Spinning-up: the case of the symbiotic X-ray binary 3A 1954+319
We present a timing and spectral analysis of the variable X-ray source \ta. Our analysis is mainly based on an outburst serendipitously observed during \inteKey Program observations of the Cygnus region in 2008 fall and on the \swift/BAT longterm light curve. Previous observations, though sparse, have identified the source to be one of only nine known symbiotic X-ray binaries, i.e., systems composed of an accreting neutron star orbiting in a highly inhomogeneous medium around an M-giant companion. The spectrum of \taabove keV can be best described by a broken power law model. The extremely long pulse period of 5.3 hours is clearly visible in the \inte/ISGRI light curve and confirmed through an epoch folding period search. Furthermore, the light curve allows us to determine a very strong spin up of \nicefrachh during the outburst. This spin up is confirmed by the pulse period evolution calculated from \swift/BAT data. The \swift/BAT data also show a long spin-down trend prior to the 2008 outburst, which is confirmed in archival \inte/ISGRI data. We discuss possible accretion models and geometries allowing for the transfer of such large amounts of angular momentum and investigate the harder spectrum of this outburst compared to previously published results.
Spinning-up: the symbotic X-ray binary 3A 1954+319
\FullConference8th INTEGRAL Workshop “The Restless Gamma-ray Universe”- Integral2010,
September 27-30, 2010
belongs to a small class of Low Mass X-ray Binaries (LMXB), in which a neutron star orbits around an M-type companion [(6)]. The orbits are typically wide, making accretion via common Roche-Lobe overflow impossible. Instead, material from the wind of the companion is accreted onto the neutron star and thereby converted into X-rays. This accretion mechanism is very common among High Mass X-ray Binaries and their O- and B-type companions with strong stellar winds, but not well investigated for late type binary systems. In the case of \tathe accretion stream onto the neutron star seems to be highly variable as irregular X-ray flares are observed as well as times where the source is hardly detected at all. Binaries hosting a neutron star and an M-type companion are commonly referred to as Symbiotic X-ray Binaries (SyXB), in analogy of Symbiotic Stars consisting of a White Dwarf and a late-type star. These systems were named after their optical spectra in which both components, the hot White Dwarf and the cold companion are seen. Such features are not seen in SyXB, as the neutron star is invisible in the optical. A close look in X-rays is thus indispensable for a description of those systems.
Clearly confirmed as SyXBs are only 7 sources today (nespoli09a, ), two more are under dispute. Only very few of the SyXB, like GX 1+4 or 4U 1700+24 have already been studied in detail with different X-ray instruments, including high-resolution CCD detectors, and much is still to be learned about these systems and the properties of the wind and the accretion mechanism. Being easily observable in X-rays SyXB provide a unique opportunity for studying the wind properties of late type stars, via X-ray variations of the emission lines and absoprtion column.
was detected in the late 1970s with Uhuru and Ariel V (forman78a, ; warwick81a, ). tweedy89a () analyzed Ginga data and proposed that the system could be a HMXB. No more detailed studies were performed until masetti06a () identified the companion as an M-star using an accurate Chandra-position, placing the system in the group of SyXB. Soon thereafter, corbet06a () discovered a 5 h period in \swift/BAT data. This lead to closer investigations by mattana06a (), who presented the first broad-band spectrum of \ta, using \inteand BeppoSAX. More spectral analysis was presented by masetti07b (), who also showed that the spectrum is consistent with the typical spectrum expected from an accreting neutron star and can be modelled using an either a cutoff powerlaw model or a thermal Comptonization model.
A strong decrease of the pulse period from 5.17 h to 5.09 h during an outburst in 2004/2005 and possible physical explanations thereof was reported by corbet08a () in \swift/BAT data. The period itself is most likely the spin period of the neutron star, as it is much too short for the expected wide orbit of the system and no other significant periods have been found between – s (corbet08a, ; masetti07b, ). It is also unlikely that the observed decrease in the period is due to the orbit of the binary system, as the mass function would then lead to impossible values of the mass of the companion, assuming a canonical mass of 1.4 for the neutron star.
In 2008 November \inteserendipitously observed a strong outburst of \ta, obtaining high quality lightcurves and spectra, which are presented here.
2 Lightcurves and Timing
The long-term daily lightcurve of \taas measured with \swift/BAT is shown in Fig. 1, left. It is clearly seen that \tashows irregular flaring with some flares being as bright as almost 100 mCrab and other episodes where it falls below the detection limit of \swift/BAT. Tickmarks above the lightcurve mark times when \tawas in the field of view of \inte/ISGRI. A relatively dense sampling of \inteobservations was performed during a flare in 2008 November when Key Programme observations of Cygnus X-1 (PI Wilms) were scheduled. As seen in the left panel of Fig. 1 this flare was one of the brightest since the monitoring of \swift/BAT, with an average flux over the outburst of between 20–100 keV, i.e., 60 mCrab and a peak flux of more than 180 mCrab. We extracted \intedata of that outburst using the standard Off-line Science Analysis (OSA) 8.0 to obtain spectra and using ii_light, as distributed with OSA 7.0, to obtain lightcurves with a time resolution of 100 s. The \inte/ISGRI lightcurve of that period is shown in Fig. 1, right. Even though \tawas in an overall bright state during that time the \intedata show that the X-ray flux is still highly variable and varies by a factor of 20. Using the epoch-folding technique (leahy83a, ), a period of 5.3 h is clearly detected in the data. Compared to the results by corbet08a (), the pulse period is significantly longer, i.e., the neutron star has spun down between 2005 and 2008.
As corbet08a () found a strong spin up trend in the 2005 flare, we split the \intelightcurve into three equally long parts and searched in each one for the pulse period separately. We found a clear spin up trend in these data, with the pulse period decreasing from 5.34 h in the beginning of the outburst to 5.26 h in the end. The results of the epoch-folding are shown in Fig. 2. From this analysis we determined the pulse ephemeris to be (during that outburst): d (mean of observation), , and . To check for consistency, we folded the lightcurve with this ephemeris to obtain a pulse profile and compared it to the measured lightcurve. As seen in the insets of the right panel of Fig. 1 the pulse profile describes the lightcurve very well, in the beginning of the outburst as well as close to the end. This is also the first time that the pulsed flux of \tais directly visible in a lightcurve, showing that \tadoes not show large pulse-to-pulse variations.
In order to put our \interesults into perspective, we updated the \swift/BAT period analysis of \ta, again using the epoch-folding technique. Using all publicly available \swift/BAT lightcurve data, local period determinations were performed for 20 d long segments. Figure 2 shows all successful \swift/BAT period measurements, that is, all data in which a period could be significantly measured. As the \swift/BAT data a impaired by a low signal-to-noise ratio, we also analyzed more \inte/ISGRI archive data. Additionally to the three data points from the 2008 November outburst we could obtain 3 more measurements of the pulse period during times with extensive coverage of \tawith \inte. All data points, \swift/BAT and \inte/ISGRI, are plotted together in the right panel of Fig. 2. As can be clearly seen the measurements for both satellites agree very well. The pulse period is slowly decreasing during quiescence but increases strongly during prominent outbursts, as expected from standard accretion theory (ghosh79a, ). One such outburst was measured by corbet08a () in 2005, the other one was caught serendipitously by \intein 2008 November. Even though in between these outbursts \tashows smaller flares, no measurable spin up is seen there. Correlating the luminosity with the change in pulse period led to no conclusive result, however, this may be due to the sparse sampling of the pulse period between outbursts.
3 Spectral analysis
The \inteobservations present the unique possibility to gain knowledge about the accretion physics during a strong flare of \ta, which was accompanied by a strong spin up of the neutron star. Due to the fact that the source is close to \inte/ISGRI’s detection limit during pulse minimum, we created a peak spectrum of only those 50% of data, which were measured at phases close to the major peak of the pulse profile, thereby improving the signal-to-noise ratio. Comparing pulse profiles at different energies above 20 keV did not reveal any strong energy dependence with pulse phase, indicating that the spectrum is not sensitive to the selected pulse phase and that the obtained spectrum is very similar to an overall phase averaged spectrum. This filtering resulted in the selection of 75 Science Windows (ScWs) with a pointing offset of to \ta, leading to a pulse peak spectrum with 445 ks exposure in ISGRI and 22 ks in JEM-X. Modelling these JEM-X and ISGRI data between 3–100 keV with the published models lead to unacceptable descriptions only. A much better description in terms of ( with 43 d.o.f.) was obtained, when using the bknpower model, a model consisting of two power-laws with different photon indices and , with the change from one photon index to the other taking place at a specific break energy . It was not necessary to apply systematics for this fit. The model was attenuated by photo absorption at lower energies, expressed as equivalent column density of neutral hydrogen atoms . Data and best fit model are shown in Fig. 3, and the parameters of the best fit model can be found in Tab. 1. The required new description of the model compared to the older spectra leads to an overall harder spectrum, as no exponential cut-off is visible anymore. The soft energy continuum has a slope which is consistent with previous published results, see, e.g., mattana06a ().
4 Outlook & Conclusions
We presented a timing and spectral analysis of \intedata of the large outburst of \tain 2008 November. We confirm the unusually long spin period and measure a value of h. During the outburst we found that the pulse period is drastically increasing, with a spin up rate as high as . This rate is a magnitude higher than the one found by corbet08a () for the 2005 outburst. As these authors already stated, a change of this magnitude can not be due to the orbital motion, as it would lead to a mass-function for the companion of . On the other hand, if we assume that the neutron star is spinning close to equilibrium, i.e., that the Alfvén radius is equal to the Keplerian co-rotating radius, then a magnetic field of G would be required. In the standard accretion picture of ghosh79a (), this assumption is justified as only close to equilibrium the torque can easily change direction. However, more sophisticated accretion models have been proposed for other sources, including retrograde spinning accretion disks (perna06a, ) and spherical wind accretion. If no stable accretion disk forms and material and momentum is accreted from the stellar wind, a torque transfer of this magnitude might be possible (see Postnov et al. elsewhere in these proceedings). Continued monitoring of \tais necessary in any case to clarify possible models and check if the source continues to spin down during quiet phases, while strongly spinning up during flares.
The pulse peak spectrum of \tawe presented is distinctly harder than the previously published spectra, but is otherwise featureless. Especially, we find no indication for a Cyclotron Resonant Scattering Feature (CRSF), as seen in many other X-ray binaries, where the neutron star possesses a magnetic field G. Such a feature is the only direct way to measure the magnetic field strength in the X-ray producing region and could clarify the size of the accretion region. If \tareally has a magnetic field on the order of G, it is not suprising that we do not see any CRSF, as the fundamental line energy would be MeV, where the source is not visible anymore.
For a more detailed presentation of the results see a forthcoming publication by Marcu et al., 2011, in prep.
- (1) R. Corbet, L. Barbier, S. Barthelmy, J. Cummings, E. Fenimore, N. Gehrels, D. Hullinger, H. Krimm, C. Markwardt, D. Palmer, A. Parsons, T. Sakamoto, G. Sato, and J. Tueller, Swift/BAT Detection of a 5 hour Period in 4U 1954+31, ATEL 797, (2006).
- (2) R. H. D. Corbet, J. L. Sokoloski, K. Mukai, C. B. Markwardt, and J. Tueller, A comparison of the variability of the symbiotic X-Ray binaries GX 1+4, 4U 1954+31, and 4U 1700+24 from Swift/BAT and RXTE/ASM observations, ApJ 675, 1424–1435 (2008).
- (3) W. Forman, C. Jones, L. Cominsky, P. Julien, S. Murray, G. Peters, H. Tananbaum, and R. Giacconi, The fourth Uhuru catalog of X-ray sources., A&AS 38, 357–412 (1978).
- (4) P. Ghosh and F. K. Lamb, Accretion by rotating magnetic neutron stars. III - accretion torques and period changes in pulsating x-ray sources, ApJ 234, 296–316 (1979).
- (5) D. A. Leahy, R. F. Elsner, and M. C. Weisskopf, On searches for periodic pulsed emission - The Rayleigh test compared to epoch folding, ApJ 272, 256–258 (1983).
- (6) N. Masetti, M. Orlandini, E. Palazzi, L. Amati, and F. Frontera, M-type giants as optical counterparts of X-ray sources 4U 1700+24 and 4U 1954+319, A&A 453, 295–299 (2006).
- (7) N. Masetti, E. Rigon, E. Maiorano, G. Cusumano, E. Palazzi, M. Orlandini, L. Amati, and F. Frontera, X-ray broad-band study of the symbiotic x-ray binary 4U 1954+31, A&A 464, 277–278 (2007).
- (8) F. Mattana, D. Götz, M. Falanga, F. Senziani, A. de Luca, P. Esposito, and P. A. Caraveo, A new symbiotic low mass X-ray binary system: 4U 1954+319, A&A 460, L1–L4 (2006).
- (9) E. Nespoli, J. Fabregat, and R. E. Mennickent, K-band spectroscopy of IGR J16358-4726 and IGR J16393-4643: two new symbiotic X-ray binaries, A&A, 516, A94 (2010).
- (10) Rosalba Perna, Enrico Bozzo, and Luigi Stella, On the Spin-up/Spin-down transitions in accreting X-Ray binaries, ApJ 639, 363–376 (2006).
- (11) R. W. Tweedy, R. S. Warwick, and R. Remillard, The 3A1954+319: a possible supergiant X-ray binary in proceedings of Two Topics in X-ray Astronomy: Vol. 1 (ESA-SP-296), (J. Hunt and B. Battrick, eds.), November 1989, pp. 661–663.
- (12) R. S. Warwick, N. Marshall, G. W. Fraser, M. G. Watson, A. Lawrence, C. G. Page, K. A. Pounds, M. J. Ricketts, M. R. Sims, and A. Smith, The ariel V (3A) catalogue of X-ray sources - I. Sources at low galactic latitude (absolute value of b less than 10 deg), MNRAS 197, 865–891 (1981).