Table Of Content

Astronomy&Astrophysicsmanuscriptno.a0535 (cid:13)c ESO2008 February2,2008 LE The pre-outburst flare of the A0535+26 August/September 2005 outburst I.Caballero1,A.Santangelo1,P.Kretschmar2,R.Staubert1,K.Postnov1,3,D.Klochkov1,A.Camero-Arranz4, M.H.Finger5,I.Kreykenbohm1,6,K.Pottschmidt7,8,R.E.Rothschild9,S.Suchy9,J.Wilms10,andC.A.Wilson5 8 0 1 Institutfu¨rAstronomieundAstrophysik,Sand1,72076Tu¨bingen,Germany 0 2 ISOC,EuropeanSpaceAstronomyCentre(ESAC),P.O.Box78,28691VillanuevadelaCanãda(Madrid),Spain 2 3 SternbergAstronomicalInstitute,119999,Moscow,Russia 4 GACE,InstitutodeCienciasdelosMateriales,UniversidaddeValencia,POBox20085,46071Valencia,Spain n 5 NationalSpaceScienceandTechnologyCenter,320SparkmanDriveNW,Huntsville,AL,35805,USA a J 6 ISDC,16Ch.d’Ećogia,1290Versoix,Switzerland 7 CRESST,UniversityofMarylandBaltimoreCounty,1000HilltopCircle,Baltimore,MD21250,USA 1 8 NASAGoddardSpaceFlightCenter,AstrophysicsScienceDivision,Code661,Greenbelt,MD20771,USA 2 9 CASS,UniversityofCaliforniaatSanDiego,LaJolla,CA92093-0424,USA 10 Dr.Remeis-Sternwarte,AstronomischesInstitutderUniversita¨tErlangen-Nu¨rnberg,Sternwartstr.7,96049Bamberg,Germany ] h Received<date>;Accepted<date> p - o ABSTRACT r t Aims.WestudythespectralandtemporalbehavioroftheHighMassX-rayBinaryA0535+26duringa‘pre-outburstflare’which s a tookplace∼5dbeforethepeakofanormal(typeI)outburstinAugust/September2005.Wecomparethestudiedbehaviorwiththat [ observedduringtheoutburst. Methods.WeanalyseRXTEobservationsthatmonitoredA0535+26duringtheoutburst.Wecompletespectralandtiminganalyses 1 ofthedata.Westudytheevolutionofthepulseperiod,presentenergy-dependent pulseprofilesbothattheinitialpre-outburstflare v andclosetooutburstmaximum,andmeasurehowthecyclotronresonance-scatteringfeature(hereafterCRSF)evolves. 7 Results.Wepresentthreemainresults:aconstantperiodP=103.3960(5)sismeasureduntilperiastronpassage,followedbyaspin-up 6 withadecreasingperiodderivativeofP˙=(−1.69±0.04)×10−8ss−1atMJD53618,andPremainsconstantagainattheendofthemain 1 outburst.Thespin-upprovidesevidencefortheexistenceofanaccretiondiskduringthenormaloutburst.WemeasureaCRSFenergy 3 ofE ∼50keVduringthepre-outburstflare,andE ∼46keVduringthemainoutburst.Thepulseshape,whichvariessignificantly cyc cyc 1. duringbothpre-outburstflareandmainoutburst,evolvesstronglywithphotonenergy. 0 Keywords.X-rays:binaries-stars:magneticfields–stars:individual:A0535+26 8 0 : v 1. Introduction (Fingeretal. 1994). The estimated distance of the system i X is 2kpc (Steeleetal. 1998). The measured pulse period of The Be/X-ray binary A0535+26 1 is a transient source, the neutron star at MJD 53614.5137 is P=103.39315(5)s r characterized by quiescent states with X-ray luminosity a (Caballeroetal.2007).Duringquiescenceaspin-downtrendof L .1036ergs−1, interrupted by normal (type I) outbursts, as- X theneutronstarhasbeenreported(Fingeretal.1996;Hilletal. sociated in general with periaston passage, when a luminos- 2007), and during giant outbursts the neutron star shows a ity L ∼1036−37ergs−1 is reached, and giant (type II) out- X strong spin-up. In the June 1983 giant outburst, a spin-up of bursts for which LX>1037ergs−1. A0535+26 was discovered ν˙∼0.6 × 10−11Hzs−1 was measured (Sembayetal. 1990). In by Rosenbergetal. (1975) during a giant outburst of luminos- the February 1994 giant outburst the spin-up reached ν˙∼1.2× ity level of L(3−7keV)∼ 1.2×1037ergs−1. Since then, five giant 10−11Hzs−1 and quasi periodic oscillations were present, con- outbursts have been detected in October 1980 (Nagaseetal. firmingthepresenceofanaccretiondisk(Fingeretal.1996). 1982), June 1983 (Sembayetal. 1990), March/April 1989 (Makinoetal. 1989), February 1994 (Fingeretal. 1994), and May/June 2005 (Tuelleretal. 2005). Following the giant out- The X-ray spectrum of A0535+26 is typically described by a phenomenological model consisting of a power law with burst in May/June 2005, two normal outbursts occurred in August/September 2005 (Finger 2005b) and December 2005 an exponentialcutoff. Two absorptionlike features, interpreted as cyclotron resonance scattering features CRSFs (fundamen- (Finger2005a). tal and first harmonic), have been observed in the spectrum ThesourceA0535+26isinabinaryorbit,withanO9.7IIIe at ∼45keV and ∼100keV (Kendziorraetal. 1994; Groveetal. opticalcompanionHDE245770(Giangrandeetal.1980),ofor- bitalperiodP =110.3±0.3daysandeccentricitye=0.47±0.02 1995). Recent observationsduringthe August/September2005 orb normaloutburstwithINTEGRALandRXTE(Kretschmaretal. Sendoffprintrequeststo:I.Caballero, 2005;Wilson&Finger2005;Caballeroetal.2007)andSuzaku e-mail:[email protected] (Teradaetal.2006)haveconfirmedthisresult.Thecentroiden- 1 Referredtoas1A0535+262inSIMBAD ergyoftheCRSFwasmeasuredduringthemainoutburstatdif- 2 I.Caballeroetal.:Pre-outburstflareofA0535+26 2005 August 2005 September Table 1. Formalfunctiondecribingthe pulse periodduringthe 29 1 8 15 22 29 outburst.ThereferenceforthethirdorderpolinomialfitisMJD 1000 103.405 53618.FromMJD53629theperiodisessentiallyconstant. 103.400 800 MJD MJD −1U) 53608.70−53613.02 53613.11−53629 PC 103.395 P(s) 103.3960(5) 103.3883(5) −1ounts s 600 103.390P(s) PP˙¨((ssss−−21)) -- (−1(.699±±3)0.×041)0×−1150−8 c e ( 400 d3P/dt3(ss−3) - (2.5±0.9)×10−20 at A r 103.385 C P 200 barycentric-correctedPCA and HEXTE light curves and cor- 103.380 rected the photonarrivaltimes for the orbitalmotion.We used Periastron the orbital ephemeris from Fingeretal. (1994) with the peri- 0 103.375 53610 53620 53630 53640 astron epoch updated to MJD 53613.0±1.3(Fingeretal. 2006, MJD 2007). We then produced pulse profiles using a constant, trial Fig.1. Left axis, dotted line: ∼3–30keV PCA mean count rate period obtained using INTEGRAL observations of the same for each available observation during the August/September outburst (Caballeroetal. 2007). For both PCA and HEXTE 2005outburst.EachdiamondindicatesoneRXTEobservation. we determined a reference time in the pulse profiles, and per- The mean was obtained averaging each individual light curve. formedaphase-connectionanalysisaspreviouslycompletedin Theerrorsshownarethesigma ofthedistributionforeachob- Caballeroetal.2007(seealsoFerrignoetal.2007).Asaphase servation. Right axis, solid line: Pulse period evolution during reference we selected the midpoint of a sharp edge present in theoutburst.Theverticallineindicatesthetimeofperiastron. thepulseprofiles,anapparentlystablefeature.Thepulseperiod is approximately constant during the initial pre-outburst flare, and a significant spin-up is measured after periastron. The pe- ferentluminositylevelsanditwasnotfoundtovarywithX-ray riodevolutionafterperiastroncanbedescribedbyathird-order luminosity. polynomial. The results of the fit are provided in Table 1. The In this Letter we focuson the August/September2005nor- period at any time during the outburst is well-described using maloutburst.Attheonsetofthisoutburst,asharp,pre-outburst thosefunctions,and±5msisaconservativeestimateoftheun- flareisobservedsuperimposedonthegradualincreasetowards certainty. The function describing the period evolution is plot- the peak flux in the main outburst. This pre-outburst flare de- ted in Fig. 1. It is consistent with the period measured using velops on a time scale of 1 day, short compared to the three- INTEGRALdatawithinuncertainties(Caballeroetal.2007). week duration of the main outburst. As can be seen from the light curve measured bySwift/BAT, the flare is one of several suchflaressuperimposedontherisingedgeofthemainoutburst 3.2.Pulseprofileevolutionduringtheoutburst (see ?).The pre-outburstflare studied here reachesa PCA flux of810countss−1PCU−1,andthemaximumfluxduringthemain Applying the measured period and its derivatives to fold light outburstis 1005countss−1PCU−1. The spectral and timing be- curves, we studied changes in the pulse profiles between the pre-outburstflare and the main outburst. Strong changesin the havioursofthesourceduringthepre-outburstflareappeartobe differentfromtheonesobservedduringthemainoutburst. pulse-profile shape, with photon energy, are observed for both the main outburst and pre-outburst flare but in quite different ways. Fig. 2 shows pulse profiles from observations (a), dur- 2. InstrumentsandObservations ingthepre-outburstflare,and(b),closetothemaximumofthe main outburst. Observation (b) was chosen because of its high RXTE (Bradtetal. 1993) observed A0535+26 between 2005 dataqualityandthepulseprofilesarerepresentativeofdataac- August 28 and 2005 September 24 in a target of opportu- quiredduringtheoutburstpeak.Thelow-energypulseprofilesin nity observation completed as part of a campaign studying ac- bothcasesshow a complexpattern,butdifferentstructuresand creting pulsars. We present observationsfrom the Proportional evolution.Athigherenergiesbothpulseprofilesshowasimpler CounterArrayPCA(2–60keV,Jahodaetal.1996)andfromthe twopeakedshape.Duringthemainoutburst,oneofthepeaksis High Energy X-ray Timing ExperimentHEXTE (20–200keV, stronglyreducedatthecyclotronenergy.Duringthepre-outburst Rothschildetal.1998).Atotalof44pointedobservationswere flarethereisasmoothevolutiontowardshigherenergieswithout completedtomonitortheoutburstwithatotalexposuretimeof achangecrossingthecyclotronenergy.Energy-dependentpulse ∼140ks.Fig.1showsthePCAlightcurveduringtheoutburstin profilesforalltheobservationsduringtheoutburstareshownin the∼3–30keVrange,obtainedbyaveragingtheindividuallight Camero-Arranz(2007)andadetailedstudywillbepresentedin curvesof all the observations.The analysisofRXTEdata was aforthcomingpaper(Camero-Arranzetal.2007). performedwithFTOOLS6.3.1andthespectralanalysiswiththe X-RaySpectralFittingPackageXSPECv11(Arnaud1996). 4. Spectralanalysis 3. Timinganalysis To model the phase-averaged,X-ray continuum, we have used a phenomenologicalmodel, a power law times an exponential 3.1.Evolutionofpulseperiodduringtheoutburst cut-off(XSPECmodelcutoffpl),whichisthesimplestcapa- We measuredthepulseperiodofthepulsarwithhighaccuracy bleofreproducingthedata.ItisdescribedbyF(E)∝E−αe−E/Efold, duringtheoutburst.Foreachavailableobservation,weextracted where α is the photonindex and E the foldingenergy.One, fold I.Caballeroetal.:Pre-outburstflareofA0535+26 3 −1U) 53610 53615 53620 M J D 53625 53630 53635 tion(b)weincludedaharmonic,cyclotronlinewiththeenergy −1A rate (counts s PC1024680000000000 a) b) 9cfih0xa%endTgacaebotlinEenfic2edycnce=eonr1ncg0etya)2i.i.n5sWssktetehaVectiosb(nteiitccslitaus-lfildmytevesfaiargslounumeirfiesdctfhaoaenrttct.Ehoecnyttcho=ru1er0ep2ol.ob5ts+−se43r..t53vhkaaettViothnaest PC a) b) during the pre-outburstflare and for the sum of those observa- 45 50 tions,aswellasthebest-fitvaluesforthemainoutburst(obser- 35 40 vation(b)inFig.2). 25 PCA 1.75−2.98keV 30 PCA 1.75−2.98keV 125 140 100 110 75 PCA 2.98−4.63keV 90 PCA 2.98−4.63keV 19.5 12 120 120 19.0 11 105 100 counts/s 111 956024000000 PPCCAA 46..6238−−68..2786kkeeVV 111 259878050005 PPCCAA 46..6238−−68..2786kkeeVV E(keV)fold111788...505 σ(keV) 1089 7 40 PCA 8.76−10.42keV 55 PCA 8.76−10.42keV 17.0 100 275 44 46 48 50 52 44 46 48 50 52 E (keV) E (keV) 80 225 cyc cyc 60 PCA 10.42−20.45keV 175 PCA 10.42−20.45keV Fig.3.Efold vsEcyc (left)andσvsEcyc (right)contourplotsfor 70 100 oneobservationnearthemaximum(dottedlines,observation(b) 50 80 in Fig. 2) and for the sum of the three available observations 30 HEXTE A 20.15−45.47keV 60 HEXTE A 20.15−45.47keV duringthepre-outburstflare(solidlines).Thecontoursindicate 12 20 χ2 +2.30(68%),4.61(90%),9.21(99%)levels. 6 min 10 0 HEXTE A 45.47−91.19keV 0 HEXTE A 45.47−91.19keV 0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0 Phase Phase Fig.2.PCAandHEXTEbackgroundsubstractedpulseprofiles mfLr(oa5−mx5i0mkoeunVme)∼oo0bf.4steh5re×voa1uti0tob3n7uerdrsgutr(sib−n)1g,,watnhitdehfpLror(e5m−-p5o0eknreiVea)so∼tbr0os.en7r5vfl×aatri1oe0n3(7ane)er,agrwst−iht1he. −1−1nts s PCU) 1680000000 53610 5361M5JD 53620 53625 Two pulse cycles are shown for clarity. In the upper panel the ou 400 PCAlightcurveisshown.Theverticallinesindicatethecorre- e (c 200 spondingobservations.Theexposuretimesare2.2ksin(a)and A rat 60 C 50 12.35ksin(b). P 55 45 Ecyc50 53610 53611 andinsomeobservationstwo,absorptionfeatureswereincluded 45 in the model to enable an accurate description of the data. We 40 modeledthemwithaGaussianopticaldepthprofile,whichmod- 2.0 ifies the continuum in the following way: F′(E)=F(E)e−τ(E), 1.5 α 1.0 where τ(E)=τe−(E−Ecyc)2/(2σ2). A Gaussian emission line was 0.5 addedto the modelto accountforthe Fe Kα fluorescenceline, 0.0 withenergyfixedat6.4keVandwidthto0.5keV.Duetoafea- V) 30 e etudrgee,inseteheRorethsisdcuhaillds eatroaul.n2d0∼06)4,.7wkeeeVxc(liundsetrdumdaetnatfaolrXeenneorgnieLs E(kfold 1200 below5keVinouranalysis. 29 1 8 We performed a spectral analysis of all observations avail- 2005 August 2005 September able and studied the evolutionof the cyclotron-lineenergyand Fig.4. Firstpanel:PCAlightcurve.Secondpanel:fundamental continuumparametersduringtheoutburst.Wemeasuredacon- CRSFenergyduringtheourbusrt.Insetshowsazoomonthepre- stant value for the centroid energy of the fundamental CRSF outburstflare.Thirdandforthpanels:photonindexandfolding duringthemainoutburst,asreportedbyCaballeroetal.(2007). energyevolution.Errorsare90%confidenceforoneparameter During the pre-outburst flare, the cyclotron-centroid energy is ofinterest(χ2 +2.7). measured at a higher value, reaching E =52.0+1.6keV, com- min cyc −1.4 pared to E =46.1+0.5keV during the main outburst (at 90% cyc −0.5 confidence). To study the significance of the change, we have Fig. 4 shows the evolution of the CRSF energy during the producedχ2 contourplots for the observationsduringthe flare outburst, as well as the evolution of the continuum parameters and during the main outburst. As an example,in Fig.3 contour photonindexαandfoldingenergyE fortheobservationsin fold plotsareshownfortheobservationclosetothemaximumofthe whichthecyclotronlinewas detected.Thecontinuumparame- mainoutburstlabeled(b)inFig.2,andforthesumofthethree tersarequitevariableatthebeginningoftheoutburst.Thespec- availableobservationsduringthepre-outburstflare.Inobserva- trumatthebeginningoftheflareappearstobesofter,withpho- 4 I.Caballeroetal.:Pre-outburstflareofA0535+26 tonindexα∼1.2,andthenbecomesharderintheflare.Therising note first of all that the timescale of the flare in their simula- ofthemainoutburstpresentsaharderspectrum,withα∼0.6.The tionsappearstobetoolong(10%oftheorbitalperiod)compared decayoftheoutburstshowsasmoothsofteningofthespectrum. totheflaredurationobserved.TheinspectionoftheSwift-BAT light-curveofA0535+26duringtheoutburststudiedhereshows that RXTE actually observed only one of a collection of short 5. SummaryandDiscussion flares in the rising part of the outburst (?). Such flares are not reproducedbytheabovesimulations.Thismodeldoesnottake In this paper, we present evidence for significant changes in intoaccountthedisk-magnetosphericinteraction. the spectral and timing parameters of the Be/X-ray binary Amoreplausibleinterpretationisthatthepre-outburstflares A0535+26, during its normal outburst in August/September couldbecausedbymagnetosphericinstabilitiesbetweentheac- 2005.Ourthreemainresultsarethefollowing: cretiondisk andthe neutron-starmagnetosphereattheonsetof 1. The pulse periodof the neutronstar appearsto be constant accretion. The instability causes the plasma that has accumu- duringthepre-outburstflare,P=103.3960(5)s,andaspin-up lated in the disk-magnetosphere,boundary-layerto rapidly fall startsatperiastron,P˙=(−1.69 ±0.04) ×10−8ss−1measured onto the neutron star surface close to the magnetic poles, but at MJD 53618. The pulse period falls exponentially at the alongdifferentfieldlinesthan(quasi)stationaryaccretion(see? endoftheoutburst. fordetails). 2. Theenergy-dependent,pulseprofilesduringthepre-outburst flare are significantlydifferentfromthose measuredfor the Acknowledgements. WethankISSI(Bern),fortheirhospitalityduringtheteam meetingsheldthere.ICthanksESAC(ESA)inMadridfortheirhospitality.This mainoutburst. researchissupportedbytheBundesministeriumfu¨rWirtschaftundTechnologie 3. During the pre-outburst flare, the fundamental cyclotron- throughtheGermanSpaceAgency(DLR)undercontractno.50OR0302. lineenergycentroidreachesE =52.0+1.6keV,significantly cyc −1.4 higherthanforthemainoutburst,E =46.1+0.5keV. References cyc −0.5 This is the first observation of a normal outburst that mea- Arnaud,K.A.1996,inAstronomicalSocietyofthePacificConferenceSeries, Vol. 101, Astronomical Data Analysis Software and Systems V, ed. G. H. suresa spin-uprate forA0535+26,providingevidencethatan Jacoby&J.Barnes,17 accretion disk is being detected during a Type I outburst. The Bradt,H.V.,Rothschild,R.E.,&Swank,J.H.1993,A&AS,97,355 measuredspin-upP˙=(−1.69 ±0.04) ×10−8ss−1,orν˙=(1.58 ± Caballero,I.,Kretschmar,P.,Santangelo,A.,etal.2007,A&A,465,L21 0.04) ×10−12Hzs−1,issmallerthanthespin-upmeasureddur- Camero-Arranz,A.2007,PhDthesis,,UniversidaddeValencia inggiantoutburstsinthepast,e.g.,ν˙∼0.6×10−11Hzs−1 inJune Camero-Arranz,A.,Wilson,C.A.,Connell,P.,etal.2005,A&A,441,261 Camero-Arranz,A.etal.2007,inpreparation 1983(Sembayetal.1990),orν˙∼1.2×10−11Hzs−1 inFebruary Ferrigno,C.,Segreto,A.,Santangelo,A.,etal.2007,A&A,462,995 1994(Fingeretal.1996). Finger,M.H.2005a,ATel,676 The pulse profiles measured during the main outburst are Finger,M.H.2005b,ATel,595 consistent with those during a giant outburst in March/April Finger,M.H.,Bildsten,L.,Chakrabarty,D.,etal.1999,ApJ,517,449 Finger,M.H.,Camero-Arranz,A.,Kretschmar,P.,Wilson,C.,&Patel,S.2006, 1989 albeit at a different luminosity level, L ∼1.26 × (23−53keV) inBulletin oftheAmerican Astronomical Society, Vol.38,Bulletin ofthe 1037ergs−1 (Kendziorraetal. 1994). A comparison between AmericanAstronomicalSociety,359 thesepulseprofilesandthoseobtainedforthepre-outburstflare Finger,M.H.,Cominsky,L.R.,Wilson,R.B.,Harmon,B.A.,&Fishman,G.J. in this paper suggest that a different mode of accretion takes 1994,inAIPConf.Proc.308:TheEvolutionofX-rayBinariese,ed.S.Holt &C.S.Day,459 placeduringtheflare(see?). Finger,M.H.,Wilson,R.B.,&Harmon,B.A.1996,ApJ,459,288 The energy of the CRSF during the main outburst is Finger,M.H.etal.2007,inpreparation consistent with the energy measured with INTEGRAL close Giangrande,A.,Giovannelli,F.,Bartolini,C.,Guarnieri,A.,&Piccioni,A.1980, to the maximum of the main outburst, E =45.9 ± 0.3keV A&AS,40,289 cyc Grove,J.E.,Strickman,M.S.,Johnson,W.N.,etal.1995,ApJ,438,L25 (Caballeroetal. 2007) and with the energy measured with Hayasaki,K.&Okazaki,A.T.2006,MNRAS,372,1140 Suzakuat the end of the main outburst, E =46.3+1.5keV (us- cyc −1.3 Hill,A.B.,Bird,A.J.,Dean,A.J.,etal.2007,MNRAS,381,1275 ing the same Gaussian model for the line, Teradaetal. 2006). Jahoda,K.,Swank,J.H.,Giles,A.B.,etal.1996,inProc.SPIEVol.2808,p. From the observed cyclotron energy, the estimated, magnetic 59-70, EUV,X-Ray,andGamma-Ray Instrumentation forAstronomy VII, fieldatthesiteoftheX-rayemissionduringthemainoutburst, OswaldH.Siegmund;MarkA.Gummin;Eds.,ed.O.H.Siegmund&M.A. Gummin,59–70 assuming a redshift of z=0.3, is B∼5.2 × 1012G, and during Kendziorra,E.,Kretschmar,P.,Pan,H.C.,etal.1994,A&A,291,L31 the pre-outburst flare B∼5.8×1012G. During the main part of Kretschmar,P.,Kreykenbohm,I.,Pottschmidt,K.,etal.2005,ATel,601 the outburst the cyclotron-line energy remains constant within Makino,F.,Cook,W.,Grunsfeld,J.,etal.1989,IAUCirc.,4769 the errors, in spite of the changes in the luminosity. This sug- Mowlavi,N.,Kreykenbohm,I.,Shaw,S.E.,etal.2006,A&A,451,187 Nagase,F.,Hayakawa,S.,Kunieda,H.,etal.1982,ApJ,263,814 geststhatthestructureoftheaccretioncolumnisdifferentfrom Postnov,K.etal.2007,inpreparation that observed in both V 0332 +53 (Mowlavietal. 2006) and Rosenberg,F.D.,Eyles,C.J.,Skinner,G.K.,&Willmore,A.P.1975,Nature, 4U 0115+63 (Tsygankovetal. 2006). The source is likely to 256,628 be in the sub-Eddingtonregime,as for Her X-1 (Staubertetal. Rothschild,R.E.,Blanco,P.R.,Gruber,D.E.,etal.1998,ApJ,496,538 2007)forwhichitisbelievednoshockhasformedintheaccre- Rothschild,R.E.,Wilms,J.,Tomsick,J.,etal.2006,ApJ,641,801 Sembay,S.,Schwartz,R.A.,Orwig,L.E.,Dennis,B.R.,&Davies,S.R.1990, tioncolumn. ApJ,351,675 Pre-outburst flares have been observed in other accret- Staubert,R.,Shakura,N.I.,Postnov,K.,etal.2007,A&A,465,L25 ing pulsars, such as 2S1845-024 (Fingeretal. 1999) and Steele,I.A.,Negueruela,I.,Coe,M.J.,&Roche,P.1998,MNRAS,297,L5 EXO2030+375(Camero-Arranzetal.2005). Terada,Y.,Mihara,T.,Nakajima,M.,etal.2006,ApJ,648,L139 Tsygankov,S.S.,Lutovinov,A.A.,Churazov,E.M.,&Sunyaev,R.A.2006, Hayasaki&Okazaki (2006) have modeled accretion disks MNRAS,371,19 around neutron stars in Be/X-ray binaries. Their SPH simula- Tueller,J.,Ajello,M.,Barthelmy,S.,etal.2005,ATel,504 tionsreproducea seriesof normaloutburstsfromthe accretion Wilson,C.A.&Finger,M.H.2005,ATel,605 diskformedabouttheneutronstaratperiastron.Insomecases, single flares are found preceding the outburst maximum. We I.Caballeroetal.:Pre-outburstflareofA0535+26 5 Table2.Best-fitvaluesfortheavailableobservationsduringthepre-outburstflare,wheretheCRSFenergyismeasuredatahigher position,forthesumoftheobservationsduringtheflare,andforoneobservationclosetothemaximumofthemainoutburst.Lis theluminosityinunitsof1037ergs−1inthe5−50keVrange.Errorsare90%confidenceforoneparameterofinterest(χ2 +2.7). min Obs.ID E (keV) σ(keV) τ E (keV) α MJDstart Exp.(ks) L χ2 /dof cyc fold red 91086-01-03-01 48.3+1.1 6.4+1.0 0.27+0.03 18.2+0.4 0.73+0.02 53610.599 1.152 0.85 1.19/170 −1.0 −1.0 −0.03 −0.4 −0.02 91086-01-03-02 50.2+1.8 8.7+1.4 0.57+0.08 18.8+1.0 0.91+0.04 53610.665 2.208 0.45 0.88/170 −1.6 −1.3 −0.08 −1.0 −0.04 91086-01-03-03 52.0+1.6 10.8+1.3 0.53+0.06 19.2+0.8 0.83+0.03 53610.729 2.560 0.57 0.90/170 −1.4 −1.1 −0.06 −0.7 −0.03 sum 49.9+0.8 8.6+0.7 0.40+0.03 18.3+0.4 0.77+0.02 53610.599 5.920 0.66 1.18/220 −0.8 −0.7 −0.03 −0.4 −0.02 91085-01-01-03 46.1+0.5 10.1+0.5 0.47+0.03 18.0+0.4 0.59+0.02 53614.7 12.352 0.75 1.19/214 −0.5 −0.5 −0.02 −0.3 −0.02 102.5 8.4+2.2 1.0+0.4(1stharmonic) −1.9 −0.2