Observations of V0332+53 during the 2015 Outburst using Fermi/GBM, MAXI, Swift, and INTEGRAL
aa r X i v : . [ a s t r o - ph . H E ] F e b MNRAS , 1–8 (2016) Preprint 21 September 2018 Compiled using MNRAS L A TEX style file v3.0
Observations of V0332+53 during the 2015 Outburst using
Fermi /GBM,
MAXI , Swift , and
INTEGRAL
Zachary A. Baum , Michael L. Cherry , James Rodi , Department of Physics and Astronomy, Louisiana State University, Baton Rouge, Louisiana 70803, USA Universite´ de Toulouse; UPS-OMP; IRAP; Toulouse, France CNRS; IRAP; 9 Av. Colonel Roche, BP 44346, F-31028 Toulouse cedex 4, France
Accepted XXX. Received YYY; in original form ZZZ
ABSTRACT
We present the lightcurves, spectra, and hardness-intensity diagram (HID) of the highmass X-ray binary V0332+53 using
Fermi /GBM,
MAXI , Swift /BAT, and
INTE-GRAL through its 2015 Type II outburst. We observe characteristic features in theX-ray emission (2-50 keV) due to periastron passages, the dynamical timescale ofthe accretion disc, and changes within the accretion column between a radiation-dominated flow and a flow dominated by Coulomb interactions. Based on the HIDand the light curves, the critical luminosity is observed to decrease by ∼ − during the outburst, signaling a decrease in the magnetic field. Key words:
X-rays: general — X-rays: binaries — Neutron Stars: individual (V + ) Outbursts from the transient X-ray pulsar V0332+53were detected in 1973 (Terrell and Priedhorsky 1984),1983 (Makishima et al. 1990a), 1989 (Makishima et al.1990b), 2004-2005 (Pottshmidt et al. 2005; Mowlavi et al.2006; Lutovinov et al. 2015; Caballero-Garc´ıa et al. 2016),and recently a large Type II outburst starting inJune 2015 (Doroshenko et al. 2016a; Tsygankov et al.2016; Caballero-Garc´ıa et al. 2016; Wijnands et al. 2016;Cusumano et al. 2016). These outbursts have typically beensplit into two categories, Type-I and Type-II outbursts.The relatively less luminous Type-I outbursts are morefrequent, occurring near periastron passages and lasting afew days. Type-II outbursts have a peak luminosity nearthe Eddington luminosity (Frank et al. 2002) and can lastfor several months. The peak X-ray intensity in the gi-ant 1973 outburst reached ∼ ∼ ∼ . × G. Observations of the 2004 outburst notedthat the energy of the fundamental cyclotron line decreaseswith increasing luminosity and that the ratio of the en-ergy of the first harmonic to that of the fundamental lineis near 2 and increases with luminosity (Tsygankov et al.2006; Nakajima et al. 2010; Lutovinov et al. 2015). Thishas been interpreted as due either to changes in theheight at which the cyclotron absorption line is produced(Becker et al. 2012) or to reflection from the neutron starsurface (Poutanen et al. 2013). Nakajima et al. (2010) andTsygankov et al. (2010) also investigated possible hystere-sis of the cyclotron absorption line energy, equivalent width,and pulsed fraction between the 2004-2005 outburst rise anddecline, which might signal a change in the physics of theaccretion column or emission. They found no sign of hystere-sis in the 2004-2005 event. However, a decrease of the lineenergy has been observed over the 2015 outburst and at- © Z. A. Baum et al. tributed to a drop in the magnetic field during the outburst(Cusumano et al. 2016).Reig (2008) and Reig and Nespoli (2013) have cre-ated color-color diagrams (CCD) and hardness-intensitydiagrams (HID) for a number of Be/X-ray binaries in-cluding V0332+53 during Type II outbursts. Similar toLMXBs, the HIDs are characterized by two different ac-cretion states with a ”horizontal” branch corresponding torelatively low accretion rates and a ”diagonal” branch corre-sponding to higher accretion. The transition between thehorizontal and diagonal branches is interpreted to occur(Becker et al. 2012) at the point at which the luminos-ity reaches the critical luminosity (i.e., where the inwardflow becomes radiation-dominated on the diagonal branchrather than governed by Coulomb interactions on the hor-izontal branch). 4U 0115+63 and V0332+53 were the onlysources where hysteresis was observed in the HID duringthe time the source was on the diagonal branch, with thespectrum becoming softer during the outburst decline (Reig2008; Reig and Nespoli 2013). The reported hysteresis in theHID for V0332+53 was observed for the 2004-2005 outburst(Reig 2008).Between January 2012 and February 2015, the op-tical brightness of BQ Cam increased by ∼ MAXI (Nakajima et al. 2015) followed by
Swift (Doroshenko et al. 2015). The event rose to a peak of ∼ . Crab in the 12-25 keV energy range after approximately 50days and decreased back to quiescence after approximately150 days. Cusumano et al. (2016) have shown that the peakenergy of the fundamental cyclotron line decreased from . + . − . keV at the start of the event to . + . − . keV af-ter 56 days and then recovered to . ± . keV after 105days, corresponding to a decrease in the surface magneticfield from . × G to . × G near the peak andthen an increase back to . × G at the end of theevent (assuming a redshift of 0.3 corresponding to a . M ⊙ neutron star with radius of 10 km). Fermi /GBM observedthe evolution of the pulsar spin frequency throughout the2015 outburst and observed the spin-up of the pulsar from ∼ . mHz at MJD 57200 to ∼ . mHz at MJD57270.We have combined data from MAXI /GSC,
Fermi /GBM,
Swift /BAT,
INTEGRAL /JEM-X, and
INTEGRAL /SPI to study the light curves, spectral shape,and spectral evolution of V0332+53 during its 2015 TypeII outburst. In Section 2, we briefly describe the instru-ments and the observations; in Section 3, we present theresults of the analysis; and in Section 4, we discuss thepossible physical interpretations. In particular, we use thehardness-intensity diagram in combination with the lightcurve to provide information about the critical luminosity,and emphasize the role of periastron passages and changesin accretion state in determining the features of the lightcurve. http://gammaray.msfc.nasa.gov/gbm/science/pulsars/lightcurves/v0332.html MAXI /GSC
The
MAXI (Monitor of All Sky X-ray Image) Gas SlitCamera (GSC, Matsuoka et al. (2009)) on the
InternationalSpace Station (ISS) , launched in August 2009, is a 5350 cm array of Xe proportional counters operating at 2 - 30 keV.The instrument has a wide field of view (6 identical unitswith 160 ° × ° FOV) and covers essentially the entire skyduring each
ISS orbit (92 min).
MAXI /GSC observationsin this work are averaged over each day, with σ source sen-sitivity ∼ mCrab. Observations are removed when thesource is observed near the edge of the field of view ofthe GSC, where sensitivity drops. This behavior was con-firmed using images from the MAXI on-demand process . MAXI observed the type-II outburst of V0332+53 during2015 from approximately periastron at the beginning of therise through the peak and then over approximately 40 dayson the decline, except for days when the source was out ofor near the edge of the field of view.
Fermi /GBM
The Gamma-ray Burst Monitor (GBM), one of the two in-struments on board the
Fermi Gamma-Ray Space Telescope ,observes steady and transient sources using earth occulta-tion (Case et al. 2011; Wilson-Hodge et al. 2012).
Fermi waslaunched in June 2008 and consists of 14 detectors: 12 NaIdetectors operating over the energy range 8 keV - 1 MeVand 2 bismuth germanate (BGO) detectors operating at 150keV - 40 MeV. The NaI detectors are used for burst and pul-sar analysis, providing time-tagged data with nominal 0.256s time resolution and 8-channel spectral resolution. In theEarth occultation mode, typically 3-4 NaI detectors view anEarth occultation within 60 ° of the detector normal vector,allowing the instrument to observe > of the sky in a sin-gle orbit and the entire sky every ∼ days. The two BGOdetectors are located on opposite sides of the spacecraft andalso view a large part of the sky in the high energy range.The Fermi /GBM data presented here have been averagedover at least one day of observations, with one-day σ sen-sitivity being ∼ mCrab. Daily averages without a signif-icant number of occultations have been removed, where thenumber of occultations required for a significant detectionare dependent on the observed source flux. Fermi /GBM ob-served the entire 2015 outburst with the exception of a fewgaps where there were not a significant number of occulta-tions of V0332+53.
Swift /BAT
The
Swift /Burst Alert Telescope (BAT) is a 5200 cm coded aperture telescope operating in the 14-195 keV rangewith a 2.0 steradian FOV, 17 arcminute resolution, and 1- 3 arcminute location precision (Barthelmy et al. 2005).Launched in Nov. 2004, BAT has been used to create a hardX-ray All-Sky Survey to a flux limit ∼ − erg cm − s − http://maxi.riken.jp/mxondem/ MNRAS , 1–8 (2016) over 70 months (Baumgartner et al. 2013). BAT accumu-lates detector plane maps approximately every five min-utes in 8 energy bands. Sky coverage for transients is of ∼ − % at > mCrab in one day (Tueller et al. 2010). Swift /BAT observed the entire 2015 outburst.
INTEGRAL
The
International Gamma-ray Astrophysics Laboratory ( IN-TEGRAL ) (Jensen et al. 2003) was launched in October2002 with a highly eccentric 3-day orbital period. With theX-ray monitor JEM-X (Lund et al. 2003), the gamma-rayspectrometer SPI (Roques et al. 2003), and the gamma-rayimager IBIS (Ubertini et al. 2003),
INTEGRAL is able tostudy sources from 3 keV −
10 MeV. For this work, datawere analyzed from JEM-X because of its low energy cover-age and SPI, the latter because of its better energy resolutionrelative to IBIS/ISGRI. During the V0332+53 outburst,
IN-TEGRAL observed the source mainly through two Target ofOpportunity (ToO) periods. The first spanned 2015 July 1702:32:50 to July 18 03:03:15 UTC (MJD 57220-57221) dur-ing the rising part of the flare, and the second from 2015 July30 11:57:51 to August 1 12:47:47 UTC (MJD 57233-57235)during the peak of the flare.
Figure 1 shows the outburst light curves observed by
Fermi /GBM at 12-25 and 25-50 keV and Swift/BAT at 15-50 keV. Figure 2 shows the
MAXI /GSC 2-4, 4-10, and 10-20keV light curves of the 2015 outburst. The 2-4, 4-10, 10-20 keV
MAXI , 12-25 keV GBM, and 15-50 keV BAT lightcurves are made up of daily flux averages normalized to Crabunits, while the 25-50 keV GBM light curve is made up of3-day flux averages also normalized to Crab units. Gaps inthe GBM data are caused by not having a significant num-ber of occultations on those days. Gaps in the
MAXI dataare due to the source leaving the
MAXI /GSC field of view.In both Figures 1 and 2, the times of periastron pas-sage are marked by dashed-vertical lines (Doroshenko et al.2016a). The outburst begins near or slightly before the pe-riastron passage at MJD . . There is an indicationthat the outburst begins slightly earlier at higher energies,although because all instruments are observing near their re-spective noise limits, the differences in sensitivity may playa role in the observed effect. If the outburst does indeed be-gin earlier in the hard X-rays, this could be related to thespectral hardening observed as V0332+53 transitions fromlow-accretion states (Tsygankov et al. 2016; Wijnands et al.2016). In the hard X-rays, the flux then increases linearly un-til there is a ”kink” at MJD . As discussed in Section4.1 below, this kink coincides with V0332+53 changing fromthe horizontal branch of its HID track to the diagonal branchand is marked with a vertical dotted/dashed line. The fluxthen increases linearly across all energy ranges until the sec-ond periastron passage at MJD . , where an uptick influx is seen in all light curves. During the outburst decline,BAT had the most complete coverage of the periastron pas-sage at MJD . , and there again appeared to be a change in the slope associated with the periastron passage.There is no noticeable kink in the flux as the source movedback to the horizontal branch of the HID at MJD aswas observed in the outburst rise.To test the hypothesis of Mowlavi et al. (2006) whichsuggested an exponential decrease in flux throughout theoutburst decline related to exponential emptying of theneutron star accretion disk, a straight line was fit to the Swift /BAT data between the time of transition to the di-agonal branch (radiation-dominated braking) at MJD and , where the flux started to plateau just before pe-riastron passage at MJD . . The reduced- χ of thislinear fit was . ( DOF). An exponential fit to the sametime interval produced a reduced- χ of . ( degrees offreedom, DOF). A straight line was also fit to the databetween the periastron passage at MJD . and thetransition to the horizontal branch at MJD , with areduced- χ of . ( DOF). An attempt to fit this intervalof the outburst decline to an exponential decrease in flux wasunsuccessful. The data are best fit by using a roughly lin-ear increase and decline rather than an exponential increaseor decline. In disagreement with Mowlavi et al. (2006), thissuggests that the dominant factors in determining outburstshape are the times of periastron and transitions betweenradiation-dominated slowing (braking) of the inward flowand slowing due to Coulomb interactions.In the BAT and GBM energy ranges, V0332+53switches between modes of braking at a lower flux level dur-ing the outburst decline than during the rise. At the lower
MAXI energy range, however, no strong statements can bemade due to a lack of observations near the beginning ofthe outburst. The decrease in flux between times of transi-tion points to a more general drop in the luminosity at thesetimes. This is, by definition, a drop in the critical luminosity,confirming the observed drop in magnetic field observed byCusumano et al. (2016).
Figure 3 shows JEM-X and SPI spectra taken during Rev-olution 1565 (MJD 57220-57221), when the V0332+53 fluxwas increasing, and during Revolution 1570 (MJD 57233-57235), when V0332+53 was near the peak of its out-burst. The spectra were fit in XSPEC v. . (Arnaud 1996)using an optically thick Comptonization model (compTTin XSPEC) while the cyclotron resonant features were fitusing gaussian absorption profiles (gabs in XSPEC). Wenote that other authors have used a powerlaw with ahigh-energy cutoff (cutoffpl in XSPEC) to fit the spec-tra of HMXBs (Mowlavi et al. 2006; Tsygankov et al. 2006;Cusumano et al. 2016). Here we have used a physically mo-tivated optically thick comptonization model which gave asimilar goodness of fit to the powerlaw with a high-energycutoff when fitting. A gaussian component was also neededto describe the iron line, which was more significant in theRevolution 1565 spectrum than in the Revolution 1570 spec-trum. JEM-X channels 8-10, 24-27, and 36-45 were ignoredin the fitting because of instrumental features located inthose channels that could not be fitted appropriately. Theseed photon temperature was fixed at 0.5 keV followingTsygankov et al. (2016), where a blackbody spectrum wasobserved with a temperature of 0.5 keV in the propeller MNRAS , 1–8 (2016)
Z. A. Baum et al.
Figure 1.
Fermi /GBM and
Swift /BAT light curves. Vertical dashed lines mark times of periastron passages. Vertical dot-dashed linesmark transitions between horizontal and diagonal branches in hardness-intensity diagram.
Date (Modified Julian Day) - k e V f l u x ( C r ab ) MAXI 2-4keVDate (Modified Julian Day) - k e V f l u x ( C r ab ) MAXI 4-10keV
Date (Modified Julian Day) - k e V f l u x ( C r ab ) MAXI 10-20keV
Figure 2.
MAXI /GSC light curves. Vertical dashed lines mark times of periastron passages. Vertical dot-dashed lines mark transitionsbetween horizontal and diagonal branches in hardness-intensity diagram. regime (after accretion had stopped) for the 2015 outburst.The best fit results are shown in Table 1.In Table 1, an anti-correlation is observed betweenluminosity and the cyclotron line energies as has beenreported for the 2005 outburst (Mowlavi et al. 2006;Tsygankov et al. 2006) and previously for this 2015 outburst(Cusumano et al. 2016). The observed fundamental line en- ergies reported in Table 1 agree well with those reported inCusumano et al. (2016) for spectra taken at similar times.Differences in the fundamental line width and first harmonicline energy may be due to differences between the
Swift and
INTEGRAL instrument responses or the slightly differentcontinuum models used (cutoffpl in Cusumano et al. (2016)and compTT in this work).
MNRAS , 1–8 (2016) E F E [ k e V * ( P ho t on s c m − s − k e V − ) ]
10 100Energy (keV)-202 ( da t a - m ode l ) / e rr o r Rev 1570Rev 1565
Figure 3.
V0332+53
INTEGRAL spectra during outburst rise (Rev 1565) and near peak (Rev 1570).
INTEGRAL
Rev. 1565 1570Start Time (SPI) 57220.098 57233.490 MJDEnd Time (SPI) 57221.175 57235.535 MJDExposure Time (SPI) . × . × sStart Time (JEM-X) 57220.106 57233.548 MJDEnd Time (JEM-X) 57221.134 57235.535 MJDExposure Time (JEM-X) . × . × sCal. Factor . + . − . . . . kT + . − . . + . − . keVOpt. Depth . + . − . . + . − . Norm. Const. . + . − . . + . − . E cyc , . + . − . . + . − . keV σ cyc , . + . − . . + . − . keV τ cyc , + − . + . − . E cyc , + − + − keV σ cyc , + − + − keV τ cyc , + − + − E Fe . + . − . . + . − . keV σ Fe . + . − . . + . − . keVNorm. Const. . + . − . . + . − . Red. χ . × − . × − ergs/cm /s Table 1.
Observation times and spectral fit results for the two
INTEGRAL observation periods. Spectral fit parameters used in thecomptt, gabs, and gauss fit models are: Cal. Factor (scaling factor used to smoothly fit the JEM-X results to SPI), kT (electron plasmatemperature), Opt. Depth (optical depth), and Norm. Const. (overall normalization) for the Comptonization spectrum; E cyc , , , σ cyc , , ,and τ cyc , , (peak energy, sigma, and optical depth for the first and second harmonic cyclotron lines); E Fe , σ Fe , and Norm. Const. (peakenergy, sigma, and normalization for the iron line); overall reduced χ ; and the fitted 2-10 keV flux.MNRAS , 1–8 (2016) Z. A. Baum et al.
X-ray colors, or hardness ratios, provide a means to quantifythe spectral shape during the outburst and are shown in theHID in Figure 4. Fluxes from
Swift /BAT and
MAXI /GSCwere used to create the (15-50keV)/(2-4keV) hardness ratiosduring the 2015 outburst. In Figure 4, the source can be seento move from the rising horizontal branch (becoming harderat low intensity), to the rising diagonal branch (becomingsofter with increasing luminosity), and then down the fallingdiagonal branch (becoming harder with decreasing luminos-ity) and finally along the falling horizontal branch (becomingsofter at low intensity) as the outburst comes to an end.The transition from the horizontal branch to the diag-onal branch is expected (Becker et al. 2012) to be the pointat which the luminosity reaches the critical luminosity (i.e.,where the inward flow becomes radiation-dominated on thediagonal branch rather than governed by Coulomb inter-actions on the horizontal branch). This occurs at approxi-mately MJD 57195 in the HID, corresponding to the changein slope of the light curve at the same time. The peak inthe light curve (near MJD 57238) corresponds to the transi-tion from the rising diagonal to falling diagonal branch. Thetransition from falling diagonal to falling horizontal branchoccurs at approximately MJD 57290, where Figures 1 and 2show slight although not conclusive indications of flatteningof the light curves.Figure 4 shows a difference in color between the riseand decay of the outburst (i.e., hysteresis in the HID), withthe source being harder during the outburst rise than duringthe fall. Despite the incomplete observations of the outburstdecline, the hysteretic behavior observed is much larger thanthe plotted statistical errors and cannot be explained by thenatural spread of the data.
Using the most recent orbital parameters fromDoroshenko et al. (2016a), specifically the times of pe-riastron, the orbital modulation of the light curve becomesapparent. The outburst begins in the 12-25 keV band justprior to periastron at MJD . . In the GBM and BATlight curves, there appears to be a flattening of the lightcurve just before the second periastron at MJD . followed by a steepening. During the decline, the BAT lightcurve shows a flattening around the third periastron atMJD . . The flux is too low at the fourth periastronto allow any definitive statement about the light curvebehavior.During the outburst rise, the difference of a few daysbetween periastron and the increases in flux can be roughlydescribed by the accreted material from the decretion disk ofBQ Cam traveling through the accretion disk of the neutronstar and reaching the X-ray emission regions. For a separa-tion between BQ Cam and the neutron star at periastronof ∼ . × km, with a dynamical timescale determinedby the neutron star mass of . M ⊙ , a delay of ∼ days isexpected (Doroshenko et al. 2016a; Negueruela et al. 1999).It is likely however that increases in the accretion rate wouldbegin before the neutron star reaches periastron. It should also be noted that the accretion disk would not extend all ofthe way to the neutron star surface due to the large magneticfield of the neutron star pulling material from the accretiondisk at the Alfven radius. In this case, the energy depen-dence of the outburst light curve would then be determinedby the details of how the accretion rate affects the soft andhard X-ray emission regions/processes. Based on the depen-dence of the hard X-ray light curve on the mode of brakingwithin the accretion column, we are led to conclude that atleast the X-rays above ∼ keV (lower limit of Fermi /GBMenergy band) must originate from within the accretion col-umn to create such a dependence (Becker et al. 2012). Nodefinitive conclusions based on Figure 2 can be made aboutthe emission region of the softer X-rays due to the scarcityof
MAXI observations near the time of transition across thecritical luminosity.Mowlavi et al. (2006) have suggested that the 3-60 keV
INTEGRAL light curve during the decline of the 2005 out-burst could be modeled by an exponential decay with a fold-ing time of 20-30 days before becoming linear near the endof the outburst. They argue that the suggested behavior issimilar to that in LMXBs, where a hot disk illuminated bythe central star empties at a rate proportional to its mass,producing an exponential decay of the emitted flux. As inMowlavi et al. (2006), it follows that when the disk coolsand is no longer uniformly heated, the flux then follows alinear decrease.If the exponential behavior was real as suggested inMowlavi et al. (2006), then one would expect to see thisthroughout the entire outburst decay. Figures 1 and 2 showa roughly linear decay between periastron at MJD . and the transition between accretion states at MJD 57290 which does not support the description of exponentially de-creasing flux throughout the outburst decay . However, there isthe possibility that the dips observed just before periastronin the GBM and BAT light curves could still be describedby an exponentially emptying accretion disk. This is sup-ported by the outburst decline having more modulation ontop of the linear decay when compared to the linear increaseduring the outburst rise, because the accretion disk wouldbe more likely to only be filled partially during the decline. In Figure 1, the flux level of V0332+53 as it transitionsacross the critical luminosity is lower during the outburstdecline than during the outburst rise. This is a signal thatthe critical luminosity has dropped, and by proxy, the mag-netic field has also decreased. This confirms the result ofCusumano et al. (2016) that based on the decrease of thecyclotron resonance line energy the magnetic field droppedduring the outburst. Combining our knowledge of the timeswhere V0332+53 crosses the critical luminosity with thespectra reported by Cusumano et al. (2016), we can eval-uate the drop in the critical luminosity. Interpolating lin-early between the spectra taken by
Swift on MJD 57193and 57201 we estimate a critical luminosity of ∼ . × erg/s on 57195. Similarly, we estimate a critical luminosityof ∼ . × erg/s on MJD 57290 using the Swift spectrataken on MJD 57277 and 57293, corresponding to a ∼ decrease. This can be compared to the results obtained byusing the fundamental cyclotron line energy to estimate the MNRAS , 1–8 (2016)
Hardness Ratio (15-50)/(2-4)0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 / s ) - k e V f l u x ( / c m MJD < 57195 (HB-rise)57195 < MJD < 57238 (DB-rise)57238 < MJD < 57290 (DB-fall)MJD > 57290 (HB-fall)
Figure 4. magnetic field and then the critical luminosity (Becker et al.2012). A fundamental line energy of ∼ . keV for MJD57195, interpolating from Cusumano et al. (2016), corre-sponds to a surface magnetic field of ∼ . × G anda critical luminosity of ∼ × erg/s. For MJD 57290 weinterpolate a fundamental line energy of ∼ . keV, mag-netic field of ∼ . × G, and a critical luminosity of ∼ . × erg/s, corresponding to a ∼ drop in the crit-ical luminosity. We note that no correction has been madehere for gravitational redshift effects.Doroshenko et al. (2016b) have noted that the spin-uprate is higher for a given luminosity during the rising phasethan during the decline. They argue that this is the oppo-site of what is expected from the connection between thetorque exerted on the neutron star and the accretion flow,and suggest as a result that the decrease in the cyclotronline energy may be due to a change in the size of the mag-netosphere rather than to a decrease in the magnetic field.On the contrary, the consistency shown above between thecritical luminosity based on the times derived from the lightcurves and the HID, and the critical luminosity determinedfrom the line energies seems to support the Cusumano et al.(2016) suggestion of a direct connection between the cy-clotron line energies and a decreasing magnetic field. Hysteresis in the HID can be seen in Figure 4, where thefalling diagonal branch is softer than the rising diagonalbranch. Similar behavior has been observed during the 2004-2005 event (Reig 2008; Reig and Nespoli 2013) despite thepoor coverage of the outburst rise. This does not appear tobe related to the decrease in magnetic field during the eventsuggested by the observed decrease in cyclotron line energyin 2015, since hysteresis was observed in the HID with no observed change in the magnetic field between the rise anddecay of the 2004-2005 event (Reig 2008; Nakajima et al.2010). The lack of a connection between the magnetic fieldand hysteresis in the HID suggests that the cause of thehysteresis is likely located outside of the central accretioncolumn.In our case, we see a difference between the spectral fitsfor
INTEGRAL
Revolutions 1565 and 1570 in the electronplasma temperature, which cools from ∼ keV to ∼ . keVnear the outburst peak. More efficient cooling of the electronplasma later in the outburst could be a possible explanationfor the hysteresis observed in Figure 4, where the hard X-rays more efficiently cool the plasma as accretion increases.The hard X-rays would be down-scattered during this pro-cess, resulting in a softer spectrum as the outburst rises tomaximum. We have analyzed data from
MAXI /GSC,
Fermi /GBM,
Swift /BAT,
INTEGRAL /JEM-X, and
INTEGRAL /SPItaken during the Type II outburst of the Be/X-ray binarysystem V0332+53 in 2015. The complex features in the lightcurve are correlated with the times of periastron passage,changes in the mode of braking within the accretion column,and the dynamical timescales of the accretion disk. In agree-ment with previous measurements, we see an anticorrelationof the fundamental cyclotron line energy with the luminos-ity. We also find that the X-ray spectra are well described byan optically thick comptonization model where the electronplasma is more efficiently cooled through Compton down-scattering as the outburst progresses. We track V0332+53along its path on a hardness-intensity diagram throughoutthe outburst, observing clear indications of hysteresis (i.e.,
MNRAS , 1–8 (2016)
Z. A. Baum et al. a softening of the spectrum) during the outburst. This be-havior is similar to that observed in V0332+53 in its 2004-2005 outburst and in 4U 0115+63 (Reig and Nespoli 2013).Combining our knowledge of the times of transition betweenaccretion states with the
Swift spectra (Cusumano et al.2016), we estimate that the critical luminosity decreases by ∼ from the rise of the outburst to the decline. This paper includes data collected by the
Fermi and
Swift missions, funded by the NASA Science Mission directorate.We especially appreciate the support from Colleen Wilson-Hodge and the other members of the GBM Earth Occul-tation team. MAXI data were provided by RIKEN, JAXA,and the MAXI team. Analysis of INTEGRAL data was sup-ported by ESA and NASA through CNES. ZAB appreciatessupport from NASA EPSCoR and the Louisiana Board ofRegents.
REFERENCES
Arnaud, K., 1996, in G. Jacoby, J. Barnes, eds, ASP Conf. Ser.Vol. 101, Astronomical Data Analysis Software and SystemsV. Astron. Soc. Pac., San Francisco, p. 17Barthelmy, S. D., et al. 2005, Space Sci. Rev., 120, 143Baumgartner, W. H., et al. 2013, ApJ, Suppl. 207Becker, P.A., et al. 2012, A&A, 544, 123cCaballero-Garc´ıa, M.D. et al., 2016, A&A, 589A, 9CCase, G. L., et al. 2011, ApJ, 729, 105Cusumano, G., et al. 2016, MNRAS, submittedDoroshenko, V. et al. 2015, The Astronomers Telegram 7822Doroshenko, V., Tsygankov, S., Santangelo, A., 2016, A&A, 589A,72DDoroshenko, V., et al. 2016, arXiv:1607.03933Frank, J., King, A., and Raine, D. J., 2002,
Accretion Power inAstrophysics: Third Edition
Goranskii, V.P., 2001, Astronomy Letters, 27, 516GKreykenbohm, I., et al. 2005, A&A, 433L, 45KJensen, P. L., Clausen, K., Cassi, C. et al. 2003, A&A, 411, L7Lund, N., Budtz-Jørgensen, C., Westergaard, N. J. et al. 2003,A&A, 411, 231LLutovinov, A.A., et al. 2015, MNRAS, 448, 2175LMakishima, K., et al. 1990, PASJ, 42, 295MMakishima, K., et al. 1990, ApJ 365, L59Matsuoka, M., et al. 2009, PASJ, 61, 999Mihara, T., et al. 1995, AAS, 18710403MMowlavi, N. et al. 2006, A&A, 451, 187MNakajima, M., Mihara, T., Makishima, K., 2010, ApJ, 710, 1755NNakajima, M. et al. 2015, The Astronomers Telegram 7685Negueruela, I., et al. 1999, MNRAS, 307, 695NPottshmidt, K., et al. 2005, ApJ, 634L, 97PPoutanen, J., et al., 2013, ApJ, 777, 115PReig, P., 2008, A&A, 489, 725RReig, P., Nespoli, E., 2013, A&A, 551A, 1RRoques, J. P., Schanne, S., von Kienlin, A. et al. 2003, A&A, 411,L91Terrell, J. and Priedhorsky, W.C. 1984, ApJ 285, L15Tsygankov, S. S. et al. 2006, MNRAS, 371, 19-28Tsygankov, S. S. et al. 2010, MNRAS, 401, 1628TTsygankov, S. S. et al. 2016, A&A, submittedTueller, J., et al., 2010, ApJS, 186, 378TUbertini, P., Lebrun, F. Di Cocco, G. et al. 2003, A&A, 411, 131LWilson − Hodge, C. A., et al. 2012, ApJS, 201, 33 Wijnands, R., Degenaar, N., 2016, MNRAS, submittedThis paper has been typeset from a TEX/L A TEX file prepared bythe author. MNRAS000