Detection of fundamental and first harmonic cyclotron line in X-ray pulsar Cep X-4

0.24 MB·

by  Gaurava K. Jaisawal

Mon.Not.R.Astron.Soc.000,000–000 (0000) Printed12January2016 (MNLATEXstylefilev2.2)
Detection of fundamental and first harmonic cyclotron line
in X-ray pulsar Cep X-4
6
1 Gaurava K. Jaisawal⋆ and Sachindra Naik†
0
Astronomy and Astrophysics Division, Physical Research Laboratory, Navrangapura, Ahmedabad - 380009, Gujarat, India
2

n
a
J

1
1
ABSTRACT

]
R We report the broad-bandspectral properties of the X-ray pulsar Cep X-4 by us-
S ing a Suzaku observationin 2014July. The 0.8-70keV spectrum was found to be well
. described by three continuum models - Negative and Positive power-law with Expo-
h
nentialcutoff(NPEX),highenergycutoffpower-lawandCompTTmodels.Additional
p
components such as a cyclotron line at ∼28 keV and two Gaussian components for
-
o ironlinesat6.4and6.9keVwererequiredinthespectralfitting.Apartfromthese,an
r additional absorption feature at ∼45 keV was clearly detected in residuals obtained
t
s fromthespectralfitting.Thisadditionalfeatureat∼45keVwasclearlyseeninphase-
a resolved spectra of the pulsar. We identified this feature as the first harmonic of the
[ fundamental cyclotron line at ∼28 keV. The ratio between the first harmonic and

1 fundamentallineenergies(1.7)wasfoundtobeindisagreementwiththeconventional
v factorof2,indicatingthattheheightsoflineformingregionsaredifferentorviewedat
7 largerangles.The phase-resolvedspectroscopyof the fundamental andfirst harmonic
3 cyclotronlinesshowssignificantpulse-phasevariationofthelineparameters.Thiscan
3 be interpreted as the effect of viewing angle or the role of complicated magnetic field
2 of the pulsar.
0
. Key words: pulsars: individual (Cep X-4) – stars: neutron – X-rays: stars
1
0
6
1
:
v 1 INTRODUCTION theharmonicsofthefundamentalcyclotronlinearedetected
i only in a few cases (Pottschmidt et al. 2012 and references
X Cyclotron resonance scattering features (CRSFs) are gen-
therein).
r erally seen in the hard X-ray spectrum (10-100 keV) of ThetransientX-raypulsarCepX-4(GS2138+56) was
a the accretion powered X-ray pulsars with surface mag-
discoveredatapositionnearthegalacticplanewithOSO-7
netic field of ∼1012 G. These are absorption like features
in 1972 (Ulmeret al. 1973). X-ray pulsations were detected
which appear due to the resonant scattering of photons
in thesource at ∼66 sduringits1988 March outburst with
with electrons in quantized Landau levels (M´esz´aros 1992).
Ginga(Makinoetal.1988;Koyamaetal.1991).Acyclotron
The energy difference between these levels depends on the
absorption line feature at ∼30 keV was discovered in 1.2-
strengthofmagneticfieldandexpressedthroughtherelation
Ecyc=11.6B12×(1+z)−1 (keV)(withoutrelativisticcorrec- t3h7ekpeuVlsraarn(gMeishpaercatreutmal.ob19ta9i1n)e.dAfnroomptGicianlgsataorbosfe1rv4a.2tiomnago-f
tion); where B12 is the magnetic field in the unit of 1012 G nitudewas identified as the companion of Cep X-4 (Roche,
andzisthegravitationalred-shift.Detectionoffundamental
Green&Hoenig1997).Detailedopticalspectroscopyshowed
CRSFinthespectraofaccretionpoweredX-raypulsarspro-
thepresenceofBalmeremissionlinesthatcharacterizedthe
videsthedirectestimationoflocalmagneticfieldoftheneu-
companion as Be star of type B1-B2V. The distance of the
tronstarsinlineformingregion.Atthesametime,thestudy
binarywasestimatedas3.8kpc(Bonnet-Bidaud&Mouchet
of the harmonics (multiples of fundamental cyclotron line)
1998).
givescrucialinformationabouttheopticaldepthoftheline-
The presence of cyclotron line at ∼30 keV was con-
forming region (Araya-G´ochez & Harding 2000; Sch¨onherr
firmed with the 2002 June RXTE observations of Cep X-
et al. 2007; Nishimura 2013). As of now, CRSF has been
4 (McBride et al. 2007). However, there was no signifi-
seeninabout20accretionpoweredX-raypulsars.However,
cant changes in cyclotron line energy with source luminos-
ity observed during these observations. The pulsar was ob-
⋆ [email protected] served with NuSTAR during the 2014 June-July outburst.
† [email protected] First observation was carried out near the peak of the out-
2 G. K. Jaisawal and S. Naik
burst whereas the second was at the declining phase of
the outburst (Fu¨rst et al. 2015). The 1-50 keV combined 1.5
spectra from Swift/XRT and NuSTAR were described by 1
Fermi-Dirac cutoff power-law (FDCUT) model along with
an absorption component at ∼30keV.Additionof asimple nsity 0.5 XIS−3 0.5−10 keV
absorption component or a pseudo-Lorentzian profile was d Inte
bfaeicleadusteooefxtphleaiansythmemceytcrlioctrnoantuarbesoofrptthieonlinfeeapturorefilpe.roApleornlyg malize 1.5
or 1
with this asymmetry profile, an absorption like feature at N
∼19 keV was also detected in the spectrum. A marginal 0.5 PIN 10−70 keV
positivedependenceofcyclotronlineenergywiththepulsar
0 0.5 1 1.5 2
luminosity was seen during theNuSTAR observations. Pulse Phase
Inthepresentwork,weusedSuzaku observation of the Figure 1. Pulse profiles of Cep X-4 obtained from XIS-3 and
pulsarduring2014outbursttostudyitsspectralproperties. HXD/PIN light curves during the Suzaku observation in 2014
Thebroad-bandcoverageandlowbackgroundcapability(up July. The presence of absorption dip in soft X-ray pulse profile
to70keV)ofdetectorsonboardSuzakuprovidedbestoppor- in 0.4-0.5 phase range can be seen. The error bars represent 1σ
tunitytoinvestigatecyclotronabsorptionlineparametersin uncertainties.Twopulsesineachpanelareshownforclarity.
binaryX-raypulsars.UsingSuzakuobservationofCepX-4,
wedetectedacyclotron lineat∼28keValongwithanother
X-ray background event file provided by the instrument
absorptionlikefeatureat∼45keV.Weinterprettheabsorp-
team.ThecosmicX-raybackgroundcorrectionwasincluded
tion lines at ∼28 keV and at ∼45 keV as the fundamental
in PIN non-X-ray background spectrum. Epoch 11 PIN re-
and first harmonic cyclotron absorption lines, respectively.
sponse file (20110601) was used in spectral analysis.
Detailsontheobservations,analysisprocedures,resultsand
conclusionsarepresentedinfollowing sectionsofthepaper.
3 RESULTS
2 OBSERVATION AND ANALYSIS Source and background light curves with 2 s and 1 s
time resolution were extracted from the barycentric cor-
ATargetofOpportunity(ToO)observationofCepX-4was rected XIS-3 and PIN event data, respectively. The χ2-
carried outwith Suzaku (Mitsudaet al. 2007) in July 01-02 maximization technique was used to estimate the pulse pe-
duringits2014June-Julyoutburst.Thepulsarwasobserved riodofthepulsar.Thepulsationwasdetectedataperiodof
for exposures of ∼60 ks with XIS-3 and ∼81 ks with HXD 66.334±0.004 s from background subtracted light curves of
duringthedecayphaseoftheoutburst.Amongthreeactive XIS-3and PIN. Above pulsation period was used to gener-
XISs (XIS-0, 1 & 3), data from the XIS-3 was used in our ate pulse profiles in soft (0.5-10 keV) and hard X-rays (10-
analysis as XIS-0 and XIS-1were exposed for ∼100 s short 70keV)andareshown inFig.1.Strongenergydependence
durations.TheXIS-3wasoperatedin“normal” clockmode ofpulseprofileswithenergycanbeseeninthesoftandhard
with “1/4” window option yielding2stimeresolution. The X-rays pulse profiles. A dip like feature in soft X-rays (top
observation was carried out in “XIS nominal” position. We panel) in 0.4-0.5 phase rangedisappeared from thehard X-
used publicly available data (Obs. ID: 909001010) in the ray pulse profile. Therefore, it is interesting to investigate
presentstudy.Calibration database (CALDB)filesreleased spectral properties of thepulsar at different pulse phases.
on 2015 January 05 (XIS) and 2011 September 13 (HXD)
wereappliedduringreprocessingofdatainHeasoft(version
6.16) analysis package. 3.1 Spectral Analysis
The‘aepipeline’taskofFTOOLSwasusedtoreprocess
3.1.1 Pulse-phase-averaged spectroscopy
theunfilteredXISandHXDevents.Cleaneventsgenerated
after the reprocessing were used in our study. Barycentric Tostudythebroad-bandspectralcharacteristics ofthepul-
correction was applied on these XIS and PIN clean events sar, phase-averaged spectroscopy was carried out by using
by using ‘aebarycen’ package. XIS data were checked for the source and background spectra accumulated from the
the attitude and pile-up effects by applying S-lang scripts XIS-3 and PIN event data. The procedure for spectral ex-
(aeattcor.sl and pile estimate.sl), respectively. We found a traction was described earlier. The 0.8-70 keV spectra, ob-
pile-up of ∼8% at the centre of XIS-3 that was reduced to tained from XIS-3and PINdataweresimultaneously fitted
64% by choosing the events only from an annulus region by using XSPEC (ver. 12.8.2) package. Appropriate back-
′′ ′′
withinnerandouterradiiof10 and180 ,respectively.The ground spectra, response matrices and effective area files
source light-curves and spectra from XIS-3 were extracted for corresponding detectors were used in the spectral fit-
byusingXSELECTpackageofFTOOLS.Backgroundlight- ting. Spectral data in 1.7-1.9 keV and 2.2-2.4 keV ranges
curves and spectra for XIS-3 were created from a circular were ignored in the fitting due to presence of known artifi-
region away from thesource.Responsematrix andeffective cial emission features in the spectrum. XIS-3spectrum was
area files for XIS-3 were generated from “resp=yes” com- binned by a factor of 6, whereas PIN spectrum was binned
mandinXSELECT.SourcelightcurvesandspectraforPIN byafactorof2upto30keV,afactorof4from30to50keV
wereextractedfromcleanedeventdatabyusingXSELECT. andafactorof6from50to70keV.Allthespectralparam-
However, HXD/PIN background light curves and spectra eters were tied during the fitting, except the normalization
were accumulated in a similar manner from simulated non- constant of detectors which were kept free.
Cep X-4 with Suzaku 3
1 Table1.Best-fittingparameters(with90%errors)obtainedfrom
Counts s keV−1−1 0.00.11 0.00.11 BackgSroouurncde+Background MCsthpooeemcdsteppilveT-ec1Tltyr&.amlMMofiodtotdeidelnlegw-l3-oi2tifhsSccotuownzmsoaibskGtiunaoaouftbsisosteihnarevnoafptaaiFnorDdntiCaotwlfUCocToecvpmyecXroliond-4tgerlionNwn2Pit0lEih1nX4ebsJl,aaucrnlkeyd--.
malized 10−3 10−3 Source body,twoGaussianandonegabscomponents.
Nor 10−4 10−4 20 (keV) 50
Parameter Model-1 Model-2 Model-3
2 NPEX*CYCLABS
χ −20 NH1a 0.78±0.03 0.77±0.03 0.71±0.02
2 FDCUT*GABS NH2b 3.7±0.6 2.6±0.3 –
χ 0 Cov.fraction 0.28±0.06 0.41±0.08 –
−2
Photonindex 1.04±0.09 – 1.01±0.05
2 NPEX*CYCLABS*CYCLABS BBtemp.(keV) – – 1.08±0.06
χ −20 BBnorm.(10−4) – – 4.8±0.6
1 10 Ecut (keV) 9.5±2.1 – 25.1±4.0
Energy (keV)
Efold (keV) – – 6.8±1.1
Figure 2.Energyspectrum ofCep X-4in0.8-70keV rangeob- CompTTT0 (keV) – 0.21+−00..621 –
tained from XIS-3 and HXD/PIN data, along with the best-fit CompTTkT (keV) – 11.6±2.3 –
modelcomprisingapartialcoveringNPEXmodel,twoGaussian CompTTτ – 4.4±0.4 –
functions for iron emission lines and two cyclotron absorption Fe line parameters
components.Thesecondandfourthpanelsshowthecontributions Energy(keV) 6.41±0.03 6.41±0.03 6.42±0.03
oftheresidualstoχ2 foreachenergybinforthepartialcovering Eq.width(eV) 32±8 27±7 34±8
NPEX continuum model with one and two cyclotron absorption Energy(keV) 6.97±0.16 6.98±0.22 6.95±0.14
components,respectively.Thethirdpanelshowstheresidualsfor Eq.width(eV) 13±6 9±6 14±7
FDCUTmodelwithoneGABScomponent. Theinsetshowsthe Cyc. line parameters
HXD/PINspectrumofthepulsarwithbackground,withoutback- Energy(Ec1)(keV) 27.5±0.4 27.7±0.4 29.6±0.5∗
groundandthesimulatedHXD/PINbackground spectrum. Width(σc1)(keV) 8.9±1.0 6.1±0.8 4.8±0.5∗
Depth(Dc1) 2.3±0.3 1.8±0.3 17+−53∗
Continuummodelsusedtodescribespectraofaccretion Energy(Ec2)(keV) 45.4±2.8 43.0±3.6 –
Width(σc2)(keV) 10.3±4.6 11.9±4.2 –
poweredX-raypulsarssuchashighenergycutoffpower-law
Depth(Dc2) 1.8±0.7 1.8±0.4 –
(White et al. 1983), cutoff power-law, NPEX (Makishima Fluxc (1-10keV) 3.1±0.5 3.1±0.7 3.1±0.4
et al. 1999), FDCUT and CompTT (Titarchuk 1994) were Fluxc (10-70keV) 5.5±1.1 5.6±1.2 5.5±1.4
used to fit the 0.8-70 keV spectrum of Cep X-4. Investigat- χ2 (dofs) 254(244) 261(244) 303(247)
ing the residuals, we added two Gaussian functions at 6.4
and 6.9 keV for iron emission lines in the source spectrum. a :Equivalenthydrogencolumndensityandb :Additional
As in case of Be/X-ray binary pulsars, a partial covering hydrogencolumndensity(in1022 atomscm−2);c :Absorption
absorption component (Paul & Naik 2011 and references uncorrectedflux(in10−10 ergscm−2 s−1.);∗ :cyclotron
therein) was used in the spectral model along with the in- parametersaregivenforgabscomponent.
terstellar absorption component. A strong absorption like
feature at ∼28 keV was clearly seen in the spectrum. Ad- to2.However,inpresentcase,theratiowasestimatedtobe
dition of aCRSF component at above energy was added to ∼1.7±0.1, which can be acceptable in current understand-
allcontinuummodels.SimultaneousspectralfittingofXIS-3 ing of the cyclotron physics. Best-fit parameters obtained
andHXD/PINdatain0.8-70keVrangeshowedthatpartial from simultaneousspectralfittingaregiveninTable1.The
coveringNPEX continuummodel, partial coveringhigh en- energy spectra of thepulsar along with all threebest-fitted
ergy cutoff power-law model and partial covering CompTT model components resemble similar absorption features in
model describe the spectrum well with acceptable values of the spectral residues. Energy spectra obtained from simul-
reduced χ2 (<1.5). taneous fitting of the XIS-3 and HXD/PIN data are shown
Apart from ∼28 keV cyclotron line, we found an ad- in Fig. 2. The second and fourth panels in figure show the
ditional absorption like feature at ∼45 keV in the residues residuals to the best-fit model with one and two cyclotron
obtainedfromallthreecontinuummodels.Thisfeaturewas lines in the continuum model, respectively. As in case of
clearly seen at same energy range and was model indepen- NuSTAR observations, a weak absorption like feature can
dent (third panel in Fig. 2). The inset in Fig. 2 shows the beseenat∼19keVinthespectralresidue(thirdandfourth
HXD/PINspectrumoftheSuzakuobservationofthepulsar panels of Fig. 2). Addition of an absorption component at
with background, without background and simulated PIN ∼19 keV, however, did not show any significant improve-
background spectrum. The absorption-like feature at ∼45 ment in the spectral fitting. Negligible strength and width
keVcan beclearly seen in thebackgroundsubtractedspec- of the ∼19 keV line against other two absorption lines at
trum. The inclusion of additional CRSF component at this ∼28and∼45keV,therefore,makesitdifficulttoacceptthe
energy in the model fitted the XIS-3 and HXD/PIN spec- earlierfeatureasthefundamentalcyclotronabsorptionline.
tra well with significant improvement in the values of χ2. Therefore, we did not include any spectral component for
Absorption feature at ∼45 keV can be considered as the thisfeature in our fitting.
first harmonic of the ∼28 keV cyclotron absorption line. To verify the ∼45 keV feature as additional absorp-
Ideally, the ratio between the energy of first cyclotron har- tionfeaturedetectedinSuzakuspectra,wefittedXIS-3and
monicandfundamentallineisexpectedtobeavaluecloser HXD/PINspectrawiththespectralmodel(FDCUTmodel
4 G. K. Jaisawal and S. Naik
χSpectral residue () 05 PPPPPhhhhhaaaaassssseeeee 00000.....01234−−−−−00000.....12345 PPPPPhhhhhaaaaassssseeeee 00000.....56789−−−−−00001.....67890 FluxDE (keV)c1c1(10 ergs cms)−10−2−1 222211.54680552 10−70 keV
2
Dc2 1.5
−5
1
20 50 50
Energy (keV) V)
e 45
Fspiegcutrraew3h.ilSepfietctterdalwritehsipdauratlisalocbotvaeinriendgfNroPmEXthceonpthinausue-mresmoolvdeedl E (kc2 40
waeabictshhoropaftiftouhnnedlpaihkmeaesfenetaiantltuercreyvcaillnostro∼of4nt0h-ae5b0psoukrlepsVatrio.rnanlgineeisatcl∼ea2r8lykesVee.nAinn E/Ec2c1 11..68
0.5 1 1.5
Pulse Phase
alongwithblack-bodyandGaussianabsorption(gabs)com-
Figure 4.Spectral parameters (with90%errors)obtained from
ponents)asusedbyFu¨rstetal.(2015)tofitspectraobtained
thephase-resolvedspectroscopyofCepX-4.Toppanelshowses-
fromSwiftandNuSTARobservationsofthesource.Though timated source flux in 10-70 keV energy range. The values of
thedatashowninFig.2ofFu¨rstetal.(2015)wereupto50 cyclotron line parameters such as depths and energies for fun-
keV, we used Suzaku data in 0.8-70 keV range to compare damental and first harmonic and ratio between harmonic and
boththeresults.Thespectralparametersobtainedfromour fundamental line energies (Ec2/Ec1) are shown insecond, third,
fittingwerefoundtobecomparable(withinerrors)withthe fourth, fifth and sixth panels, respectively. Solid circles in the
parameters obtained from the NuSTAR observations and fifth panel indicate that the cyclotron line energy was fixed for
correspondingphase-binatthephase-averagedvalue.
given in Table 1. While fitting Suzaku data with the model
used by Fu¨rst et al. (2015), an absorption like feature at
∼45 keV was also seen (third panel of Fig. 2). A careful
look at the trend of distribution of points at ∼45-50 keV troscopy, the phase-resolved spectroscopy was carried out
in the residuals obtained from the spectral fitting of Obs.1 in 0.8-70 keV range. Partial covering NPEX model with a
data(secondandthirdpanelsofFig.2ofFu¨rstetal.(2015)) cyclotron component at ∼28 keV was used to describe the
indicates a hint of presence of an absorption-like feature. phase-resolved spectrum. While fitting, the normalization
Statistical significance of ∼45 keV absorption feature constants, equivalent hydrogen column density (NH1), iron
was tested by using XSPEC script simftest as applied in line parameters and cyclotron width were fixed to the cor-
case of IGR J17544-2619 (Bhalerao et al. 2015). Including responding phase-averaged valuesas given in Table 1.
systematic uncertainty of 0.3% (15-40 keV) and 1.9% (40- Spectral residues obtained from fitting all phase-
70 keV) in PIN background model, we simulated 1000 fake resolvedspectrawithabovemodelareshowninFig.3.Pres-
spectraforpartialcoveringNPEXmodelandestimatedthe ence of an additional significant absorption feature in ∼40-
differencesinχ2(∆χ2)withoutandwith∼45keVcyclotron 50 keV range can be clearly seen in each phase intervals.
absorptionline.Themaximumvalueof∆χ2 from1000sim- Detectionofthe∼45keVabsorptionlineinphase-averaged
ulations was found to be 17.8 which is less than observed aswellasphase-resolvedspectraensuresthatthisfeatureis
∆χ2=27.2 for three degree of freedom in real data. Corre- not spurious and model dependent. This feature can be in-
sponding to this, we confirmed the significant detection of terpretedasthefirstcyclotronharmonicofthe∼28keVfun-
∼45 keV absorption feature at >4σ level. damentalline.InclusionofaCRSFatthisenergyimproved
thespectralfittingsignificantlyasincaseofphase-averaged
spectroscopy. During the fitting, widths of both cyclotron
3.1.2 Pulse-phase-resolved spectroscopy
lines were frozen at phase-averaged value to constrain the
We performed phase-resolved spectroscopy to understand absorption features. Variation of best fitted cyclotron line
the change in cyclotron absorption line parameters with parameterswithpulsephaseareshowninFig.4alongwith
the pulse phase of the pulsar. Another motivation in do- hard X-ray source flux.
ing phase-resolved spectroscopy was to study the presence Cyclotronparameterssuchasdepthandenergyofboth
of ∼45 keV absorption line as well as its dependence on the lines were found significantly variable with pulse-phase
pulse-phases of the pulsar. Detection of absorption feature with maximum values in 0.3-0.6 phase range. However,
at ∼45keVateach pulse-phasebinscan confirm thedetec- the peak values of these parameters were slightly phase
tion of first cyclotron harmonic in Cep X-4. For this, the shifted (0.1 phase) with the peaks of 10-70 keV pulse pro-
phase-resolved spectroscopy was carried out by accumulat- file and source flux. Depth of fundamental line was vari-
ing source spectra in 10 pulse-phase bins. XIS-3 and PIN ableintherangeof1.8 to2.7(∼20% of thephase-averaged
phase-resolved spectra were extracted by applying phase value). Depth of first cyclotron harmonic was found to be
filter in XSELECT. Using the same background, response marginally variable with thepulse phase of the pulsar. The
and effective area files as used in phase-averaged spec- variationoftheenergyofthefirstharmonicwasfoundtobe
Cep X-4 with Suzaku 5
double (∼8 keV) of that of the fundamental line (∼4 keV). on cyclotron lines were done by considering certain sets of
However, while computing the ratio between the energy of assumptions and geometry in line forming regions (Araya-
the first harmonic and the fundamental cyclotron line, it G´ochez & Harding 2000; Sch¨onherr et al. 2007; Mukher-
wasfoundtobein∼1.6-1.8range.Sourcefluxin10-70keV jee&Bhattacharya2012). Thesestudiespredictedthat the
rangewasfoundtofollowsimilarpatternasthepulseprofile cyclotron absorption line parameters are expected to show
in same energy band. 10-20% variation over pulse phases depending on the view-
ingangleoftheaccretion column.However,morethan30%
variation in cyclotron parameters can be explained by con-
sidering distortion in the magnetic dipole geometry of the
4 DISCUSSION AND CONCLUSIONS
pulsar (Sch¨onherr et al. 2007; Mukherjee & Bhattacharya
We report the detection of first harmonic of the cyclotron 2012).InCepX-4,bothcyclotronlineparametersarevary-
absorptionlineinCepX-4.Thoughthepulsarwasobserved ingwithin 20% overpulse-phase,which can bedescribed in
with NuSTAR duringsame outburst, data from Suzaku ob- terms of the viewing angle or local distortion in magnetic
servationwithlongerexposure(twicethatofNuSTAR)con- field. Detailed modeling of the observed variations in cy-
firmed the detection of the harmonic at ∼45 keV. The har- clotron parameters would provide useful information about
monic feature was found to be model independent and also the neutron star magnetic field geometry, inclination and
presentineachbinofthephase-resolvedspectraofthepul- beamingoremissionpatterns.However,theseworksarebe-
sar. Statistical tests on the Suzaku data also confirmed the yondthe scope of thepaper.
detection of theharmonic of thecyclotron line at ∼45 keV. In summary, we report the detection of the first har-
ThoughNuSTARhasbettereffectiveareaat∼50keVthan monic of ∼28 keV fundamental cyclotron absorption fea-
Suzaku, long exposure of the Suzaku observation detected ture at ∼45 keV in Cep X-4. This feature was clearly seen
theadditional feature at ∼45 keV. in the phase-averaged and phase-resolved spectra from the
Cyclotron absorption features in thebroad-bandX-ray Suzakuobservationin2014July.Thevaluesoftheenergyof
spectrum originate due to the resonance scattering of pho- firstharmonicwithfundamentallinewerefoundanharmonic
tons with quantized electrons in the presence of magnetic with ratio of 1.7. Parameters of both the fundamental and
field. Depending on the strength of the magnetic field, the first harmonic lines were variable within 20% with pulse-
statesofelectronsarequantizedinharmonicallyspacedlev- phase which is explained as the effect of viewing angle or
els such that the first harmonic energy is expected to be local perturbation in magnetic field of line forming region.
at twice of the fundamental energy. In the present case of We sincerely thank the anonymous referee for valu-
Cep X-4, however, the first harmonic is detected at an en- ablecommentswhichimprovedthepapersignificantly.This
ergywhichis∼1.7timesthatofthefundamentallinewhich research has made use of Suzaku data obtained through
islessthantheidealcouplingfactorof2.An-harmonicspac- HEASARCOnlineService.
ing between fundamental and harmonic lines has also been
seen in a few other X-ray pulsars and described by consid-
eringtherelativisticeffectsinphoton-electronscatteringfor REFERENCES
smallchangesintheenergyratio(M´esz´aros1992;Sch¨onherr
Araya-G´ochezR.A.&HardingA.K.,2000, ApJ,544,1067
et al. 2007). However, this may not be the only cause that
BhaleraoV.,etal.2015,MNRAS,447,2274
can produce the anharmonicity in lines. Cyclotron absorp-
Bonnet-BidaudJ.M.&MouchetM.,1998,A&A,332,L9
tion phenomenafor fundamental and harmonic lines occur- Fu¨rstF.,etal.2015, ApJ,806,L24
ring at twodifferent scale heights can havedifferent optical HeindlW.A.,etal.1999,ApJ,521,L49
depthsandintroducetheanharmonicityinthecouplingfac- KoyamaK.,KawadaM.,TawaraY.,etal.1991, ApJ,366,L19
tororlineenergyratio.Indetailedstudiesofcyclotronlines, McBrideV.A.,etal.2007,A&A,470,1065
Nishimura (2005) and Sch¨onherr et al. (2007) showed that MakinoF.,&GingaTeam,1988,IAUCirc.,4575
the increase in magnetic field within a line forming region MakishimaK.,MiharaT.,NagaseF.,TanakaY.,1999,ApJ,525,
978
can result the line ratio less than 2, as seen in 4U 0115+63
M´esz´aros P., 1992, High-energy radiation from magnetized neu-
(Heindl et al. 1999) and Cep X-4 (present work). At larger
tronstars
viewingangleµ=0.79,thelineratioisexpectedtobe1.73for
MiharaT.,etal.1991,ApJ,379,L61
polarcapradiusof1.5km(Nishimura2013). Suchdecrease
MitsudaK.etal.,2007, PASJ,59,1
in energy ratio (1.57-1.73) is possible for viewing angle of MukherjeeD.,&Bhattacharya D.,2012, MNRAS,420,720
0.52-0.79 where superimposition of large numbersof funda- NishimuraO.,2005,PASJ,57,769
mental line emerging from different heights of line-forming NishimuraO.,2013,PASJ,65,84
region shifts the energy to higher side with nearly constant PaulB.,NaikS.,2011, BASI,39,429
energy for first harmonic. In another scenario, the anhar- Pottschmidt K., et al. 2012, AIP Conf. Ser. 1427, Suzaku 2011:
monicity in thelineratio can be expecteddueto distortion ExploringtheX-rayUniverse:SuzakuandBeyond(Melville,
NY:AIP),60
or displacement from the dipole geometry of the magnetic
RocheP.,GreenL.&HoenigM.,IAUCirc.,6698
field.Inthiscase,bothCRSFsaregeneratedattwodifferent
Sch¨onherrG.,etal.2007,A&A,472,353
poles of neutron star and produce a significant phase shift
TitarchukL.,1994,ApJ,434,313
betweenbothlineparameters.Suchphase-shiftwasnotseen
Ulmer M. P., Baity W. A., Wheaton W. A. & Peterson L. E.,
in Cep X-4 (Fig. 4). 1973,ApJ,184,L117
For the first time, we present the detailed analysis of WhiteN.E.,SwankJ.H.&HoltS.S.,1983,ApJ,270,711
fundamental cyclotron line with previously unknown first
cyclotronharmonicinCepX-4.Numeroussimulationworks