Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 353, 646-654 (2000)

Previous Section Next Section Title Page Table of Contents

2. Observations

Starting 12 January 1991 the position of RX J1313.2-3259 in the sky was scanned by the ROSAT XRT with the PSPC as detector. During 20 satellite orbits with a total exposure time of 485 s the object was detected with a mean count rate of 1.8 cts s-1 and a hardness ratio [FORMULA] 1. On a finding chart produced from the COSMOS scans of the SERC-J plates at ROE (Yentis et al. 1992) a V [FORMULA] star was found to be the likely optical counterpart, at a distance of [FORMULA] 3" from the X-ray position (Fig. 1). A low resolution spectrum taken on 31 August 1991 at the ESO/MPI 2.2m telescope showed strong Balmer and helium line emission thus confirming this identification. A list of all observations collected until now can be found in Table 1.

[FIGURE] Fig. 1. V-band CCD image of RX J1313.2-3259. The X-ray error circle from the RASS refers to the 90% confidence level. The position of the optical counterpart is [FORMULA], [FORMULA], the average visual magnitude [FORMULA].


Table 1. List of observations. Columns denote: (1) and (2) the time of the observation, (3) and (4) the telescope and instrument used, (5) the type of observation (spec.: spectrophotometry, point.: X-ray pointing, phot.: photometry, sp.pol.: spectropolarimetry), (6) the spectral range or the filters used, (7) the FWHM resolution (spectroscopy only), (8) the number of spectra or V filter data obtained, (9) the time span for the observation, and (10) the exposure time for the individual measurements.

2.1. X-ray photometry and spectroscopy

From the RASS data a photon event table was extracted which covered [FORMULA] arcmin2 around the X-ray position of the source. Using the EXSAS software package provided by the MPE Garching (Zimmermann et al. 1994) for the extraction of source photons in a circle of 250" radius centered on the source and background photons in a circle of 400" radius in scan direction we obtained the light curve shown in Fig. 2 and the mean spectrum inserted in Fig. 4. Phasing of the light curve has been obtained from Eq. (2) based on optical observations (see Sect. 3.2), so phase zero corresponds to the inferior conjunction of the secondary star. The error in the period results in a phase error below 0.01, therefore the cycle count is correct. The data are sampled from nine different orbital cycles of the binary. The mean count rate obtained from the light curve is 1.94 cts s-1. In one ROSAT orbit a count rate far above the average (6.5 cts s-1) was measured. We consider this as a singular event not typical for the orbital variation. Without this point the mean count rate drops to 1.68 cts s-1. In order to show that the light curve is not dominated by changes between binary orbits we have used two different symbols in Fig. 2 for the first and second half of the observation. A period search without the point at 6.5 cts s-1 revealed a likely period of [FORMULA] min. It should be noted that at no phase did the count rate drop to zero, so the accretion area giving rise to the X-ray emission never completely vanishes from view behind the horizon of the white dwarf. The alternative of a second accretion area contributing to the X-ray flux seems to be ruled out by the results of our spectropolarimetry, which shows no change of sign in the circular flux (see Sect. 2.3).

In two subsequent pointings with the ROSAT PSPC (July 1992) and HRI (July 1994) as detectors the source was observed with count rates of 0.036 PSPC cts s-1 (hardness ratio [FORMULA]) and 0.010 HRI cts s-1 (corresponding to [FORMULA] 0.06 PSPC cts s-1). This is a reduction in the count rate by factors 50 (1992) and 30 (1994) compared to the RASS. Phase binning of these data resulted in the lightcurves shown in Fig. 2. Again we note that the site of the X-ray emission always remains in view of the observer.

Fitting the RASS spectrum with a blackbody and a thermal bremsstrahlung component together with possible interstellar absorption resulted in a blackbody temperature of 58 eV and a column density of [FORMULA] H-atoms cm-2. The uncertainties for these parameters are depicted in Fig. 3. The (unabsorbed) contributions of the two components to the total flux in the ROSAT window (0.1 to 2.4 keV) amount to [FORMULA] erg cm-2 s- 1 and [FORMULA] erg cm- 2 s-1, respectively. For the temperature of the thermal bremsstrahlung component we assumed a value of 10 keV, since this cannot be derived from the spectral data. The resulting unabsorbed spectra are shown in Fig. 4 as dotted lines. Separating the count rate into the two components one obtains 1.70 cts s-1 for the blackbody component and 0.07 cts s-1 for the thermal bremsstrahlung component. The flux ratio [FORMULA] in the ROSAT band is 0.06, integrated over all frequencies it increases to 0.16. The uncertainty in these values is about a factor 2.

[FIGURE] Fig. 2. X-ray light curve of RX J1313.2-3259. The upper panel shows the source during the RASS (first half of observation: filled circles, second half: open circles) together with the measured background (crosses), the lower panels during two subsequent pointings with the PSPC and the HRI as detectors, respectively. For the preliminary period determination the single high datum at 6.5 cts s-1 has been ignored. The phases are computed from Eq. (2) in Sect. 3.2.

[FIGURE] Fig. 3. Confidence levels for the X-ray spectral fit of RX J1313.2-3259. The 1, 2, and 3[FORMULA] ranges for the fit parameters blackbody temperature and column density are shown together with the fit result (cross).

[FIGURE] Fig. 4. Overall observed spectrum of RX J1313.2-3259. The optical spectra shown at low frequencies were obtained in Feb. 93 (high state) and in Dec. 95 (low state). In the X-ray regime the RASS data are shown as circles, the PSPC pointing data as crosses. The dotted lines give the unabsorbed blackbody and bremsstrahlung components of the X-ray spectral fit to the RASS data.

The spectral fit to the data from the PSPC pointing gave a blackbody temperature of 50 eV, an unabsorbed flux of [FORMULA] erg cm-2 s- 1 for the blackbody component and [FORMULA] erg cm-2 s- 1 for the thermal bremsstrahlung component, with large uncertainties and a high correlation between flux and temperature of the blackbody fit (see Sect. 3.4). Here we again fixed the temperature for the thermal bremsstrahlung component to 10 keV, and the column density to the value obtained during the RASS. The data are also shown in Fig. 4.

2.2. Optical photometry

Sequences of direct images in different filters were taken at several epochs (see Table 1). To demonstrate the long-term variability of RX J1313.2-3259 we have plotted in Fig. 5 the mean values of V in the different runs together with the range of variability defined by the magnitudes which bracket 90% of the measurements in the corresponding observing run. Added to this plot are V-magnitudes from spectroscopy, obtained by folding the mean spectrum of each observing run with the sensitivity of the V-filter (a flux of [FORMULA] erg cm-2 s- 1 Å-1 corresponds to zero magnitude, see Bessell 1979). The dotted lines indicate the times of the X-ray observations (RASS, PSPC pointing, and HRI pointing). Typically RX J1313.2-3259 is found at [FORMULA] and brightened only for short episodes near HJD 2 449 036 (Feb. 93) and HJD 2 450 936 (May 98).

[FIGURE] Fig. 5. Long-term light curve of RX J1313.2-3259. Photometric observations are marked with filled dots (mean values) and a vertical bar (90% range), magnitudes derived from mean spectra (less accurate) by stars. The dotted lines show the times of X-ray observations (RASS, PSPC, HRI).

The orbital variation of RX J1313.2-3259 in the V-band is displayed in Fig. 6. Different symbols identify the different observing runs which are listed at the bottom of the figure. The lightcurves are clearly of different character in the bright (Feb. 93, May 98) and faint (all other) states of RX J1313.2-3259. While the bright states show only one minimum per orbit, the variation in the faintest state (Feb. 95) almost follows a sinusoidal variation with two minima per orbit. With increasing brightness the flux first increases outside the primary minimum (near phase zero) and starts to deviate from its value around the primary minimum for the bright states only.

[FIGURE] Fig. 6. Phase-folded lightcurves at different observing times as given at the bottom of the plot. For the dotted line see text in Sect. 2.2. The phases are computed from Eq. (2) in Sect. 3.2.

It is worth mentioning that the observations in Feb. 93 were collected in six subsequent nights with two different instruments and that during this high state episode the orbital-averaged flux decreased by 0.07 mag per day, suggesting a characteristic duration of the high state of a few weeks. The orbital light curve for the Feb. 93 data in Fig. 6 (diamonds) was obtained by subtracting a linear trend from the data. The fitted light curve (dotted line) represents a sinusoidal fit to the detrended data.

2.3. Optical spectroscopy and spectropolarimetry

Time sequences of spectra at different resolutions were obtained at 9 epochs as listed in Table 1 and displayed in Fig. 5 (stars). We first display in Fig. 7 the low resolution spectra from Feb. 93 (brightest state) and from July 95 (faintest state). Both spectra clearly show the contribution of a late-type star to the flux at the red wavelengths, identifiable through its TiO absorption troughs. The emission lines of hydrogen and helium are visible in both spectra, with He II [FORMULA]4686 strongly reduced in the fainter spectrum.

[FIGURE] Fig. 7. Spectra at low resolution from two different observing runs. The brighter spectrum was taken in Feb. 93, the other is the mean of six spectra obtained in July 95.

Complete coverage of the orbit in one night could be achieved for the observation run in March 97 (intermediate state) only. Medium resolution spectra in the blue (3600-5200 Å) and in the red (6440-8360 Å) were analyzed for Doppler shifts in their emission and absorption lines. The most accurate results were obtained for the red spectra. Fitting the H[FORMULA] emission line with a double Gaussian profile splits the line into a narrow (average FWHM 5.9 Å, unresolved) and a broader (average FWHM 17 Å) component. The resulting radial velocities are plotted in Fig. 8, upper panel. The average radial velocities of three absorption lines (Na I [FORMULA]8183, 8195 and K I [FORMULA]7699) are shown in the same panel. In the lower panels of Fig. 8, the fluxes in the narrow emission component of He I [FORMULA]6678 and the absorption line of K I [FORMULA]7699 are displayed together with results from irradiation computations (see Sect. 3.2). The measured fluxes of the absorption line are very sensitive to the assumed level of the continuum near this line and may be in error by up to 30%. We also measured the fluxes in the narrow emission components of H[FORMULA], H[FORMULA], H[FORMULA], H[FORMULA], He II [FORMULA]4686, He I [FORMULA]4026, and Ca II K by fitting double Gaussians to the profiles. In Table 2 the velocity amplitudes and phases for the blue-to-red zero crossings are given for the two components together with the line flux ratios (minimum to maximum) of the narrow components. The phases were derived from the ephemeris given in Sect. 3.2. The minimum emission line flux stays finite for all of these lines. In this respect, all lines behave similarly to He I [FORMULA]6678 in Fig. 8.

[FIGURE] Fig. 8. Line properties extracted from spectra taken in Mar. 97. Upper panel: radial velocities averaged from measurements of three absorption lines (stars), from the narrow component of the H[FORMULA] emission line (filled circles), and the broad component of the H[FORMULA] emission line (open circles). The solid, dashed, and dotted lines give the sinusoidal fits to the data. Middle panel: normalized line flux of the He I [FORMULA]6678 narrow emission component together with results from irradiation computations for three different inclinations (solid: [FORMULA], dashed: [FORMULA], dotted: [FORMULA], see Sect. 3.2). Lower panel: normalized line flux of the K I [FORMULA]7699 absorption line together with results from irradiation computations for the same inclinations.


Table 2. Results from double Gaussian fits. The columns list the fitted emission line, the velocity amplitudes in km s-1 and phases for blue-to-red zero crossing both for the narrow and the broad component and the line flux ratio of minimum to maximum for the narrow component.

During the observation in Dec. 95 the source was in a low state, which allows for a separation of the spectral flux into the different contributions from the M-star, the accretion area, and the white dwarf. The results of this analysis will be discussed in Sect. 3.3. The spectra taken during another low state in Jan. 92 are not suited for this kind of analysis because they cover less than half an orbit and their flux calibrations are unreliable.

For the observation in May 98 (high state) EFOSC 2 was equipped with the Wollaston prism and a quarterwave plate, using Grism B300. The sequence of [FORMULA] spectra covers only 60% of the orbital period. Circularly polarized fluxes have been obtained by taking the difference between the two spectra produced by the Wollaston prism (Fig. 9). All spectra show circular polarization of negative sign only. We find, that two maxima of the polarized flux, at [FORMULA] 5000 Å and [FORMULA] 6600 Å, are present in all spectra except near phase 0 where the viewing angle is smallest with respect to the axis of the accretion funnel. In addition, a third maximum of the polarized flux occurs at [FORMULA] 3950 Å over the restricted phase interval of 0.3 to 0.5. We will argue below that these are the [FORMULA], [FORMULA], and [FORMULA] harmonic in a field of about 56 MG and that the different phase behavior arises from optical depth and geometric effects (Sect. 3.3).

[FIGURE] Fig. 9. Circularly polarized flux extracted from spectra taken in May 98. The grayscale plot displays fluxes between 0 (white) and [FORMULA][FORMULA] erg cm-2 s- 1 Å-1 (black), orbital phases are given on the right-hand side. The image is smoothed in order to remove the contributions from the emission lines.

Spectra taken during the other observing runs have been analyzed for radial velocities, flux contribution from the secondary star, and cyclotron emission. Because of lower resolution, incomplete orbital coverage, or less favorable observing conditions they mainly helped to reinforce and confirm the results obtained from the March 97 data and are important for excluding possible alias periods.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 2000

Online publication: December 17, 1999