A double-peaked outburst of A 0535+26 observed with INTEGRAL, RXTE, and Suzaku

0.33 MB·English

by  I. Caballero

VersionofJanuary22,2013
PreprinttypesetusingLATEXstyleemulateapjv.2/16/10
ADOUBLE-PEAKEDOUTBURSTOFA0535+26OBSERVEDWITHINTEGRAL,RXTE,ANDSUZAKU
I.Caballero1,K.Pottschmidt2,D.M.Marcu2,L.Barragan3,C.Ferrigno4,D.Klochkov5,J.A.ZuritaHeras6,S.Suchy5,J.Wilms3,
P.Kretschmar7,A.Santangelo5,I.Kreykenbohm3,F.Fu¨rst8,R.Rothschild9,R.Staubert5,M.H.Finger10,A.Camero-Arranz11,
K.Makishima12,13,T.Enoto13,W.Iwakiri14,Y.Terada14
VersionofJanuary22,2013
ABSTRACT
3 TheBe/X-raybinaryA0535+26showedanormal(typeI)outburstinAugust2009.Itisthefourthinaseries
1 ofnormaloutburstsassociated with the periastron,butisunusualbypresentinga double-peakedlightcurve.
0 Thetwopeaksreacheda fluxof ∼ 450 mCrabinthe15–50keVrange. We presentresultsofthetimingand
2
spectralanalysisofINTEGRAL,RXTE,andSuzakuobservationsoftheoutburst.Theenergydependentpulse

n profiles and their evolution during the outburst are studied. No significant differences with respect to other
a normal outbursts are observed. The centroid energy of the fundamental cyclotron line shows no significant
J variationduringtheoutburst. Aspectralhardeningwithincreasingluminosityisobserved. We concludethat

1 the source is accreting in the sub-critical regime. We discuss possible explanations for the double-peaked
2 outburst.
Subjectheadings:pulsars:individual(A0535+26)—stars:magneticfield—X-rays:binaries—X-rays:stars

]
E
H 1. INTRODUCTION ∼46 and ∼100keV (see, e.g., Caballeroetal. 2007, and ref-
erencestherein).Cyclotronlinesarecausedbyresonantscat-
. The Be/X-ray binary A0535+26 was discovered during a
h teringofphotonsoffelectronsinthequantizedLandaulevels,
giantoutburstwithArielV(Rosenbergetal.1975). Atadis-
p tance of d ∼ 2kpc, it consists of the B0 IIIe optical com- andthecyclotronlineenergyisproportionaltothemagnetic
- field strength, E = ~eB/m c = 11.6keV· B/1012G (e.g.,
o panionHDE245770(Steeleetal.1998)andapulsatingneu- cyc e
r tron star of spin period P ∼ 103s, in an eccentric or- Scho¨nherretal.2007).
t spin Some accreting X-ray pulsars present a negative cor-
s bit (e = 0.47) of period of P = 111.1days (Fingeretal.
a 2006). The system presentsdioffrberentluminosity states asso- relation between the cyclotron line centroid energy and
[ ciatedwiththeactivityoftheBestar: quiescence,withX-ray the X-ray luminosity (e.g., 4U0115+6415, V0332+53, see

v1 mluamli(ntyopsietiIe)soLuXtbu.rs1ts0,3w6eitrhglsu−m1i(nRoostihtisecshLild∼et1a0l.362−03172e)r,gnso−1r-, N20a0k6a)j.imaeOttahle.r20s0o6u;rcTessy,galinkkeovHeetraXl.-210,06G;XM3o0w4l−a1v,ietanald.
X SwiftJ1626.6−5156 show the opposite trend (Staubertetal.
6 and giant (type II) outbursts, that can reach luminosities
5 L >1037ergs−1(see,e.g.,Finger,Wilson&Harmon1996). 2007, Klochkovetal. 2012, DeCesaretal. 2012). As dis-
X cussed by Beckeretal. (2012), this bimodality is probably
8 A0535+26 presents two cyclotron resonance scattering fea-
4 tures (CRSFs or cyclotron lines) in its X-ray spectrum at duetotwodifferentaccretionregimesdependingonthecrit-
. icalluminosity Lcrit. Forsourceswith LX > Lcrit (supercriti-
1
calsources),aradiativeshockdeceleratestheinfallingmatter,
0 [email protected]
withtheheightoftheemissionregionincreasingwithincreas-
3 1LaboratoireAIM,CEA/IRFU,CNRS/INSU,Universite´ParisDiderot,
CEADSM/IRFU/SAp,91191Gif-sur-Yvette,France ing mass accretion rate, explaining the negative correlation
1
2CRESST, University of Maryland, Baltimore, MD 21250 /NASA between the cyclotronline energyand the X-ray luminosity.
:
v GSFC,AstrophysicsScienceDivision,Code661,Greenbelt, MD20771, On the other hand, for sources with L < L (subcritical
X crit
i USA sources),Coulombinteractionsstoptheinfallingmatter, and
X 3Dr.Karl Remeis-Sternwarte and ECAP, FAU Erlangen-Nuremberg,
theheightoftheemissionregiondecreaseswithincreasingac-
Sternwartstr.7,96049Bamberg,Germany
r 4ISDCDataCentreforAstrophysics,1290Versoix,Switzerland cretionrate,explainingthepositivecorrelationobserved(fur-
a
5Institut fu¨r Astronomie und Astrophysik, Sand 1, 72076 Tu¨bingen, therdetails in Beckeretal. 2012, see also Basko&Sunyaev
Germany 1976, Staubertetal. 2007). Contrary to these sources, for
6Franc¸ois Arago Centre, APC (UMR 7164 Universite´ Paris Diderot, A0535+26 no significant correlation between the cyclotron
CNRS/IN2P3, CEA/DSM, Observatoire de Paris), 13 rue Watt, 75205,
line energy and the X-ray luminosity has been observed
ParisCedex13,France
7EuropeanSpaceAstronomyCentre(ESA/ESAC),ScienceOperations (Teradaetal.2006;Caballeroetal.2007),withtheexception
Department,VillanuevadelaCan˜ada(Madrid),Spain ofaflaringepisodeduringtheonsetofanormaloutburstob-
8SpaceRadiationLab,CaliforniaInstituteofTechnology,MC290-17 served in 2005 during which the cyclotron line energy sig-
Cahill,1200E.CaliforniaBlvd.,Pasadena,CA91125,USA
nificantly increased (see Caballeroetal. 2008; Postnovetal.
9CenterforAstrophysicsandSpaceScience,UCSD,LaJolla,CA,USA
2008fordetails). Theseresultsrefertopulsephaseaveraged
10NationalSpaceScienceandTechnologyCenter,320SparkmanDrive
NW,Huntsville,AL35805USA spectroscopy; see howeverKlochkovetal. (2011) for pulse-
11 InstitutdeCie`nciesdel’Espai,(IEEC-CSIC),CampusUAB,Fac.de to-pulseanalysisin whicha positivecorrelationbetweenthe
Cie`ncies,TorreC5,parell,2aplanta,08193Barcelona,Spain fluxandthecyclotronlineenergywasobserved.
12Department of Physics, The University of Tokyo, 7-3-1 Hongo, A0535+26 showed a giant outburst and a series of nor-
Bunkyo,Tokyo113-0033,Japan
13Cosmic Radiation Laboratory, RIKEN, 2-1, Hirosawa, Wako City,
Saitama351-0198,Japan 15 See, however, Mu¨lleretal. (2012b), where it is shown that for
14Graduate School of Science and Engineering, Saitama University, 4U0115+64thepresenceofacorrelationdependsonthemodelforthecon-
255Shimo-Okubo,Sakura,Saitama338-8570,Japan tinuum.
2 Caballeroetal.
2009 July 2009 August
22 29 1 8 15 22 29 1
P [s]11−2−100Swift BAT counts cm s (15−50 keV)33..550000001.....0123450 (a) IRSNuXTzTEaEGkuRAL counts/s 112233411223456050505005050000000000000000 PPCCAA First p e28a..19k−−82.09. 5k ekVeV 111110246024683456789118000000000000000002 Decay
103.500 (b) 30 6
30 40 50MJD−5500060 70 80 121005 HEXTE B 20.4−45.6 keV 246
10 4
Fig.1.—(a)Swift/BATorbitallightcurve(15–50keV)oftheA0535+26 5 2
double-peaked outburst. The tick marks indicate the times of the INTE- 0 0
GRAL,RXTE,andSuzakuobservationspresentedinthiswork.Thevertical −5 HEXTE B 45.6−91.1 keV −2
0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0
dottedlineshowsthetimeofperiastron. (b)Pulseperiodevolutionderived phase phase
fromtheRXTEobservations.
mal ones in 2009, all of them associated with the peri- Fig.2.— RXTE PCA and HEXTE energy dependent pulse profiles for
astron passage of the neutron star (Caballeroetal. 2009). oneobservation around thefirstpeak (left, MJD55047.86, ObsID94323-
The giant outburst reached a flux of F(15−50)keV = (1.17 ± 02-03-01)andoneobservationduringthedecayoftheoutburst(right,MJD
0.04)countss−1cm−2 (∼ 5.3Crab16) at MJD 55179.0 and 55065.84,ObsID94323-03-02-01). Theenergyrangesareindicatedinthe
leftpanels,andarethesameforbothobservations. ThePCAcountrateis
lasted about 40 days. In this work we focus on a double- givenincountss−1PCU−1.Twopulsecyclesareshownforclarity.
peaked outburst that preceded the giant one. We present
INTEGRAL, RXTE, and Suzaku observations of the out- ing the decay of the outburst (MJD start 55067.96, ObsID
burst. The Swift/BAT light curve of the outburst and the 404054010). We used data from the two main instruments,
times of the observations are shown in Fig. 1 (a). The first the X-ray Imaging Spectrometer (XIS, Koyamaetal. 2007)
peakoftheoutburstreachedafluxof F(15−50)keV = (0.100± and the Hard X-ray Detector (HXD, Takahashietal. 2007).
0.004)countss−1cm−2(∼455mCrab16)atMJD55048.0,that ThedataanalysiswasperformedusingHEASOFTv6.12and
decreased to F(15−50)keV = (0.047 ± 0.002)countss−1cm−2 CALDBversions20110913forHXD,20120209forXIS,and
(∼214mCrab) at MJD 55051.0 and rose again, reaching 20110630for XRT. We used XIS data between 0.5–10 keV
F(15−50)keV = (0.104±0.005)countss−1cm−2 (∼471mCrab) for XIS 1, and data between 0.7–10 keV for XIS 0, 3. XIS
data were binned as in Nowaketal. (2011). Because of the
atMJD55058.0aroundperiastron. Theobservationsarede-
XIScalibrationuncertainties,especiallyaroundtheSiKedge
scribedinSec.2. Theresultsofthetimingandspectralanal-
at ∼1.8keV, a systematic error of 1% was assumed for the
yses are presentedin Sec. 3, and the resultsare discussedin
XISdata. ThespectralanalysiswasperformedwithXSPEC
Sec.4.
v12.7.0.
2. OBSERVATIONSANDDATAANALYSIS
3. RESULTS
We made use of INTEGRAL data collected with the
gammarayspectrometerSPI(20keV–8MeV,Vedrenneetal. 3.1. Timinganalysis
2003), the imager IBIS (15 keV–10 MeV, Ubertinietal. We studied the pulse period evolution us-
2003), and the X-ray monitorJEM-X (3–35keV, Lundetal. ing the phase-connection technique described by
2003).Threepointedobservationswereperformedaroundthe Staubert,Klochkov&Wilms (2009). We extracted PCA
secondpeak ofthe outburst(MJD start 55054.89,55057.99, light curves with 15.6ms resolution in the 3.3–42.8keV
and55060.99). TheJEM-XandSPIdataanalyseswereper- range. Barycentric and orbital corrections were applied
formed using the standard analysis package OSA v9. For to the light curves, using the ephemeris from Fingeretal.
IBIS,weusedOSAv10,thatcontainsa newenergycalibra- (2006). The evolution of the pulse period is shown in
tionforISGRI. Fig. 1 (b). A decrease in the pulse period along the out-
RXTE(Bradt,Rothschild&Swank1993)performedregu- burst is observed. A linear fit can be used to describe the
larpointedobservationsofA0535+26duringtheoutburstbe- pulse period evolution between MJD 55046.89–55057.88,
tween MJD 55046.86–55078.00(Proposal ID P94323). We during the two main peaks of the outburst. We find a
used data from the Proportional Counter Array PCA (2– pulse period of P = 103.51263± 0.00004s and spin-up of
60keV,Jahodaetal.1996)andtheHighEnergyX-rayTiming P˙ =(−0.459±0.005)×10−8ss−1atMJD55047.4.Thepulse
Experiment HEXTE (20–200keV, Rothschildetal. 1998). period during the decay of the outburst (MJD > 55057.88)
The RXTE data were analyzed using HEASOFT v6.7. We can also be described with a linear fit. We obtain a spin
restricted the PCA spectralanalysis to energiesabove5keV periodof P = 103.50886±0.00001sforMJD55056.87and
due to a feature around 5keV caused by instrumental Xe L aspin-upofP˙ =(−1.104±0.002)×10−8ss−1.
edges(Rothschildetal.2006). Making use of these pulse period values to fold the light
Suzaku performed one observation of A0535+26 dur- curves, we studied the evolution of the energy dependent
pulseprofilesusingalltheRXTEpointedobservations. Dur-
16FluxesderivedfromthedailySwift/BATtransientmonitorresultspro- ingthetwomainpeaksoftheoutburst,thepulseprofilesare
videdbytheSwift/BATteam.∼0.22countss−1cm−2correspondto1Crab. remarkablystable, showinga complexstructureatlowener-
Double-peakedoutburstofA0535+26 3
10 2
−1V JEM−X 1 −1V 10 0 XIS
−1malized counts s ke1100− 20 ISGRI SPI −1malized counts s ke1100−−21 PIN GSO
nor10−4 nor10−3
(a) 10−4 (a)
8 8
χ 0 χ 0
−8 (b) −8 (b)
5 3
χ 0 χ 0
−5 (c) −3 (c)
10 20 50 100 1 2 5 10 20 50 100
Energy [keV] Energy [keV]
Fig.3.—Left:INTEGRALspectrumofA0535+26(MJD55057.99). (a)Dataandbestfitmodel. (b)Residualsofafitincludingnoabsorptionlinesinthe
model. (c)Residualsofafitincludingonelineat∼46keVinthemodel. ThefilledcirclesrepresenttheISGRIdata. Theaveragefluxduringtheobservation
F20−100keV ∼5.0×10−9ergcm−2s−1. Right:SuzakuspectrumofA0535+26(MJD55067.96). (a),(b),(c)havethesamemeaningasintheleftpanel. The
black,red,andgreensymbolsrepresentXIS0,1,and3datarespectively.TheaveragefluxduringtheobservationF20−100keV∼0.5×10−9ergcm−2s−1.
gies,andbecomingsimplerathigherenergies. Theshapeof tion is shown in Fig. 3 (left). In this case, the inclusion of
the pulse profiles at low energiesslightly varies towards the a cyclotron line in the model yields an improvement in the
decay of the outburst. As an example, the pulse profiles for χ2 /d.o.f. from8.8/132to0.88/129. TheSuzakubroadband
red
oneobservationduringthefirstpeakoftheoutburstandone spectrum is shown in Fig. 3 (right). The inclusion of a cy-
observationduringthedecayareshowninFig.2. clotronlineinthiscaseimprovestheχ2 /d.o.f.from1.73/583
red
to1.37/580.
3.2. Spectralanalysis:
Westudiedtheevolutionofthecyclotronlinecentroiden-
We studied the broad band spectral continuum and its ergy,photonindex,and foldingenergyasa functionoftime
evolution during the outburst with RXTE,INTEGRAL, and andX-rayluminosity. Tostudytheevolutionofthedifferent
Suzaku. It can be described by a powerlaw with an expo- spectralparameters, we selected the first 16 RXTEobserva-
nential cutoff, typical foraccreting X-ray binaries (see, e.g., tions (MJD 55046.86–55065.84), which are the ones where
White,Swank&Holt 1983). We used the XSPEC model the cyclotron line was significantly detected. The first har-
cutoffpl,given by F(E) ∝ E−Γe−E/Efold, where Γ and Efold monic line at E ∼ 100keV, observed in brighter outbursts
arethephotonindexandfoldingenergyrespectively. ofthesource(e.g.,Caballeroetal.2007),isnotsignificantly
TheSuzakuspectrumwasmodeledincludinganadditional detected in the observations presented here. The results are
blackbodycomponent,photoelectricabsorptionandaGaus- shown in Fig. 4. No dependency of the cyclotron line en-
sian line in emission to account for the Fe Kα line. The ergywiththeX-rayluminosityisfound.Wecheckedthatthe
presence of the Fe Kα line is marginal, and the line en- choice of continuum model does not affect the line param-
ergy and width have been fixed to 6.4 and 0.5 keV respec- eters by fitting the continuum using an exponentially cutoff
tively.Itsinclusioninthemodelyieldsanimprovementinthe power law, a Fermi Dirac cutoff power law (Tanaka 1986),
χ2 /d.o.f from 1.43/581 to 1.37/580. The equivalent width andapowerlawwithanexponentialcutoffstartingatacutoff
ofredthe line is EW (Fe Kα)=70± 2eV. The inclusion of the energy (the highecutmodel ofXSPEC). In all three cases
blackbodycomponentinthemodelleadstoanimprovement the line parameters remain the same within their error bars
in theχ2 /d.o.f. from3.01/582to 1.37/580,givingan F-test (seealsoCaballero2009). Alinearfittothe E vs. L val-
red cyc X
significance > 99.99%. The best fit value of the black body ues gives a slope of (−3 ± 3)keV/1037ergs−1 using RXTE
temperatureiskBT =1.26+−00..0054keV.Settingthecross-sections dataonly,and(−0.9±2.0)keV/1037ergs−1 usingRXTE,IN-
and the abundances to those from Verneretal. (1996) and TEGRAL, and Suzaku data. We included inter-calibration
Wilms,Allen&McCray(2000)respectively,weobtainedan constantsforthedifferentinstrumentsintheseparateRXTE,
absorptioncolumnofNH =0.70±0.03×1022atomscm−2. INTEGRAL, and Suzakuspectral analysis. For the Suzaku
To model the RXTE spectra, we added a Gaussian emis- analysis, we fixed the constant to 1 for XIS 0 and allowed
sionlinetoaccountforresidualspresentinthe5–7keVrange. the others to vary. In the case of the GSO detector of the
Theenergyofthelinewasleftfree,andthewidthoftheline HXD, we fixed the constant to c = 0.88 (the value ob-
GSO
wasfrozentothebestfitvalueineachobservation. Theline tainedbeforeaddingthecyclotronlinetothemodel),because
energyobtainedforthedifferentobservationsvariesbetween whenunconstrained,thebestfitresultsinanunrealisticvalue
5.0and6.6keV,consistentwiththeinstrumentalXeLedgeat of c = 0.4± 0.1. By doing this, the folding energy de-
∼ 5keV and the 6.4 keV Fe Kα line. It was not possible to creaGseSsOfrom E = 44+9 to 31 ± 2keV, while the pow-
modelthetwocomponentsindependently. erlaw index andfoldthe cycl−o6tron line energy remain constant
In the INTEGRAL, Suzaku, and most of the RXTE ob- within the errors. As seen in the middle and bottom pan-
servations, a significant absorption-like feature is present in els of Fig. 4, the photon index significantly decreases with
the residuals at E ∼ 46keV, that we model using a Gaus- increasing luminosity. The folding energy remains constant
sianopticaldepthprofileτ(E)=τe−(E−Ecyc)2/(2σ2),whichmod- duringthemainpartoftheoutburst,downtoaluminosityof
ifies the continuum as F′(E) = F(E)e−τ(E). As an example, L(3−50)keV ∼ 0.26×1037ergs−1,andshowsanincreaseinthe
thebroadbandspectrumofthesecondINTEGRALobserva- lastobservationsoftheoutburst. Boththefoldingenergyand
4 Caballeroetal.
52 andNaiketal.2008),andconsistentwiththoseobtainedwith
(a) (d)
50 Fermi-GBMbyCamero-Arranzetal.(2012)duringthesame
E [keV]cyc444468 peWrioedo.bservenosignificantvariationofthecyclotronlineen-
ergywiththeX-rayluminosityduringthedouble-peakedout-
42
1.2 (b) (e) burst,andasignificantdecreaseofthephotonindexΓ(spec-
1.1
1.0 tralhardening)with increasingluminosity. A similarbehav-
Γ 0.9 ior has been observed in other outbursts of the source (e.g.,
0.8
Caballeroetal.2008;Mu¨lleretal.2012a).InaccretingX-ray
0.7
0.6 pulsars,apositiveornegativecorrelationofE withL isre-
cyc X
(c) (f) latedtodifferentaccretionregimes. Followingthetheoretical
35
E [kev]fold223050 pfoorbersdeAircv0tai5ot3ino5sn+sf2rpo6rmceosBerrneetcsekpdeorhneedtrseatloa.rL(e2cr0wit1e∼2l)l,6b.t7he8elo×cwr1itt0ihc3aa7tlelrliugmmsi−ti1,n.owTsiihtthye
1545 50 55 60 65 700.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 L(3−50)keV rangingbetween∼ 0.04and∼ 0.56×1037ergs−1,
MJD−55000 L(3−50) keV [1037 erg s−1] and therefore A0535+26 is probably accreting in the sub-
criticalregimeduringtheoutburst. Apositivecorrelationbe-
phFoitgo.n4.i—ndeLxeΓft(:b)E,vaonldutfioolndionfgtehneecrgyycloEtfrooldn(lci)newictehnttirmoied.eInneprgaynelE(cayc),(tah)e, tween Ecyc and LX is expected for sub-critical sources. The
crossesrepresenttheSwift-BATlightcurveinarbitraryunits. Right:evolu- fact that no correlation is observed might be due to the fact
tionofEcyc(d),Γ(e),andEfold(f)withthe(3–50)keVX-rayluminosity.The thatourobservationsarenearorbelowtheCoulombstopping
RXTE,INTEGRAL,andSuzakuobservationsarerepresentedwithtriangles,
limit, L ,wherelittlevariationof E with L isexpected
squares,andacirclerespectively. Inpanel(d),thedashedlinerepresentsa coul cyc X
linearfittothedata.Errorbarsareat90%confidence. (seeBeckeretal.2012,forfurtherdetails).
The double-peaked shape of the light curve could be due
the photonindexincreasewith decreasingluminosities. The to perturbations in the Be disk around the optical compan-
valuesplottedinFig.4havebeenobtainedleavingthesetwo ion. Moritanietal.(2011)showedthatA0535+26exhibited
parametersfree in the spectralfits. The increase of the pho- strongHα variability during,before, andafter the giantout-
tonindexwithdecreasingX-rayluminosityindicatesasoften- burstin 2009, and suggested that strongperturbationsin the
ingofthespectrum,whiletheincreaseofthefoldingenergy Be disk started about one cycle before the giant outburst.
withdecreasingX-rayluminosityindicatesahardeningofthe In addition, the Hα EW and V-band brightness showed an
spectrum. In order to check if the spectrum becomes softer anti-correlationbeforethe2009giantoutburst(Yan,Li&Liu
orharderastheluminositydecreases,inthelastthreeRXTE 2012;Camero-Arranzetal.2012). Thisindicatesthatamass
observationswefixedthefoldingenergytothemeanvalueof ejection event took place before the giant outburst, pro-
the first 13 observations, Efold = 18.4keV. By doing this, a ducing a low-density region in the inner part of the disk
slopeof(−0.44±0.04)/1037ergs−1 isobtainedfromalinear that could explain the double-peaked profile (Yan,Li&Liu
fit to the Γ vs. LX values, with a Pearson correlation coeffi- 2012). Interestingly, Yan,Li&Liu (2012) reported a simi-
cientof0.82.Thisresultindicatesasofteningofthespectrum larHαEWandV-bandbrightnessevolutionbeforethe1994
withdecreasingluminosity. outburst, when two double-peaked outbursts took place be-
forethegiantone(Finger,Wilson&Harmon1996). Suchan
4. DISCUSSION anticorrelation was not observed before the 2005 giant out-
We have presented the first observations of a double- burst, that took place without precursor outbursts. Double-
peaked outburst of A0535+26 with RXTE, INTEGRAL, peaked light curves have recently been observed in other
and Suzaku. The double-peaked outburst was the precur- sources, for instance GX304−1 (Nakajimaetal. 2012) and
sor of a giant one in December 2009. Two double-peaked XTE J1946+274 (Mu¨lleretal. 2012). While the spectral
outbursts were also observed before a giant one in 1994 shapeagainremainedrelativelyconstantovertheoutburstfor
(Finger,Wilson&Harmon1996),whileothergiantoutbursts the latter, this is a special case, since two outbursts per or-
donotshowanyprecursoroutbursts(Tuelleretal.2005).An- bitareobserved. InthecaseofGX304−1thedouble-peaked
otheroutburstwithapeculiarshape(twoseparatepeaks)was outburst took place before a giant one, similar to what has
observedtwoorbitalphasesafterthe2009giantoutburst(see beenobservedforA0535+26,suggestingthatdouble-peaked
Camero-Arranzetal.2012). outbursts could be indicators of upcoming giant outbursts
Wemeasuredapulseperiodof P=103.51263±0.00004s (Nakajimaetal. 2012). The double-peaked outburst in the
and spin-up of P˙ = (−0.459 ± 0.005) × 10−8ss−1 at MJD case of GX304−1 was similar to the one of A0535+26 in
55047.4,andaspin-upof P˙ = (−1.104±0.002)×10−8ss−1 termsofduration,lastingabout40days. Whileinthecaseof
at MJD 55056.87. These results are in agreement with the A0535+26thetwopeaksreacheda similarfluxlevel,inthe
values obtained with Fermi-GBM by Camero-Arranzetal. case of GX304−1the secondpeakreacheda flux two times
(2012),whoalsopresentthelong-termpulseperiodhistoryof higherthanthefirstone.Theintensitiesofthetwopeakswere
A0535+26.Wenotethatthechangesinthespin-upratealong F(15−50)keV =(0.098±0.004)countss−1cm−2(∼445mCrab)at
theoutburstcouldbedueto theoutdatedorbitalsolution. A MJD56075andF(15−50)keV =(0.196±0.008)countss−1cm−2
new orbital solution is required to explore the pulse period (∼891mCrab) at MJD 56087, with a flux between the
evolutionin more detail. A spin-upduringnormaloutbursts two peaks that dropped down to F(15−50)keV = (0.021 ±
hasalsobeenobservedinthepast(Caballeroetal.2008),pro- 0.001)countss−1cm−2 (∼95mCrab) at MJD 56082. Note
vidingevidenceforanaccretiondiskaroundtheneutronstar that the distances of the two sources are rather similar,
during normaloutbursts. The energyand luminosity depen- withGX304−1beingat∼2.4kpc(Parkes,Murdin&Mason
dent pulse profiles are remarkably stable compared to past 1980). ThegiantoutburstofGX304−1lastedabout50days,
observations of the source (see, e.g., Caballeroetal. 2007
Double-peakedoutburstofA0535+26 5
showing a first weaker peak that reached a X-ray flux of schedulingoftheobservations,andISSI(Bern)fortheirhos-
F(15−50)keV = (0.163±0.007)countss−1cm−2 (∼741mCrab) pitalityduringourcollaborationmeetings.ICthanksPhilippe
at MJD 56208. The flux then dropped down to zero (for- Laurent for the help with the INTEGRAL analysis, Yuuki
mally F(15−50)keV = (0.004±0.021)countss−1cm−2 at MJD Moritani for useful discussions, and acknowledges financial
56223),andincreasedagain,reaching F(15−50)keV = (0.406± supportfromtheFrenchSpaceAgencyCNESthroughCNRS.
0.015)countss−1cm−2 (∼ 1.85Crab) at MJD 56234. Fur- KP and DMM acknowledge support from NASA guest ob-
ther simultaneousX-ray and optical observationsof double- server grants NNXIOAJ47G for INTEGRAL cycle 6 and
peakedoutburstsareneededtobetterunderstandtheirorigin NNXIOAJ48G for Suzaku cycle 4. JW and IK acknowl-
andrelationtogiantones. edge partial funding from the Deutsches Zentrum fu¨r Luft-
undRaumfahrtundercontractnumber50OR 1113. ACAis
supportedbythegrantsAYA2009-07391andSGR2009-811,
We thank the anonymous referee for his/her useful com- aswellastheFormosaprogramTW2010005andiLINKpro-
ments, the RXTE, INTEGRAL, and Suzaku teams for the gram2011-0303.
REFERENCES
Basko,M.M.,&Sunyaev,R.A.,1976,MNRAS,175,395 Nakajima,M.,etal.,2006,ApJ,646,1125
Becker,P.A.,etal.,2012,A&A,544,A123 Nowak,M.A.,etal.,2011,ApJ,728,13
Bradt,H.V.,Rothschild,R.E.,&Swank,J.H.,1993,97,355 Parkes,G.E.,Murdin,P.G.,&Mason,K.O.,1980,MNRAS,190,537
Caballero,I.,2009, Ph.D.thesis,IAATUniversityofTuebingen Postnov,K.,etal.,2008,A&A,480,L21
Caballero,I.,etal.,2009,ATel,2337 Rosenberg,F.D.,etal.,1975,Nature,256,628
Caballero,I.,etal.,2007,A&A,465,L21 Rothschild,R.E.,etal.,1998,496,538
Caballero,I.,etal.,2008,A&A,480,L17 Rothschild,R.E.,etal.,2012, inpreparation
Camero-Arranz,A.,etal.,2012,ApJ,754,20 Rothschild,R.E.,etal.,2006,ApJ,641,801
DeCesar,M.E.,etal.,2012, ApJ,inpress Scho¨nherr,G.,etal.,2007,A&A,472,353
Finger,M.H.,etal.,2006, inBAAS,Vol.38,359 Staubert,R.,Klochkov,D.,&Wilms,J.,2009,A&A,500,883
Finger,M.H.,Wilson,R.B.,&Harmon,B.A.,1996,ApJ,459,288 Staubert,R.,etal.,2007,A&A,465,L25
Jahoda,K.,etal.,1996, inProc.SPIEVol.2808,EUV,X-Ray,and Steele,I.A.,etal.,1998,MNRAS,297,L5
Gamma-RayInstrumentationforAstronomyVII,ed.O.H.Siegmund, Takahashi,T.,etal.,2007,PASJ,59,35
M.A.Gummin,59 Tanaka,Y.,1986, inIAUColloq.89:RadiationHydrodynamicsinStarsand
Klochkov,D.,etal.,2012,A&A,542,L28 CompactObjects,ed.D.Mihalas,K.-H.A.Winkler,Vol.255,198
Klochkov,D.,etal.,2011,A&A,532,A126 Terada,Y.,etal.,2006,ApJ,648,L139
Koyama,K.,etal.,2007,PASJ,59,23 Tsygankov,S.S.,etal.,2006,MNRAS,371,19
Lund,N.,etal.,2003,A&A,411,L231 Tueller,J.,etal.,2005,ATel,504
Moritani,Y.,etal.,2011,PASJ,63,L25 Ubertini,P.,etal.,2003,411,L131
Mowlavi,N.,etal.,2006,A&A,451,187 Vedrenne,G.,etal.,2003,A&A,411,L63
Mu¨ller,D.,etal.,2012a, A&A,submitted Verner,D.A.,etal.,1996,ApJ,465,487
Mu¨ller,S.,etal.,2012b, A&A,inpress White,N.E.,Swank,J.H.,&Holt,S.S.,1983,ApJ,270,711
Mu¨ller,S.,etal.,2012,A&A,546,A125 Wilms,J.,Allen,A.,&McCray,R.,2000,ApJ,542,914
Naik,S.,etal.,2008,ApJ,672,516 Yan,J.,Li,H.,&Liu,Q.,2012,ApJ,744,37
Nakajima,M.,etal.,2012,ATel,4420