Mon.Not.R.Astron.Soc.000,000–000 (0000) Printed12January2016 (MNLATEXstylefilev2.2) Suzaku view of Be/X-ray binary pulsar GX 304-1 during Type I X-ray outbursts 6 ⋆ 1 Gaurava K. Jaisawal , Sachindra Naik† and Prahlad Epili‡ 0 Astronomy and Astrophysics Division, Physical Research Laboratory, Navrangapura, Ahmedabad - 380009, Gujarat, India 2 n a J 1 1 ABSTRACT ] R WereportthetimingandspectralpropertiesofBe/X-raybinarypulsarGX304-1 S by using two Suzaku observations during its 2010 August and 2012 January X-ray . outbursts. Pulsations at ∼275 s were clearly detected in the light curves from both h the observations. Pulse profiles were found to be strongly energy-dependent. During p 2010 observation, prominent dips seen in soft X-ray (610 keV) pulse profiles were - o found to be absent at higher energies. However, during 2012 observation, the pulse r profiles were complex due to the presence of several dips. Significant changes in the st shape of the pulse profiles were detected at high energies (>35 keV). A phase shift a of ∼0.3 was detected while comparing the phase of main dip in pulse profiles below [ and above ∼35 keV. Broad-band energy spectrum of pulsar was well described by a 1 partially absorbed Negative and Positive power-law with Exponential cutoff (NPEX) v model with 6.4 keV iron line and a cyclotron absorption feature. Energy of cyclotron 8 absorption line was found to be ∼53 and 50 keV for 2010 and 2012 observations, re- 4 spectively, indicating a marginal positive dependence on source luminosity. Based on 3 theresultsobtainedfromphase-resolvedspectroscopy,theabsorptiondipsinthepulse 2 profilescanbeinterpretedasduetothepresenceofadditionalmatteratsamephases. 0 Observed positive correlation between cyclotron line energy and luminosity, and sig- . 1 nificantpulse-phasevariationofcyclotronparametersarediscussedinthe perspective 0 of theoretical models on cyclotron absorption line in X-ray pulsars. 6 1 Key words: pulsars: individual (GX 304-1) – stars: neutron – X-rays: stars : v i X r 1 INTRODUCTION datafrom Vela5B,theorbital period of theGX 304-1 was a reportedtobe132.5d(Priedhorsky&Terrell1983).Ashell Be/X-ray binary pulsar GX 304-1 was discovered during starwithvisualmagnitudeof15wasdiscoveredintheX-ray hard X-ray sky surveys with balloon observations in 1967 error box of the neutron star and identified as the optical (McClintock et al. 1971). The X-ray source was detected companion of the pulsar (Mason et al. 1978). High resolu- with successive space missions and recognized as 3U 1258- tionopticalspectroscopyestablishedthespectralclassofthe 61,4U1258-61&2S1258-613(Giacconietal.1974;Bradtet companionasBestaroftypeB2Vnewhichisatadistance al.1977; Formanet al.1978). Usingdatafrom SAS−3ob- of 2.4±0.5 kpc(Parkes et al. 1980). servations,X-raypulsationsat∼272swerediscoveredinthe source(McClintocketal.1977).Spectralinvestigationofthe pulsar, carried out from balloon observations, showed that GX304-1wasmonitoredwithEXOSAT coveringadu- the continuum spectrum in 18-35 keV range was described rationofanexpectedoutburstin1984July/August(Pietsch by a power-law (Maurer et al. 1982). Later, a power-law et al. 1986). However, no X-ray outburst was detected dur- modelmodifiedwithhighenergycutoffwasusedtoexplain ing the EXOSAT monitoring campaign. In contrast, the the 2-40 keV continuum spectrum obtained from HEAO 1 source flux was estimated to be significantly low e.g. by a observation of thepulsar(White,Swank &Holt 1983). An- factorof25,thanthequiescentfluxlevel.Thisobservedpe- alyzing the periodicity of the X-ray outbursts in 7 years of culiarity was characterized as the X-ray “off” state of the pulsar. Long term optical monitoring of the Be companion starin 1978-1988 suggested amajor change(loss) intheBe ⋆ [email protected] † [email protected] envelop or circumstellar disk, the consequence of which is ‡ [email protected] considered as the possible cause of the X-ray “off” state in 2 G. K. Jaisawal, S. Naik and P. Epili 0.3 2010 August Outburst 0.3 2012 January Outburst V) e k 50 0.2 0.2 − 5 1 m s (−2−1 0.1 0.1 c nts u o C 0 0 5.541×104 5.542×104 5.543×104 5.544×104 5.594×104 5.596×104 Time (MJD) Time (MJD) 1.4 1.4 0.5−10 keV XIS−0 0.5−10 keV XIS−0 1.2 1.2 1 1 0.8 0.8 0.6 0.6 1.4 1.4 nsity 1.2 10−70 keV HXD/PIN 1.2 10−70 keV HXD/PIN e nt 1 1 d I ze 0.8 0.8 ali m 0.6 0.6 Nor 1.4 40−200 keV HXD/GSO 40−200 keV HXD/GSO 4 1.2 2 1 0.8 0 0.6 0 0.5 1 1.5 2 0 0.5 1 1.5 2 Pulse Phase Pulse Phase Figure 1.Swift/BAT lightcurvesofGX304-1inthe15-50keV energyband,from2010 July28(MJD 55405)to2010September 06 (MJD55445)and2011December30(MJD55925)to2012February18(MJD55975)areshowninbothsidesoffirstpanel,respectively. The arrowmarks inboth sides of firstpanel shows the date of Suzaku observations of GX 304-1 duringoutburst. Corresponding pulse profiles in 0.5-10 keV (XIS-0; second panel), 10-70 keV (PIN; third panel) and 40-200 keV (GSO; fourth panel) obtained from the backgroundsubtractedlightcurvesofboththeobservationsareshowninbothsidesoffigure.PhasezerowasarbitrarilychosenatMJD 55421.76 and55957.4326 forfirstandsecond observations, respectively. Theerrorsinthepulseprofilesareestimatedfor1σ confidence levelandtwopulsesareshownforclarity. GX 304-1 (Corbet et al. 1986; Pietsch et al. 1986; Haefner ∼4.7×1012 G. A positive correlation between cyclotron en- 1988). ergyandluminositywasseenduring2012January-February outburstwithINTEGRAL(Klochkovetal.2012).Malacaria After28yearsofquiescence,anX-rayoutburstwasde- et al. (2015) performed timing and spectral analysis of the tected from GX 304-1 with the INTEGRAL observatory in pulsar using the same INTEGRAL data. The shape of the 2008 June (Manousakis et al. 2008) after which the source pulse profiles obtained from these observations were found was found to be active in X-rays. Since then, several X-ray to be similar in 20-40 keV, 40-60 keV and 18-80 keV en- outburstshavebeendetectedin thepulsarwithSwift/BAT ergy ranges. Phase-resolved spectroscopy was carried out and MAXI (Yamamoto et al. 2009, 2012; Krimm et al. by stacking multiple spectra of different fluxes. From this 2010; Mihara et al. 2010). Using the Rossi X-Ray Tim- analysis, the cyclotron absorption line energy was found to ingExplorer (RXTE)observationsduring2010Augustout- be nearly constant (within errors) with pulse phases, ex- burst, energy and luminosity dependence of pulse profiles cept about 10% variation at one phase bin (Malacaria et were found in GX 304-1 (Devasia et al. 2011). Apart from al.2015).Theevolutionofthepulseperiod withluminosity the evolution of pulse profiles, a quasi-periodic oscillation wasstudiedbyusingdatafromMAXI/GSC,RXTE/PCA, (QPO)at∼0.125Hzwasdetectedwithharmonicsinseveral Swift/XRT and Fermi/GBM observations during a series RXTE/PCAobservationsduringthisoutburst.Thepulsar of outbursts from 2009 to 2013 (Postnov et al. 2015; Sug- spectrum in 3-30 keV range was described with a partial izaki et al. 2015). The observed pulse period variation was coveringhighenergycutoffpower-lawmodel(Devasiaetal. interpreted in terms of binary modulation along with the 2011).Duringthesameoutburstin2010August,acyclotron spinning-up of the neutron star. The orbital parameters of absorption feature at ∼54 keV was detected in the pul- the binary system were estimated to be – orbital period sar spectrum (Yamamoto et al. 2011) and the correspond- = 132.19 d, epoch at the periastron = MJD 55425, pulse ing magnetic field of the neutron star was estimated to be Suzaku view of GX 304-1 3 1.4 1.4 1.4 GSO 50−60 keV 1.2 1.2 1.2 1.2 1 1 1 1 0.8 0.8 0.8 0.6 XIS−0 0.5−2 keV 0.6 XIS−0 10−12 keV 0.6 PIN 30−35 keV 0.8 1.4 1.4 PIN 12−15 keV 1.4 1.2 1.2 1.2 1.2 1 1 1 1 0.8 0.8 0.8 0.6 XIS−0 2−4 keV 0.6 0.6 PIN 35−40 keV 0.8 GSO 60−80 keV y 1.4 1.4 PIN 15−20 keV 1.4 2 nsit 1.2 1.2 1.2 e nt 1 1 1 1 d i ze 0.8 0.8 0.8 mali 0.6 XIS−0 4−6 keV 0.6 0.6 PIN 40−45 keV 0 GSO 80−100 keV Nor 1.4 1.4 PIN 20−25 keV 1.4 4 GSO 100−120 keV 1.2 1.2 1.2 1 1 1 2 0.8 0.8 0.8 0 0.6 XIS−0 6−8 keV 0.6 0.6 PIN 45−50 keV 1.4 1.4 PIN 25−30 keV 1.4 GSO 120−150 keV 10 1.2 1.2 1.2 1 1 1 0 0.8 0.8 0.8 −10 0.6 XIS−0 8−10 keV 0.6 0.6 PIN 50−60 keV 0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2 Pulse Phase Pulse Phase Pulse Phase Pulse Phase Figure 2. Energy-resolved pulse profiles of GX 304-1 obtained from XIS-0, HXD/PIN and HXD/GSO light curves at various energy ranges,duringfirstSuzakuobservationin2010Augustoutburst.Thepresenceofabsorptiondipsinprofilescanbeseenatvariouspulse phases.Theerrorbarsrepresent1σ uncertainties.Twopulsesineachpanelareshownforclarity. period≃275.45 s,orbitaleccentricity≃0.5,a sini≃500– (0.2-600keV)intheelectromagneticspectrumbyusingtwo x 600 light-s and ω at periastron ≃ 122.5◦–130◦ (Sugizaki et sets of instruments e.g. the X-ray Imaging Spectrometers al. 2015). (XIS;Koyamaetal.2007)andHardX-rayDetectors(HXD; Inthiswork,thetimingandbroad-bandspectralprop- Takahashietal.2007).TheXISsareimagingCCDcameras erties of GX 304-1 were presented in detail by using two andcover0.2-12keVenergyrange.AmongfourXISs,three Suzakuobservationsduringitsoutburstsin2010Augustand XISs (XIS-0, XIS-2, XIS-3) are front-illuminated whereas 2012 January. Earlier, the evolution of pulse profiles up to one XIS (XIS-1) is back-illuminated. The effective areas of ∼30 keV had been reported by using RXTE/PCA obser- frontandback-illuminatedXISsare340cm2and390cm2at vations during 2010 outburst (Devasia et al. 2011). Using 1.5keV,respectively.FieldofviewofXISdetectorsis18′× ′ Suzaku observations, the evolution of pulse profiles up to 18 in full window mode. The HXD unit of Suzaku consists ∼150 keV has been presented in this paper. The observed of two sets of detectors such as HXD/PIN and HXD/GSO. changes in theshape of pulse profiles close to thecyclotron HXD/PIN is a silicon diode detector covering 10-70 keV absorption line are also being discussed. Although spectral energy range, whereas HXD/GSO is a crystal scintillator studies of the pulsar during its 2012 outburst has been re- detectorworkingin40-600keVenergyrange.Effectiveareas ported in 5-100 keV range by using INTEGRAL observa- for HXD/PIN and HXD/GSO are 145 cm2 at 15 keV and tions (Malacaria et al. 2015), Suzaku observations provide 315cm2at100keV,respectively.FieldofviewofHXD/PIN a better opportunity to investigate a detailed spectral and and HXD/GSO (upto 100 keV) is 34′×34′. timingstudyinbroadenergyrange(1-150keV).Thedetails ofobservations,dataanalysis,resultsandinterpretationsare GX304-1wasobservedwithSuzakuduringitsoutbursts described in following sections of thepaper. in 2010 August and 2012 January.Swift/BAT light curves ofthepulsarin15-50keVrange,coveringtheoutburstsare shown in top panels of Fig. 1. Arrow marks in both panels indicate the date of Suzaku observation of GX 304-1 dur- 2 OBSERVATION AND ANALYSIS ingrespective outbursts. During2010 Augustoutburst, the Suzaku, the fifth Japanese X-ray mission, was launched by Suzaku observation was carried out at the peak of outburst JapanAerospaceExplorationAgency(JAXA)on2005July whereas thesecond observation was madeduringthedecay 10 (Mitsuda et al. 2007). It offers a broad energy coverage phase of the 2012 January outburst. The first observation 4 G. K. Jaisawal, S. Naik and P. Epili 1.4 1.4 1.4 1.2 1.2 1.2 1.2 1 1 1 1 0.8 0.8 0.8 0.6 XIS−0 0.5−2 keV 0.6 XIS−0 10−12 keV 0.6 PIN 30−35 keV 0.8 GSO 50−60 keV 1.4 1.4 1.4 GSO 60−80 keV 1.5 1.2 1.2 1.2 1 1 1 1 0.8 0.8 0.8 0.6 XIS−0 2−4 keV 0.6 PIN 12−15 keV 0.6 PIN 35−40 keV y 1.4 1.4 1.4 4 GSO 80−100 keV nsit 1.2 1.2 1.2 e nt 1 1 1 2 d i ze 0.8 0.8 0.8 mali 0.6 XIS−0 4−6 keV 0.6 PIN 15−20 keV 0.6 PIN 40−45 keV 0 Nor 1.4 1.4 PIN 20−25 keV 1.4 2 1.2 1.2 1.2 1 1 1 1 0.8 0.8 0.8 0 0.6 XIS−0 6−8 keV 0.6 0.6 PIN 45−50 keV GSO 100−120 keV 1.4 1.4 GSO 120−150 keV 1.2 1.2 PIN 25−30 keV 1.5 1.5 1 1 1 1 0.8 0.8 0.6 XIS−0 8−10 keV 0.6 0.5 PIN 50−60 keV 0.5 0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2 Pulse Phase Pulse Phase Pulse Phase Pulse Phase Figure 3. Energy-resolved pulse profiles of GX 304-1 obtained from XIS-0, HXD/PIN and HXD/GSO light curves at various energy ranges, during second Suzaku observation in 2012 January outburst. The presence of absorption dips in profiles at higher energies can beseenatvariouspulsephases.Theerrorbarsrepresent1σ uncertainties.Twopulsesineachpanelareshownforclarity. was performed on 2010 August 13, in ‘HXD nominal’ po- by applying attitude correction script aeattcor.sl1. Subse- sition with an effective exposure of ∼5.1 ks and ∼12.9 ks quently, the pile-up estimation was made for XISs data by for XISand HXDdetectors, respectively. However, thesec- usingS-langscriptpile estimate.sl2.Forfirstobservation,a ondobservationwasmadeduring2012January31-February pile-upof∼22%,∼18%and∼28%wasfoundatthecenters 02 for a longer effective exposures of ∼16.5 ks for XIS and of XIS-0, XIS-1 and XIS-3, respectively. Therefore, an an- ∼58.7 ks for HXD. The second observation was performed nulusregionwithinnerandouterradiiof60′′ and180′′ was in ‘XIS nominal’ position. During both observations, the selected to reduce the effect of pile-up to 64%. During the XIS detectors were operated in “burst” clock mode with second observation, the pile-up was estimated to be 17%, ‘1/4’windowoptionyielding0.5stimeresolution.Thepub- 13% and 18% at the centers of XIS-0, XIS-1and XIS-3,re- ′′ ′′ liclyavailablearchivaldata(observationIDs:905002010and spectively.Anannularregionwith40 innerand180 outer 406060010) wereusedinthepresentstudy.Heasoftsoftware radii was used to reduce the pile-up effect to 64% for the package of version 6.12 and calibration database (CALDB) secondobservation.Theseannulusregionswereusedforthe for XIS and HXD released on 2014 February 03 and 2011 extraction of source light curves and spectra from cleaned September13, respectively,were used in thedata analysis. XISeventdata.TheXISbackgroundlightcurvesandspec- trawereextractedfromtheeventdatabyselectingacircular region away from the source. Response matrices and effec- tiveareafilesforXISswerecreatedbyusing‘xisrmfgen’and The unfiltered event data files were reprocessed by ap- ‘xissimarfgen’ tasksofFTOOLS,respectively.HXDbeinga plyingthe‘aepipeline’packageofFTOOLS.Thecleanevent non-imaging detector system, source light curves and spec- files generated after the reprocessing of XIS and HXD un- tra were obtained from cleaned HXD/PIN and HXD/GSO filtered event data were used in our analysis. The arrival eventdatabyusingXSELECT package. However,PIN and timesoftheX-rayphotonsrecordedinXISandHXDevent GSObackgroundlightcurvesandspectrawereaccumulated datawerecorrectedforsolarsystembarycenterbyapplying from simulated non-X-ray background event files provided ‘aebarycen’ task of FTOOLS. The light curves and spectra ofthepulsarwereextractedfrom clean eventdatabyusing the XSELECT package of FTOOLS. The XIS event data 1 http://space.mit.edu/ASC/software/suzaku/aeattcor.sl werecorrectedfortheeffectofthermalflexingandwobbling 2 http://space.mit.edu/ASC/software/suzaku/pile estimate.sl Suzaku view of GX 304-1 5 bytheinstrumentteam.AcorrectionforcosmicX-rayback- ground (CXB3) was also applied to PIN spectrum. In our 10 2010 August sJJpuaenlycutararanyldaonb2a0sel1yr1vsiasJt,uioHnneXs,wDre/erPsepIeNucsterivdeseplfyoo.rnFs2oe0r1fiG0leSsAOruegdleuaaststae,darneidnsp22o00n11s02e Counts s keV−1−1 0.11 a2n01d0eAffeucgtuivset oabresearvfialetsionresl.eased in 2010 May were used for malized 0.01 Nor 5 0 3 RESULTS χ −5 5 3.1 Timing Analysis 0 χ As described above, source and background light curves −5 with 1 s time resolution were extracted from barycentric 1 10 100 corrected XIS-0, PIN and GSO event data for both the Energy (keV) observations. The χ2-maximization technique was used to Figure 4. Energy spectrum of GX 304-1 in 1-130 keV energy estimate the pulse period of the pulsar. Pulsations at pe- range obtained with the XIS-0, XIS-3, PIN and GSO detectors riods of 275.45±0.05 and 274.88±0.01 s were detected in fromthefirstSuzakuobservationin2010Augustoutburst,along the source light curves obtained from the first and second withthebest-fitmodelcomprisingapartialcoveringNPEXcon- Suzakuobservations,respectively.Quotederrorsinpulsepe- tinuum model, a Gaussian function for iron emission line and a cyclotron absorption component. The middle and bottom pan- riod are calculated for 90% confidence level. The estimated elsshow thecontributions of the residualsto χ2 foreach energy pulseperiodswereusedtogeneratepulseprofilesfromback- binforthepartialcoveringNPEXcontinuummodelwithoutand groundsubtractedlightcurvesfrom correspondingobserva- withcyclotroncomponent inthemodel,respectively. tions.Pulseprofilesin0.5-10 keV(XIS-0),10-70keV(PIN) and 40-200 keV (GSO) energy ranges for both the obser- vations are shown in second, third and fourth panels of phase range (panels in second column of Fig. 2). We sug- Fig. 1, respectively. Pulse profiles were generated by using gest that thisbroadening of theminima in thepulseprofile 55421.7600 MJD as epoch (phase zero) for the first obser- is the possible cause of observed ∼0.1 phase shift in the vation where as 55957.4326 MJD was used for the second soft(0.5-10keV)andhardX-ray(10-70keV)pulseprofiles. observation. A careful inspection of Fig. 2 & 3 showed that the pulse Strong energy dependence of pulse profiles can be profiles were complex due to presence of several absorption clearly seen during both observations (Fig. 1) . Absorption features (dips) at various pulse phases and strongly energy dipsatcertainphaseswereseeninthesoftX-raypulsepro- dependent up to ∼12 keV beyond which the shape of the files (0.5-10 keV range). However, these dips disappeared profilesbecamesimple upto∼35keV.Themain dip in the from the pulse profiles in 10-70 keV range. The pulsations pulseprofilesinthisenergyrange(12-35keV)wasobserved were absent or marginally seen in 40-200 keV pulse profiles to be phase-shifted by ∼0.1 phase compared to that in soft obtainedfrom GSOlight curves.Apartfrom theenergyde- X-ray profiles. In pulse profiles beyond ∼35 keV, the main pendence,thepulseprofilesare also found tobeluminosity dip at ∼0.1 phase appeared to be filled-up gradually with dependent.ThetoppanelsoftheFig.1showthattheSuzaku increaseinenergy.Alongwiththeincreaseinthenormalized observations of the pulsar were carried out at different lu- intensity at ∼0.1 phase (Fig. 2 & 3), a significant decrease minosity levels e.g. the source was comparatively brighter inintensitywasobservedat∼0.7-0.8phaserangewhichap- during the2010 August observation than the2012 January peared as the main dip in hard X-ray pulse profiles. These observation.However,theshapeof thepulseprofilesinsoft hard X-ray pulse profiles (>40 keV) appeared to be single- andhardX-rays(secondandthirdpanels)aredifferentdue peaked and pulsations were detected up to ∼120 keV. tothepresenceofabsorptiondipsordip-likefeatures.Along The energy resolved pulse profiles of GX 304-1 were withtheenergyandluminositydependenceofthepulsepro- found to be complex during both the Suzaku observations. filesinGX304-1,aphaseshift of∼0.1(seeFig.1)wasalso Presenceofmultiplenarrow andprominentabsorption dips found between the soft (XIS) and hard (PIN) X-ray pulse wereseenuptoashighas∼50keV.Beyondthisenergy,the profiles obtained from both theobservations. profiles appeared relatively simple. Presence of prominent To investigate the evolution of pulse profiles with en- dips up to higher energies and sudden change in phase of ergy duringboth Suzaku observations, we generated energy main dip in pulse profiles beyond ∼35 keV made it inter- resolvedpulseprofilesinvariousenergybandsandareshown estingtoinvestigatethepropertiesofpulsarthroughphase- inFig.2& 3forfirstandsecondobservations,respectively. averaged and phase-resolved spectroscopy. It can be seen from Fig. 2 that a prominent and narrow absorption dip was present in pulse profiles up to ∼8 keV. Beyond this energy, the peak in the pulse profiles prior to 3.2 Spectral Analysis thedip (6 0.3 phase; left panels of Fig. 2) disappeared and 3.2.1 Pulse-phase-averaged spectroscopy broadenedtheminimainthepulseprofileto0.05-0.25pulse Phase-averaged spectroscopy was performed by using spec- tra accumulated from XIS-0, XIS-1, XIS-3, PIN and GSO 3 http://heasarc.nasa.gov/docs/suzaku/analysis/pin cxb.html dataobtainedfromboththeobservations.Earlierdescribed 6 G. K. Jaisawal, S. Naik and P. Epili Table 1.Best-fitting parameters (with 90% errors)obtained from the spectral fitting of Suzaku observations of GX 304-1 during2010 Augustand2012Januaryoutbursts.Model-1:partialcoveringNPEXmodelwithGaussiancomponent;Model-2:partialcoveringNPEX modelwithGaussiancomponentandcyclotronabsorptionline. Parameter 2010August 2012January Model-1 Model-2 Model-1 Model-2 NH1a 1.04±0.02 1.02±0.02 0.98±0.02 0.97±0.02 NH2b 13.2±1 13.7±1.2 5.7±0.5 5.3±0.5 Cov.fraction 0.35±0.02 0.32±0.02 0.25±0.02 0.23±0.02 Photonindex 0.6±0.03 0.57±0.03 0.42±0.02 0.43±0.02 Ecut (keV) 6.6±0.1 7.1±0.2 6.8±0.1 7.4±0.2 Felineenergy(keV) 6.41±0.02 6.41±0.02 6.41±0.01 6.41±0.01 Eq.widthofFeline(eV) 44±7 43±8 23±2 23±3 Cyclotronlineenergy(keV) – 53.2±0.8 – 50±1 Widthofcyclotronline(keV) – 6.5+−21..16 – 5.5+−21..68 Depthofcyclotronline – 0.8±0.1 – 0.5±0.1 Fluxc (1-10keV) 9.2±0.5 9.2±0.6 4.9±0.2 4.9±0.2 Fluxc (10-70keV) 24.5±1.5 24.4±2.0 9.6±0.5 9.6±1.1 Fluxc (70-130keV) 0.2±0.1 0.2±0.1 – – χ2 (dofs) 857(527) 638(524) 863(588) 765(585) a :Equivalenthydrogencolumndensityinthesourcedirection(in1022 atoms cm−2 units), b :Additionalhydrogencolumndensity(in1022 atomscm−2 units), c :Absorptioncorrectedfluxinunitsof10−9 ergscm−2 s−1. afactorof8from7to10keV,whereasfor2012Januaryob- 10 2012 January servation,theXISspectrawerebinnedbyafactorof5upto Counts s keV−1−1 0.11 1b405ykkaeeVVfa.ctPtooIrN50osfpk2eeVcftrroaamnfdro2am5fkabecoVttohrttoohf4e55ofbkrseoeVmrv,a5at0ifoatnocst7ow0reorkefeVb3i.nfrTnohemde malized 0.01 GteSamO.spDeacttarainwe1r.e7-g1r.o9upkeeVdaasnsdug2g.e2s-t2e.4dbkyeVinsetnreurmgyenrtaantgioens Nor were ignored from the spectral fitting due to the presence 5 of known Si and Au edge features in the XIS spectra. All thespectralparametersexcepttherelativenormalizationof χ 0 detectors were tied together duringthe fitting. −5 5 It has been found that the phase-averaged spectra of χ 0 Be/X-ray binary pulsars during outbursts have been de- scribed by power-law (e.g. GX 304-1; Maurer et al. 1982), −5 1 2 5 10 20 50 highenergycutoffpower-law(e.g.4U0115+63,4U1145-61, Energy (keV) GX 304-1; White et al. 1983), Fermi Dirac cutoff power- Figure 5. Energy spectrum of GX 304-1 in 1-70 keV energy law(e.g.X0331+53,A0535+26; Tanaka1986),NPEX(e.g. range obtained with the XIS-0, XIS-3 and PIN detectors from 4U 0115+63, X0331+53, Cep X-4; Makishima et al. 1999) the second Suzaku observation in 2012 January outburst, along continuummodels.However,recentlyithasbeenfoundthat withthebest-fitmodelcomprisingapartialcoveringNPEXcon- whileperformingphase-resolvedspectroscopyondatataken tinuum model, a Gaussian function for iron emission line and a duringX-rayoutburstsofthesepulsars,abovemodelsdonot cyclotron absorption component. The middle and bottom pan- els show the contributions ofthe residualsto χ2 foreach energy yield acceptable fit at all phase bins, specifically at phases binforthepartialcoveringNPEXcontinuummodelwithoutand of prominent and narrow absorption dips in the pulse pro- withcyclotroncomponent inthemodel,respectively. files (Naik et al. 2011; Paul & Naik 2011; Naik et al. 2013 and references therein). To investigate the changes in spec- tralparametersatallphasebins(dipandnon-dipphasesof thepulseprofiles),apartialcoveringabsorptioncomponent procedures were followed to obtain source and background has been added to above standard continuum models. Ad- spectra, response matrices and effective area files for corre- dition of this component to the continuum model resulted sponding detectors. After appropriate background subtrac- in gettingacceptable fittoall phasebinsand explained the tion, simultaneous spectral fitting was carried out by using causeofthenarrowandprominentdipsinthepulseprofiles. XSPEC v12.7 package. As the pulsar was bright during the Partial covering absorption model consists of two different firstobservation,broad-bandspectralfittingwascarriedout power-lawcomponentswithsamephotonindexbutdifferent in1-130keVrange.However,datain1-70keVenergyrange normalizations,beingabsorbedbydifferentcolumndensities were used for simultaneous fitting for second observation. TheXISspectrafrom2010Augustobservationwerebinned (NH1 & NH2), respectively (Endoet al. 2000). byafactorof4upto3keV,afactorof6from3to7keVand In the beginning, standard continuum models such Suzaku view of GX 304-1 7 eV) 54 V)10 Cyclotron line energy (k 5523 + 2010 Cyclotron line width (ke 468 + 2010 7 7.1 7.2 52 53 54 High energy cutoff (keV) Cyclotron line energy (keV) 52 V) V) 8 Cyclotron line energy (ke 455901 + 2012 Cyclotron line width (ke 46 + 2012 7.2 7.3 7.4 7.5 7.6 49 50 51 52 High energy cutoff (keV) Cyclotron line energy (keV) Figure 6.The χ2 confidence contours between highenergy cutoff, cyclotron lineenergy and width, obtained fromthe phase averaged spectrafittedbythepartialcoveringNPEXmodelwithcyclotroncomponentduring2010August(upperpanel)and2012January(lower panel)Suzakuobservations.Theinnermosttooutermostcontoursrepresent68%,90%and99%confidence levels,respectively.The“+” signindicates thebestfitvaluesforbothparameters. as high energy cutoff power-law (HECUT; White et al. was detected in spectrum during both the observations. A 1983), Fermi Dirac cutoff power-law (FDCUT; Tanaka weak iron emission edgelikefeature wasfound in theresid- 1986), NewHcut (Burderi et al. 2000), cutoff power-law, ualduring thefittingof second observation. This was mod- NPEX (Makishima et al. 1999) and Thermal Comptoniza- eled by the addition of an edge component at ∼7.7 keV in tion model (CompTT; Titarchuk 1994) were used in our thespectral model. spectral fitting to describe the continuum spectrum of GX304-1.Duetothepresenceofnarrowabsorptiondipsin thepulse profiles of thepulsar (previous section) as seen in Anabsorption likefeature at ∼54keVwasclearly seen otherBe/X-raybinarypulsars,apartialcoveringabsorption in residuals obtained from spectral fittings of both the ob- component was added to above continuum models. Among servations. Addition of a cyclotron absorption component thesemodels,partial coveringNPEXcontinuummodelwas (‘CYCLABS’ in XSPEC package) in the partial covering foundtofitthesourcespectraobtainedfromboththeobser- NPEX continuum model improved spectral fitting further vationsbetterthanallothercontinuummodels.Weselected with reduced χ2 of ∼1.5 for both the observations. The thismodeltouseinphase-averagedandphase-resolvedspec- values of reduced χ2, though acceptable, are found to be tral analysis of both theSuzaku observations of GX 304-1. large. Investigating the residuals in 1-10 keV range, we no- The NPEX continuum model is a combination of two ticed a marginal cross calibration uncertainties present be- power-lawswithpositiveandnegativeindicesandahighen- tweenXIS-1(backilluminatedCCD)andXIS-0&3(front- ergy cutoff. This model is an approximation of the unsatu- illuminated CCD) spectra. While fitting the broad-band ratedthermalComptonizationinhotplasma.Theanalytical spectra without considering data from XIS-1, the values of form of NPEX model is reduced χ2 obtained are 1.22 and 1.31 for first and sec- ond observations, respectively. Best-fit model parameters obtained from simultaneous spectral fitting of XIS-0 & 3, NPEX(E)=(N1E−α1+N2E+α2) exp − E PINandGSOdataaregiveninTable1.Theenergyspectra (cid:18) kT(cid:19) of the pulsar along with the best-fit model components are where N1, α1 and N2, α2 are the normalization and shown in Fig. 4 and 5. The middle and bottom panels in photon index of the negative and positive power laws, re- eachfigureshowtheresidualstothebest-fitmodelwithout spectively. kT represents the cutoff energy in the unit of andwiththeadditionofcyclotronabsorptionfeatureinthe keV. The photon index of positive power law is fixed at a continuum model, respectively. The χ2 confidence contours value of 2, representing Wien’s peak. During the spectral were plotted to check the dependence of the cyclotron line modeling, we found that the partial covering NPEX model energy on high energy cutoff and cyclotron width and are described the 1-130 keV (2010 observation) and 1-70 keV shown in the top and bottom panels of Fig. 6 for first and (2012observation)spectraofthepulsarwell.Inadditionto secondobservations,respectively.Wedidnotfindanystrong the continuum, iron fluorescence emission line at ∼6.4 keV degeneracy among these spectral parameters. 8 G. K. Jaisawal, S. Naik and P. Epili 1.5 1.5 y y sit 1 sit 1 n n nte XIS−0 nte XIS−0 d I 1.5 d I 1.5 e e z z ali 1 ali 1 m m or PIN or PIN N N NH20 cm)22−2 123000 Dcyclotron 12 1 0 ( 0.6 55 CoveringFraction00..24 Ecyclotron(keV) 445050 12 1−10 keV 1 nx 10 oe PhotInd0.5 s)−1 68 30 m−2 Ecut(keV) 1200 Flux0 ergs c−9 234000 10−70 keV 1 ( 0 0.5 1 1.5 2 0 0.5 1 1.5 2 Pulse Phase Pulse Phase Figure 7. Spectral parameters obtained from the phase-resolved spectroscopy of GX 304-1 during the Suzaku observation in 2010 August. The first and second panels in both sides show pulse profiles of the pulsar in 0.5-10 keV (XIS-0) and 10-70 keV (HXD/PIN) energy ranges.The values of NH2,covering fraction, power-lawphoton indexandcutoff energy(Ecut)areshowninthird,fourth, fifth andsixth panels inleftside,respectively. Thecyclotron lineparameters such as depth (third panel), lineenergy (fourthpanel), source fluxin1-10keV(fifthpanel) and10-70keV(sixthpanel) areshowninrightsideofthefigure.Solidcirclesinthe fourthpanel inright sideindicate that the cyclotron lineenergy was fixed forcorresponding phase-binat the phase-averaged value. The errorsinthe figure areestimatedfor90%confidence level. 3.2.2 Pulse-phase-resolved spectroscopy spectroscopy.Thewidthofthecyclotronabsorptionlinewas fixed at the phase averaged value to constrain the feature. Pulse profiles of GX 304-1 are found to be different even Spectralparameters obtained from fittingtheXISand PIN when compared at same energy range from both Suzaku phase resolved spectra are shown in Fig. 7 and 8 for 2010 observations. Therefore, it is interesting to perform phase- August 13 and 2012 January 31 observations, respectively. resolved spectroscopy to probe the changes in spectral pa- Pulse profiles obtained from XIS-0 and PIN event data of rameters with pulse phase and then compare with the ob- each observation are also shown in top two panels of these servationsat differentluminositylevel.Asthefirstobserva- figures. tion was for a relatively short exposure (∼13 ks for HXD) During both the observations, all the spectral parame- compared to the second observation (∼59 ks for HXD), ters showed significant variability with pulse-phase of the phase-resolved spectroscopy was carried out by accumulat- pulsar. Fig. 7 (2010 August observation) shows that the ing source spectra in 10 and 16 pulse-phase bins for first value of additional hydrogen column density (N ) varies H2 andsecond observations,respectively.ReprocessedXISand in 1-30 × 1022 cm−2 range over pulse phases. The value of PIN event data from both the observations were used to N wasincreasedfrom0.3phaseandbecamemaximumin H2 extract phase-resolved spectra by applying phase filter in 0.6-1.0 phase range. It can be seen that normalized inten- XSELECT package. Data from HXD/GSO detector were sity (XISpulse profile; top panel) was low at 0.6 phase and not used in the phase resolved spectroscopy due to lack of remain steady before reaching the minimum value at 1.0 sufficienthardX-rayphotonsineachphasebin.Background phase. Decrease in the normalized intensity of the pulsar spectraandresponsematrices usedin phase-averagedspec- during above pulse phase range and simultaneous increase troscopy were also used in phase-resolved spectral fitting. in thevalue of N confirm that theplateau like feature in H2 Simultaneous spectral fitting was carried out on phase- XISpulseprofilein0.6-1.0pulsephaserangewasduetothe resolved spectra obtained from both the observations by presence of additional absorbing material close to the pul- using partial covering NPEX continuum model along with sar.Thiscan also explain thepresenceofanabsorption dip thecyclotron absorption component and Gaussian function intheXISpulseprofileand anincreasein theN valuein H2 for the iron emission line. While fitting, the values of rela- 0.3-0.4phaserangeofthepulsar.AsincaseofN ,thecov- H2 tiveinstrumentnormalizations,equivalenthydrogencolumn eringfractionoftheadditionalabsorptionwasalsofoundto density (N ) and iron emission line parameters were fixed becomparativelyhighduringthedipandplateauregion.As H1 at the corresponding values obtained from phase-averaged expected, the values of power-law photon index and cutoff Suzaku view of GX 304-1 9 1.5 1.5 XIS−0 XIS−0 y y sit 1 sit 1 n n e e nt nt d I 1.5 d I 1.5 e PIN e PIN z z ali 1 ali 1 m m or or N N 80 1.5 NH20 cm)22−2 246000 Dcyclotron 0.51 1 (0.8 CoveringFraction000...246 Ecyclotron(keV) 445050 6 1−10 keV nx 1 5 PhotoInde0.5 xms)−2−1 34 20 Flugs c14 10−70 keV EcutkeV) 1105 0 er−9 1102 ( (1 8 5 0 0.5 1 1.5 2 0 0.5 1 1.5 2 Pulse Phase Pulse Phase Figure 8. Spectral parameters obtained from the phase-resolved spectroscopy of GX 304-1 during the Suzaku observation in 2012 January. The first and second panels in both sides show pulse profiles of the pulsar in 0.5-10 keV (XIS-0) and 10-70 keV (HXD/PIN) energy ranges. The values of NH2, covering fraction, photon index and cutoff energy (Ecut) are shown in third, fourth, fifth and sixth panels in left side, respectively. The cyclotron line parameters such as depth (third panel), line energy (fourth panel), source flux in 1-10 keV (fifth panel) and10-70 keV (sixth panel) are showninright sideof the figure. Theerrorsinthe figureare estimated for 90% confidence level. energy were high during the main dip in the XIS and PIN above phase range in the hard X-ray pulse profiles was dif- pulseprofiles.Thedepthofthecyclotronabsorptionfeature ferent. was found to be variable and single peaked. The energy of cyclotron absorption feature also showed variation over the pulse phase of the pulsar. The values of both the parame- 4 DISCUSSION AND CONCLUSIONS tersweremaximumatdipphaseofthepulsarandgradually decreased to minimum value in 0.5-0.8 phase range. Data PulseprofilesoftransientX-raybinarypulsarsarecomplex gaps in third and fourth panels in second column are due due to the presence of multiple absorption dips/features at tothenon-detectionofcyclotronfeaturesinthespectralfit- lower energies. These absorption features are strongly en- tingforcorrespondingphase-bins.Thefluxin1-10keVand ergydependent–prominentinsoftX-raypulseprofilesand 10-70 keVshows variation with pulsephaseand follows the gradually disappear at higher energies. Complex structures shape of pulse profile in respective energy bands. or absorption dips in pulse profiles originate due to the photo-electric absorption of soft X-ray photons by matter Though theshape of XISand PIN pulseprofiles of the present around the neutron star. This can be confirmed by pulsar were significantly different duringsecond Suzaku ob- investigating energy resolved pulse profiles and the evolu- servation, the variation of spectral parameters over pulse tion of spectral properties with pulse phases of the pulsar. phase was comparable. High values of absorption column Amongthe accretion powered X-ray pulsars, thepulse pro- densitywerefoundat phasesofabsorptiondipinXISpulse files of Be/X-ray binary pulsars duringoutbursts are found profile.Thevaluesofpower-lawphotonindex,cutoffenergy, to be complex. These pulsars show regular and periodic X- depth and energy of cyclotron absorption line, source flux ray outbursts (Type I) that are associated with the perias- in 1-10 keV and 10-70 keV were followed similar pattern tronpassageoftheneutronstar.Duringperiastronpassage, as seen during the 2010 August Suzaku observation of the the neutron star captures copious amount of matter from pulsar.Thoughthenumberofabsorption dipsandshapeof thecircumstellar disk of theBe companion star and under- thepulseprofilesofthepulsarduringboththeobservations goesX-rayoutbursts.TypicalX-rayluminosityofthepulsar weredifferent,presenceofadditional matterasthecauseof during Type I outburst is ∼1036−37 erg s−1 (Negueruela et absorptiondipsinthesoftX-rayprofilesissupportedbythe al.1998).IncaseofGX304-1,TypeIX-rayoutburstsoccur findings in the present work. However, low value (65×1022 ataperiodof132.5d.Usinginstrumentswithgoodtimeres- cm−2)ofabsorptioncolumndensityat∼0.1-0.2phaserange olutionandwide-bandenergycoveragecapabilitiesonboard (Fig. 2 & 3) suggested that the cause of absorption dip at Suzaku, we carried out a detailed study of the pulsar dur- 10 G. K. Jaisawal, S. Naik and P. Epili ing two Type I outbursts. We also performed pulse-phase 2013). This is interpreted as due to the presence of dense resolvedspectroscopyofGX304-1tostudytheevolutionof additional absorbers at various narrow pulse phase bins of spectral parameters during theseoutbursts. the pulsar. Though, dip-like features in GX 304-1 (present work) appeared up to ∼50 keV, the low value of additional columndensityatdipphasessuggeststhatabsorptionisnot 4.1 Pulse Profiles thecause of thesedips in hard X-ray pulseprofiles. Pulse profiles of GX 304-1 appeared to be different dur- Energy resolved pulse profiles of GX 304-1, obtained ing 2010 August and 2012 January Type I outbursts. Dur- from both the Suzaku observations revealed a significant ing2010 Augustobservation, theprofileswere complex due phase-shift (∼0.35) of the main dip in profiles at energies to the presence/absence of peaks and dips at several pulse below and above ∼35 keV. The observed phase-shift of the phases in soft and hard X-ray energy ranges. Due to this, main dip happened to occur at an energy close to the cy- a phase shift of ∼0.1 was visible between the soft and hard clotron absorption line energy in GX 304-1. Such type of X-ray pulse profiles (Fig. 1 & 2). However, the shape of effects e.g. phase-shifts (lags) or significant variations in the pulse profiles of the pulsar during 2012 January out- pulse profiles close to the cyclotron absorption line energy burst were significantly different compared to 2010 August are also seen in two other Be/X-ray binary pulsars such as outburst. Number of absorption dips in the pulse profiles V 0332+53 (Tsygankov et al. 2006) and 4U 0115+63 (Fer- was more during 2012 observation. The strength of these rigno et al. 2011). Using a numerical study on the effect dips was also prominent compared to the earlier observa- ofcyclotronresonantscatteringinhighlymagnetizedaccre- tion. These dips were present in the pulse profiles up to tion powered X-raypulsars, Sch¨onherret al. (2014) showed higher energies unlike the 2010 August observation, where that a strong change in the pulse profile is expected at the the dips were seen in the pulse profiles up to ∼10 keV. cyclotron absorption line energy. This change is attributed Pulse-phase resolved spectroscopy during both the obser- to the effects of angular redistribution of X-ray photons by vations revealed the presence of additional matter at cer- cyclotronresonantscatteringinastrongmagneticfieldcom- tain phases, causing the absorption dips in the soft X-ray bined with relativistic effects. In GX 304-1, we observed a pulseprofiles.During2012observation,morenumberofnar- significantchangeinthephaseofmaindipinthepulsepro- row absorption dips in the pulse profiles suggest the pres- file close to the cyclotron line energy. This detection, along ence of dense narrow streams of matter around the neu- with thereportedresultsfrom V0332+53 and 4U0115+63 tronstar. However,during2010 observation,relatively sim- supports the results obtained from the numerical study of pler profile with single absorption dip and a plateau phase Sch¨onherret al. (2014). suggests that the matter distribution around the neutron starisdifferentandrelatively simple.Theluminosityofthe 4.2 Spectroscopy pulsar during 2010 Suzaku observation was estimated to be higher(2.3×1037ergss−1)thanthe2012Suzakuobservation Broad-band spectra of accretion powered X-ray pulsars are (1×1037ergss−1).Thepeakluminosityof2010Augustout- describedbyseveralcontinuummodelssuchasHIGHECUT, burstwasalsofoundtobehighcomparetothe2012January FDCUT, cutoff power-law, NPEX, CompTT. Along with outburst (first panels of Fig. 1). As both the outbursts are the continuum model, additional components such as ab- Type I X-ray outbursts, observed difference in theshape of sorption due to matter present in the interstellar medium, the pulse profiles must be due to the difference in the sub- blackbody/bremsstrahlung for soft X-ray excess, Gaussian sequent mass accretion rate. functions for emission lines and cyclotron resonance scat- AsincaseofGX304-1,anothertransientBe/X-raybi- tering features (CRSF) are also needed to explain the ob- nary pulsar EXO 2030+375 was also observed at the peaks served spectrum. In case of several Be/X-ray binary pul- of 2007 May and 2012 May Type-I outbursts with Suzaku. sars, a partially absorption component is being used to de- The pulse profiles of EXO 2030+375 during these two out- scribe the presence of several absorption features (dips) at burstswere significantly different (Naik et al. 2013; Naik & certain phases in the pulse profiles (Paul & Naik 2011 and Jaisawal 2015). Duringtheluminous outburstin 2007 May, references therein). Detection of CRSF in the broad-band the pulse profile of EXO 2030+375 consisted of several ab- pulsar spectrum provides direct estimation of the magnetic sorption dips as seen in the pulse profiles of GX 304-1 dur- fieldofX-raypulsars.However,studiesofCRSFatdifferent ing2012Januaryoutburst.However,duringthelessintense pulsephasesofthepulsarscanrevealimportantinformation 2012Mayoutburst,thepulseprofileswererelativelysmooth. aboutthemagneticfieldgeometryaroundtheneutronstar. DifferenceinthemassaccretionrateduringthesetwoTypeI Therefore, broad-band spectroscopy of data obtained from outbursts was interpreted as the cause of different shape of the observations with high spectral capability instruments pulseprofilesinEXO2030+375. Suchtypeofpulseprofiles on-board Suzaku is an appropriate tool to understand the with multiple dips are also seen in other Be/X-ray binary properties of accretion powered X-ray pulsars. pulsarssuchasA0535+35(Naiketal.2008),GROJ1008-57 Thispaperreportsthebroad-bandphase-averaged and (Naik et al. 2011), 1A 1118–61 (Maitra et al. 2012 and ref- phase-resolved spectroscopy of GX 304-1 by using two erences therein). The dips in pulse profiles of these pulsars Suzaku observations during its Type I outbursts. The es- are originated due to the absorption of emitted radiation timated values of galactic equivalenthydrogen column den- by matter close to the neutron star. In most of the cases, sity in the source direction (N ) was same (within errors) H1 the dips in the pulse profiles are seen only in soft X-rays. during both the observations. However, the values of addi- However, there are a few Be/X-ray pulsar in which the ab- tionalcolumndensity(N -localtothepulsar),werefound H2 sorptiondipsareseenatsamephaseinthepulseprofilesup to be different during both the observations. The value of tohigherenergy(∼70keV)e.g.EXO2030+375(Naiketal. N washighduringthe2010Augustobservationcompared H2