3 Source classes

In Sect. 3.1 we discuss HRI sources which were identified either with sources already known from literature or with candidates which were found in former X-ray studies and in PSPC data (HP99b). Section 3.2 deals with new classification of HRI sources based on their X-ray properties.

  
3.1 Source identification

For 97 HRI sources out of 138 which were also detected by the PSPC the HRI observation yielded smaller positional error circles and consequently more accurate source positions compared to the PSPC results. Therefore for several sources likely optical counterparts could be determined which was not possible only with PSPC data. 94 HRI sources were identified with known objects in the LMC, foreground stars, or background objects mainly based on their position (see Sect.2.3). As they comprise different source types X-ray properties characteristic for each source class could be derived from HRI and PSPC data. Table 2 lists HRI sources with identification.

HP99b have shown that extent and extent likelihood as well as the hardness ratios measured by the PSPC have characteristic values for different source classes and can be used as classification criteria.

Figure 5: Source extent and extent likelihood of HRI sources in the LMC. SNRs are marked with open squares, known point sources with open triangles. Crossed symbols are candidates for SNRs or point sources known from literature, filled symbols are new classifications

In Fig.5 extent and extent likelihood of the HRI sources are shown. The extent was calculated in the maximum likelihood algorithm and so gives the value resulting from fitting Gaussians. Thus in some cases it may not be the extent of the whole source but only of knots which were found within the extended source. Identified SNRs, marked with open squares, are distributed in the region with large extent and high extent likelihood. Crossed squares indicate known SNR candidates and filled squares sources newly classified as SNR candidates in this work. Point sources have lower extent likelihood unless they were extremely bright like AB Dor (No. 180), LMC X-1 (No. 311), or RX J0439.8-6809 (No. 4) where the deviation of the point spread function from the assumed Gaussian profile becomes significant.

3.1.1 Foreground stars

By cross-correlating the HRI source catalogue with SIMBAD and TYCHO catalogues and using the finding charts presented by Schmidtke et al. (1994, hereafter SCF94), Cowley at al. (1997, hereafter CSM97), and Schmidtke et al. (1999, hereafter SCC99) 39 sources were identified with Galactic foreground stars (Table 2). Most of them could also be identified with the help of UBV photometry results presented by Gochermann et al. (1993) and Grothues et al. (1997). On DSS-images there are point sources as very likely optical counterparts at the positions of these HRI sources within the error circle.

Based on hardness ratios of the PSPC observations two point sources were suggested as foreground star candidates by HP99b (Nos. 189 and 349). They were detected in PSPC images and their hardness ratios are within the range characteristic of stars (HP99b). DSS images show an optical point source within the improved HRI error circle in both cases.

Figure 6: Flux ratio log( $f_{\rm x}/f_{\rm opt}$) as a function of hardness ratio 1. Open squares are SNRs, open lozenges stars, open hexagons XBs, open triangles AGN, and asterisks SSSs. Crossed symbols are already known candidates and filled symbols are new classifications

3.1.2 Supernova remnants

Most SNRs in the LMC are extended X-ray sources which could be resolved by the HRI. They typically show extents of about 5 $^{\prime\prime}$ - 20 $^{\prime\prime}$ and high extent likelihood (> 10.0). A total of 24 known SNRs were observed by the HRI, four HRI sources are identified with known SNR candidates (Nos. 50, 231, 310, and 315). For both Nos. 231 and 310 the measured hardness ratios are typical for SNRs. No. 50 has a harder X-ray spectrum with HR1 = 1.00 $\pm$ 0.10 and HR2 = 0.34 $\pm$ 0.07.

3.1.3 Supersoft sources

SSSs have very soft X-ray spectra and so far seven SSSs have been discovered in the LMC (HP99b). Two of them were sources of the Einstein LMC survey (Long et al. 1981) and five were found with the help of the ROSAT PSPC. In the HRI pointings five LMC SSSs listed in Table 2 were observed and detected with high existence likelihood.

3.1.4 X-ray binaries

Characteristic for most X-ray binaries is the hard X-ray spectrum and flux variability. In HRI observations nine bright sources could be identified with well known massive X-ray binaries (HMXB). The point source RX J0532.7-6926, here No. 238, has been suggested to be a low mass X-ray binary (LMXB) candidate by Haberl & Pietsch (1999a, hereafter HP99a) and was also detected by the HRI. In HP99a a lightcurve with PSPC and HRI measurements is presented and variability is discussed in detail. Between 1990 and 1993 the source showed an exponential intensity decay.

  

Table 2: Identified HRI sources in the LMC

$\parbox{250mm}{ \begin{tabular}{rrrrrccrrccl} \hline\noalign{\smalls... ...5V HD 40156 \\ \noalign{\smallskip} \hline \noalign{\smallskip} \end{tabular}}$ $\parbox{250mm}{ \vspace{2mm} \noindent N... ...eviations for references in square brackets are given in the literature list. }$

3.1.5 AGN

Nine known background AGN with redshifts between 0.06 and 0.44 (SCF94; CSM97; Crampton et al. 1997) were re-identified in the HRI pointings. Because of its positional coincidence with the radio source PKS 0552-640 and its hardness ratios measured by the PSPC the HRI source No. 389 was classified as AGN candidate (No. 37 in HP99b). On the DSS frame an optical source with m_B = 16.3 within the HRI error circle is identified as the most likely optical counterpart.

 

Table 2: continued

$\parbox{250mm}{ \begin{tabular}{rrrrrccrrccl} \hline\noalign{\smalls... ... & HMXB RX J0544.1--7100 [HP99b] \\ \noalign{\smallskip} \hline \end{tabular}}$

 

Table 2: continued

$\parbox{250mm}{ \begin{tabular}{rrrrrccrrccl} \hline\noalign{\smalls... ....12 & AGN? PKS 0552--640 [HP99b] \\ \noalign{\smallskip} \hline \end{tabular}}$

  

Table 3: Classified HRI sources

$\parbox{250mm}{ \begin{tabular}{rrrrrccrrrccl} \hline\noalign{\small... ... XB$>$\space or $<$ AGN$>$\space \\ \noalign{\smallskip} \hline \end{tabular}}$

  
3.2 New classifications

The extensive detection list produced from the HRI pointings towards the LMC allowed us to search for new candidates for different source types. In the course of studying the newly discovered HRI sources the following parameters were of prime importance: count rates, source extent, extent likelihood, flux variability, and counterparts in other wavelengths.

In addition to these X-ray properties we calculated the X-ray to optical flux ratio of HRI sources, for which possible optical counterparts could be found. The flux ratio was computed according to the equation $\log(f_{\rm x}/f_{\rm opt}) = \log$ (3 $\cdot$ HRI counts/s $\cdot\ 10^{-11}$) + 0.4 m_B + 5.37 (Maccacaro et al. 1988; HP99b). The relation used for PSPC observations in HP99b was applied here for HRI sources converting the HRI count rates to PSPC count rates by multiplying by the factor of 3 which is typical for hard sources. B magnitudes from the USNO-A1.0 Catalogue produced by the United States Naval Observatory (Monet 1996) were used. For several sources the optical counterpart could not be determined uniquely. In such a case the magnitude of the brightest optical object within the error circle was used resulting in lower limits for log( $f_{\rm x}/f_{\rm opt}$). For the SNRs log( $f_{\rm x}/f_{\rm opt}$) in general gives no quantitative information, but is an indicator that this source class is bright in X-ray (log( $f_{\rm x}/f_{\rm opt}$) > -1).

As one can see in Fig.6 stars in general have negative log( $f_{\rm x}/f_{\rm opt}$), for AGN it is around zero, and for SSSs and XBs it is mostly positive in particular when they were observed in their X-ray active phase. Combination of $f_{\rm x}/f_{\rm opt}$ and the hardness ratios provides a tool to exclude foreground stars.

Newly discovered HRI sources which are suggested as candidates for different source classes in this work can be found in Table 3 and are discussed in the following.

3.2.1 SNR candidates

Investigating the extent five HRI sources (Nos. 197, 284, 288, 307, 338) not classified with the help of PSPC observations are suggested as SNR candidates as their extent is larger than 8 $^{\prime\prime}$ (see Fig.5). Since they were not detected by the PSPC because of short exposure times there is no spectral information about these sources which might be crucial for further improvement of the classification.

  
3.2.2 Sources classified as stellar

For 11 HRI sources probable optical counterparts were found within the error circle which are all bright ( $m_{B} \leq 12.5$), and their log( $f_{\rm x}/f_{\rm opt}$) is negative (< -2.0). For this reason these sources are classified as stellar objects, and in particular the brightest objects are likely foreground stars. Four sources were also observed by the PSPC (Nos. 90, 135, 217, 313), but as the errors of their hardness ratios are large, no spectral information is given.

The lightcurve of No. 300 shows a strong decrease of the X-ray emission with a factor of 10 in 2 years indicating that the HRI observations were performed after an emission maximum. The point source in the optical DSS image at the HRI position is very likely the optical counterpart with a B magnitude of m_B = 12.4 according to the USNO-A1.0 Catalogue and log( $f_{\rm x}/f_{\rm opt}) = -2.75$.

3.2.3 LMC stars as candidates for high mass X-ray binaries

Two X-ray point sources detected by the HRI were identified with known LMC O and B stars (No. 328, Sanduleak 1970, m_B = 18.8 and No. 332, Brunet et al. 1975, m_B = 13.6) because of the positional coincidence. With HRI data no variability investigations could be carried out for these X-ray sources, though there exist many pointings in their direction, because they were both detected only once and in other pointings the upper limits were too high for this purpose. But their identification with optically selected LMC stars allows us to classify them as candidates for high mass X-ray binaries.

Figure 7: Identified HRI sources are plotted on a grey scale PSPC image (0.1 - 2.4 keV) of the LMC. Squares are SNRs, circles are foreground stars, double squares SSSs, crossed squares XBs, crossed circles AGN. Candidates from literature are included for each source class

Figure 8: The distribution of unidentified HRI sources and new source classifications is shown. Unidentified HRI sources are plotted as dots, squares are sources classified as SNR candidates, circles as stellar sources, crossed squares as XB candidates and double circles as candidates for XB or AGN

   
3.2.4 Sources with hard X-ray spectrum: Candidates for AGN or X-ray binary

With the help of the hardness ratios and other characteristics measured by the HRI like flux variability or $f_{\rm x}/f_{\rm opt}$three HRI sources which were also detected by the PSPC could be classified as candidates either for X-ray binary or for AGN.

The point source No. 49 shows significant flux variations, as it is shown in Fig.3, and has a hard and/or highly absorbed X-ray spectrum (HR1 = 1.00 $\pm$ 0.71, HR2 = 0.26 $\pm$ 0.16). On the DSS image a likely optical counterpart with a B magnitude of 16.4 (according to the USNO-A1.0 Catalogue) is found. Therefore this source has been classified as a candidate either for an X-ray binary or AGN.

Sources Nos. 230 and 364 are further candidates for X-ray binary or AGN as they have a hard and/or absorbed X-ray spectrum (HR1 = 1.00 $\pm$ 0.35, HR2 = 1.00 $\pm$ 0.98 and HR1 = 1.00 $\pm$ 0.21, HR2 = 1.00 $\pm$ 0.60 respectively). Since source No. 230 has a small positional error a probable optical counterpart can be found on the DSS image. This counterpart is faint ( m_B = 22.6), and we obtain a high log( $f_{\rm x}/f_{\rm opt}$) of 1.56. For source No. 364 there is a relatively faint optical source ( m_B = 18.2) within the error circle which might be the counterpart (log( $f_{\rm x}/f_{\rm opt}) = 0.43$).

Another nine sources detected by the HRI were identified with sources in the PSPC catalogue (HP99b) showing a hard X-ray spectrum. But from the HRI observations no additional information could be obtained. Thus the HRI sources are simply classified as hard X-ray sources because of the hardness ratios of their PSPC detections.

3.3 Source distribution

Due to the high spatial resolution of the HRI many sources could be detected both in the outer regions and in the optical bar region of the LMC. In Fig.7 HRI sources identified with known objects and known candidates are plotted on a grey scale PSPC image (0.1 - 2.4 keV) of the LMC (from HP99b). The sources are located in different regions of the LMC and show no spatial preferences, it is not only background AGN or foreground stars and candidates which are distributed over the whole LMC region. There are still more than 250 non-identified point sources which are homogeneously distributed in all LMC regions which were covered by ROSAT HRI pointings as it is shown in Fig.8. In contrast, in PSPC observations not many additional sources could be detected in the regions with strong diffuse emission, because the lower spatial resolution hindered in distinguishing between extended and point like emission (HP99b).

The HRI allows to study the extent of the sources to scales of arcseconds. Therefore SNR candidates could be found not only in regions without surrounding diffuse emission. Four out of five newly suggested SNR candidates are located in regions with diffuse emission between 30 Dor and LMC X-1 (see Fig.8).

Within and around the optical bar region several new stellar sources and candidates for X-ray binary or AGN were found.