4 Discussion

4.1 V4641 Sgr - summary

Based on the available RXTE data of the Galactic Center scans, ASM monitoring and RXTE pointed observation the overall picture of the V4641 Sgr evolution can be summarized as follows:

1.

During approximately a half of a year after its discovery V4641 Sgr remained in the "quiet'' state (Fig. 1) with X-ray luminosity at the level of $L_{\rm X}\sim 10^{36}$ erg/s (flux $\sim$10-20 mCrab) and a soft spectrum (Fig. 4).

2.

A "landmark'' in V4641 Sgr evolution was a dip which started on Sep. 2 and lasted for $\sim$10 days (Fig. 1). The source luminosity droped by a factor of $\ga$10. After the dip the source spectrum changed dramatically (Fig. 5) and the optical emission increased by $\Delta m_V\sim$ 2 (Kato et al. 1999). Interestingly, the apparent size of the radio jet can be reconciled with the constrains on the binary system inclination angle only assuming that the ejection occurred before Sep. 6. It seems plausible to associate the jet ejection with the X-ray dip.

3.

On Sep. 14.9 a period of flaring activity started (Fig. 1, lower panel, Fig. 12). Although, due to fastness of the events, the data are rather sparse, comparison of the optical and X-ray light curves (Fig. 12) suggests sufficiently smooth evolution of the optical light, probably reflecting change of the bolometric luminosity and/or mass accretion rate. The X-ray flux, on the contrary, changed in a rather irregular fashion. Three bright X-ray flares were detected during rising and decaying parts of the optical light curve, with a deep minimum at the epoch of the maximal optical/bolometric luminosity. The source spectrum (Fig. 5) during the X-ray minimum had a strong emission line at $\approx$6.6 keV with equivalent width of $\approx$2.4 keV, indicating dominant contribution of the reflected/reprocessed emission. During the brightest flare the source reached flux level of 12.2 Crab corresponding to the 1-10 keV luminosity of $\sim$(3-4) $\times$ 10³⁹ erg/s or exceeded Eddington luminosity for a 10 $M_{\odot}$ black hole.

4.

The final phase of the third flare was studied by collimated instruments aboard RXTE. The spectral evolution during the flare can be understood in terms of absorption by ionized medium of varying column density. Strong fluorescent line of iron present in the spectra had smaller fractional rms than the continuum flux and was delayed with respect to the continuum by $\sim$25-50 s.

5.

In the period of low absorption the source demonstrated a hard spectrum typical for black holes in the low spectral state (Fig. 9). Its luminosity, however, exceeded by a factor of $\sim$10 and $\sim$2-3 respectively the Cyg X-1 luminosity in the hard and the soft spectral states.

6.

An orbital modulation of the X-ray flux during the period of quiescence was detected with fractional amplitude of $\sim$30%. The phase and energy dependence of the orbital modulation is consistent with obscuration by the matter in the vicinity of the optical companion, in which case the inclination of the binary system should be close to $\sim$65-70 $\hbox{$^\circ$ }$.

Figure 12: The light curves of V4641 Sgr in optical (VSNET data, http://www.kusastro.kyoto-u.ac.jp/vsnet/Xray/gmsgr.html) and X-ray energy bands. The RXTE/ASM points are shown by the open circles, the RXTE/PCA points by open squares.

4.2 Super-Eddington accretion and optically thick envelope

Earlier, we have suggested (Revnivtsev et al. 2002), that the optical data collected during the period of flaring activity of Sgr, can be naturally understood assuming formation of an optically thick warm ( $T\sim 10^5$K) envelope/outflow enshrouding the central source. The envelope is a direct consequence of significantly super-Eddington accretion and disappears when the mass accretion rate decreases below the Eddington value. Such an envelope, being optically thick at the optical wavelengths due to free-free processes and in the X-ray band due to absorption by the metals and Compton scattering, absorbs and re-emits bulk of the central source luminosity. In such a picture, smooth behavior of the optical light reflects evolution of the bolometric luminosity and, possibly, of the mass accretion rate. The bright X-ray flares occurred on the rising and decaying part of the optical light curve are a result of the changes in geometry and/or optical depth of the envelope.

At the peak of the optical light (which likely corresponded to the peak of the bolometric luminosity and the maximum in the optical depth of the envelope) the central source was almost completely obscured, which caused deep minimum in the X-ray flux. The weak X-ray emission observed in the scanning observation could be a result of reprocession of the primary X-rays by surrounding warm Compton thick gas. This is in good qualitative agreement with the actually observed spectrum (Fig. 5) which is extremely hard and has very strong fluorescent line of iron with equivalent width of $\approx$2.4 keV. The line centroid energy, $E\sim6.7$ keV requires significant ionization of iron. An alternative explanation of the observed large equivalent width of the line could be thermal emission from optically thin gas in the base of the jet, similarly to interpretation of the spectra of SS433. The picture could be clarified if the hard X-ray data was available.

The only broad band spectroscopical data available is that of the RXTE pointed observation performed in the end of the outburst. Coincidentally, it caught the tail of a short flare occurred during decaying part of the outburst. The spectral evolution observed by RXTE can be understood as an effect of absorption by ionized gas with the column density decreasing with time. This picture is further supported by the study of variability of the iron K$_{\alpha}$emission. We found that fractional rms of the iron K$_{\alpha}$flux is smaller than that of the surrounding continuum (Fig. 10). This result can be easily understood as smearing of the variations on the time scales shorter than the light crossing time of the reprocessing media. In addition, the finite light crossing time of the reprocessing media should cause time delay of the reprocessed emission with respect to the primary flux - in a good agreement with the observed behavior (Fig. 11). The amplitude of the observed time delay, $\tau\sim50$ s, corresponds to the linear size of $\sim$10¹² cm, i.e. is of the order of the Roche lobe of the binary system, as should be expected.