Astron. Astrophys. 336, 855-877 (1998)

2. ROSAT observations and preliminary analysis

2.1. HRI data

The field surrounding IC 4329A was observed with the ROSAT HRI on the 14th of January 1995 in essentially two separate observations for a total 15.5 ks. The data appeared essentially very clean, and only 200 seconds of data were removed on the basis of very high ( ct s^-1) and very low ( ct s^-1) count rates, and on large values of atomic oxygen column density along the line of sight.

Source detection and position determination were performed over the full field of view ( diameter) with the EXSAS local detect, map detect, and maximum likelihood algorithms (Zimmermann et al. 1992). Images of pixel size were used for the source detection. During the source detection procedure, only events detected in the raw HRI channels 2-8 were used, thus reducing the UV/cosmic ray background.

Sources accepted as detections were those with a likelihood L 8, and this gave rise to a number of sources. Probabilities P, are related to maximum likelihood values L, by the relation P. Thus a likelihood L of 8 corresponds to a Gaussian significance of 3.6 (Cruddace et al. 1988; Zimmermann et al. 1994). The source searching algorithms had difficulty in reliably detecting sources close to the bright central feature. Sources that were `detected' here generally appeared to be extended (i.e. they had a large value of extension likelihood), while possessing only a small number of counts. This is very suggestive of them being merely spurious features situated in the wings of the bright central feature. These sources were excluded from the analysis, leaving the 17 listed in Table 1.

Table 1. X-ray properties of point sources detected with the HRI (see text). Tabulated fluxes assume a 5 keV thermal bremsstrahlung model and a hydrogen column density of  cm^-2 (see text for conversion factors for different temperatures/models).

In an effort to improve the accuracy of the HRI source positions, the RA and Dec of five bright sources (H2, H7, H15, H16 and H17) were compared with the APM finding charts of Irwin et al. (1994). The sources associated with the two central galaxies IC 4329A (H11) and IC 4329 (H5), although they appeared to be very coincident, were not used in the re-aligning process, because of the extended nature of both their optical and X-ray emission. A very small offset of in right ascension and in declination is observed.

Table 1 lists the 17 detected HRI sources as follows: source number (column 1, prefixed by a `H' to distinguish from PSPC-detected sources, see Sect. 2.2), corrected right ascension and declination (columns 2 and 3), error on the source position (column 4, including a systematic attitude solution error), likelihood of existence (column 5), net counts and error (column 6) count rates and errors after applying deadtime and vignetting corrections (column 7), and 0.1-2.4 keV flux, assuming a 5 keV thermal bremsstrahlung model (column 8). Count rates of the HRI-detected point sources can be converted into fluxes (and into luminosities), assuming other additional spectral models, e.g. thermal bremsstrahlung and Raymond & Smith hot plasma models (with cosmic abundances) at temperatures of 0.3 keV and 3.0 keV. HRI Conversion factors (in units of erg cm^-2 cts^-1 s^-1) for these models, from count rate (in s^-1) into flux (corrected for Galactic absorption), are as follows: Thermal bremsstrahlung: 11.4 (0.3 keV) and 5.3 (3.0 keV); Raymond & Smith hot plasma: 11.0 (0.3 keV) and 5.2 (3.0 keV). As a final point, only source H5 is flagged as extended (at a likelihood of 173), with a FWHM of .

Fig. 1 shows the inner 2525 arcminute field of view (where all of the 17 sources listed in Table 1 are visible). Sources coincident with both IC 4329A (H11, with sources H9, H10, H12 and H13 in close proximity) and IC 4329 (H5) are visible, as is a further bright source (H1), to the southwest of IC 4329A.

Fig. 1. Grey-scale image of the X-ray flux seen with the ROSAT HRI over the central field. The image was constructed with a pixel size of , and smoothed with a Gaussian of FWHM . The point sources, as given in Table 1, are enclosed by boxes and numbered.

To investigate whether any residual, low surface brightness emission is visible, an adaptive filtering technique was used to create an intensity-dependent smoothed image. The signal-to-noise level of low surface brightness emission is boosted by smoothing lower and lower intensity sections of the image using Gaussians of progressively larger FWHM. Photons (again, from raw channels 2-8) were binned into an image of pixel size . Pixels of amplitude 1 (2,3,4,5,6,7,8) were smoothed with a Gaussian of FWHM . Pixels of amplitude greater than 8 remained unsmoothed, thus ensuring that the bright point sources were not smoothed into the background regions. The resultant image is shown as a contour map overlayed on an optical image in Fig. 2 (the optical image is taken from the U.K. Schmidt plate digitised sky survey).

Fig. 2. ROSAT HRI map of the IC 4329A field obtained using an adaptive filtering technique (see text), overlayed on a digitized sky survey image. Only channels 2-8 are used. The contour levels are at 2, 3, 5, 9, 15, 31, 63, 127, 255, 511, 1023 and 2047 ( being cts s-1 arcmin-2) above the background ( cts s-1 arcmin-2).

2.2. PSPC data

The field surrounding IC 4329A was observed with the ROSAT PSPC for a total of 8.23 ks, in five separate observation intervals between the 12th and the 14th of January 1993.

Source detection and position determination were performed over the full field of view as with the HRI data, using the EXSAS local detect, map detect, and maximum likelihood algorithms (Zimmermann et al. 1992). As with the HRI analysis, it is very doubtful whether many of the sources initially detected close to the bright central source, associated with IC 4329A, are true detections. Again, as with the HRI, those sources close to the bright central source with a large value of extension likelihood and a low number of counts were excluded from the analysis, leaving 22 sources.

As with the HRI analysis, an effort was made to correct the attitude solution. Comparisons were made between the positions of 4 bright point sources (P3, P11, P12, P14) and the APM finding charts (Irwin et al. 1994). The positions of the bright sources associated with the central galaxies were not used in the re-aligning, on account of their extended nature. An offset (purely in the north-south direction) of is seen.

22 sources were detected with a likelihood L 10 over the entire PSPC field of view. The 17 sources detected within the central area are listed below in Table 2 as follows: source number (column 1, prefixed by a `P' for PSPC), corrected right ascension and declination (columns 2, 3), error on the source position (column 4, including a systematic attitude solution error), likelihood of existence (column 5), net counts and error (column 6) count rates and errors after applying deadtime and vignetting corrections (column 7), and the (0.1-2.4 keV) flux, assuming a 5 keV thermal bremsstrahlung model (column 8). The only source flagged as extended is P6 (corresponding to HRI source H5; see Table 1), at a likelihood of 186, and with a FWHM of .

Table 2. X-ray properties of point sources detected with the PSPC (see text). Tabulated fluxes assume a 5 keV thermal bremsstrahlung model and a hydrogen column density of  cm^-2 (see text for conversion factors for different temperatures/models).

Fig. 3 shows a broad band (channels 8-235, corresponding approximately to 0.08-2.35 keV) contour image of the central region close to the centre of the field of view. All of the sources listed in Table 2 are visible including sources associated with both IC 4329A (P8) and IC 4329 (P6). Also shown in Fig. 3 are three smaller images, again showing the central emission, selected over three separate spectral bands - the `soft' band (channels 8-41), the `hard 1' band (channels 52-90) and the `hard 2' band (channels 91-201).

Fig. 3. ROSAT PSPC maps of the IC 4329A field in the broad (channels 8-235, corresponding approximately to 0.08-2.35 keV) band (main picture) and in the soft (channels 8-41), hard 1 (channels 52-90) and hard 2 (channels 91-201) bands (three smaller pictures). The contour levels in each figure are at 2, 3, 5, 9, 15, 31, 63, 127, 255, 511, 1023, 2047 and 4095 ( being (broad), (soft), (hard 1) and (hard 2) cts s-1 arcmin-2) above the background ( (broad), (soft), (hard 1) and (hard 2) cts s-1 arcmin-2). Source positions, as given in Table 2, are marked on the broad band image.

Comparison of Fig. 3 with Fig. 1 shows many sources visible in both the HRI and in the PSPC fields of view. Of the 9 HRI sources with a high value () of detection likelihood, every one has a PSPC counterpart. The converse is also true; all 8 of the PSPC sources with detection likelihoods greater than 27 (that lie within the HRI field of view), have a HRI counterpart. We can be very confident therefore, that all of these features are genuine point sources. The joint HRI/PSPC properties of these bright sources, that we concentrate on below, are summarised in Table 3 as follows: HRI source number (column 1), PSPC source number (column 2), offset (in arcsec) between the (attitude-corrected) HRI and PSPC positions (column 3), count rates and errors, after applying deadtime and vignetting corrections, for the HRI (column 4), the PSPC (column 5), and the PSPC analysis of M95 (column 6, - note that a different technique has been used here - see M95). Fluxes, corrected for Galactic absorption, are given in column 7 (HRI) and column 8 (PSPC) assuming in both cases a 5 keV thermal bremsstrahlung spectrum. It may appear that some discrepancy exists between the PSPC- and HRI-calculated fluxes, especially in the brighter sources. This could be due either to time variability (discussed in the next section) or to the assumption of the wrong spectral model (discussed in Sect. 3). The final column (column 9) of Table 3 gives the nearest bright optical counterpart, using the APM finding charts of Irwin et al. (1994); type (S-stellar, G-galaxy, F-faint), B magnitude, and offset (from the HRI position) in arcsec.

Table 3. X-ray properties of point sources detected with both the HRI and the PSPC (see text). Tabulated fluxes for both the HRI and PSPC assume a 5 keV thermal bremsstrahlung spectrum, assuming a hydrogen column density of  cm^-2. Note that for H1-P3, H5-P6 and H11-P8, further detailed spectral analysis has been performed (see Sect. 3). Note also that the PSPC source P4 is resolved by the HRI into two separate sources, H2 and H4.

2.3. Time variability study of point sources

A time variability study was performed for all of the sources detected in either the ROSAT HRI and PSPC fields of view, with special attention paid to the nine brightest sources (eight in the PSPC - see above) detected with both instruments.

For the HRI-detected sources, the complete observation was binned into five observation blocks (of between 2 and 4 ks). A maximum likelihood search at the source positions given in Table 1 was performed for each of the five observation blocks, the vignetting and deadtime corrected count rates (and errors) calculated within a cut radius of the PSF FWHM at the source positions. Where a source was not detected with a likelihood (corresponding to a Gaussian significance of ), a upper limit to the count rate was calculated. An essentially identical procedure was followed for the PSPC data, the observation binned again into five observation blocks (of ks each). A cut radius of the PSF FWHM was used. For the very bright source IC 4329A, the timing analyses described above were repeated for both the HRI and PSPC data, the observations being binned into 40 observation blocks.

Figs. 4 and 5 shows the results of this analysis for all the sources (except for the bright IC 4329A source) detected with both the HRI and the PSPC (Fig. 4 shows the results for the HRI, Fig. 5 for the PSPC). The layout of the figures is such that the coincident sources appear at the same position (hence H2 and H4 being coincident with P4). The results of the timing analysis for IC 4329A are shown in Fig. 6 (both HRI and PSPC).

Fig. 4. ROSAT HRI lightcurves of the eight bright point sources detected with both the HRI and the PSPC (excluding IC 4329A) from Table 3. Observation times are 1.5, 4.1, 4.1, 3.9, and 1.7 ks. Count rates are shown by filled squares with 1 error bars, and 2 upper limits are shown by open squares. Dashed lines indicate the average count rate calculated over the complete observation.

Fig. 5. ROSAT PSPC lightcurves of the seven bright point sources detected with both the HRI and the PSPC (excluding IC 4329A) from Table 3. Observation times are 1.6, 1.6, 1.6, 1.7, and 1.8 ks. Count rates are shown by filled squares with 1 error bars, and 2 upper limits are shown by open squares. Dashed lines indicate the average count rate calculated over the complete observation.

Fig. 6. ROSAT HRI (top) and PSPC (bottom) lightcurves of IC 4329A. The Observations were split into s (HRI) and s (PSPC) bins. Count rates are shown by filled squares with 1 error bars, and the dashed lines indicate the average count rate calculated over the complete observation.

Note that we are here interested in variations in the count rate from these sources, and not in the absolute values of these count rates, as these have been calculated earlier (Table 3). Hence we have been able to use small cut radii to avoid contamination from neighbouring sources and concentrate at the very centres of each source. Differences in the calculated count rates between these two methods appear to be negligible, except in the HRI cases involving the very bright, extended sources, where reductions in count rate, from those given in Table 3, can be seen. This is to be expected, given the methods involved.

It appears that little in the way of any temporal variability exists for any of the bright sources (including that associated with IC 4329A), and this is borne out by fitting the light curves to constant flux values. Lightcurve H2, in fact, shows the largest deviation from consistency, a constant flux level fitting the data with a reduced of 2.6. The H2 lightcurve therefore is variable only at the 2.1 significance level (or at likelihood of 3.42). Thus no significant variability is observed within any of the detected sources.

© European Southern Observatory (ESO) 1998

Online publication: July 27, 1998
helpdesk.link@springer.de