Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 343, 455-465 (1999)

Previous Section Next Section Title Page Table of Contents

3. Analysis and results

3.1. The lightcurves

All observed MCCP and MEKASPEK lightcurves are presented in Fig. 3. Only B-band lightcurves are plotted since the very blue continuum of RXJ0019 in combination with the quantum efficiency of the detectors and the color temperature of the comparison provides the best S/N value in B. All observations are plotted individually to emphasize the high variability of this object. A plot including all observed lightcurves superposed according to the orbital period can be found in Meyer-Hofmeister et al. (1998) in their Fig. 1. Also compare with Fig. 2 in Will & Barwig (1996). The following features can be observed:

[FIGURE] Fig. 1. Minima observed in 1996. The minima occur statistically around phase [FORMULA]. The depth and the shape of the minima varies. All lightcurves were approximated by splines.

1. RXJ0019 shows a deep, broad minimum ([FORMULA] = [FORMULA] in UBV on Oct. 12, 1996, Table 4), lasting about 0.55 of a complete orbital phase. As can be seen in Fig. 1 the depth and shape of the minima varies and the timings of the minima are not regular. This explains the systematic difference of the orbital period of Will & Barwig (1996), which lies out of the error bars of other periods given in the literature (e.g. Greiner & Wenzel 1995 or Matsumoto 1996): our dataset was too short to eliminate the statistical effects. With the epoch of the photometric minimum given by Greiner & Wenzel (1995) and the very accurate determination of the minima observed at the Wendelstein observatory since 1992 we derive the following orbital period:


The timings of the minima used for this calculation can be found in Table 5 and in Will & Barwig (1996) in their Table 2.


Table 4. Magnitudes [FORMULA], [FORMULA] and orbital modulation [FORMULA] of RXJ0019.8 on Oct 12,1996.


Table 5. Timings of primary minima (HJD)

2. A secondary minimum appears around phase [FORMULA]. The amplitude of the secondary minimum is [FORMULA]. The following maximum at [FORMULA] is slightly fainter than the first maximum. Sometimes the second maximum is even completely smeared out and after a flat portion of the lightcurve the ingress to the main minimum begins (e.g. Oct 22, 1995 and Dec 15, 1996 in Fig. 3).

3. We observe humps and step-like features during ingress and egress of the main minimum. A period analysis of these features, when apparent in the lightcurves, reveals a period of approximately 1.8 h (e.g. Oct 11,1996 and Oct 13,1996). In contrast to the lightcurves showing steps and humps there are also lightcurves with a rather flat appearance (e.g. Nov 09,1996 or Dec 12,1996). We assume that RXJ0019 changes between two states: an "excited optical state" where humps and steps appear quasi-periodically with a period of 1.8 h and a "quiet optical state", where the lightcurves show no additional features. RXJ0019 can change from one state into the other from night to night. A probable explanation might be a short-term variation in the mass accretion rate (Meyer-Hofmeister et al. 1998).

4. Matsumoto (1996) did not observe any color variations in his lightcurves. We also cannot find any color variations in [FORMULA], [FORMULA] and [FORMULA]. But in [FORMULA] we detect a weak flux increase symmetrically to [FORMULA] with an amplitude of [FORMULA]m [FORMULA]. This variation in [FORMULA] can only be observed when the main minimum appears at midnight hours. In the evening and in the morning the natural reddening of the sky prevents an observation of this color variation. Therefore, we observed this color variation only on Oct 12, Nov 3 and Nov 9, 1996 (Fig. 2). In every night the width and the amplitude of this variation are similar. All other portions of the color curves from phase [FORMULA] to 0.9 are flat and show no variations.

[FIGURE] Fig. 2. Color Variations on Nov 9, 1996. Symmetrically to phase 0.0 an increase in the [FORMULA] flux can be observed with an amplitude of [FORMULA] [FORMULA]. The zero point of the vertical scale is arbitrary.

[FIGURE] Fig. 3. MCCP and MEKASPEK B-band lightcurves between September 1992 to October 1997. Three lightcurves from 1992 and 1995 have also been included (Will & Barwig, 1996). Note that in 1995 the brightness of RXJ0019 is slightly lower. All lightcurves differ significantly from each other. Also note the quasi-periodic 1.8h humps on Sept 21,1992, Oct 11,1996, Oct 13,1996. Increased scatter in the lightcurves is due to very poor observing conditions.

3.2. The mean orbital spectra

High-resolution spectra of RXJ0019 have recently been obtained by Tomov et al. (1998) and Becker et al. (1998). We also present high resolution mean orbital spectra (in Fig. 4) in the blue (3700...5000 Å) and in the red spectral range (6300...8300 Å). Similar to other SSS (e.g. Crampton et al. 1996 and Southwell et al. 1996) the very blue spectrum is dominated by He II and Balmer lines. The Balmer series can be traced up to H12. All transitions of the He II (n,3),(n,4) and (n,5) series can be observed when not blended by Balmer lines. No He I can be detected indicating a high ionisation state.

[FIGURE] Fig. 4. Mean orbital spectrum in the blue spectral range range from 4000Å to 5000Å and in the red spectral range from 6400Å to 7250Å. All Balmer lines show P-Cygni absorption on their blue wings. The base of He II [FORMULA] 4686 is very broad and asymmetric but the jet lines are not separated from the main emission. The jet lines (labeled S1 and S2) at [FORMULA] and [FORMULA] are clearly visible. [FORMULA] is considerably blended by two He II emission lines.

Additionally, we see some higher ionization emission features such as O VI [FORMULA] 3811. The C III -N III [FORMULA] 4640 - 4660 emission complex contributes to the blue wing of the He II [FORMULA] 4686.

On the blue wings of the Balmer lines P-Cygni profiles are visible in all observed lines. The H8 to H12 transitions can only be detected by their corresponding P-Cygni absorption between [FORMULA] to 0.30 (Fig. 5). We do not observe these absorptions after [FORMULA]. At the other Balmer lines the P-Cygni absorption shows an orbital modulation that almost disappears between phase [FORMULA] and 0.9 (Fig. 6), indicating that there might be a directed fast wind in this system. The velocity of the wind amounts to [FORMULA] 590 km s-1 with the blue absorption wing extending to [FORMULA] 900 km s-1. The He II emission lines are not truncated by P-Cygni profiles.

[FIGURE] Fig. 5. Phase-resolved, uncalibrated spectra showing the variability of the emission lines between 3700 Å and 4100 Å. The first spectrum from [FORMULA] is dominated by P-Cygni absorptions at the Balmer lines up to H12, which cannot be seen at later phases. Marks at the bottom indicate the Balmer series up to H13.

[FIGURE] Fig. 6. Phase-resolved spectra showing the variability of the emission lines between 4500 Å and 4900 Å. The P-Cygni absorption at [FORMULA] almost disappears between [FORMULA]. Also note the change of the base of He II [FORMULA] 4686 on the red wing of the emission line.

Two emission features, which are clearly above the S/N ratio at [FORMULA] 4500, 4930, could not be identified with any reasonable ion species.

The Balmer lines [FORMULA], [FORMULA] and [FORMULA] have broad bases and show satellite lines symmetrically around the main emission. All bright Balmer lines are blended by He II emission lines, in particular the two Helium lines [FORMULA] 6527, 6560 significantly contribute to [FORMULA] and He II [FORMULA] 4859 to [FORMULA]. Therefore, substructures in these lines are blurred by the Helium emission, whereas in [FORMULA], [FORMULA] and [FORMULA] similar complex substructures of different components can be detected. For illustration the trailed [FORMULA] spectrum is included in Fig. 10.

In He II [FORMULA] 4686 we only observe a broad base. The adjacent satellite components are not clearly separated from the main emission and are asymmetric (Fig. 4 and 6). Tomov et al. (1998) and Becker at al. (1998) argue that these satellite lines are the spectral signatures of jets originating near the white dwarf. These collimated high velocity outflows have recently also been reported by Quaintrell & Fender (1998) from their infrared spectroscopy.

Transient jets seem to be a common feature among the SSS. Jets have also been detected in RX J0513.9-6951 (e.g. Southwell et al. 1996) and recently in RX J0925.7-4758 (Motch 1998). The observed projected high outflow velocities of these sources, which are seen at very low inclination, indicate an origin of the jet near the white dwarf. Therefore, the jet velocity is close to the escape velocity of the central object (see Livio 1998) and can be used for estimating the inclination of the system.

The projected velocities of the jet lines (measured at maximum of their emission profiles) of RX J0019.8 are rather low. We measure 920 km s-1 for S1 and 805 km s-1 for S2. The asymmetry can be explained by the P-Cygni absorption, which alters the appearance of the jet emission on the blue side significantly. The low velocities may be due to a high system inclination (Tomov et al. 1998) if the jet originates near the white dwarf. A medium to high inclination is also consistent with the deep eclipse lightcurves from our photometry.

We note, that no spectral features of a secondary star are detected in our high resolution spectra.

3.3. Radial velocities

The usual procedure of measuring radial velocities in complex emission lines is fitting Gaussians to their line profiles. Measuring radial velocities of the emission lines of the RXJ0019 spectra is difficult. First, the Balmer emission lines are truncated by the phase dependent P-Cygni profiles and blended by He II emission lines which will alter the system velocity and semi amplitude considerably. Second, the asymmetric profile of the satellite lines will cause systematic errors in the velocity determinations. Furthermore, the measured velocities are only of the order of the spectral resolution of the spectrograph. Therefore, we restricted our investigation to the strongest emission lines [FORMULA] and He II .

For the pixel rows of the central parts of our trailed spectra we fitted Gaussians to the emission line profile. In order to suppress the influence of the noise and the satellite lines, the central region of the Gaussians was given a bigger weight than the adjacent spectral range (Bevington & Robinson 1992). After that a sine-fit was applied to the fitted radial velocity V and the observed semi-amplitude [FORMULA], using the following relation:


where [FORMULA] is the system velocity, [FORMULA] is the radial semi-amplitude of the white dwarf (assuming these emission lines originate near the primary component), and [FORMULA] and [FORMULA] indicate the phase and the phase offset of the binary system, respectively.

As we do not have blue calibration spectra from the first observing night our velocity determinations of He II were restricted to the spectra obtained in the last night.

There is a systematic shift in our velocity measurements: measurements from a trailed spectrum show a continuous drift to higher velocities. The reason for this is that the CCD chips are slightly warped. We account for this effect by fitting sky lines in the spectra from which we calculate the distortion. But the effect could not be eliminated in the blue spectral range, as there were only two weak sky lines which could not be used for a reasonable fit. Therefore, no correction was made in the blue spectral range. It should be noted that in spite of this systematic error our residuals are still low (Figs. 7 and 8).

[FIGURE] Fig. 7. Radial velocity curve determined by fitting Gaussians to the trailed spectra of [FORMULA]. The upper panel shows the residuals, the lower the distribution of the data. The solid line is the sine-fit to the data. For comparison the radial velocities from Tomov et al. (1998) are also plotted as asterisks (different symbols of our measurements refer to different trailed spectra).

[FIGURE] Fig. 8. Radial velocity curve determined by fitting Gaussians to the trailed spectra of He II . The upper panel shows the residuals, the lower the distribution of the data. The solid line is the sine-fit to the data. As the CCD is slightly warped we measure a systematic shift of the radial velocities within a single spectrum (see text) to higher velocities.

The sine-fits to the radial velocity curves of [FORMULA] and He II are given in Figs. 7 and 8, respectively. The deduced velocities are listed in Table 6. It is obvious, that there is a substantial difference between the radial velocities in [FORMULA] and He II : the semi-amplitude in He II is considerably higher. The difference in the radial velocity between the two lines can be explained by the blurred [FORMULA] emission (see Sect. 3.2). Our result for the radial velocity of He II [FORMULA] 4686 is [FORMULA] km s-1, which is in accordance with other values given in the literature (Becker et al. 1998, Beuermann et al. 1995).


Table 6. Orbital parameters for RXJ0019. [FORMULA] is the amplitude of the radial velocity and [FORMULA] denotes the system velocity in km s-1. [FORMULA] denotes the phase offset between photometric and spectroscopic phase (see Text). [FORMULA] designates the errors of the mentioned parameters. The system velocities are not consistent within their error bars. See text for an explanation.

We also observe a considerable difference in the system velocity between [FORMULA] and He II in our data. This effect is caused by the P-Cygni profiles: the blue wing of the [FORMULA] emission is truncated by the P-Cygni absorption and shifts the maximum of emission to slightly longer wavelengths. Thus, the system velocity in [FORMULA] differs notably from the velocity measured in He II . Compared with other values published in literature (Becker et al. 1998) our derived system velocities are much higher, especially for He II . For a reliable determination of the system velocity a dataset covering only a single orbital period is probably too short to eliminate statistical effects, which are caused by the varying influence of the jet emission. For comparison we show in Fig. 7 the velocities given in Tomov et al. (1998) in their Table 1. It is obvious that their system velocity for [FORMULA] is even higher. But due to the systematic shift in our distorted spectra our measured system velocity should not be given too much weight.

Photometric minimum and spectroscopic phase zero do not coincide. The photometric minimum occurs 0.11 and 0.17 in phase before spectroscopic phase zero for He II and [FORMULA], respectively. A similar shift for He II was already observed by Beuermann et al. (1995).

For further investigations we adopt our results of the semi-amplitude for the rest of this paper. The system velocities for the Doppler tomography are taken from Becker et al. (1998).

As the jet lines of [FORMULA] mimic the orbital velocity modulation of He II (Fig. 10), we assumed that the radial velocity modulation of He II is related to the primary and therefore represents the motion of the white dwarf. Knowing the orbital period and amplitude of the radial velocity variations of the mass accretor we can deduce the mass function of the secondary star. For the calculation we use the data from the He II emission line. The mass function is determined by


where [FORMULA] is the orbital period, [FORMULA] is the radial velocity of the primary, [FORMULA] and [FORMULA] denote the mass of the white dwarf and the secondary, respectively, and i is the inclination.

The mass function is obviously very low. If the emission lines of He II are indeed related to the compact object we can deduce the mass of the companion. In Fig. 9 we plot our derived mass function for various inclinations i. Because of the arguments given in Sect. 3.2, we assume that the inclination is reasonably high. We conservatively estimate that the inclination ranges between [FORMULA]. If the compact object is a white dwarf with [FORMULA], then we obtain [FORMULA] for the mass of the companion. The companion star is almost certainly a low-mass object. This could explain why we do not see any spectral features of the companion in our high resolution spectra. As low mass functions are also known from other SSS (e.g. Crampton et al. 1996, Hutchings et al. 1998), a low mass companion star may be common among this class of objects. A possible evolutionary scenario for low-mass secondaries in SSS is proposed by van Teeseling & King (1998).

[FIGURE] Fig. 9. Primary mass of RXJ0019 for various inclinations, i, plotted from the mass function [FORMULA]. For any reasonable inclination it is clear that the companion must be a low-mass star.

[FIGURE] Fig. 10. Zoomed grey scaled representations of continuum subtracted, photometrically calibrated, high resolution spectra of [FORMULA], [FORMULA], [FORMULA] (left side), He II [FORMULA] 4686 and He II [FORMULA] 4542 (right side) in v, [FORMULA]-coordinates. Not covered phases are marked with empty rows. In [FORMULA] the jets are clearly visible and show an orbital motion similar to the He II emission lines, whereas the main emission shows almost no orbital motion. The blue side of the Balmer lines are truncated by the P-Cygni absorption. In [FORMULA] complex substructures can be detected. The fragmented line appearing in the [FORMULA] spectra at a velocity of 1200 km s-1 is a remnant of a night sky line. Within the Helium lines a weak S-wave component is visible.

3.4. Doppler tomography

Doppler tomography is an useful tool to extract further information on the emission line origin from trailed spectra. This indirect imaging technique, which was developed by Marsh & Horne (1988), uses the velocity of emission lines at each phase to create a two-dimensional intensity image in velocity space coordinates [FORMULA]. The Doppler map can be interpreted as a projection of emitting regions in accreting binary systems onto the plane perpendicular to the observer's line of sight. The Doppler map is a function of the velocity [FORMULA], where the X-axis points from the white dwarf to the secondary and the Y-axis points in the direction of the secondary's motion.

An image pixel with given velocity coordinates [FORMULA] produces an S-wave with the radial velocity


where [FORMULA] and [FORMULA] denote the system velocity and the phase, respectively.

To accomplish this, a linear tomography algorithm, the Fourier-filtered back-projection (FFBP) is used, which is described in detail by Horne (1991). The resulting Doppler map (or tomogram) is displayed as a grey-scale image. To assist in interpreting Doppler maps we also mark the position of the secondary star and the ballistic trajectory of the gas stream. Assuming a Keplerian velocity field, emission originating in the inner parts of the disk has a larger velocity and thus appears in the outer regions of the map. Therefore, an image in such a representation is turned inside-out. For the tomogram analysis we restricted our investigation to the two Balmer lines [FORMULA] and [FORMULA] and the emission lines He II ([FORMULA] 4686) and He II ([FORMULA] 4542). No reasonable Doppler maps could be produced for other emission lines because of their poor S/N ratios.

Before computing the tomograms we subtracted the underlying continuum from the individual emission lines, since their line flux is the quantity needed to produce these maps. This was done by subtracting from each pixel row of the trailed spectra the corresponding median of the intensity. We achieved a FWHM resolution of 150 km s-1 in the central regions of the tomograms, whereas at higher velocities ([FORMULA] 800 km s-1) the resolution of the maps suffers a considerable degradation. Linear structures in the high velocity regions (Fig. 11) of the maps are sampling artifacts (aliasing streaks) which are not taken into account in the further interpretation. For a detailed discussion about sampling artifacts see e.g. Robinson et al. (1993).

[FIGURE] Fig. 11. Doppler-maps of [FORMULA], [FORMULA], He II [FORMULA] 4686 and He II [FORMULA] 4542 in velocity space ([FORMULA]), using the Fourier-filtered back-projection technique. The emission mainly originates in a "disk" centered at the center of mass, which extends to velocities of only 350 km s-1. The maximum of the emission of the Helium lines is shifted to [FORMULA] -100 km s-1 and [FORMULA] -100 km s-1. The dim ring in [FORMULA] is caused by the jet emission. From the distribution of the emission in the Doppler maps it is clear, that the emitting material is not located in an accretion disk.

In Fig. 10 we present the trailed spectra of the emission lines in V, [FORMULA]-coordinates; the Doppler tomograms are shown in Fig. 11. The schematic overlays in the Doppler maps represent the Roche-lobe of the companion star and the gas stream. They are a function of [FORMULA] and [FORMULA]. From our discussion in Sect. 3.3 we have taken 1[FORMULA] for the white dwarf and a 0.5[FORMULA] companion star. The center of mass and the location of the white dwarf are respectively marked by a cross and a point below the Roche lobe. The ballistic stream is represented by an arc originating from the secondary's Roche lobe at the inner Lagrangian point. This arc is marked every 0.1[FORMULA] (open circles, [FORMULA] is the distance from the center of the primary to the inner Lagrangian point) as it accelerates towards the compact object.

In the Balmer emission lines the blue side is truncated by the P-Cygni absorption. No S-wave structures can be detected at [FORMULA] and [FORMULA]. The jet lines of [FORMULA] clearly show an orbital motion which is in phase with the motion of He II . The main emission component of [FORMULA] reveals almost no orbital motion (compare Table 6). Therefore, we assume that the jet and the main emission do not have the same spatial origin.

The trailed spectra of both He II lines show more details: in each spectrogram a weak S-wave component within the main emission can be detected.

The Doppler tomograms (Fig. 11) of our data immediately reveal that the emission line distribution does not resemble a typical accretion disk. The usual shape of an accretion disk would appear as a dark extended ring in the inverse grey-scaled Doppler images, e.g. compare with Doppler maps of IP Peg obtained with the same technique in Wolf et al. (1998). Our observed projected velocities are smaller than 350 km s-1. For a binary system with parameters mentioned above and a disk radius of [FORMULA] cm ([FORMULA] is the volume radius of the Roche lobe of the white dwarf) one finds a Kepler velocity of 384 km s-1 at the rim of the disk. As the Kepler velocity increases rapidly within the accretion disk, the observed emission cannot originate within the disk if the inclination is high. For medium inclinations an origin only at the rim of the disk is possible. This conclusion is valid for a huge variety of system parameters (Fig. 12). Low inclinations are excluded (see discussion in Sect. 3.2).

[FIGURE] Fig. 12. Kepler velocity at the rim of the accretion disk at [FORMULA] ([FORMULA] is the volume radius of the Roche lobe of the white dwarf) over mass ratio q for various masses MWD of the white dwarf and inclinations [FORMULA] and [FORMULA]. For the observed velocities it is clear that the emission lines cannot originate within the accretion disk. For medium inclinations an origin only at the rim of the disk is possible.

The dim ring in the map of [FORMULA] at a velocity of [FORMULA] km s-1 is due to the emission from the jet lines. The Balmer lines originate at locations with almost no or only very low velocities. All of the observed flux is symmetrically distributed around the center of mass extending to velocities of roughly 350 km s-1.

The intense emission maximum is located near the center of mass. Part of this emission might come from the irradiated secondary, but as the maximum of the emission is clearly shifted to [FORMULA] km s-1 an irradiated secondary probably cannot account for all of the observed flux. The Doppler map of [FORMULA] looks similar. Again, the whole flux including the intense maximum is distributed symmetrically around the center of mass.

The Doppler maps from the Helium lines are slightly different. Comparable to the Balmer emission the observed flux originates at locations with very low velocities. The intensity distribution as a whole is distributed roughly symmetrical around the center of mass. But the emission maximum of the Helium lines is not centered at the center of mass. The coordinates of this intense maximum are shifted to approximately [FORMULA] [FORMULA] -100 km s-1 and [FORMULA] [FORMULA] -100 km s-1 in both maps. This emission maximum does not coincide with the gas stream trajectory. But due to its position in velocity space the emission might originate in the elevated and radially extended accretion disk rim as proposed by Meyer-Hofmeister et al. (1998). Thus, the Helium emission might partly be related to the outermost accretion disk rim. The He II ([FORMULA] 4686) map also indicates emission from regions near the white dwarf, which supports our assumptions in Sect. 3.3 for the estimation of the mass of the secondary.

Another indication about where this material might be located is given by the fact that the emission is roughly centered on the center of mass and not on the primary as would be expected for a typical accretion disk. This material might perhaps orbit the binary system around the mass center far outside of both Roche lobes. In a very simplified estimate one finds a distance of [FORMULA] cm for the above given binary parameters and a supposed velocity of 150 km s-1 for the emitting circum-binary material. This is about 3.5 times the distance of the binary separation.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1999

Online publication: March 1, 1999