Swift/XRT suggests a faint soft state in the 2015 outburst of V4641 Sgr

ATel #7904; A. Bahramian (Alberta), D. Altamirano (Southampton), C. Heinke (Alberta), M. Middleton (Cambridge), G. Sivakoff, B. Tetarenko (Alberta) on behalf of a larger XRB collaboration
on 12 Aug 2015; 15:03 UT
Credential Certification: Arash Bahramian (bahramia@ualberta.ca)

Tweet

V4641 Sgr is a black hole X-ray binary with frequent outbursts over the last decade (e.g., ATel #1796, #2785, #5803). MAXI detected a new outburst of this source (ATel #7858), later confirmed with a Swift/XRT Photon Counting (PC) mode observation (ATel #7874). We have monitored V4641 with Swift/XRT in Windowed Timing (WT) mode daily for the last 9 days. The source is clearly detected in all observations, with X-ray luminosities (0.5-10 keV) between 5e35 and 5e36 erg/s (assuming the models below, and a distance of 6.2 kpc; Macdonald et al. 2014, ApJ, 784, 2). For a black hole mass of 6.4 M_solar, this corresponds to Eddington luminosity fractions of about 0.06 to 0.6%.

We extracted spectra from all the Swift/XRT observations performed during this outburst (2015 August 2-11), using the 0.3-10 keV band for PC mode and 0.6-10.0 keV band for WT mode. As the observation is piled-up in the PC mode observation, we excluded the piled-up region following the XRT pile-up thread (http://www.swift.ac.uk/analysis/xrt/pileup.php). Due to the high source count rates of ~ 2-20 cnt/s, we only extracted grade 0 events for spectral analysis in all WT mode observations. We notice rapid variations between observations; between two observations separated by 4 hours (MJD ~ 57245), we see an increase in count rate by a factor of 5.

We first fit all spectra simultaneously with an absorbed powerlaw (tbabs*pegpwrlw), assuming Wilms et al. (2000, ApJ 542, 914) abundances and Verner et al. (1996, ApJ, 465, 487) cross sections. We tied N_H to a single value across observations and let the photon index vary. This fit yields N_H = 5e21 cm^-2 and photon indices of 1.6 to 1.8. However, this fit shows a very high chi-squared (1268 for 391 d.o.f) and strong correlated trends in residuals in all observations. Untying N_H does not resolve this issue. Thus, this model seems inappropriate for these data.

We then fit all spectra with an absorbed disk-blackbody model (tbabs*diskbb), tying N_H between observations while letting the inner disk temperature and model normalization vary. This gives a significantly better fit (chi-squared = 482 for 391 d.o.f). This fit yields N_H = (2.3+/-0.1)*1e21 cm, varying inner disk temperatures between 1.4 and 1.7 keV, and varying "apparent" inner disk radii between ~ 1 and 3 km (assuming a distance of 6.2 kpc and inclination angle of 70 degrees; Macdonald et al. 2014). If confirmed, a soft state is rather unusual for a LMXB below 1% Eddington luminosity. However, the black hole X-ray binary 4U 1630-47 has also been observed with a soft spectrum at an Eddington fraction of 0.008% (Tomsick et al. 2014, ApJ, 791, 70).

Inner disk radii values of <~3 km are too small for the innermost stable circular orbit of a maximally rotating 6 M_solar black hole. Perhaps we are only seeing a portion of the inner disk. This system has a high inclination (Macdonald et al. 2014), and has shown a wide range of absorbing column densities in the literature (e.g., Tomsick et al. 2014; Morningstar et al. 2014, ApJL, 786, 20). Our absorbing column density is significantly lower than these values, except for Beppo-SAX observations that suggest nearly no absorption (in't Zand et al. 2000, A&A, 357, 520). We note that the E(B-V) of this source (Macdonald et al. 2014) indicates a foreground N_H of (3.2+/-1.6)*1e21 cm^-2, consistent with our derived column density (following Bahramian et al. 2015, MNRAS, 452, 3475).

Analysing all publicly available MAXI data, we have detected 8 outbursts or flares in V4641 Sgr (Tetarenko, B. in prep), only 3 of which are also detected by Swift/BAT. Many of these flares are outbursts with typical peak (bolometric) luminosities of a few times 10^36 ergs/s, and durations lasting between ~6-328 days. If V4641 Sgr often undergoes soft spectrum outbursts, this may explain the relative lack of Swift/BAT detections of this source.

We thank the Swift team for rapid scheduling of our observations.