGL J2039.6 − . Based upon its γ -ray characteristics, 3FGL http://tevcat.uchicago.edu/ − − ∼
80% of the MSPs detected by the
Fermi -LAT are in binary systems, one can expectthat would be a binary MSP.We investigated this scenario through a multi-wavelength observation campaign (X rays,ultraviolet, optical, infrared) of 3FGL J2039.6 −
2. Observations and Data Reduction2.1. Target selection
Recently, we developed an advanced classification code (Salvetti et al. 2013) that canrecognise different classes of γ -ray pulsars, e.g. young/middle-aged pulsars and MSPs. Thiscode uses a statistical predictive method based on Artificial Neural Network (ANN) tech-niques to quantify the probability of a given source to be MSP-like on the basis of its γ -raytemporal and spectral characteristics. The method is based on an advanced hierarchicalANN architecture consisting of simple neural networks applied in sequence to discriminatepulsar-like from AGN-like objects in first place and disentangle MSPs from young/middleage pulsars in second place. Such a method correctly classifies 84% of the identified MSPs,while the false positive fraction is lower than 10% (Salvetti et al., in preparation). We thenapplied the optimized hierarchical neural network to all unassociated − We carried out an
XMM-Newton observation of the 3FGL J2039.6 − γ -ray error box(Programme ID: 0720750301), which started on 2013 October 10 at 09:43:18 UT (revolution2534) and lasted 44.6 ks. The pn detector (Struder et al. 2001) of the European PhotonImaging Camera (EPIC) instrument was operated in Extended Full Frame mode, with a 5 –time resolution of 200 ms over a 26 ′ × ′ Field-of-View (FoV), while the Metal Oxide Semi-conductor (MOS) detectors (Turner et al. 2001) were set in Full Frame mode (2.6 s timeresolution on a 15 ′ radius FoV). The thin optical filter was used for the pn while a mediumone was used for the MOS cameras. We retrieved the Observation Data Files (ODF) from the XMM-Newton
Science Archive and used the most recent release of the XMM-Newton
ScienceAnalysis Software (SAS) v14.0 to analyze them. We performed a standard data processing,using the epproc and emproc tools, and screening for high particle background time intervals(e.g., De Luca et al. 2005). Our analysis revealed no significant contamination from softprotons. After the standard data processing, the good, dead-time corrected exposure timewas 41.6 ks for the pn and 43.2 ks for the two MOS detectors.
In the optical/near-infrared (near-IR), we observed the 3FGL J2039.6 − ′ , r ′ , i ′ , z ′ bands in the optical and in the J, H, K s bands in the near-IR. The observations were split into sequences of 18, 18, and 17 exposuresper day, each consisting of four 115 s dithered exposures in the optical and forty eight10 s dithered exposures in the near-IR. The observations were executed in grey time withairmass between 1.12 and 1.28 and mean seeing of 1.0 ′′ . Single dithered exposures werereduced (bias subtraction, flat-fielding, distortion correction) and stacked using standard IRAF tasks implemented in the GROND pipeline (Kr¨uhler et al. 2008; Yoldas et al. 2008).The astrometry calibration was computed on single exposures against stars selected from theUSNO-B1.0 catalogue (Monet et al. 2003) in the optical bands and the 2MASS catalogue(Skrutskie et al. 2006) in the near-IR bands, yielding an accuracy of 0 . ′′ δ = − ◦ observed inthe first night under photometric conditions. From the calibrated images we extracted a gridof secondary photometric calibrators for direct on–the–frame calibration on the subsequentnights. In the near-IR, the photometric calibration was computed against 2MASS starsidentified in the GROND field of view. The accuracy of the absolute photometry calibration http://xmm.esac.esa.int/xsa/ IRAF is distributed by the National Optical Astronomy Observatories, which are operated by the As-sociation of Universities for Research in Astronomy, Inc., under cooperative agreement with the NationalScience Foundation. ′ , r ′ , i ′ , z ′ bands, 0.03 magnitudes in J and H, and 0.05 in theK s band.In addition to GROND, we used serendipitous JHK s images of the 3FGL J2039.6 − ),consists of three sequences of 16 consecutive exposures of 15 s each in the J band and of 7.5s each in both the H and K s bands. In the optical and near-ultraviolet (near-UV) we used U,UVW1 ( λ = 2910 ˚A; ∆ λ = 1180˚A) and UVM2 ( λ = 2310 ˚A; ∆ λ = 710˚A) images from the XMM-Newton
Optical Monitor (OM; Mason et al. 2001) obtained in parallel to our obser-vations, with exposure times of 1900, 2700, and 3080 s, respectively. We also used archivalUVW2 ( λ = 2055 ˚A; ∆ λ = 557˚A) images from the Swift
UltraViolet and Optical Telescope(UVOT; Roming et al. 2005) performed on February 21 2011 (OBSID=00041479002), con-sisting of five exposures for a total integration time of 3587 s. The OM and UVOT datawere processed and calibrated using the SAS tool omichain and the
HEASOFT softwarepackage, respectively.
3. Data analysis and results
To search for possible counterparts of 3FGL J2039.6 − α =20 h m . s
32 and δ = − ◦ ′ . ′′ ′ .6 and 2 ′ .4, respectively, and a positionangle of 74 . ◦
95, measured East of North. We started from our
XMM-Newton observationsto find potential X-ray counterparts to 3FGL J2039.6 − γ -raysource. In particular, since we expect that 3FGL J2039.6 − http://casu.ast.cam.ac.uk/ For our X-ray analysis, we selected only 0 − −
12 fromthe two MOS detectors with the default flag mask . The source detection in the 0.3–10 keVenergy range was run simultaneously on the event lists of each of the EPIC-pn and MOSdetectors using a maximum likelihood fitting with the SAS task edetect chain invokingother SAS tools to produce background, sensitivity, and vignetting-corrected exposure maps.The final source list includes 90 X-ray sources from both the pn and MOS detectors, witha combined pn+MOS detection likelihood greater than 10, corresponding to a significanceabove 3 . σ . Figure 1 shows the 0.3–10 keV exposure-corrected XMM-Newton
FoV obtainedcombining the images of the EPIC-pn and MOS detectors. We focused our analysis on the 16X-ray sources detected within, or close to, the 95% confidence position error ellipse of 3FGLJ2039.6 − For each EPIC detector we extracted the source photons using an extraction radius of20 ′′ , while we extracted background photons from source-free regions in the same CCD chipas the source, with radii of 50 ′′ –120 ′′ . For each detector, we used the SAS task specgroup torebin all the extracted spectra and have at least 25 counts for each background-subtractedspectral channel and generated ad hoc response matrices and ancillary files using the SAStasks rmfgen and arfgen . For each source, we fitted simultaneously the pn and MOS spectrausing XSPEC v12.8, forcing the same set of parameters and considering three differentspectral models: a power-law (PL), well suited for both AGN and pulsars, an apec (AP)for stellar coronae, and a black-body (BB) for the pulsar thermal component. In all cases,the hydrogen column density N H was left as a free parameter. For each emission model wecomputed the 90% confidence level error on the spectral parameters. When the best-fit N H values were comparable to zero, we assumed the measured uncertainties to determine the3 σ confidence level upper limit. Spectra with very low counts were fitted after fixing the N H to the estimated value along the line of sight (5 × cm − ; Dickey & Lockman 1990), orfixing either the photon index or the temperature to typical values, i.e. Γ X = 2 and kT = 3 . kT = 0 . χ =52.32 (42 degrees of freedom, 8 –d.o.f.) for the former and χ =16.06 (22 d.o.f.) for the latter. For the remaining 13 sources,at least two different models were required to obtain an acceptable fit with null hypothesisprobability > χ =34.9 (40 d.o.f.). From the F-test (Bevington1969), we computed a 0.0003 probability that adding a thermal component to the modelwould produce a chance improvement to the fit. Since this probability corresponds to a ∼ σ significance only, hereafter we ignored the absorbed BB plus PL spectral model. Since formost X-ray sources the measured counts are too few to clearly discriminate among differentspectral models, we checked whether we could extract qualitative spectral information froman hardness ratio (HR) analysis (Marelli et al. 2014). However, we found that the observedHRs are compatible with different spectral models and, therefore, are not constraining. In order to detect possible time variability during the
XMM-Newton observation, wegenerated standard light curves from the pn-data for all the 16 X-ray sources in Table 1.Starting from the source and background regions described in Section 3.1.2, we extractedsource+background and background light curves, respectively, with the SAS task evselect and combined them into a background-subtracted light curve with the task epiclccorr . Thistask also corrects the time series for vignetting, bad pixels, chip gaps, quantum efficiency,dead time, exposure and good time intervals. For each source we generated a light curvewith time binning of 2500 s, or multiple of this value, to have at least 25 counts per bin andwe ran a χ test to evaluate the variability significance. Only Source 3 is characterized bya significant variability during the observation ( χ = 66 .
18, with 16 d.o.f.), with a chanceprobability of 4 . × − ( > σ ). After combining the data from all the three EPIC camerasthe Source 3 light curve shows a more apparent variability (Figure 2, left), with a chanceprobability of 7 . × − , whereas the other X-ray still show no evidence of significantvariability.The Source 3 light curve also hints at a possibly periodic modulation. We convertedphoton arrival times to the Solar system Barycentric Dynamical Time (TBD) with the SAStask barycen and used the FTOOL task efsearch to find the best period in the light 9 –curve through a maximum χ test. We folded the light curve with periods ranging from100 to 43000 s, the latter comparable to the length of the XMM-Newton observation. Wefound the best-fit period at 0.2245 ± χ of 104.5 (9 d.o.f.) gives a chance probabilityof 6 . × − , accounting for the number of trials, and makes the periodicity statisticallysignificant ( ∼ . σ ). The presence of a periodic signal at the corresponding frequency of ∼ × − Hz was independently confirmed by the power spectrum produced with theFTOOL powspec . We note that the best-fit X-ray period of Source 3 is comparable to abouthalf the length of the
XMM-Newton observation, so that the observed periodicity might bespurious. We examined the light curves of other comparably bright X-ray sources detected inthe whole
XMM-Newton
FoV and found that none of them showed evidence of periodicity atany time scales. Nonetheless, the fact that the length of the
XMM-Newton observation onlycovers ∼ ± − . We checked whether the foldedX-ray light curve of Source 3 varied as a function of the energy and whether the X-rayspectrum changed as a function of the phase. In both cases, however, the available statisticsis not sufficient to highlight significant differences in the energy-resolved light curves and thephase-resolved spectra.We looked for archival X-ray images of the 3FGL J2039.6 − Suzaku (Mitsuda et al. 2007) on October 28, 2010 for a total exposuretime of 21.5 ks (OBSID 705028010). We extracted source counts from a circle of radius1 . ′ XMM-Newton coordinates of Source 3 and background counts from anearby, source-free 2 ′ -radius circle using the HEASOFT(v.6.16) tool xselect and summedthe spectra from all the XIS cameras with the mathpha , addarf and addrmf tools. Weobtained 350 counts, of which ∼
50% are from the source. A fit with a PL gives a nullhypothesis probability of 0.06 (4 d.o.f.), with N H < × cm − (90% upper limit), photonindex Γ=1.3 ± +0 . − . × − erg cm − s − , fully compatible with the XMM-Newton results. The
XMM-Newton count-rate and spectral parameters (Table 1) are also compatible with the non-detection above the3 σ threshold ( ∼ × − erg cm − s − ) of Source 3 in the short Swift /XRT images ofTakeuchi et al. (2013), taken in 2010 and 2011. Therefore, we find no evidence of variabilityof Source 3 on time scales of three years. 10 –
We cross-matched the positions of all the 16
XMM-Newton sources in Table 1 withthe source catalogues obtained from the GROND observations. We performed the sourcedetection on the single-band GROND exposures using the starfind tool in
IRAF , matchedthe source catalogs over the different observations, and checked for variable sources againstthe median of all observations. Object photometry was computed using the task daophot in IRAF and the airmass correction was applied using the standard atmospheric extinctioncoefficients for the La Silla Observatory. For the cross-match we used a radius obtained bycombining the statistical 1 σ uncertainty on the X-ray source centroid plus the 90% confidencelevel systematic error associated with the absolute accuracy of the XMM-Newton aspectsolution, which is 1 . ′′ . The uncertainty on the absolute astrometry of theGROND images (0 . ′′
3) is much lower than the
XMM-Newton one, and is accounted for byour choice of the matching radius. There are only six
XMM-Newton sources (Source 3, 13,22, 24, 40, 76) with at least a candidate optical or near-IR counterpart in the GROND data(Figure 4).We also cross-correlated the
XMM-Newton source list with the VISTA, OM, and UVOTsource catalogues. The source detection and photometry in the VISTA images were carriedout as a part of the CASU pipeline (Sectn. 2.3). For the OM images, the source detectionand photometry were carried out with the SAS tasks omdetect and ommag , respectively, andfor the UVOT images with the HEASOFT taks uvotdetect and uvotsource . As before, thematching radius accounted for the uncertainty on the absolute astrometry of the used sourcecatalogues, i.e. . . ′′ . ′′ . ′′ XMM-Newton sources aresummarised in Table 2. Only for the candidate counterparts to Source 3, 40, and 76 we havean adequate spectral coverage in at least the optical and near-IR. Calibration technical note XMM-SOC-CAL-TN-0018
11 –
Among the six
XMM-Newton sources with a possible GROND counterpart, only Source3 is associated with a clearly variable object ( χ = 1232, with 51 d.o.f.). Figure 5 (left) showsthe multi-band light curves of this object for the three nights spanned by the GROND obser-vations. As seen, the light curves seem modulated, with an amplitude of ∼ ′ band. In particular, the modulation seems to be periodic and feature a double-peakedprofile (night 1) with the two peaks only partially seen in night 2 and 3, likely owing tothe different sampling of the light curve. This modulation is also seen in the r ′ , i ′ , and z ′ light curves, with both shape and amplitudes similar to the g ′ one, and is also recognisedin the J and H-band light curves, and very marginally in the K-band one. This suggeststhat the observed modulation is real and not due, for instance, to possible problems withthe photometry in a given filter. As a check, for all filters we extracted the light curves ofseveral stars of comparable brightness detected in the GROND FoV but found no evidenceof such a modulation. This confirms that it is not due to random effects, such as variationsin the sky conditions (transparency, background, moon illumination), or systematic effects,such as variations in the encircled flux due to the fixed size of the photometry aperture withrespect to the seeing disk. Furthermore, the fact that the modulation seems to be periodicargues against the possibility that it is produced by any of such effects and implies that isassociated with an intrinsic star variability.We computed the probability that the association between Source 3 and its candidateGROND counterpart is due to a chance coincidence. We computed the probability as P =1 − exp( − πρr ), where r is the matching radius used for Source 3 (1 . ′′
8) and ρ is the density ofstellar objects in the GROND field, regardless of their brightness, measured in the co-additionof all g’-band exposures. For a stellar density ρ ∼ . − we estimated achance coincidence probability P ∼ .
02, which makes a chance coincidence unlikely.To confirm the existence of a periodic modulation of the Source 3 candidate counterpart,we carried out a periodicity analysis based on the Generalized Lombe-Scargle periodogrammethod (Lomb 1976; Scargle 1982; Zechmeister & Kuerted 2009). For cross-checking pur-pose, we also used the phase dispersion minimisation technique (PDM; Stellingwerf 1978)using the pdm code in
IRAF and the
Period04 software package (Lenz & Breger 2005),which is especially dedicated to the statistical analysis of large astronomical time seriescontaining gaps (see Figure 5). All methods indicate the presence of a periodicity with aperiod of ∼ ∼ × − ( ∼ σ ). The best period was found at0.22748 ± ± XMM-Newton data for Source 3(0.2245 ± − − − The optical/near-IR light curves of the Source 3 counterpart folded at the correspondingbest-fit periods are shown in Figure 5 (right). A double-peak light curve is clearly visible,with a main peak and a secondary peak, separated in phase by ∼ .
5. Per each band, wedetermined the phase of the two peaks by fitting a Gaussian function to their profiles toprecisely compute their phase separations and errors. The peak phase separation remainsconstant in the optical (0 . ± . . ± . φ =0.2–0.5), the secondary peak ( φ =0.7–0.9), the “bridge” ( φ =0.5–0.7), and the“off-peak” ( φ =0.0–0.2 and φ =0.9–1.0). The shape of the light curve is similar in all bands,but the profile of the modulation changes from the optical to the near-IR (Figure 6, left).In particular, the primary peak becomes broader and its amplitude decreases, from ∼ ′ band to ∼ ∼ XMM-Newton and GROND observations (MJD=56885–56887 andMJD=56575, respectively) corresponds to a maximum time difference of 312 d. The ac-curacy on the best-fit optical period derived from the GROND observations is 0.00043 d(Sectn. 3.2.2), which corresponds to a phase uncertainty of ∼ . XMM-Newton observation would bring an uncertainty of ∼ ∼ . ∼ .
45 in the optical).
We computed the time-average multi-band photometry of the Source 3 counterpart fromthe GROND data and obtained g ′ =19.40 ± ′ =18.71 ± ′ =18.59 ± ′ =18.52 ± ± ± ± XMM-Newton /OM with AB magnitudes U=21.26 ± UV W =21.88 ± σ limiting magnitude of 21.99 and in the Swift /UVOT UVW2 image down to a limiting magni-tude of 23.34 (AB). In the near-IR, we also identified the Source 3 counterpart in the VISTAimages, with AB magnitudes J=18.24 ± ± s =18.64 ± . We found that the VISTA magnitudes (epoch 2010.6)are all compatible with the GROND ones (epoch 2014.7), after accounting for the differencebetween the K and K s filters, which excludes long-term variability on year time scales fromthe Source 3 counterpart.We used the time-averaged g ′ , r ′ , i ′ , z ′ -band magnitudes of the Source 3 counterpart as areference for its classification by analysing its location in the observed (i.e. not corrected forthe reddening) colour-magnitude (CM) and colour-colour (CC) diagrams of the field. Thediagrams are shown in Figure 7, where the location of the field stars is shown by the blackfilled circles and that of the Source 3 counterpart as a red filled triangle. Red and greentriangles correspond to the location computed from the single-image photometry. In order toreject outliers and include only high-confidence measurements, we plotted only field stars forwhich at least 20 measurements per filter were available and with σ < .
08. We comparedthe observed CM and CC diagrams with simulated stellar sequences computed from theBesan¸con models (Robin et al. 2004) for different stellar populations and distance valuesup to 15 kpc. The simulated sequences are shown in Figure 7 as the grey scale map. Thedark grey regions in the CM diagrams correspond to a likely distance range (200 . d . ± ′ -band magnitude of the Source 3 counterpart.The distance to Source 3 is unknown a priori. The upper limit on the hydrogen columndensity derived from the fits to the X-ray spectrum of Source 3 ( N H < × cm − ; Table 1)indicates a distance lower than ≈ . N H ofHe, Ng & Kaspi (2013). Without a parallax measurement, the distance to Source 3 cannot beprecisely constrained from kinematic measurements of its optical counterpart. The NOMADcatalogue (Zacharias et al. 2005) gives a proper motion of µ α cos ( δ ) = 14 ± − and µ δ = − ± − in right ascension and declination, respectively. This correspondsto a spatial velocity of 101 +32 − × (d/1 kpc) km s − . If we equate it to the median of thetransverse velocity distribution of MSPs ( ∼
108 km s − , with a standard deviation of ∼ − ) computed from the Australia National Telescope Facility (ATNF) Pulsar Catalogue (Manchester et al. 2005) we obtain a distance of ∼ N H . Any determination of a lower limit on the distance is more uncertain.Again, if Source 3 were a binary MSP, the distance distribution of known binary MSPs fromthe ATNF Pulsar Catalogue gives a probability of ∼ . d .
900 pc) is reasonable. http://casu.ast.cam.ac.uk/surveys-projects/vista/technical/filter-set
15 –We also used the upper limit on the N H to infer an interstellar extinction along the lineof sight E ( B − V ) < . E ( B − V ) = 0 . ∼ We built the optical/near-UV/near-IR spectrum of the Source 3 counterpart using theavailable multi-band photometry (Sect. 3.3.2). In all cases, we used as a reference themeasured AB magnitudes to compute the spectral fluxes at the filter peak wavelengths. Asa reference for the interstellar extinction correction we used a maximum extinction valueof E ( B − V ) = 0 . N H (Predehl & Schmitt 1995) obtained with the fit with a PL spectral model (Sect. 3.1.2).We fitted the spectrum with both a single and a double BB spectral model. However,we found that the optical/near-UV/near-IR data cannot be simultaneously fitted by a singleBB and that a double BB is required to fit the entire spectrum ( χ = 20 .
6, 6 d.o.f.). Theinferred temperatures are T H ∼ T C ∼ T H ∼ T C ∼ ∼ XMM-Newton /OM. As done above, we fittedthe four phase-resolved spectra using a two-BB model, considering both null and maximuminterstellar extinction. However, we did not find evidence of a significant spectrum evolutionacross the different phase intervals. This is partially ascribed to the fact the spectra areless constrained at shorter wavelengths without the flux measurements in the U and UVW1bands.
4. Discussion
The optical and X-ray emission of Source 3, modulated at a common periodicity of ∼ . β Lyr type (Geske et al. 2006). However, the system would be extremely peculiareven for these classes of binaries. The orbital period would be one of the shortest ever ob-served; the spectral type one of the latest; the asymmetry and separation of the peaks in theoptical light curves hard to explain. The second scenario, sounds more plausible. Moreover,when compared to the γ -ray flux of 3FGL J2039.6 − F γ = (1 . ± . × − ergcm − s − , the 0.3–10 keV unabsorbed X-ray flux of Source 3 (F X = 10 . +0 . − . × − ergcm − s − ) would give a γ –to–X-ray flux ratio F γ /F X ≈ − l = 341 . ◦ and b = − . ◦ , like most MSPs. We note thatthe NOMAD proper motion of Source 3 in Galactic coordinates, µ l = − ± − and µ b = − ± − , would suggest that it is moving towards the Galactic centre from 17 –its present location. This, however, would not argue against an MSP identification sinceseveral MSPs in the ATNF catalogue have a negative proper motion in Galactic latitude.Being older than a Gyr, MSPs are indeed expected to orbit in the Galactic potential andperiodically move away and towards the plane. Interestingly, if ascribed to an orbital motion,the period of the observed optical flux modulation ( ∼ γ -ray pulsars by the Fermi -LAT are either BWsor RBs, the possible identification of Source 3 as a BW/RB system would, then, concur tomake it the most likely X-ray counterpart to 3FGL J2039.6 − X = 1 .
36, and its X-ray luminosity in the 0.3–10 keV energyband is L X ∼ erg s − d kpc , where d kpc is the distance in units of kpc. Both the X-rayluminosity and photon index of Source 3 are in general agreement with those of RB/BWMSPs (Gentile et al. 2014; Roberts 2014), although its relatively hard X-ray spectrum wouldpoint more at a RB than a BW. As we noted in Sectn.3.1.2, a thermal component might bepresent in the X-ray spectrum of Source 3. However, further observations are necessary toclearly discriminate between a purely non-thermal and a composite spectral model.The X-ray light curve of Source 3 can be explained assuming a binary MSP scenario,the emission from the intrabinary shock is expected to be modulated at the orbital period.Recent studies suggest that the X-ray modulation may be due to synchrotron beaming,Doppler boosting of the flow within the shock, or obscuration by the companion (Bogdanovet al. 2011; Gentile et al. 2014; Roberts 2014). The shape of the X-ray light curve stronglydepends on the geometrical and physical parameters of the binary system but, on average, itis characterized by an overall increase of a factor of ∼ ∼ . . . − − − − − − γ -ray error ellipse of 3FGL J2039.6 − DM = 200 pc cm − (Camilo et al.2015). No pulsations were found, but this is not unexpected, because RBs are very elusivetargets in radio. In fact, the intrabinary material ablated from the star causes strong andvariable scattering and absorption of radio waves. The radio detection of the pulsations oftenrequires several dedicated long observations (Ray et al. 2013; Ray et al., in preparation). Standard models for the RB optical light curves based on tidal distortion and pulsar ir-radiation (e.g., Thorstensen & Armstrong 2005) cannot fit well neither the asymmetric peaksnor the peak separation. Therefore, we built a simple three-dimensional model including anadditional component related to the asymmetric irradiation of the companion star to fit thelight curve of the Source 3 counterpart. In this process, we considered only the optical lightcurves because are those with the highest signal–to–noise. 19 –In order to probe the RB scenario and estimate its physical parameters we built a simplethree-dimensional model of a RB binary system with very few free parameters and fit it tothe observed optical light curves of Source 3. In this model, the shape of the companion staris approximated by a sphere and a tangent cone pointing to the neutron star. The cone ismeant to account for the tidal deformation of the star as it approaches filling its Roche lobe.By locking the star rotation to the orbital motion we assumed two different brightnessesfor night and day, a characteristic commonly found in RB systems. The asymmetry in thetwo peaks of the optical light curve implies some asymmetry in the physical system thatproduces it. Therefore, we allowed a tilt angle between the cone axis and the day/nightseparator line, as measured on the orbital plane. As for other RBs, we assumed a perfectlycircular orbit, reducing the number of free parameters in the Kepler’s orbital parameter space.To summarize, our model accounts for four geometrical parameters: the star deformation(distance of the cone tip from the star center in star radius units); the night/day asymmetry(angle between the cone axis and the line of sight to the star); the orbital inclination (anglebetween the line of sight and the orbital plane); the epoch of quadrature (when the projectedaxis of companion star orbit lies perpendicular to the line of sight). Furthermore, for eachband we have two free parameters: the light curve normalization and the brightness ratiobetween night and day.We fitted our model to the data, including both statiscal and systematic errors, andfound good qualitative agreement for a narrow range of parameters. The overall goodnessof the fit turns out to be 197.3 (196 d.o.f.). The configuration obtained from the best fitto the model implies a deformation of 1.519 ± . ◦ ± . ◦ . Theorbital inclination of the binary system obtained from the best fit is 48 . ◦ ± . ◦ , whereasthe epoch of quadrature is at MJD 56884.9667 ± ± ± ± ± night and T day , respec-tively. A brightness ratio implies a relation between these two temperatures, as a functionof the temperature itself. This relation weakly depends on the interstellar extinction, whichcan only be constrained by our upper limits on the N H . We computed these relations foreach band, adding in quadrature the uncertainty associated with the interstellar extinctioncorrection. The values of temperature ratios are compatible in the four bands, at 1 σ , onlyfor a value of the day-side temperature T day < day − T night T day = 8 . × − (cid:18) T day (cid:19) − . × − (cid:18) T day (cid:19) (1)The large value of the asymmetry, ∼ ◦ , is implied by the different levels of minimaand maxima in the light curves. Similar features are observed in other RBs, like PSRJ1628 −
5. Conclusions
We carried out multi-wavelength observations of the unidentified
Fermi -LAT source3FGL J2039.6 − XMM-Newton and GROND. We detected a likely X-ray counter-part (Source 3) within the γ -ray error box of 3FGL J2039.6 − X = 1 . ± .
09) that is indicative of strong magnetosphericemission. The upper limit on the hydrogen column density inferred from the X-ray spectralfit, N H < × cm − , imply a distance probably lower than 1 kpc. The X-ray lightcurve of Source 3 features a modulation with a period of 0.2245 ± ∼ .
3. Using the GROND data, we found an optical counterpartto Source 3 that features an asymmetric, double-peaked, light curve and a flux modulationwith a period of 0.22748 ± − ∼ .
5, suggest that Source3/3FGL J2039.6 − γ -ray photons and searchfor the MSP pulsations. A timing solution, that can be extended back in time to the launchof Fermi in 2008, will provide even tighter constraints on the orbital parameters and theirevolution. This has a potentiially huge scientific payoff in terms of fundamental physics(Romani et al., 2012; Pletsch & Clark, 2015)The research leading to these results has received funding from the European Commis-sion Seventh Framework Programme (FP7/2007-2013) under grant agreement n. 267251.This work was supported by the ASI-INAF contract I/004/11/0, art.22 L.240/2010 for theproject “Studio di sorgenti di alta energia con Swift”. CD acknowledges support throughEXTraS, funded from the European Union’s Seventh Framework Programme for research,technological development and demonstration under grant agreement no 607452. Part of thefunding for GROND (both hardware as well as personnel) was generously granted from theLeibniz-Prize to Prof. G. Hasinger (DFG grant HA 1850/28-1).
Facilities:
XMM-Newton, GROND, Swift, Suzaku, VISTA
REFERENCES
Abdo, A.A., et al., 2009, Science, 325, 840Abdo, A.A., et al., 2010, ApJS, 188, 405Abdo, A. A., et al. 2013, ApJS, 208, 17Acero, F., et al., 2015, ApJS, 218, 23Ackermann, M., et al., 2012, ApJ, 753, 83Atwood, W. B., et al., 2006, ApJ, 652, 49Atwood, W. B., et al., 2009, ApJ, 697, 1071Bassa, C. G., Patruno, A., Hessels, J. W. T., et al., 2014, MNRAS, 441, 1825 22 –Bevington, P. R., 1969, Data Reduction and Error Analysis for the Physical Science (NewYork: McGraw-Hill)Bogdanov, S., Archibald, A. M., Hessels, J. W. T., et al., 2011, ApJ, 742, 97Breeveld A. A., et al., 2010, MNRAS, 406, 1687Breton, R. P., van Kerkwijk, M. H., Roberts, M. S. E., et al., 2013, ApJ, 769, 108Camilo, F, Kerr, M., Ray, P. S., et al., 2015, ApJ, 2015, ApJ, 810, 85Dalton, G. B., Lewis, I. J., Bonfield, D. G., et al., 2006, SPIE, 6269, 4De Luca, A., Caraveo, P.A., Mereghetti, S., Negroni, M., Bignami, G.F., 2005, ApJ, 623,1051Dickey, J. M., Lockman, F. J., 1990, ARA&A, 28, 215Drake, A. J., Djorgovski, S. G.; Mahabal, A., et al., 2009, ApJ, 696, 870Emerson, J. P., Irwin, M. J., Lewis, J., et al., 2004, Optimizing Scientific Return for Astron-omy through Information Technologies, eds. P. J. Quinn & A. Bridger, Proc. of theSPIE, Vol. 5493, pp. 401Emerson, J., McPherson A., Sutherland W., 2006, The Messenger, 126, 41Fitzpatrick, E. L., 1999, PASP, 111, 63Gentile, P. A., Roberts, M. S. E., McLaughlin, M. A.., et al., 2014, ApJ, 783, 69Gilliland, R. L., Baliunas, S. L., 1987, ApJ, 314, 766Geske, M. T., Gettel, S. J., McKay, T. A., 2006, ApJ, 131, 633Greiner, J., et al., 2008, PASP, 120, 405He, C., Ng, C.-Y., Kaspi ,V. M., 2013, ApJ, 768, 64Hui, C. Y., Hu, C. P., Park, S. M., et al., 2015, ApJL, 801, 27Kaplan, D. L., Stovall, K., Ransom, S. M., et al., 2012, ApJ, 753, 174Kataoka, J., et al., 2012, ApJ, 757, 176Kong, A. K. H., et al., 2012, ApJ, 747, L3 23 –Kong, A. K. H., et al., 2014, ApJ, 794, L22Kr¨uhler, T., et al., ApJ, 685, 376Leahy, D. A., 1987, A&A, 180, 275LLenz P., Breger M. 2005, CoAst, 146, 53Li, M., Halpern, J. P., Thorstensen, J. R., 2014, ApJ, 795, 115Lomb, N. R. 1976, ApSS, 39, 447Manchester R. N., Hobbs G. B., Teoh A. & Hobbs M., 2005, AJ, 129, 1993Marelli, M., De Luca, A., Caraveo, P. A., 2011, ApJ, 733, 82Marelli, M., Harding, H., Pizzocaro, D., De Luca, A., Wood, K. S., Caraveo, P. A., Salvetti,D., Saz Parkinson, P. M., Acero, F., 2014, ApJ, 795, 168Marelli, M., Mignani, R. P., De Luca, A., Saz Parkinson, P. M., Salvetti, D., Den Hartog,P. R., Wolff, M. T., 2015, ApJ, 802, 78Mason K. O., et al., 2001, A&A, 365, 36Mauch, T., et al., 2003, MNRAS, 342, 1117Mirabal, N., Fr´ıas-Martinez, V., Hassan, T., Fr´ıas-Martinez, E., 2012, MNRAS, 424, 64Nolan, P.L. et al., 2012, ApJS, 199, 31Orosz, J. A., van Kerkwijk, M. H., 2003, A&A, 397, 237Page M. J., et al., 2012, MNRAS, 426, 903Petrov, L., Mahony, E. K., Edwards, P. G., Sadler, E. M., Schinzel, F. K., McConnell, D.,2013, MNRAS, 432, 1294Pletsch, H. J., et al., 2012, Science, 338, 1314Pletsch, H. J., et al., 2013, ApJ, 779, L11Pletsch, H. J., & Clark, C. J., 2014, ApJ, 795, 75Pletsch, H. J., & Clark, C. J., 2015, ApJ, 807, 718Poole T.S., et al. 2008, MNRAS, 383, 627 24 –Predehl P. & Schmitt J.H.M.M. 1995, A&A, 293, 889Ray, P. S., et al., 2013, ApJ, 763, L13Roberts, Mallory S. E., 2013, IAU Symposium, 291, 127Robin A.C., Reyl´e C., Derri´ere S., Picaud S., 2004, A&A 416, 157Romani, R. W. & Shaw, M. S., 2011, ApJ, 743, L26Romani, R. W., 2012, ApJ, 754, L25Romani, R. W., et al., 2012, ApJ, 760, L36Romani, R. W., Filippenko, A. V., Silverman, J. M., 2012, ApJ, 760, 36Romani, R. W., Filippenko, A. V., Cenko, S. B., 2014, ApJ, 793, L20Roming, P. W. A., et al. 2005, Space Sci. Rev., 120, 95Scargle, J. D. 1982, ApJ, 302, 757Schroeder, J., Halpern, J., 2014, ApJ, 793, 78Stappers, B. W., van Kerkwijk, M. H., Bell, J. F., Kulkarni, S. R., 2001, ApJ, 548, L183Stellingwerf, R. F., 1978, ApJ, 224, 953Strader, J., et al., 2014, ApJ, 788, L27Struder, L., Briel, U., Dennerl, K. et al. 2001, A&A, 365, L18Skrutskie M. F., et al. 2006, AJ, 131, 1163Takeuchi, Y., Kataoka, J., Maeda, K., Takahashi, Y., Nakamori, T., Tahara, M., 2013, ApJS,208, 25Thorstensen, J. R. & Armstrong, E., 2005, AJ, 130, 759Turner, M.J.L., Abbey, A., Arnaud, M. et al. 2001, A&A, 365, 27van Staden, A., 2015, MNSSA, 74, 22Yoldas, A. K. et al., 2008, GAMMA-RAY BURSTS 2007: Proceedings of the Santa FeConference. AIP Conference Proceedings, Volume 1000, pp. 227York D.G., et al., 2000, AJ, 120, 1579 25 –Zacharias, N., Monet, D. G., Levine, S. E., Urban, S. E., Gaume, R., Wycoff, G. L., 2005,Bulletin of the American Astronomical Society, Vol. 36, p.1418Zechmeister, M., & Kuerted, M. 2009, A&A, 496, 577
This preprint was prepared with the AAS L A TEX macros v5.2.
Table 1: Spectral properties of the
XMM-Newton sources detected within, or close to, the error ellipse of 3FGLJ2039.6 − power-law (PL), apec (AP), and black-body (BB).Source ID J2000 coord. Counts rate Spectral NH Γ X Flux [0 . − keV ] VariabilityRA Dec [ ◦ ] (stat. err. a ) 10 − cts/s model 10 cm − − erg cm − s − χ (d.o.f.)3 309.8956 -56.2861 (0.3 ′′ ) 23.78 ± < .
04 1.36 ± +0 . − . ′′ ) 10.76 ± < .
16 1.33 +0 . − . +0 . − . ′′ ) 4.90 ± < .
26 2.07 +0 . − . +0 . − . ′′ ) 6.51 ± < .
24 1.54 +0 . − . +0 . − . ′′ ) 7.84 ± < .
77 1.80 +0 . − . +1 . − . ′′ ) 4.44 ± < .
46 1.86 +1 . − . +0 . − . ′′ ) 5.19 ± < .
05 2.37 +2 . − . +1 . − . ′′ ) 4.21 ± < .
91 1.80 +0 . − . +1 . − . ′′ ) 3.03 ± b +0 . − . ′′ ) 2.46 ± b +0 . − . ′′ ) 2.00 ± < .
21 1.11 +3 . − . +1 . − . ′′ ) 1.78 ± b < .
42 2.33 +1 . − . +0 . − . ′′ ) 1.58 ± b +0 . − . ′′ ) 1.10 ± b +0 . − . ′′ ) 1.24 ± b +0 . − . ′′ ) 1.62 ± · · · · · · · · · · · · a Here we report only the 1 σ statistical error, the 1 σ systematic error is about 1.5 ′′ for each X-ray source. b Owing to the low number of counts, we fixed the photon index (Γ X ) to 2 for a PL model and the temperature( kT ) to 3.5 and 0.2 for an AP and a BB model respectively.Note. — Results of the spectral analysis of the XMM-Newton sources detected within an error circle of1.5 times the 95% confidence error ellipse of 3FGL J2039.6 − σ . Table 2: Magnitudes of the optical, near-IR, and near-UV counterparts to the
XMM-Newton sources detected within,close to, the error ellipse of 3FGL J2039.6 − ′′ which accounts for both the statistical and systematic errors on the XMM-Newton source coordinates (see Table1) and the accuracy of the absolute astrometry of the optical, near-IR, and near-UV images. All magnitudes are in theAB system.ID GROND VISTA OM UVOTg’ r’ i’ z’ J H K J H K s U UVW1 UVM2 UVW23 19.40 18.73 18.72 18.61 18.53 18.15 18.34 18.24 18.24 18.64 21.26 21.8811 20.33 20.02 20.7313 22.61 20.70 21.2719 20.6422 19.37 17.8818.02 17.96 17.84 17.93 17.9724 24.2129 20.83 20.3940 21.95 21.22 21.45 22.92 19.82 19.85 20.10 22.0872 19.6660 19.3376 19.64 18.36 18.63 18.20 17.65 17.66 17.77 17.63 17.58 17.9021.52 19.36 18.84 18.6423.9788 20.93 21.31 28 –Fig. 1.— 0.3–10 keV exposure-corrected
XMM-Newton image of the 3FGL J2039.6 − ′′ . The 95%confidence error ellipse of 3FGL J2039.6 − ′′ radius and labeled as in Table 1, whereasother X-ray sources detected in the FoV are plotted with a radius of 10 ′′ . The colour of thecircles correspond to the likelihood of the source detection (DET ML): DET M L <
25 (red),25 < DET M L <
50 (magenta), 50 < DET M L <
100 (yellow),
DET M L >
100 (green). 29 –Fig. 2.—
Left : Background-subtracted light curve combining data from the 3 EPIC camerasfor Source 3 in the 0.3–10 keV energy range, sampled with a bin time of 2500 s.
Right : Samebut folded around the best period of 0.2245 days normalized to the average source intensity.In both panels, error bars are reported at 1 σ . 30 –Fig. 3.— Binned spectrum of Source 3, the brightest source among the most probable candi-date X-ray counterparts to 3FGL J2039.6 − − XMM-Newton sources detected within, or close to, the 3FGL error ellipse,here represented by the blue ellipse. In all cases the circle radius has been arbitrarily set to5 ′′ for a better visualisation. 32 – g' −0.15−0.10−0.050.000.050.100.15 < g ' > - g ' r' −0.15−0.10−0.050.000.050.100.15 < r ' > - r ' i' −0.15−0.10−0.050.000.050.100.15 < i ' > - i ' z' −0.15−0.10−0.050.000.050.100.15 < z ' > - z ' J −0.2−0.10.00.10.2 < J ' > - J ' H −0.3−0.2−0.10.00.10.20.3 < H ' > - H ' (Time - 5885) [MJD] K Phase −1.0−0.50.00.51.0 < K ' > - K ' Fig. 5.—
Left : Multi-band light curves of the optical counterpart to
XMM-Newton
Source 3.
Right : Light curves folded at the best-fitting period. The axis on the right are magnitudesrelative to the mean. The oscillations in magnitude in the same phase bins in the i’ and z’bands are likely due to fringing. Only statistical errors are plotted. The vertical ticks are thesystematic errors associated with the accuracy of the photometric calibration (Sectn. 2.3).The vertical dashed lines define the main peak ( φ =0.2–0.5), the secondary peak ( φ =0.7–0.9),the “bridge” ( φ =0.5–0.7), and the “off-peak” ( φ =0.0–0.2 and φ =0.9–1.0) regions. 33 – g ' - r ' g ' - i ' g ' - z ' r ' - i ' Phase r ' - z ' Phase i ' - z ' Fig. 6.— Colors of the Source 3 counterpart as a function of phase. Only statistical errorsare plotted. The vertical ticks are the systematic errors in the color determination associatedwith the accuracy of the photometric calibration (Sectn. 2.3). Different colors correspond todifferent nights, i.e. night 1 (red), night 2 (blue), night 3 (green). The vertical dashed linescorresponds to the four regions defined in Figure 5. 34 –Fig. 7.—
Top : Observed CMDs for the 3FGL J2039.6 − Bottom : Observed CC diagram. In all panels, the location offield stars is indicated by the black filled circles, whereas that of the optical counterpart ofthe
XMM-Newton
Source 3 is indicated by the red filled triangle. The filled green trianglesindicates the counterpart location computed from the photometry computed on the singleimage. Stellar sequences simulated from the Besan¸con models for different values of distanceare shown in light and dark grey. In the CM diagrams the dark grey regions correspond todistance values 200 < d <
900 pc, whereas in the CC diagram they correspond to magnitudeswithin ± Phase g ' r ' i ' z ' M a g n i t u d e g' bandr' bandi' bandz' band Fig. 8.— Multi-band light curve of the optical counterpart of the
XMM-Newton
Source 3.g ′ , r ′ , i ′ and z ′′