2. The BeppoSAX observation
After the optical outburst of U Sco was reported (Schmeer et al. 1999) a target of opportunity observation of U Sco was performed with the BeppoSAX X-ray satellite. According to the calculations of Kato (1996), supersoft (SSS) X-ray emission is predicted to be observed 10-60 days after the optical outburst. The 50 ks exposure observation was performed during 1999 March 16.214-17.425, 19-20 days after the optical outburst. Here we report the first results of an analysis of the mean X-ray spectrum observed during this observation.
The scientific payload of BeppoSAX (see Boella et al. 1997a) comprises four coaligned Narrow Field Instruments including the LECS (Parmar et al. 1997) and MECS (Boella et al. 1997b). U Sco was detected with mean LECS and MECS net count rates, after background subtraction, of and respectively. The source was not detected in the high-energy non-imaging instruments. The X-ray flux varies by a factor of 1.5 during the observation possibly due to orbital variations or a rise in flux.
The combined LECS and MECS spectrum was first fit with a simple blackbody spectral model. The fit is unacceptable with a of 72 for 10 degrees of freedom (dof). We then added absorption edges due to highly ionized species of N and O expected in the hot atmosphere of a steadily nuclear burning WD. The edge energies were fixed at 0.55 keV, 0.67 keV, 0.74 keV, and 0.87 keV, corresponding to the Lyman edges of N VI , N VII , O VII , and O VIII , respectively. Only the NVI , NVII , and OVIII edges were detected at high significance with absorption depths of 4.3, 2.4, and 5.6, respectively. The OVII edge is not detected and the 90% confidence upper limit to its absorption depth is 1.6. The is 12 for 6 dof. However, other interpretations of the spectral shape above 0.8 keV appear to be more likely (see below and the discussion). We independently fitted the edge energies of the N VI and N VII features and derived 90% confidence ranges of 0.524-0.555 keV and 0.630-0.669 keV, respectively and an absorbing hydrogen column density .
WD atmosphere spectra have been shown to deviate strongly from simple blackbodies (e.g., Hartmann & Heise 1997). The use of sophisticated WD atmosphere model spectra is required. We applied a non-LTE WD atmosphere spectral model grid assuming a very massive (log(g)=9.5) WD with cosmic CNO abundances (see e.g., Hartmann et al. 1999). The fit was unacceptable at energies 0.8 keV. We added an optically thin thermal component (Raymond & Smith 1977), hereafter RS, to the model. Such a component may be due to a strong wind from the WD atmosphere and has been observed in the classical nova Cyg 1992 (Balman et al. 1998). The fit was still unacceptable with a of 23.4 for 8 dof. We also fitted the observed spectrum with two optically thin RS components. We found that the fit was not acceptable with a of 55.8 for 8 dof.
When the CNO cycle is active then the N/C and O/C ratios are strongly modified. A strong enrichment of N with respect to C is expected as N is involved in the slowest reaction. We calculated log g=9.5 non-LTE WD atmosphere spectral models with He and CNO (number) abundances (, , , with respect to helium) according to values determined from optical/UV studies of the nova ejecta of U Sco (Williams et al. 1981). In addition, we applied a hot optically thin thermal component. We found that with these assumptions the fit was acceptable with a of 10.7 for 8 dof. The best-fit atmospheric temperature is (90% confidence), the atmospheric radius is , and the bolometric luminosity . For the optically thin component we derive a temperature, kT, of 0.22-0.52 keV and an emission measure, EM, of assuming that He is enriched and N/C enhanced. The absorbing hydrogen column density is . This value is larger than the galactic absorption in the direction of U Sco of (see Introduction) indicating a substantial intrinsic absorption.
© European Southern Observatory (ESO) 1999
Online publication: June 6, 1999