2000ApJ...539..445S

We have obtained optical spectra of the soft X-ray transient GRO J1655-40 during different X-ray spectral states (quiescence, high-soft, and hard outburst) between 1994 August and 1997 June. Characteristic features observed during the 1996-1997 high-soft state were: (1) broad absorption lines at Hα and Hβ, probably formed in the inner disk; (2) double-peaked He II λ4686 emission lines, formed in a temperature-inversion layer on the disk surface, created by the soft X-ray irradiation; and (3) double-peaked Hα emission, with a strength associated with the hard X-ray flux, suggesting that it was probably emitted from deeper layers than He II λ4686. The He II λ4686 line profile appeared approximately symmetric, as we would expect from a disk surface with an axisymmetric emissivity function. The Balmer emission, on the other hand, appeared to come only from a double-armed region on the disk, possibly the locations of tidal density waves or spiral shocks. The observed rotational velocities of all the double-peaked lines suggest that the disk was extended slightly beyond its tidal radius. Three classes of lines were identified in the spectra taken in 1994 August-September, during a period of low X-ray activity between two strong X-ray flares: broad absorption, broad (flat-topped) emission, and narrow emission. We have found that the narrow (single-peaked or double-peaked) emission lines cannot be explained by a conventional thin accretion disk model. We propose that the system was in a transient state, in which the accretion disk might have had an extended optically thin cocoon and significant matter outflow, which would also explain the systematic blueshift of the narrow emission lines and the flat-topped profiles of the broad emission lines. After the onset of a hard X-ray flare the disk signatures disappeared, and strong single-peaked Hα and Paschen emission was detected, suggesting that the cocoon became opaque to optical radiation. High-ionization lines disappeared or weakened. Two weeks after the end of the flare, the cocoon appeared to be once again optically thin.