For all X-ray positions in our sample, with the exception of a few positions
for which obvious optical counterparts were known from SIMBAD, NED or other
astronomical data bases (see below), we obtained optical images and
spectra. These observations were carried out at the 2.15m telescope of the
Guillermo Haro Observatory which is located near Cananea, Sonora,
Mexico and which is operated by INAOE. For this purpose a dedicated focal
reducer camera, the
Landessternwarte Faint Object Spectrograph and Camera (LFOSC)
was constructed at the LSW. The instrument was attached to the Cassegrain
focus of the telescope and allows direct CCD imaging,
filter photometry, and multi-object spectroscopy. It is
equipped with an EEV P8603 CCD detector ( pixels, 22
m
linear pixel size). With a
field of view the image scale is 1
pixel
.
For direct imaging we used Johnson B and Cousins R filters (Bessell
1979). In this observing mode a limiting
magnitude of can be reached
within 900
. Observations in B and R were used to derive the
colours for a rough first object classification.
For the spectroscopic observations we used masks with circular holes of
3 projected diameter in the focal plane of the telescope.
These masks were produced directly from
the CCD images using a computer controlled drilling device. In addition
to the object holes each mask contains holes producing spectra of the
sky background near the objects. Alternatively, a slit mask can be inserted
in order to perform normal
long slit spectroscopy. The spectra are oriented in E-W
direction. Perpendicular to the direction of dispersion, i.e. in N-S
direction, a minimum separation of the holes of
at least 5
was
found to be necessary to avoid overlapping spectra. Therefore more
than one spectral exposure
was required if several candidate objects separated by less than 5
in N-S direction were present in the field. Figure 4 (click here) shows an example
for such a field.
Two different grisms giving reciprocal linear dispersions of
250Åmm and 360Åmm
, respectively, were used.
For holes 2
5 to the east of the field centre
(which is also the position of the long slit) the spectral intervals covered
were 4000-7200 Å and 4200-8800 Å, respectively.
With a hole diameter of 3
the spectral resolution
was about 13 Å
and 18 Å, respectively, i.e. a resolving power of
at central wavelengths.
For flat fielding and wavelength calibration of the spectra built-in halogen, and neon and xenon lamps were used. Flux standard stars were observed for flux calibration of the spectra. The spectra were reduced with a ESO-MIDAS based software package available at the Landessternwarte.
Accurate sky subtraction was achieved by using the intensity of the
night sky emission line of [OI]5577 for calibrating small differences
in the throughput of the individual mask holes.
For the classification of stellar counterparts we observed a grid of
spectroscopic standard stars with spectral types between O and late M, and
luminosity classes I, III, and V using both grisms. Tests showed that with
these standard stars
a spectral classification with an accuracy of better than 5 subclasses
can be obtained for objects brighter than about 19
with
40
exposure time. Therefore, even the optically faintest X-ray
luminous coronal emitters expected in our subsample could be identified with
LFOSC. Counterparts with emission lines, as e.g. AGN,
could be identified as faint as
to 21
.
Figure 3: R band image of the field around the X-ray source
RXJ1207.7+3148. The 90%
SASS error circle (radius 35 ) is indicated. Many faint and diffuse
objects,
most likely distant galaxies, are visible. The brightness of the faintest
visible objects is
. For the two brightest objects near the
source position, designated by "A'' and "B'', spectra were obtained.
The remaining
objects are too faint for spectroscopy. Object "A'', whose spectrum is shown
in the lower panel, appears to be a galaxy with
20.3
and
22.3
. Absorption features most likely due to MgI b and NaI D
with a redshift of
are visible.
Residuals of the night sky lines are marked by "ns'', "x'' is a cosmic.
Object "B'' is a 16th magnitude F-type star. This star is visually too
faint to be a plausible counterpart. On the B image no object brighter than
about 22
was visible making a QSO as counterpart unlikely.
The galaxy "A'' is
visually too faint to be a plausible counterpart of the X-ray source
(see Sect. 4.3 (click here)).
Hence, the most likely identification of
this X-ray source is a distant cluster of galaxies
Figure 4: R image of the X-ray source RXJ0747.3+6822. The 90%
SASS error circle is indicated (radius 44 ).
With two exposures spectra of all objects within 60
radius
around the RASS position could be obtained. "r'' denotes reflex images of
bright stars in the field.
In the lower panel the spectrum of object "A''
is displayed which is the likely counterpart of the RASS source. It is a Sy 1
galaxy with
18
at redshift z = 0.120
Figure 5:
R image of the position of the X-ray source RXJ0403.5+0837. In the the 90%
SASS error circle (radius 44 ) several possible candidates for the
optical counterpart are visible. The objects observed spectroscopically are
designated by S1, "2'', "5'', and "8''. The bright
object S1 is a 13th magnitude G to K-type star. Object #5 with
18.4
is a QSO with H
+[OIII ] and MgII
at redshift z = 0.589.
H
partly falls into the atmospheric band at 7600Å.
Each of the two objects could be the X-ray source or at least contribute to
the observed X-ray flux (see text). The remaining objects are faint stars
which can be excluded for being the counterpart of the X-ray source
Examples of observations obtained with LFOSC are shown in Figs.
3 (click here), 4 (click here), and 5 (click here). Usually possible counterparts within a
circle of about 50 to 60
radius around the X-ray position were
observed (see below). This
is typically the 3
error circle as calculated by the standard ROSAT
reduction software (SASS) (see Paper I). The average SASS 90% error circles
which are indicated in the figures are of the order of 40
to
45
. As noted below, our identifications discussed in Sect.
5 (click here) showed that the true error circles are, in fact, smaller.
The observations collected in Cananea were supplemented by
spectroscopic observations with
the 2.2m telescope at ESO, La Silla, and with the 72 cm Waltz telescope at
the Landessternwarte Heidelberg. In March 1996 we observed part of the
sources in area III at the ESO/MPIA 2.2m telescope. These observations were
obtained with the EFOSC2 spectrometer
which was equipped with a Thomson pixel CCD chip (ESO CCD
#19). The spectral resolution obtained with grism #1 and the 1
slit
(cf. ESO Users Manual)
was
Å and hence comparable to that achieved with the lower
resolution grism of
LFOSC. A sample of bright stellar counterparts
previously identified with LFOSC were observed with higher spectral
resolution between March and October 1993
at the 72 cm Waltz telescope in order to study the spectra in more detail.
A Boller & Chivens spectrograph attached to the Nasmyth focus and equipped
with an EEV P8603 CCD chip with
22
m pixels was used.
The grating with 1200 lines mm
yielded a reciprocal linear dispersion
of 44 Å
, and (with a 2.4
slit) a
spectral resolution of about 2Å. The observed spectral region was
6250-6750 Å.
Because of varying weather conditions most of the direct images could not be
directly
calibrated photometrically. Therefore we started an additional observing
program to obtain secondary photometric sequences for the fields around each
X-ray source. These photometric observations are being carried out in Cananea
and at the Calar Alto 1.23m telescope in Spain. Details will be described
elsewhere. At this time and throughout the present
paper we use photometry taken from the HST Guide
Star Catalog
(Lasker et al. 1988) (GSC) and from the APM catalogue
(Irwin & McMahon 1992). For the APM photometric data
V magnitudes can be
estimated from O and O-E by using the relation by
(Irwin & McMahon 1992) and the colour transformation
determined for the POSS plates by Humphreys et al. (1991)
who found O-B to be nearly independent of B-V to within 0.3
for
. Likewise, an estimate for R can be obtained from E and
O.
Since Sep. 1990 we observed more than 800 RASS positions and we obtained in this way more than 3500 spectra of suspected optical counterparts. Meanwhile, observations or literature identifications (SIMBAD and NED databases) exist for nearly all sources in the count-rate and area limited complete subsample described above. In the next section we discuss the method of optical identification.