8 Discussion

The uncertain distance makes it difficult to assess the peak luminosity accurately, but the low galactic latitude places SAX J1711.6-3808 in the galactic disk and diminishes the likelihood of large distances through association with the Galactic halo. For a canonical distance of 8.5 kpc the 1-200 keV unabsorbed luminosity would be $(4\pm1)~10^{37}$ erg s^-1. For masses in excess of 1.4 $M_\odot$ the implied luminosity is below 25% of the Eddington limit. The source would have to be at a distance at least twice that to the Galactic center for the peak luminosity to be near Eddington. Despite the lack of coverage of the onset, the available data suggest that the true peak flux of the outburst was not much higher than observed because the source flux remained on a plateau for a month after the discovery before it started to decay.

Wijnands & Miller (2002) announce a decoupling between the 3 to 20 keV luminosity and the state transitions in the sense that the soft component appears at intermediate luminosities instead of at the peak as would be expected. They consider it unlikely that this may be explained by the evolution of the flux above 20 keV and conclude that the decoupling is a significant physical effect. However, the picture becomes different when the flux below 3 keV is also taken into account. Our analysis shows (Fig. 11) that there is an important bolometric correction from below 3 keV. In fact, the soft excess contains so much flux that its appearance marks the peak flux for the whole outburst in the 1-20 keV band. While keeping in mind that a substantial fraction of the bolometric flux may be outside the 1-20 keV band, as evidenced by the NFI data, the conclusion is justified that there is no hard evidence for a decoupling of the bolometric flux and state transitions.

The 6 keV emission feature in SAX J1711.6-3808 is characterized by 1) a centroid energy of 6.3-6.7 keV; 2) a FWHM of about 2.6 keV and an equivalent width of 0.3 to 0.8 keV; 3) little evidence for asymmetries or structure; 4) no evidence for line shape variability during the outburst when the line flux declines by one order of magnitude; and 5) temporary factor-of-2 to 3 increase in the equivalent width simultaneous with a strong increase of a disk black body component at substantially lower energies.

The broadness is exceptional though not unprecedented among neutron star low-mass X-ray binaries. Asai et al. (2000) recently analyzed Fe-K line emission properties of 20 such LMXBs as measured with ASCA with relatively high energy resolution (2% at 6 keV for SIS and 8% for GIS), and found such lines in half of all cases. The FWHM in those detections ranges up to 0.7 keV. This is a confirmation of earlier EXOSAT work compiled by Gottwald et al. (1995) who find widths of up to $\sim$1 keV. Recently, two cases were reported of Fe-K emission lines in neutron star systems with FWHM in excess of 1 keV: Ser X-1 with a FWHM of $2.3\pm0.33$ keV (Oosterbroek et al. 2001), and GX 354-0, with a width of $3.0\pm0.7$ keV (Narita et al. 2001).

While unusual for NS systems, the line width has relatively more precedents among BH systems. There are at least 7 cases: the transient micro-quasar XTE J1748-288 (Miller et al. 2001), the transients GX 339-4 (Feng et al. 2001), XTE J1550-564 (Sobczak et al. 2000), GRO J1655-40 (Tomsick et al. 1999), 4U 1543-47 (van der Woerd et al. 1989) and XTE J2012+381 (Campana et al. 2002), and the persistently bright Cyg X-1 (Barr et al. 1985; Fabian et al. 1989; Miller et al. 2002). A recent BeppoSAX observation of Cyg X-1 (Frontera et al. 2000) in its low/hard state shows a Gaussian profile centered on 6.2 keV with a FWHM of 2.9 keV and an equivalent width of 0.35 keV. These values are similar to those for SAX J1711.6-3808. The width and prominence of the line in Cyg X-1 are less when the source is in the soft state.

A black hole nature of SAX J1711.6-3808 would also be consistent with the lack of coherent oscillations in the PCA data and the lack of type-I X-ray bursts in all X-ray data. The total exposure time on the active SAX J1711.6-3808 is about 412 ksec. For the exhibited range of luminosities (assuming a canonical distance of 8.5 kpc), this should have been ample opportunity to detect a type-I burst if the compact object would have been a neutron star (e.g., In 't Zand 2001). Of all 24 X-ray binary transients within 20$^{\rm o}$ from the Galactic center that WFC detected persistent emission from, seven failed to exhibit type-I X-ray bursts: GRO J1655-40, GRS J1737-31, GRS 1739-278 XTE J1748-288, XTE J1755-324, SAX J1819.3-2525 and SAX J1711.6-3808. Apart from the latter, these are either dynamically confirmed black hole candidates (Bailyn et al. 1995; Orosz et al. 2001) or suspected black holes on arguments different from the lack of type-I bursts (see Cui et al. 1997 for GRS J1737-31, Borozdin et al. 1998 for GRS 1739-278, e.g. Miller et al. 2001 for XTE J1748-288, and Goldoni et al. 1999 for XTE J1755-324).

It is tempting to explain the line broadness as relativistic broadening, in analogy to such features in a number of active galactic nuclei, most notably in MCG-6-30-15 by Tanaka et al. (1995) and NGC 3516 by Nandra et al. (1999; for a recent review, see Fabian 2001). The line energies in those cases are about 6.5 keV, and the equivalent widths between 0.3 and 0.6 keV. If we do this, our data suggests that we are viewing this system nearly edge on. If the inclination angle is really that high, one would have expected orbital signatures ([partial] eclipses, dips) in the light curve, but none were detected. This may have been chance coincidence, since our data coverage is not exhaustive (i.e., 4.7 days net exposure time which compares to BHC orbital periods of 0.2 to 6.5 d).

In analogy to other Galactic stellar black hole X-ray transients, the relatively low luminosity, the hard continuum spectrum in combination with a weak black body component, and the power density spectrum show that this source was primarily observed in the so-called low/hard state (e.g., Tanaka & Lewin 1995) when the inner edge of the optically thick accretion disk is thought to be much further out than the innermost stable circular orbit (6$R_{\rm g}$ for a non-spinning compact object). Therefore, if the broadening of the Fe-K line is relativistic in nature, either it is unrelated to the accretion disk or the low/hard state is unrelated to the radius of the inner edge of the accretion disk. Both these possibilities seem unlikely.

An alternative to the relativistic interpretation of the line broadening is Comptonization of the line photons by the $kT_{\rm e}=26$ keV plasma that is measured through the continuum. The expected width is $\sigma_E/E=E\tau^2/m_{\rm e}c^2$. The width is consistent with the measured value if $\tau=3.8$. This value is in rough agreement with the measurements. Broadening due to Comptonization seems more likely than due to relativistic orbital motion because it would naturally explain the line flux increase at the time of the soft excess increase. When a soft excess appears, the inner edge of the accretion disk is thought to move closer in (due to an increased mass accretion rate). As a result, the emission from the disk becomes harder. This may yield higher Fe-K fluxes. The lack of change in line energy and broadness (compare Figs. 5 and 9) indicates that the plasma cloud is not related to the accretion disk but located in a spherical geometry around the compact object.

Another alternative is Comptonization in approaching and receding outflows. Fender (2001) shows that low/hard state black hole X-ray binaries like 1E 1740.7-2942 and Cyg X-1 exhibit radio emission that is thought to arise from collimated outflows which is supported in some cases by spatially resolved observations. A correlation between hard X-ray and radio flux suggests that these outflows do not have large bulk Lorentz factors, unless both fluxes are beamed by the same factor. This is in line with the width of the Fe-K feature in SAX J1711.6-3808 which would imply Lorentz factors up to at most 2.