2. Observations and analysis
2.1. The ROSAT all-sky survey (RASS) observations
RX J2115.7-5840 was observed during the RASS for a total of 294 sec. The source was scanned 13 times with exposures ranging from 16.2 to 25.9 sec. It showed a modulation of the X-ray flux by 100% reaching a peak countrate of 1.2 s-1. The mean survey countrate was s-1 and the mean hardness ratio HR1 , where H and S are the counts in the ROSAT hard (0.4 - 2.4 keV) and soft (0.1 - 0.4 keV) bands, respectively. Folded over the most likely optical period, the X-ray light curve shows a clear on/off behaviour with length of the X-ray bright phase extending for 50% of the orbital cycle.
With only 102 detected photons the X-ray spectrum of RX J2115.7-5840 is not very well constrained. It can be fitted using a single bremsstrahlung component ( keV) with X-ray spectral flux at 1 keV of 2 photons cm-2 s-1 keV-1. The spectrum is very weakly absorbed, cm-2, and even compatible with zero absorption.
2.2. Spectroscopic observations
An R -band CCD image of the field of RX J2115.7-5840 is reproduced in Fig. 1. Only one possible optical counterpart of the X-ray source lies within the positional error circle of the RASS X-ray observations. A total of 21 low-resolution spectra of this star (12 Å FWHM, integration time 300 sec) and one with intermediate resolution (3 Å FWHM, integration time 600 sec) were obtained with the EFOSC2 spectrograph at the ESO/MPG 2.2m-telescope on October 23, 1996, between UT 0:19 and 3:30.
The flux-calibrated low-resolution spectra were folded through Johnson BVR filter curves in order to derive broad-band optical light curves. These are suspected to be accurate within 0.5 mag (absolute) but variability and colours can be determined with much higher accuracy of approximately 0.1 - 0.2 mag.
The B -band light curve does not show any significant variation, the V -band light curve displays marginal variability at a level of 0.2 mag, but the R -band light curve shows a pronounced orbital hump with full amplitude of about 0.5 mag centered on HJD 245 0379.6231. The optical R -band light curve as derived from our low-resolution spectra is shown together with the variation in and the radial velocity variation of the main emission lines in Fig. 2. Two more spectra (not shown in Fig. 2) were recorded during the preceding faint and bright phases, respectively. Using all the spectra, a photometric period of 113 min was derived by a period search based on Scargles (1982) algorithm.
Mean bright- and faint-phase spectra with low spectral resolution are shown in Fig. 3 and the one spectrum with higher resolution is shown in Fig. 4. RX J2115.7-5840 shows the typical features of a magnetic cataclysmic binary with strong emission lines of the H-Balmer series (including the strong Balmer jump in emission), HeI, HeII, and the CIII /NIII Bowen blend at 4640/50 Å. The lines exhibit a phase-dependent asymmetry and display pronounced radial velocity variations. We have determined the radial velocity of 5 main emission lines (H , H , H , HeI 5876, HeII 4686) by fitting single gaussians. No significant difference between the radial velocity variations of the different lines was found. We, therefore, show the average radial velocity of the 5 lines in the lower panel of Fig. 2. Although the emission lines are slightly asymmetric, our resolution is not high enough in order to discern between possible (likely) multiple emission components. Although not reflecting the change of the radial velocity variations very well, we used a sine approximation in order to estimate the spectroscopic period. The best fit gives min, in agreement with the period derived from the photometric variations of our spectra, and consistent with Vennes et al. (1996) spectroscopic period.
CCD photometry of RX J2115.7-5840 was undertaken at SAAO over 7 nights, from 1996 September 3 to 9. The observations were made on the 0.75-m telescope with the UCT CCD camera, employing a Wright Instruments blue-sensitive EEV CCD chip operated in frame transfer mode. Observations were mostly conducted without a filter, except for those done on 7/8 September, for which alternately a B and I filter were employed. Details of the observations appear in the observing log (Table 1), suffice to say that integration times were either 20 or 30 sec for the filterless data, with no dead time between frames. The observations were obtained in both photometric and non-photometric conditions, while seeing was equally as variable, from good () to poor (2-3 arcsec). The scale of the CCD is 0.37 arcsec in normal mode (which was used for the majority of the observations), and twice that value for 2 2 prebinning mode, which was used when the seeing was poor. The field of the UCT CCD on the 0.75-m telescope is .
After the usual flat-fielding and bias subtraction, batch mode DoPHOT routines (Mateo & Schechter 1989) were used to obtain both PSF profile-fitted and aperture magnitudes. Brighter stars on the frames were used as comparison stars, and differential magnitudes derived. The rms scatter of the corrected comparison stars was typically 0.006 magnitudes for the filterless photometry, and 0.01 mag for the B & I data. RX J2115.7-5840 exhibits a large degree of variability, up to mag in I, 1 mag for filterless ("white-light"), and much less () for B. The pronounced "hump" seen in the longer wavelength light curves (I & filterless) is not seen in the B light curve (see Fig. 5). Furthermore, there are substantial night-to-night changes in the light curves, evidenced by a less pronounced hump, or even a double hump on the first night (see Fig. 6).
White-light photopolarimetry of RX J2115.7-5840 was undertaken on the SAAO 1.9-m telescope using the UCT Polarimeter (Cropper 1985) on 1997 September 15/16 and November 10/11 and 11/12. The relative faintness of the object precluded any filtered observations. The instrument was run in the so-called "Stokes mode", simultaneously measuring linear and circular polarization every 180 s, and the intensity every 10 s. Polarization standards were observed at the end of each night in order to derive the instrumental waveplate offsets, which are constant to from night to night. Sky measurement were obtained every 15 to 20 min, and the sky background values interpolated with a polynomial spline before subtraction from stellar intensity data arrays. The observations were obtained in photometric conditions with 1 arcsecond seeing, except for the last night (11/12 Nov), for which the seeing was worse ().
RX J2115.7-5840 shows clearly detectable circular polarization, ranging from to (Fig. 7). The accuracy of our linear polarization measurement was 3% and within this accuracy no clear detection of the linear polarisation is seen. The various circular polarization curves are quite dissimilar, indicating that large changes in the accretion geometry between the different observations took place. The most simple curve at JD 398 is reminiscent of a one-pole accreting AM Herculis star with extended self-eclipses of the accretion region by the revolving white dwarf. The large positive circular polarization seen for a rather short phase interval at JD 342 then indicates that also a second accretion region on the opposite hemisphere (with opposite circular polarization polarity) became active.
The light curves recorded simultaneously displays no pronounced orbital hump, and resemble those seen at JD 330 and 331 (see Fig. 6).
© European Southern Observatory (ESO) 1997
Online publication: April 20, 1998