We present Gemini spectroscopy for 21 candidate optical counterparts to X-ray sources discovered in the Galactic Bulge Survey (GBS). For the majority of the 21 sources, the optical spectroscopy establishes that they are indeed the likely counterparts
. One of the criteria we used for the identification was the presence of an Ha emission line. The spectra of several sources revealed an Ha emission line only after careful subtraction of the F or G stellar spectral absorption lines. In a sub-class of three of these sources the residual Halpha emission line is broad (> 400 km/s) which suggests that it is formed in an accretion disk, whereas in other cases the line width is such that we currently cannot determine whether the line emission is formed in an active star/binary or in an accretion disk. GBS source CX377 shows this hidden accretion behaviour most dramatically. The previously-identified broad Ha emission of this source is not present in its Gemini spectra taken about 1 year later. However, broad emission is revealed after subtracting an F6 template star spectrum. The Gemini spectra of three sources (CX446, CX1004, and CXB2) as well as the presence of possible eclipses in light curves of these sources suggest that these sources are accreting binaries viewed under a high inclination.
We obtained VLT/FORS2 spectra of the optical counterparts of four high-luminosity (L_X >= 10^40 erg/s) ULX candidates from the catalog of Walton (2011). We first determined accurate positions for the X-ray sources from archival Chandra observations a
nd identified counterparts in archival optical observations that are sufficiently bright for spectroscopy with an 8 meter telescope. From the spectra we determine the redshifts to the optical counterparts and emission line ratios. One of the candidate ULXs, in the spiral galaxy ESO 306-003, appears to be a bona fide ULX in an HII region. The other three sources, near the elliptical galaxies NGC 533 and NGC 741 and in the ring galaxy AM 0644-741, turn out to be background AGN with redshifts of 1.85, 0.88 or 1.75 and 1.40 respectively. Our findings confirm the trend of a high probability of finding background AGN for systems with a ratio of log(F_X/F_opt) in the range of -1 to 1.
