aa r X i v : . [ a s t r o - ph . H E ] J u l Astronomy & Astrophysics manuscript no. 11636shu c (cid:13)
ESO 2018November 13, 2018
Outbursts from IGR J17473-2721
Shu Zhang , Yu-peng Chen , Jian-Min Wang , , Diego F. Torres , Ti-Pei Li , Laboratory for Particle Astrophysics, Institute of High Energy Physics, Beijing 100049, China Theoretical Physics Center for Science Facilities (TPCSF), CAS. ICREA & Institut de Ci`encies de l’Espai (IEEC-CSIC), Campus UAB, Facultat de Ci`encies, Torre C5-parell, 2aplanta, 08193 Barcelona, Spain Center for Astrophysics,Tsinghua University, Beijing 100084, ChinaReceived / Accepted
ABSTRACT
Aims.
IGR J17473-2721 was discovered by
INTEGRAL as a neutron star low mass X-ray binary. To date, two outburstshave been detected in X-rays by
RXTE /ASM and
SWIFT : the one occurring in 2005 was weak and was characterizedby a low/hard state spectrum; the one of March 2008 was strong and showed a 6-step evolution in its flux. We investigatetheir evolution, emphasizing the later outburst.
Methods.
We analyzed all available observations carried out by
RXTE on IGR J17473-2721 during its later outburst.We analyzed as well all the available
SWIFT /BAT data on this source.
Results.
The flux of the latter outburst rose in ∼ one month and then kept roughly constant for the following ∼ twomonths. During this time period, the source was in a low/hard state. The source moved to a high/soft state withinthe following three days, accompanied by the occurrence of an additional outburst at soft X-rays and the end of thepreceding outburst in hard X-rays. During the decay of this soft outburst, the source went back to a low/hard statewithin 6 days, with a luminosity 4 times lower than the first transition. This shows a full cycle of the hysteresis intransition between the hard and the soft states. The fact that the flux remained roughly constant for ∼ two months attimes prior to the spectral transition to a high/soft state might be regarded as the result of balancing the evaporationof the inner disk and the inward accretion flow, in a model in which the state transition is determined by the mass flowrate. Such a balance might be broken via an additional mass flow accreting onto the inner disk, which lightens the extrasoft outburst and causes the state transition. However, the possibility of an origin of the emission from the jet duringthis time period cannot be excluded. The spectral analysis suggests an inclined XRB system for IGR J17473-2721. Sucha long-lived preceding low/hard state makes IGR J17473-2721 resemble the behavior of outbursts seen in black holeX-ray binaries like GX 339-4. Key words.
X-rays: individual: IGR J17473-2721
1. Introduction
X-ray binaries (XRBs) are generally classified as black hole(BH) or neutron systems according to the type of the com-pact star. The latter class is further grouped into the so-called Z and atoll sources, according to their behavior inspectral and time evolution. Z sources are usually brighterthan the atoll ones, due to a higher accretion rate. The typ-ical luminosity of atoll sources is generally less than 0.1 × the Eddington limit, while for Z sources it is > × theEddington limit (Hasinger and van der Klis 1989). Atollsources are mostly studied at their outbursts, during whichthe spectrum evolves from the island state to the bananastate , which is analogous to the BH counterparts: the hardand the soft states, respectively. The spectral evolution ofboth types of XRBs can be quite similar, as observed by RXTE and
BeppoSAX . Those outbursts that (1) stay onlyin the hard state; (2) have a hard outburst preceding a softone; (3) show hysteresis in the transition between the hardand the soft states are particularly interesting.Meyer et al. (2000) believed that the change in spectralstate is related to the modifications in the accretion flow
Send offprint requests to : Shu Zhang
Correspondence to : [email protected] and its balance by the evaporation at the inner disk. At thestart of the outburst, the accretion rate is low and the disk iscold and optically thick. The accretion flow in the inner diskregion evaporates to form hot plasma or an ADAF region,to produce hard X-rays via Comptonizing the seed pho-tons from the surface of the neutron star or the boundarylayer. The sources stay in the low/hard state. An increasein accretion flow toward the compact star might allow thehot plasma to be cooled and form an inner disk. As a re-sult, the ADAF region is suppressed and the source movesinto the high/soft state. Mayer-Hofmeister et al. (2005) ar-gued that the hysteresis could be a natural result of thecooling/heating of the corona by the photons produced atdifferent mass accretion rates. Since the last decade, theoutbursts of low mass X-ray binaries (LMXBs) have beenobserved mainly by the high sensitivity detectors on boardthe satellites
RXTE and
BeppoSAX in soft and hard X-rays. However, since the neutron star has a solid surfaceand magnetic field, spectral modeling the outburst in neu-tron star LMXBs could be more complicated than thosewith a BH. So far, two main paradigms exist to model out-bursts: the Eastern model (Mitsuda et al. 1989) and theWestern model (White et al. 1988). They differ in how thethermal and Comptonized components are dealt with. The
Shu Zhang et al.: Outbursts from...
Eastern model considers the thermal emission from the diskand the Comptonization from hot plasma around the neu-tron star, while the Western model considers the thermalemission from the boundary layer and the Comptonizationfrom the disk. Although Done et al. (2007) showed thatthe fast variability of the hard component of the energyspectrum could be evidence against the Western model,in practice the models are usually hard to unambiguouslyresolve from the observational data, e.g. the existence ofthe so-called degeneracy in models. For example, the hardcomponent of the outbursts as observed in NS LMXBsAql X-1 and 4U 1608-52 by
RXTE is well described bya broken power law model (Lin et al. 2007), while duringthe 1998 outburst of 4U 1608-52 is described by thermalComptonization (Gierlinski et al. 2001).Since the satellites like
INTEGRAL and
SWIFT beganto produce data in 2002 and 2004, more atoll sources, forexamples, IGR J00291+5934 (Eckert et al. 2004; Linares2008a) and SWIFT J1756.9-2508 (Krimm et al. 2007;Linares et al. 2008b), have been discovered in hard X-rays.The hard X-ray source IGR J17473-2721 was discovered asan additional member of the sample by the IBIS/ISGRItelescope on board of INTEGRAL during the ultra deepOpen Program observations of the Galactic center carriedout on March 17-28 and April 9-21, 2005 (Grebenev et al.2005). Following this discovery, observations were carriedout at other wavelengths as well, i.e. at soft X-rays by
Chandra with an exposure of 1.1ks, and in the infrared atthe Ks band by the
PANIC camera of the 6.5 m MagellanI telescope with an exposure of 270 seconds.Historically, the first outburst from IGR J17473-2721 was observed in 2005, with a flux level of ∼ ± × − erg/cm /s at 0.5-10 keV, and a locationat (Ra,Dec)=(17 47 18.06, -27 20 38.9) (Juett et al. 2005).The second outburst was observed three years later by SuperAGILE at 17-25 keV on March 26, 2008 (Del Monte etal. 2008). According to a joint analysis of the intensity andthe temporal and spectral properties, IGR J17473-2721 wasrecognized as a type-I burster. This was later confirmed by a
SWIFT observation carried out on March 31 2008, with anexposure of 4.1 ks. A 100-second type-I burst was detectedat 0.01-100 keV with a temperature ∼ +0 . − . keV and abolometric flux ∼ × − erg/cm /s (Altamirano et al.2008a). From SWIFT observations, a column density of ∼ N H =3.8 +0 . − . × cm − and a spectral index of 1.68 +0 . − . were derived (Altamirano et al. 2008a). The source was es-timated to be at a distance of ∼ ∼ × erg/s (Altamiranoet al. 2008a). The following observations were performedby INTEGRAL and
SWIFT in hard X-rays on April 1 and8, 2008. These observations show clearly that a very largeburst occurred in the source (Kuulkers et al. 2008; Baldovinet al. 2008). Noticeably, the
INTEGRAL /JEMX detectedanother 50-second type-I burst of a peak flux ∼ ∼ ± RXTE results identified thesource as an atoll source (Altamirano et al. 2008b).Since both outburst events have been well monitored by
SWIFT in hard X-rays and by
RXTE in soft X-rays, weinvestigate the details of the outburst in these two neigh-boring X-ray bands. We aim to investigate the evolutionof the flux in hard and soft X-rays during the most recentoutburst.
2. Observations and Results
ASM is one of the three detectors on-board the
RXTE satellite (Gruber et al. 1996), which has been used to trackthe long-term behavior of the source in the energy band 1.5-12 keV since February 1996. The target source was usuallyobserved several times per day, with so-called dwells of 96seconds duration each. The extracted source lightcurves areavailable at the energy bands of 1.5–3, 3–5, 5–12, and 1.5–12keV. For IGR J17473-2721, we take the daily average at theenergy band 1.5-12 keV (see Fig. 1). The ASM lightcurveshows that two outbursts occurred from the source. Thefirst one (Fig. 2) lasted for 95 days ( ∼ MJD 53495-53590)and reached a peak flux of 6 ct/s ( ∼
80 mcrab). The fluxrose in the first 35 days and then decayed in the follow-ing 60 days. The source became active again three yearslater, with a start time at around MJD 54555. This hugeoutburst showed a more complex, 6-step evolution in flux,and accordingly six time intervals are defined in Fig. 2. Theflux rose for ∼ one month (interval I) and then kept roughlyconstant for the next ∼ two months (interval II). In the fol-lowing three days, the flux in soft X-rays suddenly jumpedto ∼
30 ct/s ( ∼
400 mcrab, interval III) and then decayed(intervals IV and V).
SWIFT carries the Burst Alert Telescope (BAT,Barthelmy et al. 2005), which has a rather large field ofview of 1 . −
150 keV energy band. This makes it possible for asource to be monitored daily in hard X-rays. The dataproducts are therefore lightcurves in the 15-50 keV energyband and are publicly available . The BAT lightcurve tracesIGR J17473-2721 back to February 12, 2005 (see Fig. 1),and covers the two outbursts as observed by ASM. In hardX-rays of 15-50 keV, the first outburst has a similar profileto that seen in the soft X-rays (Fig. 2). The peak flux isabout 0.035 ct/cm /s ( ∼
153 mcrab). The second outbursthas a profile resembling that in soft X-rays until ∼ MJD54634 (Fig. 2, intervals I and II). Later, while in soft X-rays the source had an additional huge outburst, in hardX-rays the preceding outburst had almost ceased. The fluxdropped in hard X-rays within three days from ∼ /s ( ∼
330 mcrab) to ∼ /s ( ∼
44 mcrab)(interval III). The flux remained at that low level (intervalIV) during the decay of the soft X-ray outburst until MJD54670, when a new outburst in hard X-rays formed within 6days (interval V). In the time interval VI (Fig. 2), this newoutburst has an average flux ∼ /s at 15-50keV, and ∼ The hardness ratios of the two outbursts are constructedwith a flux ratio 15-50 keV/1.5-12 keV, and are shown inFig. 3. It is obvious that the spectrum remained hard (e.g.in the low/hard state) for the whole first outburst periodand as well for the first ∼
80 days of the second outburst(intervals I and II). We performed a linear fit to the hard-ness ratio and show this in Fig. 3 for both the fit and the See the
SWIFT /BAT transient moni-tor results provided by the
SWIFT
Team at http://swift.gsfc.nasa.gov/docs/swift/results/transients/ hu Zhang et al.: Outbursts from... 3 residual. It seems that for the first outburst the spectrumslightly softens along the outburst evolution, although thedata quality is not high enough to be certain of this. For thesecond outburst, the spectrum becomes soft within threedays, with the emission dominated by the soft X-rays (inter-val III). Accompanying this change is an additional suddenoutburst in soft X-rays and the cessation of the precedingoutburst in hard X-rays (interval III). Such a feature is typ-ical for LMXB in their spectral transitions from a low/hardto a high/soft state. Along with the decay in soft X-rays,the source stayed in a high/soft spectral state (interval IV)until the transition in interval V. Within 6 days, the sourcehad returned to a low/hard state at a relatively low lu-minosity level (interval VI). This shows a full cycle of thehysteresis in the transition between the hard and the softstates.
During the long outburst of IGR J17473-2721, 93
RXTE /PCA observations were made, with the identifier(OBSID) of proposal number (PN) 93442, available in thedata archive. These observations contain ∼
225 ks of ex-posure on the source, and are scattered over the wholeevolution of the outburst (see Fig. 2). These observationssmoothly cover the last 5 time intervals II-VI. The analy-sis of PCA data is performed using Heasoft v. 6.2. We filterthe data using the standard
RXTE /PCA criteria. Only thetop layer of PCU2 (in the 0-4 numbering scheme) has beenused for the analysis, because only the PCU2 was 100% onduring the observation, and we use the time intervals withthe following constraints: elevation angle > ◦ , and point-ing offset < . ◦ . Elevation is the angle above the limb ofthe Earth and the pointing offset is the angular distancebetween the pointing and the source. The background fileused in the analysis of PCA data is the most recent oneavailable from the HEASARC website for bright sources,pca bkgd cmbrightvle eMv20051128.mdl. Data from clus-ter 1 of the HEXTE system are used to produce the spectra.An additional 1% systematic error is added to the spec-tra because of calibration uncertainties, if not otherwisespecified. The spectra are fitted with XSPEC v12.3.1 andthe model parameters are estimated with a 90% confidencelevel.We investigate the spectrum by fitting the data fromeach of the 5 time intervals as shown in Fig. 2. For thethermal component, we used the multicolor disk black bodymodel (diskbb) and the black body model (bbodyrad),where the soft X-rays are supposed to come from the diskand the boundary layer or surface of the neutron star, re-spectively. For the hard component, we performed trialswith the thermal Comptonization model (comptt), the cut-off power law model (cutoffpl), and the broken power lawmodel (bkn). For all the trials of the models, the central en-ergy of the iron line is fixed at 6.4 keV, and the absorptionis fixed at 3.8 × atoms/cm (Altamirano et al. 2008a).The energy band is adopted at 3-30 keV for PCA and at 30-100 keV for HEXTE. A constant is added into the model toaccount for the different normalization between PCA andHEXTE.The normalization of the disk black body is describedas ( R in /D ) cos ( θ ), where the R in in units of km, D thesource distance in units of 10 kpc, and cos( θ ) the disk in-clination. Generally, for XRB systems, an observation of an eclipse means that the accretion disk is edge-on, andtherefore puts stronger constraints on the disk inclination.Since for IGR J17473-2721 no eclipses are observed, we canconstrain the normalization factor in the disk black bodymodel. If we take the companion star to be of solar size ( ∼ ∼ ◦ in order toavoid having an eclipse in the lightcurves. By putting thesource as D ∼ in as greater than10 km, we have the lower limit of the normalization of thedisk black body as 15-30 or even larger. In our model fittingwe remove those where the normalization of the disk blackbody is well below 30. Also, we do not consider models witha reduced χ larger than 1.2.We find that for the time interval II only the modelconsisting of the black body and the cutoff power law canfit the data well (see Fig. 4 for the spectrum and Table 1for the parameters). The reduced χ is 1.17 with 71 dofs.Replacing the black body by a disk black body results inthe normalization of ∼ χ is too large to be acceptable( ∼ χ ∼ χ ∼ χ ∼ χ ∼ χ is about1.01 with 72 dofs (see Fig. 4 for the spectrum and Table 1for the parameters). In the other trials, the reduced χ val-ues are acceptable but there are unconstrained parameters.For the time interval VI, there are three possible models:a black body plus broken power law ( χ ∼ χ ∼ χ ∼ Shu Zhang et al.: Outbursts from...
The flux ratio of 4-6.4 keV/3-4 keV is defined as the softcolor whereas the ratio 9.7-16 keV/6.4-9.7 keV is definedas the hard color. Using them, we produce the color-colordiagram (CCD) in Fig. 5. Such a color definition is con-sistent with that adopted in Gladstone et al. (2007). Theoutburst evolution is clearly seen in this diagram; it hasa typical profile of an atoll source. During the hard statebefore the additional soft outburst, the source remains onthe island, and then moves to the banana branch in thefollowing soft state; after the transition to the hard stateagain, the source returns to the island branch, showing afurther feature unlike those generally seen in atoll sources.Along the evolution of the outbursts, most atoll sourceshave CCDs moving from the island branch to the lower leftpart of the banana branch and then to the right part of thebanana branch (Done et al. 2007; Hasinger & van der Klis1989).To further investigate the CCD, we follow the analy-sis in Gladstone et al. (2007) on the
RXTE atoll samples.We perform spectral fittings to the intrinsic soft and hardcolors, which are free from absorption and independent ofthe instrument response. The resulted CCD and the color-luminosity diagrams are shown in Fig. 6, which supportsthat IGR J17473-2721 should belong to the diagonal groupas defined in Gladstone et al. (2007) for the atoll source clas-sification. During outburst decay back to the hard state, thetrace in the color-color diagram is vertical at a luminosityaround 0.03-0.05 L
Edd (for distances of 3.9-5.4 kpc). Thisoccurs for the first atoll source of the diagonal group, withthe full cycle of hysteresis observed during the outburst.
3. Discussion
In the outbursts of IGR J17473-2721, we have seen al-most the entire range of state evolution so far detected inother atoll sources: the preceding hard state, the soft stateand the hysteresis in state transitions between the hardand the soft. By comparing the outburst evolution betweenhard and soft X-rays for the newly discovered neutron starLMXB system IGR J17473-2721, we have found some otherinteresting behavior. The first of its outbursts was ratherweak and the source stayed in the low/hard state. So didthe second outburst for the time period prior to the firststate transition. This resembles the small subset of LMXBtransients like GS 1354-64, GRO J0422+32, GRS 1719 andXTE J1118+480, which were observed during an outburstto have a spectrum remaining in a low/hard state with notransition to a high/soft state (e.g. Brocksopp et al. 2001).An explanation of this attributes the spectral transitionto the change in accretion mass rate (Meyer et al. 2000).Along with the increasing mass flow rate, the evaporationand inward mass flow become dynamically balanced and,as a result, the truncated inner disk moves closer to thecompact object. Such a procedure is responsible for the lu-minosity rise of the outburst, and the source stays for mostof the time in a low/hard state. Given an individual LMXBsystem, when the mass flow rate exceeds the critical value,the balance between evaporation and inward flow will bedestroyed. A direct result of this is the ceasing of the out-burst in hard X-rays and increasing emission dominatedby the soft X-rays, e.g. to form a so-called high/soft state. Otherwise the source stays in a low/hard state for the wholelife of the outburst (Meyer-Hofmeister 2004).For IGR J17473-2721, instead of forming a sharp peakat the end of the flux rise, the flux does not change muchfor the next two months in a low/hard state. This might beregarded as evidence for a steady mass flow moving inwardand an ADAF filled inner region remaining for quite a longtime. The mass flow rate did not increase during this timeperiod to reach the critical rate for the spectral transition.An even more interesting feature is that, after the long-lived low/hard state, the source finally moved to a high/softstate. Such a transition requires the balance to be upset viaadditional mass flow accreting into the inner region.We notice that the outburst of IGR J17473-2721 in hardX-rays is a long-live plateau rather than a sharp peak, typi-cal of the previous samples like Aql X-1. The entire outburstevolution of IGR J17473-2721 resembles the 1998 outburstof the BH XRB GX 339-4. GX 339-4 stayed in a hard statefor 500 days prior to transition to a soft state within roughly100 days. The entire outburst lasted for at least 400 days, ifthe start time is the transition to soft state (Yu et al. 2007,Belloni et al. 1999). The outburst of IGR J17473-2721 ismore similar to GX339-4 based on their long-lived preced-ing outburst in hard X-rays. To date, Aql X-1 is among thefew NS to have an outburst similar to GX339-4 by showingthe hysteresis, but the obvious difference of forming a sharppeak in the preceding hard X-ray outburst limits its sim-ilarity. Therefore, the long-lived preceding hard outburstand the clearly established hysteresis with the luminositydiffering by a factor of four between the state transitions,make IGR J17473-2721 resemble the BH XRBs in outburstmore than the other few known atolls.As it is currently known, neutron star LMXBs havespectral states comparable to those in BH XRBs, whenin outburst. The state transitions and the outburst evo-lution of both atolls and BH XRBs are likely described in asimilar scenario, where the accretion rate and the balancebetween inward accretion flow and the evaporation of theinner disk play an important role. Although the spectralchanges are obvious for LMXB in outburst, their modellingis more complicated than in BH XRBs due to the uncer-tainty attributed to the emission from the surface of theneutron star or its boundary layer. The atoll sources Aql X-1 and 4U 1608-52 are examples: they have been observed sofar (by
RXTE ) in more than 20 outbursts. Lin et al. (2007)find that, instead of the classic Comptonization model forthe hard component, a broken power law model works bet-ter both for the hard and the soft states. For 4U 1608-52,its 1998 outburst was well studied as well by Gierlinski andDone (2001), who found that the non-thermal component ofthe spectrum can be well represented in a Comptonizationplus Compton reflection model. These findings indicate thedifficulty in unambiguously modeling the spectrum of theatolls when in outburst. For IGR J17473-2721 the spec-trum of the high/soft state can be well represented usingthe model consisting of a black body and cutoff power law.That the hard component for most of the outburst can befitted by the non-Comptonization model in its the bananabranch is consistent with the spectral fits of Aql X-1 and4U 1608-52 in outburst using a broken power law model asthe hard component (Lin et al. 2007). The broken powerlaw component can be considered as Comptonization undercomplex conditions or in combination with other radiationprocess (Lin et al. 2007). We notice that the data for the hu Zhang et al.: Outbursts from... 5 long-lived hard state preceding to the hard/soft transitioncan only be fitted by using a model with the components ofthe black body and the cutoff power law. Unlike the othertime intervals (III-VI) where a spectral degeneracy exists,the thermal Comptonization model can be simply rejectedfrom the obviously unacceptable χ . This may provide evi-dence that the hard X-rays in this long-lived hard state maynot be the Comptonization emission and may instead origi-nate in a jet. We notice as well that for IGR J17473-2721 thedisk black body model is likely not appropriate for almostthe whole evolution of the outburst; those normalizationsof the black body used for fitting the soft component in thespectrum are relatively larger during the hard states thanduring the soft states. These features may be understood ina largely inclined XRB system: due to the large inclinationof the disk, the projection of the disk emission area to theline of sight becomes small and hardly visible in the energyspectrum. The soft component of the spectrum may comefrom the area near the surface of the neutron star, and besurrounded by the corona region of the hot plasma whichis optically thin in the hard state.In summary, the newly discovered neutron star LMXBIGR J17473-2721 showed interesting features in its outburstevolution, with a long-lived plateau hard outburst followedby a spectral transition to the high/soft state and a fur-ther increase of luminosity during the high/soft state. Inthe model of a truncated accretion disk this might be in-terpreted as a consequence of a steady mass flow movinginward, a truncated inner disk and an ADAF filled innerregion staying for a long time during the outburst evolu-tion of IGR J17473-2721. The possibility of an origin ofthe emission from the jet during this time period also can-not be excluded. The spectral analysis suggests an inclinedXRB system for IGR J17473-2721. The full cycle of hys-teresis together with a long-lived plateau in the outburstlight curve during the hard state, preceding the hard/softspectral transition, make IGR J17473-2721 more analogousto the BH XRBs than any other atolls observed previously. Acknowledgements.
This work was subsidized by the NationalNatural Science Foundation of China, the CAS key Project KJCX2-YW-T03, and 973 program 2009CB824800. J.-M. W. thanks theNatural Science Foundation of China for support via NSFC-10325313,10521001 and 10733010. DFT acknowledges support from grants AYA2006-00530 and CSIC-PIE 200750I029.
References
Altamirano D., Degenaar N., and Markwardt C. 2008a, ATel 1459Altamirano D.,Galloway D., Chenevez J., et al. 2008b, Atel, 1651Baldovin Saavedra C., Chenevez J., Beckmann V., et al. 2008, ATel1468Barthelmy S.D., Barbier L.M., Cummings J.R., et al. 2005, Space Sci.Rev., 120, 143Belloni T., Mendez M., van der Klis M., et al. 1999, ApJ, 519, L159Brocksopp C., Jonker P. G., Fender R. P., et al. 2001, MNRAS, 323,517Del Monte E., Evangelista Y., Feroci M., et al. 2008, ATel 1445Done C., Gierlinski M., & Kubota A. 2007, A&ARv, 15, 1Eckert D., et al. 2004, ATel, 352Gierlinski M., & Done C. 2001, MNRAS, 337, 1373Gladstone J., Done C., & Gierlinski M. 2007, MNRAS, 378, 13Grebenev S.A., Molkov S.V., Sunyaev R.A. 2005, ATel 467Gruber D. E., Blanco P. R., Heindl W. A., et al. 1996, A&AS, 120,641Hasinger G., & van der Klis M. 1989, A&A, 225, 79Juett A.M., Kaplan D.L., Chakrabarty D., et al. 2005, ATel 521Krimm H.A., et al. 2007, ATel, 1105Kuulkers E., Shaw S., Beckmann V., et al. 2008, ATel 1461 Lin, D., Remillard, R.A., & Homan, J. 2007, ApJ, 667, 1073Linares M. 2008, astro-ph/0809.3368Linares M., Rudy W., van der Klis M., et al. 2008, ApJ, 677, 515Meyer-Hofmeister E. 2004, A&A, 423, 321Meyer-Hofmeister E., Liu B.F., & Meyer F. 2005, A&A, 432, 181Meyer F., Liu B.F., & Meyer-Hofmeister E. 2000, A&A, 361, 175Mitsuda K., Inoue H., Nakamura N., & Tanaka Y. 1989, PASJ, 41, 97White N.E., Stella L., & Parmar A.N. 1988, ApJ, 324, 363Yu W.F., Lamb F.K., Fender R., & Van der Kilis M. 2007, ApJ, 663,1309
Shu Zhang et al.: Outbursts from... R a t e ( A S M c t / s ) Burst 1 Burst 21.5-12 keV
MJD (days)53500 54000 54500 ) R a t e ( BA T c t / s / c m -0.0500.050.1 Fig. 1.
ASM lightcurve (1.5-12 keV, upper panel) and
SWIFT /BAT lightcurve (15-50 keV, lower panel) covering thetwo outbursts of IGR J17473-2721. The dashed lines show the start and the end of each outburst. R a t e ( A S M c t / s ) a ) R a t e ( BA T c t / s / c m b R a t e ( A S M c t / s ) I II III V VIIV c MJD (days)54550 54600 54650 54700 ) R a t e ( BA T c t / s / c m d Fig. 2.
Flux evolution of the first outburst as observed in ASM (panel a) and in
SWIFT /BAT (panel b), and of thesecond outburst as observed in ASM (panel c) and in
SWIFT /BAT (panel d). The dashed lines in panels c and d denotesix steps in flux evolution. The available PCA observations are marked at the top the panel c (diamonds). hu Zhang et al.: Outbursts from... 7 H a r dn ess r a t i o R es i du a l -0.0100.01 H a r dn ess r a t i o I II III IV V VI
MJD (days)54550 54600 54650 54700 R es i du a l -0.0100.01 Fig. 3.
Evolution of the hardness ratio (15-50 keV/1.5-12 keV) of the first outburst (upper panel) and of the secondoutburst (lower panel). The linear fits are shown by the solid lines and the residuals are given in the lower part of eachfigure. no r m a li z ed c oun t s s (cid:358) k e V (cid:358)
105 20 50 (cid:358) (cid:39) S (cid:70)
105 200.010.1110100 no r m a li z ed c oun t s s (cid:358) k e V (cid:358)
105 2002 (cid:39) S (cid:70)
105 20 50 105 20 50Energy (keV) 105 20 50Energy (keV) cutoffpl + bbodyradTime Interval IIHEXTEHEXTE HEXTEHEXTE PCAPCA PCAPCAPCA PCA cutoffpl + bbodyradTime Interval IVTime Interval Vbknpow Time Interval VIbknpow+bbodyrad Time Interval VIcomptt+bbodyradcutoffpl + bbodyradTime Interval III
Energy (keV) (cid:358) Fig. 4.
The spectra and the best fit models for each of the five time intervals of the outburst.
Shu Zhang et al.: Outbursts from...
Soft color H a r d c o l o r Graph
Bursts
IIIIIIV V VI
Fig. 5.
The CCD of IGR J17473-2721. The hard color (9.7-16 keV/6.4-9.7 keV) and the soft color (4-6.4 keV/3-4 keV)are calculated by using the PCA count rate. The lines show the track in flux evolution of the outburst. 16 Type I burstsare shown at the region with large values of soft color.
Soft colour1.4 1.6 1.8 2 H a r d c o l ou r Graph II III
IV V VI
Edd /L bol L -1 H a r d c o l ou r Graph
II IIIIVVVI
Fig. 6.
The intrinsic CCD (left) and the color-luminosity (right) diagram of IGR J17473-2721. The hard color (9.7-16 keV/6.4-9.7 keV) and the soft color (4-6.4 keV/3-4 keV) are calculated by using the energy spectrum fit to thePCA/HEXTE data. The luminosity is estimated at 3-30 keV at a distance of 5.4 kpc. The five time intervals are markedwith the numbers II-VI. hu Zhang et al.: Outbursts from... 9
Table 1.
The results from fitting PCA/HEXTE data with different models.
Model Parameters (units) Time IntervalII III IVbbodyrad + kT bb (keV) 1 . +0 . − . . +1 . − . . +0 . − . N bb . +41 . − . . +1 . − . . +1 . − . cutoffpl Γ 1 . +0 . − . . +0 . − . . +0 . − . E cut (keV) 38 . +2 . − . . +0 . − . . +0 . − . N cutoffpl . +0 . − . . +0 . − . . +0 . − . σ F e (keV) 0.6 +0 . − . ± ± N F e +7 − ± +2 − χ ν /dof 1.17/71 0.83/47 1.10/47Model Parameters (units) Time IntervalV VIbbodyrad + kT bb (keV) 0 . +0 . − . N bb . +52 . − . bkn Γ . +0 . − . . +0 . − . E bkn (keV) 12 . +1 . − . . +6 . − . Γ . +0 . − . . +0 . − . N bkn . +0 . − . . +0 . − . σ F e (keV) 0.8 ± ± N F e ± ± χ ν /dof 1.01/72 1.03/70bbodyrad + kT bb (keV) 0 . +0 . − . N bb . +94 . − . Comptt kT seed (keV) 1 . +0 . − . kT e (keV) 21 . +4 . − . τ T . +0 . − . N comp . +0 . − . σ F e (keV) 0.7 ± N F e ± χ ν /dof 1.07/70Note: All the fits have the iron line fixed at 6.4 keV and absorption at 3.8 × atoms/cm . By putting the source at a distanceof 5.4 kpc, the luminosities calculated at 3-30 keV are 19.7 × erg/s for time interval II, 24.9 × erg/s for time interval III,17.6 × erg/s for time interval IV, 5.7 × erg/s for time interval V, 5.6 × erg/s for time interval VI. N F e is the ironline normalization in units of 10 − ph cm − s − . The other normalizations N cutoffpl and N bkn are in units of cm − s − keV − . N bb is the black body normalization proportional to the surface area, defined as R km /D , where R km is the surface radius inkm and D10