Geometry of X-ray sources in accreting black-hole binaries

According to the long-dominant paradigm, the accretion disc in the hard state is truncated at a radius >> ISCO whereas it reaches the ISCO in the soft state. This explains many observed phenomena, e.g., spectral and variability differences between the states and transition from quiescence in transients. On the other hand, there have been many claims that the disc extends to the ISCO in the hard state, and the X-ray source has been postulated to be a lamppost very close to the horizon. I will discuss the current evidence for and against the disc truncation and the lamppost geometry, and their implications. If the lamppost model were correct, most of the produced photons would be trapped by the black hole, and the source luminosity as measured at infinity would then be much larger than that observed. Also, the luminosity measured in the local frame would be >> that observed, due to the photon trapping, time dilation and redshift, and T_e would be significantly higher than that observed. I will also present results of a study of off of the X-ray spectra of the hard state of GX 339-4 observed by XMM-Newton. These results show the truncation radius to be ISCO for all the data sets. These radii also agree with the independent determination of De Marco et al. (2015) based on soft X-ray lags.