Astron. Astrophys. 344, 459-471 (1999)

2. The sample of supersoft sources in M31

2.1. The SW-sample

15 firm candidate supersoft sources have been found in the 1991 ROSAT PSPC observations of M31 by Supper et al. (1997), cf. Greiner et al. (1997) by applying to the ROSAT PSPC hardness ratio HR1 which is defined as

with S = counts in channel 11-41 (roughly 0.1-0.4 keV), H = counts in channel 52-201 (roughly 0.5-2.1 keV). The selection criterion for a supersoft source is:

A 16-th supersoft source, a recurrent transient has been discovered by White et al. (1994). We call this sub-sample of 16 sources the Supper-White (SW) sample.

It has been shown by Kahabka (1998) that the individual spectral parameters giving information on the white dwarf masses of these 16 M31 supersoft sources can be constrained if the ROSAT PSPC hardness ratios HR1 and HR2 and the count rate as given in the catalog of Supper et al. (1997) are taken into account. The definition of HR2 is

with H1 = counts in channel 52-90 (roughly 0.5-0.9 keV), H2 = counts in channel 91-201 (roughly 0.9-2.0 keV). The hardness ratios HR1 and HR2 and the count rates have been compared with theoretical values derived using non-LTE white dwarf atmosphere spectra. As a result we found that for all these 16 sources the white dwarf masses were quite large .

In the present work non-LTE models of white dwarf atmospheres are used (Hartmann & Heise (1997)) extending to effective temperatures as low as . Absorbing hydrogen columns and effective temperatures have been determined in a - plane from the overlap of the 90% confidence parameter regions as determined from the HR1, HR2 and count rate constraints in that plane. In order to calculate the HR1-effective temperature planes somewhat reduced errors (0.85 errors) have been used. This has the effect of bounding the hydrogen column to which is consistent with a minimum due to the galactic foreground column. In Table 1 values are given for the SW-sample taking these new constraints into account. These values differ only slightly from those derived in Kahabka (1998). Count rates are given for the broad (0.1-2.4 keV) band. It should be noted that although the standard deviations for some sources are very large (which might suggest that these sources are not detected significantly) all these sources have been detected significantly in the soft (0.1-0.4 keV) band (see Table 5 of Supper et al. 1997). The high standard deviations in the 0.1-2.4 keV band are due to the fact that in this band most of the counts are background ones from the 0.4-2.4 keV range.

Table 1. ROSAT PSPC count rates (0.1-2.4 keV), hardness ratios HR1, from non-LTE white dwarf atmosphere models M4 and M5 derived absorbing hydrogen columns ), effective temperatures , white dwarf masses , index and tentative identification from the Supper et al. (1997) catalog (a = foreground star, e = SNR, * = bulge source).
Remarks:
quality flag [1] = full overlap of HR1, HR2, CPS contours, [2] = medium overlap of HR1 and CPS contours and overlap of HR2 contours considering uncertainties, [3] = full overlap of HR1 and CPS contours but no overlap of HR2 contour possibly due to source confusion in the hard band for the SW-sample of supersoft sources in M31 (Supper et al. 1997, White et al. 1994, cf. Greiner et al. 1997) and for the C-sample, [H] = histogram flag, entry in -histogram, sources identified as (a) or (e) have no entry in the black histogram, index and tentative identification from the Supper et al. (1997) catalog (a = foreground star, e = SNR, * = bulge source).

White dwarf masses are determined under the assumption that the source is on the stability line of surface hydrogen burning (cf. Iben 1982). The source may even be on the plateau of the Hertzsprung-Russell diagram with radius expansion. Then its mass would be even larger. This method has been applied to the Beppo-SAX observation of CAL87 and CAL83 and reasonable white dwarf masses of and have been derived respectively (Parmar et al. 1997, 1998).

2.2. The complimentary sample (C-sample)

The SW sample cannot be complete as it has been shown that supersoft sources are expected to cover a much wider range in HR1 (Kahabka 1998). Actually all values of HR1 in the range are possible in case the hottest (most massive) and more strongly absorbed () sources are included. Sources lying deep inside the galaxy disk or even located below the galaxy disk are expected to be at least in part even more strongly absorbed and are not covered by the selection criterion used for the SW sample. It may well be that part of this population is detectable but in order to investigate this point the calculations of model atmosphere spectra have to be extended to values in excess of the present upper bound of .

The SW-sample per definition has no correlation with either a foreground star nor a supernova remnant. We define a complementary sample (the C-sample) as the sample covering a much wider range of candidates fulfilling the conditions:

The C-sample comprises 26 objects and is given in Table 1. It turns out to contain 4 objects correlating with foreground stars and 4 with supernova remnants. If all identifications are correct then this sample reduces to 18 objects. We introduce quality flags (1=high, 2=medium and 3=low) to qualify the overlap of the HR1, HR2 and count rate constraints in the -plane. "High" means that all three contours overlap, "medium" means that the HR1 and the count rate contour overlap and the HR2 contour overlap within 3-, "low" means that the HR1 and the count rate contour overlap and the HR2 contour does not overlaps within 3-. Especially objects C31 and C33 which have quality flags L and M respectively and correlate with SNRs may be discarded. Object C36 shows the characteristics of a perfect candidate (all the H1, H2 and CPS contours overlap) and a correlation with a SNR may be by chance. C18, C20, C22 and C34 correlate with foreground stars but show contour overlap. If they are indeed stars then their temperatures must be very low (possibly M stars).

A discussion of the individual candidate sources of the C-sample is beyond the scope of this article. Interestingly source C26 is located in the bulge of M31 and (if the classification is correct) could harbor a very massive white dwarf very similar to the SWt transient (cf. Table 1). This source may be recurrent or/and very luminous. The latter point is confirmed by the high detected count rate of (cf. the SWt transient has a very similar count rate of ).

© European Southern Observatory (ESO) 1999

Online publication: March 18, 1999
helpdesk.link@springer.de