EDP Sciences
Free access
Issue
A&A
Volume 509, January 2010
Article Number L3
Number of page(s) 4
Section Letters
DOI http://dx.doi.org/10.1051/0004-6361/200913517
Published online 12 January 2010
A&A 509, L3 (2010)

LETTER TO THE EDITOR

Swift monitoring of the new accreting millisecond X-ray pulsar IGR J17511-3057 in outburst

E. Bozzo1 - C. Ferrigno1,2 - M. Falanga3 - S. Campana4 - J. A. Kennea5 - A. Papitto6,7

1 - ISDC data centre for astrophysics, University of Geneva, Chemin d'Écogia 16, 1290 Versoix, Switzerland
2 - IAAT, Abt. Astronomie, Universität Tübingen Sand 1, 72076 Tübingen, Germany
3 - International Space Science Institute (ISSI) Hallerstrasse 6, 3012 Bern, Switzerland
4 - INAF - Osservatorio Astronomico di Brera, via Emilio Bianchi 46, 23807 Merate (LC), Italy
5 - Department of Astronomy and Astrophysics, 525 Davey Lab, Pennsylvania State University, University Park, PA 16802, USA
6 - Universitá degli Studi di Cagliari, Dipartimento di Fisica, SP Monserrato-Sestu, KM 0.7, 09042 Monserrato, Italy
7 - INAF - Osservatorio Astronomico di Cagliari, Poggio dei Pini, Strada 54, 09012 Capoterra (CA), Italy

Received 21 October 2009 / Accepted 28 November 2009

Abstract
Context. A new accreting millisecond X-ray pulsar, IGR J17511-3057, was discovered in outburst on 2009 September 12 during the INTEGRAL Galactic bulge monitoring programme.
Aims. To study the evolution of the source X-ray flux and spectral properties during the outburst, we requested a Swift monitoring of IGR J17511-3057.
Methods. In this paper we report on the results of the first two weeks of monitoring the source.
Results. The persistent emission of IGR J17511-3057 during the outburst is modelled well with an absorbed blackbody ( $kT \sim 0.9$ keV) and a power-law component ($\Gamma$$\sim 1$-2), similar to what has been observed from other previously known millisecond pulsars. Swift also detected three type-I Xray bursts from this source. By assuming that the peak luminosity of these bursts is equal to the Eddington value for a pure helium type-I X-ray burst, we derived an upper limit to the source distance of $\sim$10 kpc. The theoretically expected recurrence time of the bursts according to the helium burst hypothesis is 0.2-0.9 days, in agreement with the observations.

Key words: X-rays: binaries - pulsars: individual: IGR J17511-3057

1 Introduction

Accreting millisecond pulsars (AMSP) are neutron stars (NS) that accrete mass from a low-mass ($\ll$$M_{\odot}$) companion star and that show coherent pulsations at their millisecond spin period (Bhattacharya & van den Heuvel 1991). Since the discovery of the first AMSP in 1998, SAX J1808.4-3658 (Wijnands & van der Klis 1998), 11 other AMSPs were discovered (see e.g.; Altamirano et al. 2009; Casella et al. 2008; Altamirano et al. 2008; Wijnands 2006, for reviews). All the AMSP are hosted in low-mass X-ray binaries (LMXBs) with typical orbital periods of either about 40 min or a few hours, and exhibit a transient X-ray emission with bright outbursts (1036-1037 erg/s) occurring on a time scale from two to more than ten years, and a typical duration of a few weeks (see e.g., Wijnands 2006). However, longer activity episodes have also been recorded (XTE J1807-294, see e.g., Falanga et al. 2005a), and one source, HETE J1900.1-3455, has never returned in quiescence since its discovery in 2005 (Kaaret et al. 2006; Galloway 2007). So far, only AMSP with orbital periods of a few hours have exhibited type-I X-ray bursts (the only exception being IGR J00291+5934). This is believed to stem from the composition of the material that is accreted onto the NS that in turns depend on the nature of the donor star (see e.g., Galloway & Cumming 2006, and references therein). The broad band X-ray (0.5-200 keV) spectra of AMSPs are generally fit by a model consisting of one or two soft thermal components and a comptonized hard component. The two soft components are associated with the radiation from the accretion disk and from the heated hot spots on the NS surface. The hard emission is likely to be produced by thermal Comptonization of the soft photons emitted by the NS surface in the hot accretion column above the NS hot spots (Gierlinski et al. 2002; Papitto et al. 2009; Poutanen & Gierlinski 2003; Gierlinski & Poutanen 2005; Falanga et al. 2005a; Ibragimov & Poutanen 2003; Patruno et al. 2009; Falanga et al. 2007,2005b).

In this letter we report on the first $\sim$2 weeks of monitoring with Swift of the 12th newly discovered AMSP in outburst, IGR J17511-3057. We discuss the evolution of the source X-ray flux and spectrum from the onset of the outburst, when the source was discovered, up to the beginning of the outburst decay. We also detected and analysed three type-I X-ray bursts.

Table 1:   Swift observation log of IGR J17511-3057.

1.1 The source IGR J17511-3057

IGR J17511-3057 was discovered on 2009 September 12 during the INTEGRAL Galactic bulge monitoring program (Baldovin et al. 2009; Kuulkers et al. 2007). The source was detected in both the IBIS/ISGRI and JEM-X mosaics, and its 3-100 keV energy spectrum was modelled by using a power law with index $2.0\pm0.2$. The corresponding flux was $1.1\times 10^{-9}~{\rm erg~s^{-1}~cm^{-2}}$. A position within $2^\prime$ was determined, which allowed identification of IGR J17511-3057 as a new 245 Hz accreting pulsar by means of the RXTE PCA bulge scan data (Markwardt et al. 2009). Using this instrument, the source was initially detected on 2009 September 11, but not recognized as a new object due to the proximity of two previously known sources, XTE J1751-305 and GRS 1747-312, the first of which is a 435 Hz X-ray millisecond pulsar. The RXTE spectrum could be described by an absorbed power-law model with photon index 1.8 and a 2-10 keV flux of $4 \times 10^{-10}~{\rm erg~s^{-1}~cm^{-2}}$. A Swift ToO observation was carried out on 2009 September 13, and it allowed for a preliminary characterization of the soft (0.5-10 keV) X-ray spectrum and the discovery of the first thermonuclear type-I X-ray burst from the source (Bozzo et al. 2009). Burst oscillations at $\sim$245 Hz in other type-I X-ray bursts from this source were reported by Watts et al. (2009). Further follow-up observations of IGR J17511-3057 were carried out later with XMM-Newton (Papitto et al. 2009b) and RXTE (Riggio et al. 2009). The latter provided the most precise ephemeris of the source and yielded a pulse frequency of 244.83395157(7) Hz, an orbital period of 12487.5126(9) s, and an $a \sin(i)/c$ value of 275.194(3) lt-ms. The mass function of the system is thus 0.00107025(4) $M_{\odot}$, giving a minimum companion mass of 0.13 $M_{\odot}$ (assuming an NS mass of 1.4 $M_{\odot}$ and errors at 1$\sigma$ c.l. in the last digit). The first accurate source position was determined through a Chandra observation at $\alpha_{\rm J2000} =17^{\rm h}$51$^{\rm m}$08 $\hbox{$.\!\!^{\rm s}$ }$66 and $\delta_{\rm J2000}=-30{\hbox{$^\circ$ }}$57 $\hbox{$^\prime$ }$41 $\hbox{$.\!\!^{\prime\prime}$ }$0 (1 $\sigma$ error of $\sim$0.6 $\hbox{$.\!\!^{\prime\prime}$ }$, Nowak et al. 2009), and allowed for identifying its infrared counterpart (Torres et al. 2009). Radio observations at this position did not result in any detection (Miller-Jones et al. 2009).

\begin{figure}
\par\includegraphics[angle=-90,width=7.2cm,clip]{13517fig1.ps}
\end{figure} Figure 1:

Swift /XRT long-term light curve of the outburst of IGR J17511-3057 (time bin 1000 s, start time 2009 Sept. 13 at 20:00:41 UTC). The arrows in the plot mark the time of the type-I X-ray bursts observed with Swift. The two inserts show a zoom of the first two type-I X-ray bursts (time bin is 1 s). The time on the X-axes of these inserts is measured from the t0 of the bursts. These are 2009 Sept. 14 00:50:27 and 2009 Sept. 15 at 17:17:19 (UTC), respectively. The third type-I X-ray burst (C) is reported in Fig. 2.

Open with DEXTER

2 Data analysis and results

To monitor the X-ray flux of IGR J17511-3057 in outburst, we requested a 2 ks daily observation for the first week, and then two other observations of 5 ks were scheduled during the second week when we noticed that the source X-ray flux was already decreasing after the beginning of the outburst (see Fig. 1). A complete log of the observations is provided in Table 1. We processed all the Swift data by using standard procedures (Burrows et al. 2005) and the latest calibration files available (caldb v. 20090407). The Swift/XRT data were analysed both in window-timing (WT) and photon-counting (PC) modes (processed with the XRTPIPELINE v.0.12.3). We used Swift /BAT data accumulated only in EVENT mode, as the statistics of the data in SURVEY mode were too poor to provide any significant constraint on the source high-energy emission (15-150 keV).

\begin{figure}
\par\includegraphics[angle=-90,width=7cm,clip]{13517fig2.ps}
\end{figure} Figure 2:

The brightest type-I X-ray burst observed with Swift. The upper panel shows the BAT light curve (15-25 keV), whereas in the lower panel we reported the XRT light curve (0.5-10 keV). The time bin of the BAT and XRT/WT (XRT/PC) light curves is 1 s (5 s). The start time of the burst is 2009 Sept. 30 18:31:57 (UTC).

Open with DEXTER

Filtering and screening criteria were applied by using FTOOLS (Heasoft v.6.6.3). We extracted source and background light curves and spectra by selecting event grades of 0-2 and 0-12, respectively, for the WT and PC modes. Exposure maps were created through the XRTEXPOMAP task, and we used the latest spectral redistribution matrices in the HEASARC calibration database (v.011). Ancillary response files, accounting for different extraction regions, vignetting, and PSF corrections, were generated by using the XRTMKARF task. All PC data were affected by a strong pile-up, and corrected according to the technique developed by Vaughan (2006). We used the XRTLCCORR task to account for this correction in the background-subtracted light curves.

During the three type-I X-ray bursts we also checked a possible pile-up of the XRT/WT data, caused by the high source count rate. We extracted the source spectrum during the brightest part (6 s) of the three events by adopting a box-shaped extraction region in which we progressively excluded the first and then two, three, and four inner central pixels[*]. These spectra were then fit with an absorbed black body model. We did not notice any significant pile-up in the spectral properties of the source at the peaks of the first two bursts. Only in the third burst (burst ``C'', see Fig. 2), we noticed that the XRT/WT data of the first two seconds of the burst were affected by a relatively strong pile-up. However, the time interval over which these data were accumulated is too short to apply the correction method described above and obtain a usable spectrum. Therefore, we discarded these first two seconds of observation. The peak luminosity we derive below for the third burst might thus have been somewhat higher than the value we reported, yielding a conservative upper limit on the distance we provide in Sect. 3.

We performed a Fourier analysis of the persistent emission and of the type I X-ray burst emission of IGR J17511-3057 in order to search for the known 245 Hz periodicity of the source. We corrected the photons' arrival times for the orbital motion of the source according to the solution given by Riggio et al. (2009). During the type-I X-ray bursts, we could not detect any periodicity up to 283 Hz given a detection threshold of 5$\sigma$. This translates into an upper limit on any fractional amplitude of the pulsed signal of 0.23 (at 3$\sigma$ confidence level), where the binning effect of the temporal resolution of Swift /XRT in WT mode (1.8 ms) has been taken into account (van der Klis 1989). A similar analysis carried out on the persistent emission of the source only led to a marginal 5$\sigma$ detection of pulsations, corresponding to an amplitude $\gtrsim$0.07 (3 sigma c.l.). We note that the sensitivity to periodic signals is hampered here by the lower number of photons collected by Swift /XRT than by, e.g. RXTE/PCA, with which the pulsations and the burst oscillations were detected from this source (see Sect. 1). In Fig. 1 we show the source light curve during about two weeks of monitoring.

2.1 Persistent emission

We extracted the source spectrum in the 0.5-10 keV energy band for each observation in Table 1 excluding the time intervals in which a type I X-ray burst was detected. We excluded about $\sim$100 s before and after the bursts in the observations 00031492001 and 003. In those observations in which the XRT/WT data were accumulated during two or three different revolutions of the satellite, we extracted the X-ray spectrum separately to search for spectral variations. We indicated these different spectra in Table 1 by using the notation ``1'', ``2'', and ``3''. The XRT/WT spectra with the higher statistics (observation ID: 00031492001, 003, 004, 006, 007) could not be fit by only using a simple absorbed power-law ($\Gamma$$\sim$1-2) model (reduced $\chi^2$$\sim$1.5-1.9, d.o.f. in the range 112-447), and the addition of a second component was required by the data. According to previous studies of AMSPs in outburst, we tried a model comprising an absorbed power law (PL) plus a black body ( BBODYRAD in XSPEC, hereafter BB) or a disk black body ( DISKBB in XSPEC) component. We found that the latter choice would need a highly improbable physical explanation to account for the large, inner accretion radius (a few hundred km) implied by the DISKBB model. As the spectra collected at higher statistics for this and similar sources show two soft components arising from the disk and the NS surface (see Sect. 1), we interpret the soft component detected with Swift as originating in the NS hotspots. Our interpretation is strengthened as it provides an area for the BB-emitting region that is fully compatible with the NS surface. This is shown in Table 1 (errors at 90% c.l.). We checked that the fits to the Swift /XRT spectra cannot be improved significantly by introducing a Comptonization model instead of a simple power law above $\sim$2 keV. We tried a COMPTT model in XSPEC with different values of the seed photon temperature. The estimated column density is $\sim$0.5- $1 \times 10^{22}$ cm-2, compatible with the Galactic absorption in the direction of the source (Dickey & Lockman 1990). The 0.5-10 keV X-ray flux of the source decreased slowly from $5.0 \times 10^{-10}$ erg cm-2 s to $2.0 \times 10^{-10}$ erg cm-2 s during the first two weeks of observation.

Table 2:   The three type-I X-ray bursts parameters.

2.2 Type I X-ray bursts

We also detected in the Swift observations three type-I X-ray bursts. The light curves of these bursts are reported in Figs. 1 and 2. The start time of the third burst was determined by Swift /BAT (XRT started observing the source only about $\sim$5 s later). We performed a time-resolved spectral analysis of the three bursts by accumulating the XRT/WT spectra in intervals of different durations (from 1 to 10 s), depending on the source count rate. This time-resolved analysis did not reveal a clear signature of a photospheric radius expansion (PRE) in any of the three bursts (Lewin et al. 1993). The relevant parameters for each burst are reported in Table 2. Here the peak flux of each burst was determined from a BB fit to the spectrum of the initial 6 s of each burst (fixing the $N_{\rm H}$ at 0.6, see Table 1). The persistent spectrum determined from the closest XRT/WT observation to each of the burst was used as a background in the fit. We indicated with $\tau_{\rm lc}$ the decay time of the burst measured by fitting the observed light curve with an exponential function, and $\tau$ is the duration of the burst (see e.g. Lewin et al. 1993).

3 Discussion and conclusions

In this letter we reported on the results of the first $\sim$2 weeks of the Swift monitoring of the newly discovered AMSP IGR J17511-3057 in outburst. Regarding the persistent emission of this source, the higher statistics spectra measured with XRT/WT showed that two different components were required to fit the data. These comprise a BB emission that we interpreted as being produced onto the NS surface (as suggested by the size of the measured radius compatible with a hotspot origin), and a power-law component that is most likely caused by the comptonization of the soft emission in the NS accretion column (see Sect. 1). This spectrum qualitatively agrees with observations of other AMSPs in outburst and with the preliminary results of the XMM-Newton observation of IGR J17511-3057 reported by Papitto et al. (2009b). At odds with their analysis, we could not detect the softest component of the spectrum that was modelled with a multicolor disk blackbody (temperature of $0.13~\pm~~0.1$ keV). This most likely stems from the lower statistics and the short exposure time of the XRT spectra compared to what is obtained with XMM-Newton (exposure time $\sim$71 ks).

From the analysis of the type-I X-ray bursts, we can estimate an upper limit on the source distance. In principle, determination of the distance can be obtained only when a burst undergoes a PRE; in this case, it is assumed that the bolometric peak luminosity of the source is saturated at the Eddington limit, $L_{\rm Edd} \approx 3.8 \times 10^{38}~{\rm erg}~{\rm s}^{-1}$ (as empirically derived by Kuulkers et al. 2003). Unfortunately, in the case of IGR J17511-3057, our time-resolved analysis of the bursts could not detect any evidence of a PRE. From the measured orbital period (several hours) and the mass function of IGR J17511-3057, we can argue that the companion star in this system is most likely a hydrogen-rich brown dwarf, as suggested in the case of SAX J1808.4-3658 (Bildsten & Chakrabarty 2001).

Similar to the case for this source, we thus expect that the time elapsed between different bursts is long enough to allow the hot CNO burning to deplete the accreted hydrogen. Therefore, the type I X-ray bursts of IGR J17511-3057 are most likely produced by the ignition of pure helium. Under the above assumptions, the peak luminosity of the brightest burst (C) can then be considered to be (at most) the Eddington value $L_{\rm Edd} \approx 3.8 \times 10^{38}~{\rm erg}^{-1}$, and the resulting upper limit on the source distance is $d = 10.1 \pm 0.5$ kpc. For comparison, the theoretical value of this distance found by assuming an helium atmosphere and canonical NS parameters (1.4 solar mass, and a radius of 10 km; see e.g., Lewin et al. 1993), is $8.9 \pm 0.4$ kpc. By using the above results, we can also evaluate the theoretically expected recurrence time of the bursts. With a distance of d=10 kpc, the persistent unabsorbed 0.1-100 keV flux of the source at the timeof the bursts (A), (B), and (C) would translate into a bolometric luminosity of $L_{\rm pers}\approx 3.7, 4.2, 1.1 \times 10^{37}$ erg s-1, respectively.

With these values at hand, we can estimate the local accretion rate per unit area, $\dot{m}$, through the relation $L_{\rm pers}=4\pi R^2\dot m(GM/R)/(1+z)$ (with z = 0.31 the NS gravitational redshift; see e.g., Lewin et al. 1993). The ignition depths, $y_{\rm ign}$ for the three bursts can be calculated by using the equation $E_{\rm burst}=4\pi R^2y_{\rm ign}Q_{\rm nuc}/(1+z)$, where $E_{\rm burst}=4\pi d^2f_{\rm b}$, $f_{\rm b}$ is the fluence of the burst (see Table 2), and $Q_{\rm nuc}\approx 1.6$ MeV corresponds to the nuclear energy release per nucleon for complete burning of helium to iron group elements (Wallace & Woosley 1981). We obtain $y_{\rm ign} = 2.2$,2.4, $4.0 \times 10^8$  ${\rm g\ cm^{-2}}$ for the bursts (A), (B), and (C), respectively. For the above values of the local accretion rates and ignition depths, the expected recurrence time of the bursts is about $\Delta t=(y_{\rm ign}/\dot m)(1+z) \simeq 0.2$-0.9 days. This agrees with the recurrence time measured from the XMM-Newton (two bursts in $\sim$71 ks, Papitto et al. 2009b) and INTEGRAL observations (4 bursts in $\sim$200 ks carried out from 2009 September 16 to 2009 September 19, Falanga et al. 2009, in preparation).

Several more Swift observations of IGR J17511-3057 have already been planned. The results of this long-term monitoring campaign will be reported elsewhere (Campana et al. 2009, in preparation).

Acknowledgements
We thank N. Gehrels and the Swift team for the rapid schedule of the observations of IGR J17511-3057, and an anonymous referee for useful comments. C.F. has been supported by grant DLR 50 OG 0601.

References

Footnotes

... pixels[*]
see also http://www.swift.ac.uk/pileup.shtml

All Tables

Table 1:   Swift observation log of IGR J17511-3057.

Table 2:   The three type-I X-ray bursts parameters.

All Figures

  \begin{figure}
\par\includegraphics[angle=-90,width=7.2cm,clip]{13517fig1.ps}
\end{figure} Figure 1:

Swift /XRT long-term light curve of the outburst of IGR J17511-3057 (time bin 1000 s, start time 2009 Sept. 13 at 20:00:41 UTC). The arrows in the plot mark the time of the type-I X-ray bursts observed with Swift. The two inserts show a zoom of the first two type-I X-ray bursts (time bin is 1 s). The time on the X-axes of these inserts is measured from the t0 of the bursts. These are 2009 Sept. 14 00:50:27 and 2009 Sept. 15 at 17:17:19 (UTC), respectively. The third type-I X-ray burst (C) is reported in Fig. 2.

Open with DEXTER
In the text

  \begin{figure}
\par\includegraphics[angle=-90,width=7cm,clip]{13517fig2.ps}
\end{figure} Figure 2:

The brightest type-I X-ray burst observed with Swift. The upper panel shows the BAT light curve (15-25 keV), whereas in the lower panel we reported the XRT light curve (0.5-10 keV). The time bin of the BAT and XRT/WT (XRT/PC) light curves is 1 s (5 s). The start time of the burst is 2009 Sept. 30 18:31:57 (UTC).

Open with DEXTER
In the text


Copyright ESO 2010