Supergiant Fast X–ray Transients in outburst: new Swift observations of XTE J1739302, Igr J175442619, and IGR J084084503
We report on new X–ray outbursts observed with Swift from three Supergiant Fast X–ray Transients (SFXTs): XTE J1739302, IGR J175442619, and IGR J084084503. XTE J1739302 underwent a new outburst on 2008, August 13, IGR J175442619 on 2008, September 4, while IGR J084084503 on 2008, September 21. While the XTE J1739302 and IGR J084084503 bright emission triggered the Swift/Burst Alert Telescope, IGR J175442619 did not, thus we could perform a spectral investigation only of the spectrum below 10 keV. The broad band spectra from XTE J1739302 and IGR J084084503 were compatible with the X–ray spectral shape displayed during the previous flares. A variable absorbing column density during the flare was observed in XTE J1739302 for the first time. The broad-band spectrum of IGR J084084503 requires the presence of two distinct photon populations, a cold one (0.3 keV) most likely from a thermal halo around the neutron star and a hotter one (1.4–1.8 keV) from the accreting column. The outburst from XTE J1739302 could be monitored with a very good sampling, thus revealing a shape which can be explained with a second wind component in this SFXT, in analogy to what we have suggested in the periodic SFXT IGR J11215–5952. The outburst recurrence timescale in IGR J175442619 during our monitoring campaign with Swift suggests a long orbital period of 150 days (in an highly eccentric orbit), compatible with what previously observed with INTEGRAL.
keywords:X-rays: binaries: individual (XTE J1739302, IGR J175442619, IGR J084084503)
The discovery of a new class of Galactic bright X–ray transients, composed of a compact object and an OB supergiant companion, the so-called Supergiant Fast X–ray Transients (SFXTs), is one of the most intriguing results obtained by the INTEGRAL satellite (Sguera et al. 2005, Negueruela et al. 2006). Since its launch in October 2002, the Galactic plane monitoring performed with INTEGRAL/IBIS led to the discovery of several new sources (Bird et al., 2007), some of which were characterized by short flares reaching 10–10 erg s, and later optically associated with blue supergiant stars (e.g. Masetti et al. 2006). A couple of members of the class were discovered years before 2002, with other satellites, and later re-discovered with INTEGRAL and firmly classified as SFXTs: XTE J1739–302 (Smith et al., 1998), now considered the prototype of the class, and the X–ray pulsar AX J1841.0–0536 (Bamba et al., 2001). SFXTs display a broad band spectral shape similar to that of accreting X–ray pulsars (Walter et al., 2006), and a high dynamic range in X–rays (up to four or five orders of magnitude) down to a quiescent luminosity at 10 erg s (in’t Zand, 2005). Although X–ray pulsations have not been discovered, to date, in all the members of the class, the spectral similarity with the accreting X–ray pulsars suggests that all, or at least most of them, host neutron stars. Their peculiar transient behaviour is still waiting for a convincing theoretical explanation, although all the physical mechanisms proposed to date are mainly related to the structure of the supergiant wind and/or to the properties of the accreting neutron star (see Sidoli 2009 and references therein for a recent review).
The first systematic monitoring of the X–ray activity of SFXTs has been performed with Swift, during a campaign (still in progress) which started in October 2007, with the main aim of studying the long-term properties of a sample of SFXTs: XTE J1739–302/IGR J17391–3021, IGR J17544–2619, IGR J16479–4514 and AX J1841.0–0536/IGR J18410–0535 (Sidoli et al. 2008b, hereafter Paper I).
One of the important results of these Swift observations is the discovery that SFXTs do not spend most of the time in quiescence, as previously thought, but in an intermediate level of emission at around 10–10 erg s, displaying a hard X–ray spectrum and a frequent low intensity flaring activity, with a dynamic range of more than one order of magnitude (Paper I).
During the Swift monitoring, about 1–2 bright outbursts per year per source have been caught (IGR J164794514, Romano et al. 2008c, hereafter Paper II; IGR J175442619 and XTE J1739302, Sidoli et al. 2009, hereafter Paper III), leading to the study of the first broad band spectrum from these sources, observed simultaneously from 0.3 to 60 keV. The results of a Swift monitoring of a multi-flaring X–ray activity in July 2008 from a fifth SFXT, IGR J084084503, has been reported by Romano et al. (2009a).
In this paper we report on the latest three outbursts caught by Swift from three SFXTs, already announced to the scientific community: XTE J1739–302 (2008, August 13; Romano et al. 2008b), IGR J17544–2619 (2008, September 4; Romano et al. 2008a). IGR J084084503 (2008, September 21; Mangano et al. 2008).
|Source||Sequence||Instrument||Start time (UT)||End time (UT)||Exposure|
|/Mode||(yyyy-mm-dd hh:mm:ss)||(yyyy-mm-dd hh:mm:ss)||(s)|
|XTE J1739302||00319963000||BAT/event||2008-08-13 23:47:14||2008-08-14 00:05:22||1088|
|00319963000||XRT/WT||2008-08-13 23:55:55||2008-08-14 00:29:09||1688|
|00319963000||XRT/PC||2008-08-14 00:03:19||2008-08-14 00:04:34||75|
|00319964000||BAT/event||2008-08-14 00:08:56||2008-08-14 00:28:58||728|
|00030987070||XRT/WT||2008-08-14 01:23:10||2008-08-14 13:03:50||1207|
|00030987070||XRT/PC||2008-08-14 04:36:11||2008-08-14 13:17:16||10714|
|00030987071||XRT/PC||2008-08-15 00:00:34||2008-08-15 00:16:57||983|
|00030987072||XRT/PC||2008-08-16 03:21:09||2008-08-16 05:08:57||1379|
|00030987073||BAT/PC||2008-08-17 01:44:09||2008-08-17 02:01:57||1068|
|00030987074||XRT/PC||2008-08-18 13:04:08||2008-08-18 13:21:56||1068|
|00030987075||XRT/PC||2008-08-19 13:09:42||2008-08-19 13:27:58||1095|
|00030987076||XRT/PC||2008-08-20 05:14:06||2008-08-20 05:31:56||1071|
|00030987077||XRT/PC||2008-08-21 00:30:30||2008-08-21 15:08:55||1173|
|00030987078||XRT/PC||2008-08-22 00:36:19||2008-08-22 02:21:55||1093|
|00030987079||XRT/PC||2008-08-23 03:55:11||2008-08-23 05:41:56||1229|
|00030987080||XRT/PC||2008-08-24 21:44:19||2008-08-24 23:31:57||1332|
|00030987081||XRT/PC||2008-08-25 21:50:21||2008-08-25 23:35:56||1274|
|IGR J084084503||00325461000||BAT/event||2008-09-21 07:54:11||2008-09-21 09:16:38||903|
|00325461000||XRT/WT||2008-09-21 08:03:54||2008-09-21 14:05:06||1159|
|00325461000||XRT/PC||2008-09-21 09:16:31||2008-09-21 14:24:59||4116|
|00030707013||XRT/PC||2008-09-23 15:53:39||2008-09-23 17:36:57||1109|
|00030707014||XRT/PC||2008-09-25 18:15:07||2008-09-25 19:57:54||1094|
|00030707015||XRT/PC||2008-09-26 18:10:01||2008-09-26 18:26:37||995|
|00030707016||XRT/PC||2008-09-27 16:39:46||2008-09-27 18:19:57||216|
|00030707017||XRT/PC||2008-09-28 16:56:30||2008-09-28 17:01:26||295|
|00030707018||XRT/PC||2008-09-29 10:36:24||2008-09-29 17:02:57||521|
|IGR J175442619||00035056061||XRT/WT||2008-09-04 00:12:40||2008-09-04 00:25:03||217|
|00035056061||XRT/PC||2008-09-04 00:12:45||2008-09-04 00:26:55||632|
1.1 Previous observations of the three SFXTs
XTE J1739302 was discovered by in August 1997 (Smith et al., 1998), reaching a peak flux of 3.610 erg cm s (2–25 keV). The optical counterpart is an O8 Iab(f) star (Negueruela et al., 2006) located at 2.7 kpc (Rahoui et al., 2008). ASCA observations (Sakano et al., 2002) allowed a constraint on the quiescent emission at a level of 1.110 erg cm s. The source showed bright outbursts, reaching 300 mCrab, observed with IBIS/ISGRI in 2003 March, and 2004 March (Sguera et al., 2006). Other flares observed with INTEGRAL have been reported by Walter & Zurita Heras (2007) and Blay et al. (2008). More recently, it triggered the Swift Burst Alert Telescope (BAT) on 2008 April 8, when a bright flare was caught, reaching an X–ray luminosity of 310 erg s (0.5–100 keV; Paper III). Since October 2007 the source has been monitored with Swift/XRT 2–3 times a week, showing a high variability in its flux even outside outbursts (Paper I).
IGR J175442619 was discovered with INTEGRAL on 2003 September 17 during a short flare reaching 160 mCrab (18–25 keV; Sunyaev et al. 2003). A observation performed in 2004 caught both the quiescence level and the onset of an outburst (in’t Zand, 2005), translating into a dynamic range as large as 10. The optical counterpart is an O9Ib star (Pellizza et al., 2006) located at 3.6 kpc (Rahoui et al., 2008). Several more bright flares have been observed with INTEGRAL in 2003, 2004 and 2005 (Grebenev et al. 2003, Grebenev et al. 2004, Sguera et al. 2006, Walter & Zurita Heras 2007, Kuulkers et al. 2007, and Ducci et al. 2008), Two outbursts were detected with the Swift satellite, on 2007 November 8 (Krimm et al., 2007) and on 2008 March 31 (Sidoli et al., 2008a), 144 days apart. Fainter activity at a level of 30–40 mCrab (20–60 keV) from IGR J175442619 was also observed on 2007 September 21, with IBIS/ISGRI on-board INTEGRAL (Kuulkers et al., 2007). IGR J175442619 is one of the four SFXTs we have been monitoring with Swift/XRT since October 2007 (Paper I).
IGR J084084503 was discovered on 2006 May 15, when a bright flare reaching a peak flux of 250 mCrab in the 20–40 keV energy band was caught by INTEGRAL (Götz et al., 2006). Analysis of archival INTEGRAL observations of the source field showed that IGR J084084503 was previously active on 2003 July 1 (Mereghetti et al., 2006). A refined position with Swift/XRT (Kennea & Campana, 2006) allowed to associate the source with a O8.5Ib(f) supergiant star, HD 74194, at a distance of about 3 kpc (Masetti et al., 2006). Three additional flares observed with INTEGRAL and Swift were studied by Götz et al. (2007). A new outburst from IGR J084084503 was caught on 2008 July 5 by Swift/BAT and then followed up at softer energies with Swift/XRT (Romano et al., 2009a). In that occasion the source displayed a multiple flaring activity (the XRT lightcurve showed three bright flares in excess of 10 s). The properties of the flares and of the times of the outbursts suggested an orbital period of 35 days (Romano et al., 2009a).
2 Observations and Data Reduction
As a response to a first BAT trigger from XTE J1739302 on 2008-08-13 at 23:49:17 UT (image trigger 319963), Swift executed an immediate slew and was on target in 390 s; a second trigger (319964) occurred while XTE J1739302 was in the XRT field of view on 2008-08-14 at 00:12:53 UT. The bright flare of IGR J175442619 was instead discovered as part of the yearly monitoring with Swift/XRT, starting on 2008-09-04 at about 00:19:00 UT. The Swift/BAT did not trigger on it. IGR J084084503 triggered the BAT on 2008-09-21 at 07:55:08 UT (image trigger 325461). Swift slewed immediately and the NFI were on target in 147 s.
Table 1 reports the log of the Swift observations of the outbursts used for this work. The XRT data were processed with standard procedures (xrtpipeline v0.12.1), filtering and screening criteria by using FTOOLS in the Heasoft package (v.6.6.1). We considered both WT and PC data, and selected event grades 0–2 and 0–12, respectively (Burrows et al. 2005). When appropriate we corrected for pile-up by determining the size of the PSF core affected by comparing the observed and nominal PSF (Vaughan et al., 2006), and excluding from the analysis all the events that fell within that region. Background events were extracted in source-free annular regions, centred on the source. Ancillary response files were generated with xrtmkarf, and they account for different extraction regions, vignetting, and PSF corrections. We used the latest spectral redistribution matrices (v011) in CALDB.
The BAT data were collected in event mode for several hundred seconds after the triggers of XTE J1739302 and IGR J084084503, as detailed below, while IGR J175442619 was not detected. The BAT data were analysed using the standard BAT analysis software distributed within FTOOLS. BAT mask-weighted spectra were extracted over the time intervals simultaneous with XRT data when possible and response matrices were generated with batdrmgen. For our spectral fitting (XSPEC v11.3.2) we applied an energy-dependent systematic error.
All quoted uncertainties are given at 90% confidence level for one interesting parameter unless otherwise stated. The spectral indices are parameterized as , where (erg cm s Hz) is the flux density as a function of frequency ; we adopt as the photon index, (ph cm s keV). Times in the light curves and the text are referred to their respective BAT triggers with the exception of IGR J175442619 which did not trigger the BAT, thus the start time was set at the beginning of the observation.
3 Analysis and Results
3.1 Xte J1739302
The complete light curve of the bright flaring of XTE J1739302 as observed with Swift/XRT on 2008 August 13 is reported in Fig. 9c, while the first part (2000 s) of the observation (WT data, observation 00319963000), is expanded in Fig. 1, where a soft (below 2 keV) and a hard (above 2 keV) light curves are reported, together with their hardness ratio. Since the source hardness appears to be variable, we performed a time resolved spectroscopy extracting eight XRT/WT spectra as shown in Fig. 1. These spectra could be adequately fitted both with an absorbed power law model and with an absorbed single blackbody (see Table 2 for the results, spectra from WT 1 to WT 8). There is a clear time variability of the absorbing column density (by a factor of 3), whereas the spectral shape (photon index, , or the blackbody temperature, kT) remain constant, within the uncertainties (see Fig. 2). In particular, spectra WT 3 and WT 5 are the hardest and the softest one, respectively, thus demonstrating that the hardness ratio variability in Fig. 1 is due to a variable absorption into the line of sight. As final tests, we fixed the photon index =1.15 (average value) and then re-fitted the eight spectra. This still resulted in a variable absorbing column density. We then fixed the absorbing column density to an mean value of 510 cm and refitted the spectra. Those spectra where the absorption were previously found to be very different from this mean value, resulted in unacceptable fits.
Figure 3 shows the comparison between the time resolved spectroscopy of the XRT/WT data (Table 2) of the August 2008 outburst and two more spectral analyses: the out-of-outburst emission (Paper I) and the spectroscopy of a previous flare from XTE J1739302 (Paper III). There is no evidence for a spectral change with the source flux, nor for a correlation of the absorbing column density with the source flux. The absorption is higher during the rising phase of the bright flare.
|(s since trigger)||( cm)||(keV)||(km)|
Unabsorbed 1–10 keV flux in units of erg cm s.
Assuming a distance of 2.7 kpc.
A high energy spectrum (BAT) was also available, but only simultaneously to the XRT/WT spectrum n. 1. A joint fit of XRT/WT and BAT spectra was performed including constant factors to allow for normalization uncertainties between the two instruments (always constrained to be within their usual ranges). A single power law is unable to describe the broad band spectrum. We then tried models usually adopted to describe the X–ray emission from accreting pulsars, resulting in the spectral parameters listed in Table 3: a cutoff power law (E), and two kind of Comptonization models. The best deconvolution of the broad band spectrum has been obtained with these latter models: a Comptonization of seed photons (with a temperature kT) in a hot plasma (with electron temperature kT) as described by compTT in xspec (Titarchuk, 1994), or by bmc (Titarchuk et al., 1996).
The bmc model is the sum of a blackbody (BB) plus its Comptonization, the latter obtained as a consistent convolution of the blackbody itself with the Green’s function of the Compton corona. The bmc model is not limited to the thermal Comptonization case (as e.g. compTT) and accounts also for dynamical (bulk) Comptonization due to the converging flow. Similarly to the ordinary bbody xspec model, the normalization of bmc is the ratio of the source luminosity to the square of the distance (in units of 10 kpc). The free parameters of the bmc model (apart from the normalization) are the black-body (BB) colour temperature, , the spectral index and the logarithm of the illuminating factor A, . The parameter indicates the overall Comptonization efficiency related to an observable quantity in the photon spectrum of the data. The lower , the higher the efficiency i.e. the higher the energy transfer from the hot electrons to the soft seed photons. The parameter is an indication of the fraction of the up-scattered BB photons with respect to the BB seed photons directly visibile. In the extreme cases, the seed photons can be completely embedded in the Comptonizing cloud (none directly visible, , e.g. ) or there is no coverage by the Compton cloud (A 1, e.g. ) and we observe directly the seed photon spectrum (equivalent to a simple BB, with no Comptonization).
The XTE J1739302 broad band spectrum fitted with the compTT model is shown in Fig. 4. The estimated X–ray luminosity during the flare is 3.810 erg s (0.1–100 keV), at a source distance of 2.7 kpc.
As can be seen in Table 3, the parameters describing the properties of the Comptonizing corona (be they the temperature and optical depth in compTT or the illumination parameter logA in bmc) are not constrained. This is expected given the poor statistics of the high energy part of the spectrum. Nevertheless, the models applied do give a first order description of the physical processes involved in this system (see Discussion).
Absorbing column density is in units of cm.
High energy cutoff (E), electron temperature (kT), seed photons temperature (kT) and the blackbody color temperature kT are all in units of keV.
Unabsorbed 0.1–100 keV flux is in units of erg cm s.
kT is the blackbody color temperature of the seed photons, is the spectral index and Log(A) is the illumination parameter (see Sect. 3.1 for details).
Assuming a spherical geometry.
3.2 Igr J175442619
A new bright flare from IGR J175442619 was caught on 2008 September 4 with Swift, and it was preceded by intense activity in the previous few days observed during the INTEGRAL Galactic bulge monitoring program (Romano et al., 2008a), reaching about 50 mCrab (18–40 keV) on 2008 August 30. The flare was caught by Swift/XRT, only thanks to the on-going monitoring campaign just targeted on the source, while Swift/BAT did not trigger on it. The XRT light curve (Fig. 5) shows a peak exceeding 20 s, brighter than the previous one observed with Swift on March 31, at 20:53:27 UT (Paper III).
The total PC spectrum resulted in an exposure time of 632 s and a source net count rate of 0.67 s, while the WT data, extracted with a net exposure of 217 s, resulted in an average count rate of 8.2 s. Fitting the two spectra separately with simple models (an absorbed power law or a blackbody) resulted in the parameters listed in Table 4. A black body is a better fit to the WT data than a single power law, which produces systematic positive residuals around 1 keV. The resulting black body radius at an assumed source distance of 3.6 kpc is R= km. More complex models are not required by the data. We also fit together PC and WT spectra, adopting free normalization constant factors between the two spectra. The best fit obtained with a blackbody model of the joint PC plus WT data is reported in Fig. 6.
|joint fit WT + PC||Pow|
Unabsorbed 1–10 keV flux in units of erg cm s.
Assuming a distance of 3.6 kpc.
3.3 Igr J084084503
The IGR J084084503 Swift/XRT light curve during the new outburst detected on 2008-09-21 is reported in Fig. 7 in two energy ranges, together with their hardness ratio. Since the hardness ratio was quite constant along the XRT/WT observation, we extracted a total spectrum. It resulted in a net exposure time of 1159 s with an average count rate of 28.3 s. The fit to the 0.7–10 keV WT spectrum with an absorbed power law model is unacceptable (=1.309 for 630 dof). A significantly better fit can be obtained either with a cutoff power law (=1.187 for 629 dof) or with a power law model together with a blackbody (=1.116 for 628 dof). We note that an absorbed blackbody model is the worst fit to the WT data, resulting in a reduced 3.6. The peak flux of 2.5 erg cm s translates into an X–ray luminosity of 2.5 erg s (at 3 kpc).
The spectral parameters resulting from these fits are reported in Table 5. A second total spectrum from the fainter emission observed in PC mode has been also investigated, yielding a spectrum with a net exposure of 4100 s, and a fainter rate of 0.26 s. A fit with a single absorbed power law results into a softer spectrum than the brighter emission observed in WT mode (see Table 5 for the PC spectral results).
|(s since trigger)||( cm)||(keV)||(km)|
Unabsorbed 1–10 keV flux in units of erg cm s.
Assuming a distance of 3 kpc.
A BAT spectrum could also be extracted at the beginning of the observation, with a net exposure time of 200 s and a source count rate of 35.8 s (14–60 keV). Fitting the BAT and XRT simultaneus spectra with a single power law resulted in an unacceptable reduced of 2.38 (293 dof). A very good fit can be obtained with a cutoff power law (cutoffpl in xspec; =1.029, for 292 dof), obtaining a hard spectrum with a photon index of 0.5, a cutoff at 13 keV, and a luminosity of 10 erg s at 3 kpc. We also tried other deconvolutions of the broad band spectrum, i.e. Comptonization emission models (thermal and with bulk motion) or a power law plus a blackbody model, but they always yielded unacceptable fits with structured residuals at high energies. We next adopted more complex models for the continuum, as a bmc model modified with a high energy cutoff. This resulted in a better fit than the single absorbed bmc model, but structured residuals still appear below 10 keV. The best fit is obtained adding a blackbody model to a bmc modified with a high energy cutoff (highecut in xspec). A summary of all the spectral parameters reported in Table 6, while the best deconvolution of the broad band IGR J084084503 emission is shown in Fig. 8.
Unlike XTE J1739302, a cut-off is clearly needed in the spectrum of IGR J084084503. Indeed bmc alone (that is a non-attenuated power-law at high energies) does not fit the data well and the multiplicative factor highecut is needed. We note that the fit clearly points to two distinct photon populations (0.3 keV and 1.5-2 keV) but the current statistics does not allow us to constrain which of the two is seen directly as a blackbody and which provides part of the seed photons for Comptonization (hence we include twice the same model in Table 8, one per configuration. See the Discussion section for possibile interpretations).
Absorbing column density is in units of cm.
High energy cutoff (E), e-folding energy (E) and blackbody temperatures are in units of keV.
Unabsorbed 0.1–100 keV flux is in units of erg cm s.
kT is the blackbody color temperature of the seed photons in the bmc model, is the spectral index and Log(A) is the illumination parameter.
The radius of the bbody model is in units of km at an assumed source distance of 3 kpc.
3.4 Timing analysis
We performed a timing analysis on the three sources to investigate for the presence of X-ray pulsations. A test (Buccheri et al., 1983) on the fundamental harmonics was applied on the events collected in each WT mode sequence (see Table 1) searching in the frequency range between 0.005 and 100 Hz with a frequency resolution of 1/T Hz, where T is the duration of the WT segment. The resulting power spectrum does not reveal any significant deviations from a statistically flat distribution. Data collected in PC mode were not analyzed because of their much lower statistics content.
4 Discussion and Conclusions
We report on three new outbursts from three different Supergiant Fast X–ray Transients, XTE J1739302, IGR J175442619, and IGR J084084503, observed with Swift.
All these three sources were observed in outburst with Swift in the past, thus allowing a proper comparison between the spectral properties of the different flares.
For IGR J175442619 only the spectrum below 10 keV is available. Compared with the emission previously observed (Paper III) it displays a more absorbed and softer emission (single power law model). A similar behaviour from IGR J175442619 has been recently reported by Rampy et al. (2009) during a Suzaku observation catching the source during a long outburst (at least three days of accreting phase) in March 2008 (the same reported in Paper III). The XRT/WT spectrum is compatible with spectral parameters reported for the segment n.6 of the XIS observation (Rampy et al., 2009). A good fit to the IGR J175442619 XRT spectrum can also be obtained with a black body model, resulting in a temperature of 1–2 keV, and in a black body radius (at 3.6 kpc) of about 1–1.5 km, compatible with an origin in the neutron star polar cap.
XTE J1739302 displays for the first time a variable absorption column density within a flare. This behaviour was also observed in IGR J084084503 during the multiple flaring activity caught by Swift in July 2008 (Romano et al., 2009a). A higher absorption is observed during the rise to the flare peak, so it is possibly due to the accumulation of matter onto the compact object, instead of obscuring material into the line of sight located far away from the central X–ray source. The average column density is intermediate between the out-of-outburst emission (Paper I) and that displayed during the previous bright flare reported in Paper III. No correlation of the spectral parameters with the source flux can be found. A variability on short timescales in the obscuring column density of a similar amount was observed also during an outburst from IGR J175442619 with Suzaku (Rampy et al. 2009). This sudden absorption episode was interpreted as due to the transit of a foreground dense cloud of matter passing in front of the compact object, and as the first direct evidence at X-rays for dense clumps of matter in the supergiant wind. The simultaneous spectroscopy of the XRT and BAT data of XTE J1739302 can be adequately performed using a simple bmc model (Table 3): a significant fraction () of the initial black body (BB) seed photons ( keV) is efficiently up-scattered (1.3) by the Comptionizing plasma. The radius associated with the BB component of the bmc model is about 1.6 km (at 2.7 kpc), thus consistent with a polar cap origin. The model compTT allow us to quantify the physical conditions of the Comptonizing plasma, since it returns the plasma temperature and optical depth , instead of the Comptonization efficiency . As shown in Table 3, the temperature could not be constrained and this is consistent with the fact that the data could be fit well by the bmc model that indeed has no cut-off in its spectral shape. These two results point to the fact that the current statistics and data coverage do not require a cutoff, although they cannot exclude it. The reason why the cutoffpl model fits the data much better than a simple power law resides in the fact that the residuals using a simple power law show an excess around 1–2 keV, whereas the curved shape of the cutoffpl linked to the interplay with the absorbing column density can describe the data in a satisfactory way. We note that this excess is naturally taken into account in the physical models, compTT and bmc, where a 1–2 keV BB seed photon population is obtained. Unfortunately the current data-set does not allow us to investigate the evolution of the high energy part of the spectrum, so little can be currently said about the possible evolution of the Comptonizing medium. For the remaining part of the outburst, only the softer part below 10 keV is available. Nevertheless, it can be seen from Table 2 that a BB is more suitable to fit the Swift/XRT data rather than a power law, consistent with what is obtained in the overall XRT+BAT spectrum.
A comparison of the IGR J084084503 broad band spectrum extracted from the bright flare observed in September 2008 with the ones observed with Swift in July 2008 and in October 2006 (Romano et al., 2009a) reveals that the new spectrum is more similar to that observed in 2006 (very low absorption, a hard photon index, and a similar high energy cutoff at around 10–15 keV). Unlike XTE J1739302, the spectrum of IGR J084084503 could not be fit with a single Comptonization model, two additive components were needed: a blackbody plus a bmc model (the latter with high energy cutoff). This could be due to the very low absorption at low energy with respect to the other SFXTs studied here that required an additional component to take into account for the softer part of the spectrum. The presence of the cutoff implies that the overall spectrum is the result of thermal Comptonization (the bmc model alone has a non-attenuated power law shape). The low value of the parameter obtained (0.4, ) is typical for saturated Comptonization.
The seed photon temperature for the thermal Comptonization bmc component and the BB temperature could not be linked to the same value in the fitting process and indeed resulted in two clearly different photon populations, a cold one at about 0.3 keV and a hotter one at 1.4–1.8 keV. The current data did not allow us to establish in a solid way, which one of these two populations is seen directly as a BB and which one ends up being seed photons for the thermal Comptonization. As can be seen in Table 6 both scenarii are statistically acceptable. In one case we obtain a cold (0.3 keV) BB of about R=12 km directly visible (few percent of the total flux), together with a hotter photon population (1.4 keV) thermally Comptonized (the dominant component), part of which is directly visible (logA=0.8). This could depict a thermal halo around the neutron star [0.3 keV, as in Ferrigno et al. (2009)] together with BB seed photons from the accreting material, part of which is directly visible (e.g at the column boundaries) and part is thermally Comptonized (from the accreting matter). This scenario is consistent with what observed for XTE J1739302 (Table 3) with the thermal cold halo buried in the high column absorption (an order of magnitude higher than for IGR J084084503).
In the second case we obtain a hotter BB (1.9 keV, possibly from the base of the accreting column, R1 km) directly visible, accounting for about 30% of the total flux, together with a colder plasma (0.3 keV) embedded in a thermally Comptonizing medium (logA3) such as an atmosphere confined by multi-polar or crustal components of the magnetic field [e.g. Ferrigno et al. (2009) and references therein].
With the information at hand, we cannot exclude either of these scenarii. The spectra of HMXBs have been generally described by phenomenological models and this work is one of the few cases where two distinct spectral components linked to two different physical conditions have been observed [see also Ferrigno et al. (2009)].
4.2 Search for periodicities
IGR J175442619 was previously observed in outburst with Swift in two occasions: on 2007 November 8 and on 2008 March 31 (Paper III). This implies that the three bright flares are spaced by 144 and 157 days, respectively. We note however that in IGR J175442619 the flare occurred on 2008 September 4 was preceded by intense activity in the previous few days during the observations part of the INTEGRAL Galactic bulge monitoring program performed on 2008, August 30, reaching a flux of about 50 mCrab (18–40 keV, Romano et al. 2008a). This seems to indicate an outburst phase which began a few days before the BAT trigger, suggesting an outburst duration of several days (an outburst lasting at least 3 days has also been caught with Suzaku (Rampy et al., 2009)): this could imply a periodic occurrence of the bright flaring activity, every 150 days. If the X–ray bright flares are triggered periodically during the periastron passage in an eccentric binary, the orbital period is probably about 150 days in IGR J175442619. This is consistent with previous findings with INTEGRAL, where a possible outburst recurrence timescale of 165 days has been suggested (Walter et al., 2006).
From the times of previous IGR J084084503 flares, we suggested (Romano et al., 2009a) that a double-periodicity outburst recurrence of 11 days and 24 days was present, thus consistent with the picture where the outbursts are triggered when the compact object, along its orbit, crosses twice an inclined second component of an outflowing dense wind, confined along a preferential plane (e.g. the supergiant equator), inclined with respect to the orbital plane. On the other hand, the last flare from IGR J084084503 did not occur at the right times predicted by these double periodicities (the nearest outburst was predicted to occur on 2008, September 13, instead of 2008 September 21). This could indicate that either these derived periodicities are actually wrong (and the flaring activity is not periodic but sporadic), or that another mechanism is at work when producing this latter kind of outburst: a possible explanation is that, while the previous outbursts were produced when the neutron star crossed twice the denser wind component, this latter outburst was triggered when the neutron star approached the periastron passage, accreting matter from the polar wind, in an eccentric orbit.
The three sources analysed here do not show any evidence for X–ray pulsations (see Sect. 3.4).
4.3 SFXTs as a class
During the last outburst, a very good sampling of the XTE J1739302 light curve was possible (the best light curve during an outburst from this source, to date; see Fig. 9c). We can compare the Swift/XRT light curve of XTE J1739–302 with the X-ray luminosity predicted by a Bondi-Hoyle accretion onto the neutron star from a spherically symmetric homogeneus wind for different values of the orbital period and eccentricity. We assume for the supergiant a stellar mass of 33 M, a radius of 23 R (Vacca et al., 1996), a beta-law for the supergiant wind with , a wind terminal velocity of 1900 km s, a wind mass loss rate of 10 M yr and a temperature of the stellar wind of 10 K. We find that for any choice of the orbital period and eccentricity the decline of the light curve observed with Swift is too rapid compared with the calculated light curve, even adopting high values for the orbital period and eccentricity. Since a spherical distribution of wind matter is not able to explain the observed shape of the X–ray light curve, a possible explanation is the presence of a second outflowing wind component, denser than the supergiant polar wind, which is crossed by the neutron star along its orbit, in analogy to what we proposed for IGR J11215–5952 (Sidoli et al., 2007). Alternatively, if we consider the clumpy wind scenario, from the duration of the bright part of the flare in XTE J1739302 (0.6 days) and its luminosity (10 erg s), we can derive an accreted mass of 410 g and a size of 10 cm (Walter & Zurita Heras, 2007), which corresponds to more than 6 supergiant radii, thus making very unlikely that it is a very large single clump ejected by the supergiant (Walter & Zurita Heras, 2007). Instead, it could be alternatively explained with a huge gas stream composed by several smaller clumps.
In Fig. 9 we compare some of the outbursts from SFXTs as observed during our monitoring campaign. In particular the three outbursts discussed here are shown in panels c, e and g from XTE J1739302, IGR J175442619 and IGR J084084503, respectively. One could be tempted to conclude, from this comparison, that different types of outburst are present, even in the same source. It is actually not possible to compare, for example, the two outbursts from IGR J084084503 (last panels in Fig. 9): the several upper limits to the flux in the declining part of the outbursts are compatible with the source detections during the previous IGR J084084503 outburst (Fig. 9f), thus it is not possible to conclude that the new outburst from IGR J084084503 (Fig. 9g) is shorter than the previous one (Fig. 9f). On the other hand, the light curve from XTE J1739302 (Fig. 9c) allows us to perform a proper comparison with the outburst from the periodic SFXTs IGR J11215–5952 (Fig. 9a; Sidoli et al. 2006, Romano et al. 2007): these two sources appear to undergo similar outbursts with a similar duration. Interestingly, Rampy et al. (2009) report on a long outburst activity from IGR J175442619 during the March 2008 outburst, with a rise time much longer than what observed with Chandra during an X-ray flare in 2004 (in’t Zand, 2005) in the same source. These authors suggest that different kinds of outbursts can occur in the same SFXT.
From the duration of single well sampled short flares in IGR J084084503 we derived an orbital period of about 35 days (Romano et al., 2009a) adopting an expansion law for the clump sizes as the clump is accelerated far away from the supergiant donor, in the framework of an inhomogeneous wind (Romano et al., 2009a). Adopting this same expansion law, and from the observed durations of the short flares which compose the outburst light curves in XTE J1739302 and IGR J175442619, we can derive the distance of the accreted clump from the supergiant star, as we did in IGR J084084503 (Romano et al., 2009a). More details will be reported in Ducci et al. (2009, in preparation). Fitting with a Gaussian a few well sampled flares in XTE J1739302 we obtain a FWHM of 260 s and 390 s, while in IGR J175442619 the observed flare has a duration of 220 s (FWHM). From the clump expansion law reported in Romano et al. (2009a) we derive a distance of the compact object from the supergiant donor in these two sources as follows: in the range from 6.810 cm to 1.2510 cm for XTE J1739302, while near to 6.810 cm for IGR J175442619 (assuming a supergiant mass of 33 M and a stellar radius of 23 R). Assuming a circular orbit these distances translate into orbital periods ranging from 20 to 50 days in XTE J1739302 and around 20 days for IGR J175442619. This latter orbital period is not consistent with that of 150 days, suggested by the outburst recurrence in IGR J175442619. This discrepancy can be easily reconciled if the orbit in this SFXT is highly eccentric.
Jain et al. (2009) recently reported on the discovery of an orbital period of 3.3 days from the SFXT IGR J164794514. This orbital periodicity is even shorter than that displayed by several persistent HMXBs with supergiant companions. This poses serious problems to the different physical mechanisms proposed for SFXTs, implying an orbital separation of about 210 cm (assuming a supergiant mass of 30 M), which is about 1.2 stellar radii, thus well inside the region where the highest wind clumps number density is expected, and a persistent X–ray emission is predicted (Negueruela et al., 2008). The very different orbital periods discovered in SFXTs to date (ranging from 3.3 days to 165 days) possibly point to different kinds of SFXTs (different mechanisms at work in different sources, and possibly in the same source, as discussed above).
LS thanks INAF-IASF Palermo and PR thanks INAF-IASF Milano, where some of the work was carried out, for their kind hospitality. We thank Valentina La Parola and Cristiano Guidorzi for helpful discussions. We thank the Swift team for making these observations possible, the duty scientists, and science planners. This work was supported in Italy by ASI contracts I/023/05/0, I/088/06/0 and I/008/07/0, and partially by the grant from PRIN-INAF 2007, “Bulk motion Comptonization models in X-ray Binaries: from phenomenology to physics” (PI M. Cocchi). This work was supported at PSU by NASA contract NAS5-00136. HAK was supported by the Swift project. DNB and JAK acknowledge support from NASA contract NAS5-00136.
- Bamba et al. (2001) Bamba A., Yokogawa J., Ueno M., Koyama K., Yamauchi S., 2001, PASJ, 53, 1179
- Bird et al. (2007) Bird A. J., Malizia A., Bazzano A., Barlow E. J., Bassani L., Hill A. B., 2007, ApJS, 170, 175
- Blay et al. (2008) Blay P., Martínez-Núñez S., Negueruela I., Pottschmidt K., Smith D. M., Torrejón J. M., Reig P., Kretschmar P., Kreykenbohm I., 2008, A&A, 489, 669
- Buccheri et al. (1983) Buccheri R., et al., 1983, A&A, 128, 245
- Burrows et al. (2005) Burrows D. N., et al., 2005, Space Science Reviews, 120, 165
- Ducci et al. (2008) Ducci L., Sidoli L., Paizis A., Mereghetti S., 2008, ArXiv e-prints
- Ferrigno et al. (2009) Ferrigno C., Becker P. A., Segreto A., Mineo T., Santangelo A., 2009, ArXiv e-prints
- Götz et al. (2007) Götz D., Falanga M., Senziani F., De Luca A., Schanne S., von Kienlin A., 2007, ApJL, 655, L101
- Götz et al. (2006) Götz D., Schanne S., Rodriguez J., Leyder J.-C., von Kienlin A., Mowlavi N., Mereghetti S., 2006, Astron. Tel., 813
- Grebenev et al. (2003) Grebenev S. A., Lutovinov A. A., Sunyaev R. A., 2003, Astron. Tel., 192
- Grebenev et al. (2004) Grebenev S. A., Rodriguez J., Westergaard N. J., Sunyaev R. A., Oosterbroek T., 2004, Astron. Tel., 252
- in’t Zand (2005) in’t Zand J. J. M., 2005, A&A, 441, L1
- Jain et al. (2009) Jain C., Paul B., Dutta A., 2009, ArXiv e-prints 0903.5403
- Kennea & Campana (2006) Kennea J. A., Campana S., 2006, Astron. Tel., 818
- Krimm et al. (2007) Krimm H. A., et al., 2007, Astron. Tel., 1265
- Kuulkers et al. (2007) Kuulkers E., et al., 2007, Astron. Tel., 1266
- Mangano et al. (2008) Mangano V., et al., 2008, Astron. Tel., 1727
- Masetti et al. (2006) Masetti N., Bassani L., Bazzano A., Dean A. J., Stephen J. B., Walter R., 2006, Astron. Tel., 815
- Masetti et al. (2006) Masetti N., et al., 2006, A&A, 449, 1139
- Mereghetti et al. (2006) Mereghetti S., Sidoli L., Paizis A., Gotz D., 2006, Astron. Tel., 814
- Negueruela et al. (2006) Negueruela I., Smith D. M., Harrison T. E., Torrejón J. M., 2006, ApJ, 638, 982
- Negueruela et al. (2008) Negueruela I., Torrejón J. M., Reig P., Ribo M., Smith D. M., 2008, ArXiv e-prints 0801.3863
- Pellizza et al. (2006) Pellizza L. J., Chaty S., Negueruela I., 2006, A&A, 455, 653
- Rahoui et al. (2008) Rahoui F., Chaty S., Lagage P.-O., Pantin E., 2008, ArXiv e-prints 0802.1770
- Rampy et al. (2009) Rampy R. A., Smith D. M., Negueruela I., 2009, ArXiv e-prints 0904.1189
- Romano et al. (2008a) Romano P., et al., 2008a, Astron. Tel., 1697
- Romano et al. (2008b) Romano P., et al., 2008b, Astron. Tel., 1659
- Romano et al. (2008c) Romano P., et al., 2008c, ApJL, 680, L137 (Paper II)
- Romano et al. (2009a) Romano P., et al., 2009a, MNRAS, 392, 45
- Romano et al. (2009b) Romano P., et al, 2009b, Astron. Tel., 1920
- Romano et al. (2007) Romano P., Sidoli L., Mangano V., Mereghetti S., Cusumano G., 2007, A&A, 469, L5
- Sakano et al. (2002) Sakano M., Koyama K., Murakami H., Maeda Y., Yamauchi S., 2002, ApJS, 138, 19
- Sguera et al. (2005) Sguera V., et al., 2005, A&A, 444, 221
- Sguera et al. (2006) Sguera V., et al., 2006, ApJ, 646, 452
- Sidoli (2009) Sidoli L., 2009, Adv. Sp. Res. in press, arXiv:0809.3157v2
- Sidoli et al. (2006) Sidoli L., Paizis A., Mereghetti S., 2006, A&A, 450, L9
- Sidoli et al. (2007) Sidoli L., Romano P., Mereghetti S., Paizis A., Vercellone S., Mangano V., Götz D., 2007, A&A, 476, 1307
- Sidoli et al. (2009) Sidoli L., et al., 2009, ApJ, 690, 120 (Paper III)
- Sidoli et al. (2008a) Sidoli L., et al., 2008a, Astron. Tel., 1454
- Sidoli et al. (2008b) Sidoli L., et al., 2008b, ApJ, 687, 1230 (Paper I)
- Smith et al. (1998) Smith D. M., Main D., Marshall F., Swank J., Heindl W. A., Leventhal M., in ’t Zand J. J. M., Heise J., 1998, ApJL, 501, L181
- Sunyaev et al. (2003) Sunyaev R. A., Grebenev S. A., Lutovinov A. A., Rodriguez J., Mereghetti S., Gotz D., Courvoisier T., 2003, Astron. Tel., 190
- Titarchuk (1994) Titarchuk L., 1994, ApJ, 434, 570
- Titarchuk et al. (1996) Titarchuk L., Mastichiadis A., Kylafis N. D., 1996, A&AS, 120, C171
- Vacca et al. (1996) Vacca W. D., Garmany C. D., Shull J. M., 1996, ApJ, 460, 914
- Vaughan et al. (2006) Vaughan S., et al., 2006, ApJ, 638, 920
- Walter & Zurita Heras (2007) Walter R., Zurita Heras J., 2007, A&A, 476, 335
- Walter et al. (2006) Walter R., et al., 2006, A&A, 453, 133