Astron. Astrophys. 349, 588-594 (1999)

3. Results

In Fig. 1 we show the optical lightcurve, provided by the American Association of Variable Star Observers and the Variable Star Network , of VW Hyi at the time of our X-ray observations. These optical observations show that our first BeppoSAX observation was obtained during an ordinary outburst that peaked on Sep 24, whereas observations 2-6 were obtained in quiescence. The last ordinary outbursts preceding our first BeppoSAX observation was observed by the AAVSO to peak on Sep 8; the first outburst observed after our last BeppoSAX observation was a superoutburst that started on Nov 5 and lasted until Nov 19.

Fig. 1. Optical and X-ray lightcurves of VW Hyi. The BeppoSAX data have been accumulated in bins of 1024 and 1536 seconds for the LECS and the MECS respectively. The BeppoSAX MECS lightcurve is shown for the full energy range 1.5-10 keV, and also for the hard energy range only 5-10 keV. The six observation intervals can clearly be distinguished. The first interval coincides with the decline from the optical outburst. Indicated by the dotted lines are the average count rates of the combined observations 4-6

3.1. Lightcurve

In Fig. 1 we also show the count rates detected with the BeppoSAX LECS and MECS. For the latter instrument we show the count rates separately for the full energy range 1.5-10 keV, and for the hard energies only in the range 5-10 keV. In both LECS and MECS the count rate is lower during the outburst than in quiescence. In quiescence the count rate decreases significantly between our second and third (only in the MECS data), and between the third and fourth observations (both LECS and MECS data), but is constant after that (see Table 1).

Table 1. Observation dates, exposure times and background subtracted count rates for the BeppoSAX LECS (0.1-10 keV) and MECS (1.5-10 keV)

The MECS count rate decreases during our first observation, when VW Hyi was in outburst, as is shown in more detail in Fig. 2. This decrease can be described as exponential decline with d. The count rates in the LECS are compatible with the same decline, but the errors are too large for an independent confirmation. The count rates at lower energies, 0.1-1.5 keV, are compatible with both a constant value and the exponential decay during our first observation.

Fig. 2. Close-up of the MECS observation 1 lightcurve. The solid curve shows the fitted exponential decay

3.2. Spectral fits

We have made spectral fits to the combined MECS and LECS data for each of the six separate BeppoSAX observations and computed the luminosities assuming a distance of 65 pc to VW Hyi (see Warner 1987). As expected on the basis of earlier work, described in the introduction, we find that the observed spectra cannot be fitted with a single-temperature plasma. The combination of spectra of optically thin plasmas at two different temperatures does provide acceptable fits. The parameters of these fits are listed in Table 2, and their variation between the separate observations is illustrated in Fig. 3. The need for a two-temperature fit is illustrated in Figs. 4 and 5 for the outburst spectrum of observation 1 and for the quiescent spectrum of the combined observations 3-6: the low temperature component is required to explain the excess flux near 1 keV. The Fe-K emission line near is clearly present in our data, and is due to hydrogen or helium like iron from the hot component of the plasma. The LECS data in observations 3-6 are poorly fitted above keV which is probably due to calibration uncertainties of the instrument (Fiore et al. 1999). We fix at , the best-fit value of the combined observation 3-6. (Fixing at , which was found by Polidan et al. (1990), does not change the fit parameters, except for the chi-squared values of observations 2, 3 and 3-6 which become slightly worse; 98, 111 and 158 respectively.)

Table 2. Fit results for a two-temperature plasma model and for a cooling flow model. The errors indicated are the 90% confidence intervals. The emission measures and luminosities have been calculated assuming a distance to VW Hyi of 65 pc (see Warner 1987)

Fig. 3. a-c Fit parameters with 90% confidence intervals for a two-temperature plasma model plotted against the observation number. The hot component is indicated by open circles. The parameter ranges for the spectrum of the combined observations 3-6 are indicated by the grey areas

Fig. 4. Count rate spectra of the LECS/MECS observation 1 (top panel) and of combined observations 3-6 (lower panel). The best spectral fits for a single-temperature spectrum and for a two-temperature spectrum are shown as dashed lines and solid lines, respectively. The excess due to the Fe-L emission line complex is made visible in the one-component fit. This excess is filled up by adding a second, cooler, plasma component. The residuals of the two-component fits are indicated as well

Fig. 5. In the top and middle panel are plotted the hot and cool plasma components used in the fit of observation 1, on a linear scale. In the bottom panel are plotted, on a logarithmic scale, the high (dashed line) and low (dashed-dotted line) temperature photon spectra and their sum (solid line), folded with a Gaussian representing the BeppoSAX spectral response function. This demonstrates that the Fe-M line emission near 0.1 keV and the Fe-L line emission near 1 keV of the cool component contributes significantly to the total photon spectrum

The temperature of both the cool and the hot component of the two-temperature plasma is higher during quiescence than during the outburst, increasing from respectively 0.7 keV and 3.2 keV in outburst to 1.3 keV and 6 keV in quiescence. The temperatures immediately after outburst - in our second observation - are intermediate between those of outburst and quiescence. The emission measure (i.e. the integral of the square of the electron density over the emission volume, ) of both the cool and the hot component of the two-temperature plasma is also higher in quiescence; immediately after outburst the emission measure of the hot component is higher than during the later phases of quiescence. The temperatures and emission measures of the two-temperature plasma are constant, within the errors, in the later phases of quiescence of our observations 3-6. For that reason, we have also fitted the combined data of these four observations to obtain better constraints on the fit parameters (see Table 2). Note that the decrease of the count rate between observations 3 and 4, mentioned in Sect. 3.1, is significant even though it is not reflected in the emission measures and luminosities of the two components separately. This is due to the combined spectral fitting of the LECS and the MECS, since the decrease in count rate is less significant for the LECS. Moreover, the errors on the count rates are much smaller than those on the emission measures ( respectively).

We fit the first 31 ksec and the next 46 ksec of the outburst spectrum (1a and 1b) separately. Both fits are good with . From the fit results we compute the MECS and ROSAT PSPC count rates. The results are shown in Table 3. We have only indicated the temperature and emission measure of the hot component since the cool component is responsible for the iron line emission outside the MECS bandwidth and does not have a large impact upon the continuum emission. Note from Table 3 that the decay in count rate is entirely due to the decrease of the emission measure.

Table 3. The spectral parameters for the first 31 ksec and next 46 ksec of the outburst spectrum. From these values the MECS and ROSAT PSPC count rates are predicted

To compare our observations with the results obtained by Wheatley et al. (1996) we consider next the cooling flow model (cf. Mushotzky & Szymkowiak 1988) for our observations 1, 2 and 3-6. In this model the emission measure for each temperature is restricted by the demand that it is proportional to the cooling time of the plasma. The results of the fits are shown in Table 2. Note that these results are not better than the two-temperature model fits. Due to the poor statistics of the LECS outburst observation we cannot constrain the lower temperature limit. The MECS is not sensitive to this temperature regime at all. A contour plot of the upper and lower temperature limits for the combined quiescent observations 3-6 is shown in Fig. 6. The boundaries of the low temperature in Fig. 6 are entirely determined by the Fe-L and Fe-M line emission; for a low temperature of keV the contributions to the line flux integrated over all higher temperatures exceeds the observed line flux. For a low temperature of keV there is not sufficient line flux left in the model. The boundaries of the high temperature are determined by the continuum slope; for a high temperature of and keV the model spectrum is too soft and too hard respectively to fit the data.

Fig. 6. Confidence contours plotted as a function of the upper and lower temperature for a cooling plasma during quiescence. See text. The contours represent the 68, 90 and 99% confidence levels of the fit to the observations 3-6

© European Southern Observatory (ESO) 1999

Online publication: September 2, 1999
helpdesk.link@springer.de