Fermi Variability Study of the Candidate Pulsar Binary 2FGL~J0523.3-2530
aa r X i v : . [ a s t r o - ph . H E ] S e p Draft version June 27, 2018
Preprint typeset using L A TEX style emulateapj v. 5/2/11
FERMI
VARIABILITY STUDY OF THE CANDIDATE PULSAR BINARY 2FGL J0523.3 − Yi Xing , Zhongxiang Wang , C.-Y. Ng Draft version June 27, 2018
ABSTRACTThe
Fermi source 2FGL J0523.3 − Fermi
Large Area Telescope in the 0.2–300 GeV energyrange. Long-term, yearly variability from the source has been found, with a factor of 4 flux variationsin 1–300 GeV. From spectral analysis, we find an extra spectral component at 2–3 GeV that causesthe source brightening. While no orbital modulations have been found from the
Fermi data over thewhole period of 2008–2014, orbital modulation in the source’s > Fermi observation. Our results support the millisecond pulsar binary natureof 2FGL J0523.3 − Subject headings: binaries: close — stars: individual (2FGL J0523.3 − INTRODUCTION
Since the
Fermi Gamma-ray Space Telescope waslaunched in 2008 June, the main instrument on-board—the Large Area Telescope (LAT) has been continuouslyscanning the whole sky every three hours in the energyrange from 20 MeV to 300 GeV, discovering and mon-itoring γ -ray sources with much improved spatial reso-lution and sensitivity comparing to former γ -ray tele-scopes (Atwood et al. 2009). In 2012, using Fermi /LATdata of the first two-year survey, a catalog of 1873 γ -ray sources was released by Nolan et al. (2012) as the Fermi /LAT second source catalog (2FGL). Among the γ -ray sources, 575 of them are not associated with anyknown astrophysical objects (Nolan et al. 2012). For thepurpose of identifying the nature of these unassociatedsources, many follow-up studies, such as classifying their γ -ray characteristics (Ackermann et al. 2012), search-ing for radio pulsars (Ray et al. 2012), and observingat multi-wavelengths (Takeuchi et al. 2013; Acero et al.2013), have been carried out.The source 2FGL J0523.3 − − Fermi
LAT First Source Catalog (Abdo et al. 2010).
Swift imaging of the field has revealed a candidate X-ray coun-terpart (Takeuchi et al. 2013). While radio searches fora pulsar have failed (Guillemot et al. 2012; Petrov et al.2013), optical imaging and spectroscopy recently havediscovered orbital modulations from the X-ray counter-part, with a period of 16.5 hr (Strader et al. 2014). Thesource is located at a high Galactic latitude G b = − . ◦ γ -ray luminosity of ∼ . × erg s − (assumingsource distance d = 1 . Shanghai Astronomical Observatory, Chinese Academy ofSciences, 80 Nandan Road, Shanghai 200030, China Department of Physics, The University of Hong Kong, Pok-fulam Road, Hong Kong with a 0.8 M ⊙ companion. Furthermore, this binarycould be another so-called “redback” system, which isclassified as an eclipsing MSP binary that contains a rel-atively massive ( & . M ⊙ ), non-degenerate companion(Roberts 2013). The ablation of the companion by pul-sar wind from the MSP produces matter in the binary,which would eclipse radio emission from the pulsar atcertain orbital phases.We were intrigued by this Fermi source because wenote that it is located in the blazer region, along withthe Crab pulsar, in the curvature–variability plane of
Fermi bright sources (see Figure 4 in Romani 2012),which suggests possible variability from this source. Ithas recently been learned that the prototypical red-back system PSR J1023+0038 (Archibald et al. 2009)has shown γ -ray variability due to its temporary accre-tion activity (Stappers et al. 2014; Patruno et al. 2014;Takata et al. 2014). The newly identified redback sys-tem XSS J12270 − γ -ray emission (Hill et al.2011). Given the discovery of the orbital period of2FGL J0523.3 − Fermi data for this source, aiming to study the source’s γ -ray flux variations, determine its high-energy proper-ties, and establish the similarities to the redback systemsPSR J1023+0038 and XSS J12270 − Swift
X-ray data available for the source, we also conducted theX-ray data analysis. In this paper, we report the resultsfrom the analyses. OBSERVATION
LAT is the main instrument onboard
Fermi . It is a γ -ray imaging instrument which makes all-sky survey inan energy range from 20 MeV to 300 GeV (Atwood et al.2009). In our analysis, we selected LAT events from the Fermi
Pass 7 Reprocessed (P7REP) database inside a Xing et al.
Fig. 1.—
TS maps of a 2 o × o region centered at 2FGL 0523.3 − Fermi data. Panels (b) and (c) are 1–300 GeV maps during the time interval Iand II, respectively, that are defined in § − σ error circle centered at the best-fit position. The large circlein Panel (c) indicates the 2 σ error circle of the source during time interval II. o × o region centered at the catalog position of 2FGLJ0523.3 − ◦ , which prevents the Earth’s limb contam-ination, and during good time intervals when the qualityof the data was not affected by the spacecraft events. DATA ANALYSIS AND RESULTS
Source Identification
We included all sources within 16 ◦ centered at the po-sition of 2FGL J0523.3 − Fermi − > v9r23p5 , and extracted the Test Statis-tic (TS) map of a 2 ◦ × ◦ region centered at the po-sition of 2FGL J0523.3 − − L /L ), where L and L are the maximum likelihood values for a modelwithout and with an additional source at a specifiedlocation, respectively, and is a measurement of the fit Fig. 2.— − improvement for including the source. Generally theTS is approximately the square of the detection signif-icance of a source (Abdo et al. 2010). The γ -ray emis-sion near the center was detected with TS ≃ ∼ σ detection significance. We ran gtfindsrc inthe LAT software package to find the position of the γ -ray emission in this region and obtained a positionof R.A.=80 . ◦
83, Decl.= − . ◦
49, (equinox J2000.0), with1 σ nominal uncertainty of 0 . ◦
02. The catalog positionof 2FGL J0523.3 − . ◦
83, Decl.= − . ◦ . ◦ − . ◦ . ′′
2, Monet et al. 2003).The optical position is ∼ . ◦
03 from the best-fit position,but within the 2 σ error circle.The > − γ -ray emission), respectively. The results are given inTable 1. The source modeled with the power-law spec-trum was found to have spectral index Γ = − . ± .
04 and a TS pl value of ≃ − . ± .
1, cutoffenergy E c = 4 . ± . exp value of ≃ ∼ σ significance(estimated from p TS exp − TS pl = √ Long-term Variability Analysis
To investigate the variability of 2FGL J0523.3 − γ -ray light curvesat different energy bands ( > > > > Fig. 3.— γ -ray spectra of 2FGL J0523.3 − left panel ), and during the high (red data points) andlow (blue data points) states ( right panel ). The exponentially cutoff power law obtained from maximum likelihood analysis for the wholedata is shown as black dashed curves, and for the low state data is plotted as the blue dashed curve. A Gaussian function (red dottedcurve) can be added to the spectrum to describe the extra component at 2-3 GeV. and TS curve and plot them in Figure 2. The low energyTS curve generally has low values, similar to those of the > − photons cm − s − , consistent withbeing a constant within the uncertainties.Based on the TS curve, we defined six time intervals.In interval II and IV, the TS curve is flat with valuesof <
20, while in interval I and III, the TS curve hasvalues of 30–80. In intervals V and VI, the TS values aremostly low but with weak variations in a range of 10–30. To confirm the variations seen in the light curves, wefurther extracted the > ≃
170 and ≃
60 inthe two maps, indicating approximately 10 σ significancefor the flux variation between the two time intervals.We noted that as shown in panel (c) of Figure 1 (whenthe source was dim), the TS peak appears to have an off-set from the best-fit position. We determined the sourceposition for time interval II and found that the positionis consistent with the best-fit position within 2 σ . Wefurther checked the TS maps when the source was dim(time intervals IV, V, and VI), and the TS peaks all ap-pear to have small offsets from the best-fit position butin different directions. In our analysis, this Fermi sourceis consistent with being a point source, and no signs ofextended emission or an additional source were found.We concluded that the apparent offsets are probably dueto under-estimated uncertainties for the source position.
Spectral Analysis
The γ -ray spectrum of 2FGL J0523.3 − γ -ray spectrum is shown in Figure 3 and the values at eachbin are given in Table 2, in which the spectral points withTS greater than 4 were kept. The cutoff power law model is also displayed in Figure 3. The model does not describethe low-energy data points well, as two data points areapproximately 2 σ away from the cutoff power-law model(black dashed curve in Figure 3).To search for differences in the source’s emission duringthe ‘high’ and ‘low’ states shown in Figure 2, and thushelp understand the cause of the flux variation, we ex-tracted γ -ray spectra of 2FGL J0523.3 − ≥ <
30) states are plotted in Figure 3. Theflux values are given in Table 2. By comparing the twospectra, a component at the 2–3 GeV energy range dur-ing the high state is present. We therefore further mod-eled the low state emission with a cutoff power law, andfound Γ = 1 . ± . E c = 6 . ± . Aexp [ − ( E − E ) / σ ], where A =(4 ± × − erg cm − s − , E = 2 . ± . σ = 0 . ± . χ ,which is 3.2 (for 6 degrees of freedom) when compar-ing the high-state spectrum with the cutoff power law,is improved to 1.0 (for 3 degrees of freedom) when theGaussian component is included. Timing Analysis
Around the frequency 1.68195 ± . × − Hz,which was determined from optical radial velocity mea-surements by Strader et al. (2014), orbital modulationswere searched in γ -ray emission of the source. The searchwas performed to the LAT data within 1 . ◦ − gtpsearch in theLAT software package was used. The optical positionwas used for the barycentric correction to photon arrivaltimes. Different energy ranges of > > > > Fig. 4.—
TS maps (2–300 GeV) of a 2 o × o region centered at 2FGL 0523.3 − left panel ) and superior conjunction ( right panel ), respectively. The data are from 2012-10-01 to 2014-04-02. Symbols are thesame as in Figure 1. in the low state marginal signals were seen, but none ofthem were sufficiently convincing as the H -test values forthe signals were approximately 10.Given the uncertainty of the orbital period, we consid-ered that in time intervals V and VI, the optical timingresults are reliable (note that the binary orbit was deter-mined from optical observations during from 2013-10-01to 2014-01-10; Strader et al. 2014) and the source wasmostly in the low state. We thus searched for periodicsignals during a slightly longer time period from 2012-10-01 to the end of the LAT data we analyzed. Folding > H -test value of 15 was found. Following Wu et al. (2012),we also made two TS maps over two half orbital phasesto confirm the detection of the orbital flux variations.Phase I is the half of the orbit centered at the superiorconjunction (when the secondary is behind the primarystar), and Phase II is the other half centered at the in-ferior conjunction. The TS maps are shown in Figure 4.The > γ -ray emission from 2FGL J0523.3 − ≃
90 and ≃
20, respec-tively, at the source position.We also searched for the periodic signals in the sameenergy range and over the same time period. A signalwith H test value of ≃
18 ( ∼ σ detection significance)at the frequency of 1.682246 × − Hz was found. Thisfrequency is within the 5 σ error range of the opticalorbital value. The folded light curve is shown in Fig-ure 5, where phase zero is set at the superior conjunc-tion (MJD 56577.14636, given by Strader et al. 2014).The source was brighter during the phase of 0.25–0.55(Phase II is 0.25–0.75). Spectra during the on-peak andoff-peak phases were obtained, but due to limited num-bers of photons, the uncertainties on the flux data pointsare too large to allow any further detailed analysis.No attempt was made to search for millisecond spinsignals from the primary star, since it is difficult andcomputing-intensive to find from blind searches of Fermi γ -ray data, and thus far only one MSP has been found from blind searches (Pletsch et al. 2012). We note thatto search for the spin signal from the putative MSP, thelow-state time periods should be considered, since emis-sion during the time would primarily come from the pul-sar (see Section 4 below). Swift X-ray Data Analysis
The source 2FGL J0523.3 − Swift on 2009 Nov 12 (ObsID: 00031535001) and on 2013Sep 17 (ObsID: 00032938001) for 4.8 and 14.4 ks, respec-tively. We analyzed the photon counting mode data fromthe X-ray Telescope. The data were processed by thestandard pipeline, and in both observations, an X-raysource is clearly detected at the optical position. Us-ing a standard extraction aperture of 20 pixels (= 47 . ′′ ± ± . ± . × − and 4 . ± . × − cts s − , respec-tively. Given the large uncertainties, these two valuesare formally consistent. The source was too faint andthe two observations were too short (comparing to theorbital period) to be searched for orbital modulation.We extracted the source spectrum from the mergeddata set using the aperture mention above. The spectralanalysis was carried out in XSPEC in the 0.3–7 keV energyrange, using the telescope response files provided by thecalibration team. For the spectral fit, we employed the C-statistic (cstat in
XSPEC ) to perform unbinned likelihoodanalysis due to the low number of counts. We first triedan absorbed power-law model, but found a very smallabsorption column density. This is not surprising sincethe source is at a high Galactic latitude and the totalGalactic column density in the direction is only 1 . × cm − (Kalberla et al. 2005). We therefore did notinclude absorption in the final fit and obtained a photonindex of Γ = 1 . ± . χ value of 0.96. The energy flux is 1 . × − erg cm − s − in 0.3–7 keV. Xing et al. Fig. 5.— left panel: H -test values at trial frequencies resulting from the 2012-10-01 to 2014-04-02 data. The frequency with the highest H -test value is marked by an arrow. The error bar indicates the 5 σ error range of the optical orbital period. Right panel: folded light curve( > H -test value frequency indicated in the left panel. DISCUSSION
From our analysis of the
Fermi data for2FGL J0523.3 − γ -ray flux variations over approximately 5.5 yr Fermi observation time. Spectral analysis of the data duringthe high and low states indicates that emission fromthe source in the latter is well described by an expo-nentially cutoff power law, which is typical for pulsaremission. Comparing to the MSPs detected with
Fermi (Abdo et al. 2013), the cutoff energy is among thehighest but within the uncertainties (the highest valuewith smaller uncertainty is E c = 5 . ± γ -ray flux from the pulsar bi-nary was also 10 times larger than before, a factor of 2times higher than what was seen in 2FGL J0523.3 − γ -ray brighteningseen in PSR J1023+0038 (Takata et al. 2014). However,the optical light curve of 2FGL J0523.3 − Fermi observa-tion time, but does not show any signs of irradiationof the companion or additional optical emission from adisk. We also note that the two sets of X-ray data weretaken both during the low state, which may explain theconsistent faintness of the source during the two X-rayobservations.We have detected orbital modulation in γ -ray emissionof 2FGL J0523.3 − Fermi observation. Unfortunately dueto the limited photon counts, we were not able to obtainany spectral information about the differences betweenthe on-peak and off-peak emission. Thus far the proto-typical black window pulsar binary B1957+20, in whicha degenerate, low-mass companion is under strong irradi-ation by pulsar wind from an MSP (Fruchter et al. 1988),is the only compact binary detected with orbital γ -raymodulation (Wu et al. 2012). Similarly, Phase II of thisbinary was found to be brighter, due to excess emissionat the > γ -ray emission from MSP bi-naries, although the detailed physical processes are dif-ferent. In any case, the modulation arises due to changesof the viewing angle to the intrabinary γ -ray producingregion as the binary rotates. The high cutoff energy seenin 2FGL J0523.3 − > − Swift
X-ray luminosity was1.9 × erg s − (at distance 1.1 kpc), lower but compa-rable to that of PSR J1023+0038 and XSS J12270 − − . ◦
02 error circle of
Fermi sources, it can be difficult to exclude any possible con-tamination from other sources within the error cir-cle. For example, in the XSS J12270 − γ -ray emission(and possible flux variations) detected (Hill et al. 2011;Bogdanov et al. 2014). We note that in the SIMBADdatabase, none of the known nearby sources in the fieldcould possibly be associated with 2FGL J0523.3 − . ◦
037 away in theireffort to search for possibly associated radio sources with
Fermi objects. The majority of the high Galactic
Fermi objects ( Gb ≥ ◦ ) are associated with Active Galac-tic Nuclei (AGN; Nolan et al. 2012), and they are strongvariable sources. However no evidence supports the pres-ence of an AGN in the field. Emission from AGN gener-ally has a power-law form, arising due to IC scatteringby high-energy electrons in jets from AGN (Abdo et al.2009). Their spectral features (e.g., Williamson et al.2014) are clearly different from that seen in the high stateof 2FGL J0523.3 − γ -ray emitting AGN in the field is really low.On the basis of the AGN counts distribution study inAbdo et al. (2009), the number of AGN with Fermi γ -ray fluxes greater than 10 − ph cm − s − is 0.064 deg − .For the 0 . ◦
06 radius (3 σ ) error circle region, the proba-bility of having at least one such AGN is only 6.8 × − .Given these reasons, it is not likely that the variabilityis caused by the existence of an unknown AGN.As a summary, we have detected orbital modulation in > γ -ray emission of 2FGL J0523.3 − γ -ray source has also been detected, and the flux in-creases are due to the presence of an extra emission com-ponent at 2–3 GeV. The origin of the component as wellas the variability is not clear. In the near future, multi-wavelength studies of the source and source field shouldbe conducted, aiming to detect any correlated flux vari-ations at optical/X-ray energies and help determine theorigin of the high-energy component and its variability.We thank the anonymous referee for useful suggestionsand Liang Chen for helpful discussion about Fermi prop-erties of AGN.This research was supported by supported by ShanghaiNatural Science Foundation for Youth (13ZR1464400),the National Natural Science Foundation of China(11373055), and the Strategic Priority Research Program“The Emergence of Cosmological Structures” of the Chi-nese Academy of Sciences (Grant No. XDB09000000).Z.W. is a Research Fellow of the One-Hundred-Talentsproject of Chinese Academy of Sciences.
REFERENCESAbdo, A. A., et al. 2009, ApJ, 700, 597—. 2010, ApJS, 188, 405—. 2013, ApJS, 208, 17Acero, F., et al. 2013, ApJ, 779, 133Ackermann, M., et al. 2012, ApJ, 753, 83Archibald, A. M., Kaspi, V. M., Bogdanov, S., Hessels, J. W. T.,Stairs, I. H., Ransom, S. M., & McLaughlin, M. A. 2010, ApJ,722, 88Archibald, A. M., et al. 2009, Science, 324, 1411Atwood, W. B., et al. 2009, ApJ, 697, 1071Bassa, C. G., et al. 2014, MNRAS, 441, 1825Bednarek, W. 2014, A&A, 561, A116Bogdanov, S., Archibald, A. M., Hessels, J. W. T., Kaspi, V. M.,Lorimer, D., McLaughlin, M. A., Ransom, S. M., & Stairs, I. H.2011, ApJ, 742, 97Bogdanov, S., Patruno, A., Archibald, A. M., Bassa, C., Hessels,J. W. T., Janssen, G. H., & Stappers, B. W. 2014, ApJ, 789, 40Fruchter, A. S., Stinebring, D. R., & Taylor, J. H. 1988, Nature,333, 237Gentile, P. A., et al. 2014, ApJ, 783, 69Guillemot, L., et al. 2012, MNRAS, 422, 1294Hill, A. B., et al. 2011, MNRAS, 415, 235Kalberla, P. M. W., Burton, W. B., Hartmann, D., Arnal, E. M.,Bajaja, E., Morras, R., & P¨oppel, W. G. L. 2005, A&A, 440,775Monet, D. G., et al. 2003, AJ, 125, 984 Nolan, P. L., et al. 2012, ApJS, 199, 31Patruno, A., et al. 2014, ApJ, 781, L3Petrov, 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, 1314Ray, P. S., et al. 2012, ArXiv e-printsRoberts, M. S. E. 2013, in IAU Symposium, Vol. 291, IAUSymposium, ed. J. van Leeuwen, 127–132Romani, R. W. 2012, ApJ, 754, L25Roy, J., Bhattacharyya, B., & Ray, P. S. 2014, The Astronomer’sTelegram, 5890, 1Schinzel, F. K., Petrov, L., Taylor, G. B., Mahony, E. K.,Edwards, P. G., & Kovalev, Y. Y. 2014, ArXiv e-printsStappers, B. W., et al. 2014, ApJ, 790, 39Strader, J., Chomiuk, L., Sonbas, E., Sokolovsky, K., Sand, D. J.,Moskvitin, A. S., & Cheung, C. C. 2014, ApJ, 788, L27Takata, J., et al. 2014, ApJ, 785, 131Takeuchi, Y., Kataoka, J., Maeda, K., Takahashi, Y., Nakamori,T., & Tahara, M. 2013, ApJS, 208, 25Williamson, K. E., et al. 2014, ApJ, 789, 135Wu, E. M. H., Takata, J., Cheng, K. S., Huang, R. H. H., Hui,C. Y., Kong, A. K. H., Tam, P. H. T., & Wu, J. H. K. 2012,ApJ, 761, 181
Xing et al.
TABLE 1Binned likelihood analysis results for 2FGL J0523.3 − Spectral model Flux/10 − Γ E c TS(photon cm − s − ) (GeV)Power law 11.5 ± ± · · · ± ± ± a ± ± ± a The results are from analyzing the low state data (see Section 3.3).
TABLE 2Flux measurements for 2FGL J0523.3 − E F low /10 − F high /10 − F total /10 − (GeV) (erg cm − s − ) (erg cm − s − ) (erg cm − s − )0.13 3.2 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Note . — F = E dN/dEdN/dE