Astron. Astrophys. 319, 655-663 (1997)

2. New observations of G 18.95-1.1

2.1. X-ray observations

The pointed X-ray observations of the SNR G 18.95-1.1 were obtained 1991 March 30-31 with the PSPC detector aboard the ROSAT observatory. The effective collecting area including the PSPC efficiency is about at on-axis and decreases slightly with increasing field angle. The angular resolution is on-axis and increases to a few arcminutes for larger field angles. The energy resolution is . A detailed description of the mirror assembly and the PSPC can be found in the papers by Aschenbach (1988), Beckstette el al. (1988), Pfeffermann et al. (1986), and Trümper (1983) respectively.

The effective observation time was . Fig. 1 shows a grey-scale and a contour map of the X-ray emission. As was already pointed out by Aschenbach et al. (1991) based on ROSAT survey data, the dominant X-ray features delineate a partial shell of diameter. Because of the substantially longer exposure time the new data are of higher sensitivity than the previous observations. From the new data the X-ray analogon of the radio bar is partly detected (Fig. 9, Sect. 3.2). A point-like X-ray source appears at or , which has not yet been identified.

Fig. 1. Grey scale and contour map of the soft X-ray emission of G 18.95-1.1 as observed with the ROSAT satellite. The first contour and grey scale level is at . Further contours (grey levels) are apart (pixel size = ).

The spectral data of the entire source are shown in Fig. 2 and Fig. 3. In Fig. 2 the data are fitted to a thermal spectrum following Raymond & Smith (1977). The best fit () is obtained for and an interstellar absorption equivalent to . The corresponding observed X-ray flux density is . If corrected to the flux density is .

Fig. 2. Pulse height spectrum of the total emission from G 18.95-1.1 and the best fitting spectrum to the thermal emission of Raymond-Smith models with cosmic element abundances.

Fig. 3. Pulse height spectrum of the total emission from G 18.95-1.1 and the best fitting spectrum to a power law.

A similarly good fit () is obtained for a low temperature plasma: and . The corresponding X-ray flux density is and respectively.

The best fit () to a power law is shown in Fig. 3. The fit shows a reasonable spectrum but the corresponding photon index of () is very low if compared with for the Crab nebula. The fit to the data of G 18.95-1.1 seems to be unrealistic.

Assuming a photon index of -2.1, a combination of thermal and power law emission results in reasonable fits only for a fraction of the power law emission less than about .

We have also considered various regions of G 18.95-1.1 separately. In no case a nonthermal contribution was detected. The fits to a thermal spectrum do not show any significant variation from the mean value for the temperature as well as for the column density.

The interstellar column density derived from the X-ray spectral data may be compared with the HI column density towards G 18.95-1.1 derived from HI line observations at 21 cm wavelength with the 100-m telescope by Braunsfurth & Rohlfs (1984) (see Fürst et al. (1989) for details). We integrated the HI line temperature up to a distance of 2 kpc and converted the temperature to column density according to

(Kerr 1968),

where at the distance of . The result is shown in Fig. 4. It has been pointed out by Fürst et al. (1989) that G 18.95-1.1 is probably associated with a cavity in the interstellar HI distribution. From Fig. 4 we also derive an increase of the column density from at the centre to at the outer boundary of G 18.95-1.1. The difference of just 10% is much too small to become visible in the X-ray data (about 1/3 of the error quoted above). The HI line data also show that a column density of as is obtained for the low-temperature fit places G 18.95-1.1 beyond 15 kpc, which is the alternative kinematical distance of the radial velocity of at about Galactic longitude. Fürst et al. (1989) already excluded this large distance, since in that case the object would have very unusual properties. Hence, the low-temperature fit is not considered further on in this paper. Contrary to the low temperature fit, the high-temperature solution results in a HI column density, which equals the HI line column density at a kinematic distance of within the errors.

Fig. 4. Grey scale map of the HI column density (the grey scale starts at (white) and runs in steps of . Overlayed are contours of the soft X-ray emission (contours start at and are apart).

In Table 1 we have summarized the parameters.

Table 1. Results from pointed ROSAT Observations

2.2. Radio observations at 10.55 GHz

The first radio observations of G 18.95-1.1 at 10 GHz have been obtained by Fürst et al. (1985) with the Nobeyama 45-m telescope. These observations had about 2:07 of angular resolution (HPBW) and did not include the linear polarization. The integrated flux density was underestimated due to some missing large scale emission (Fürst et al. 1989). We used the Effelsberg 100-m telescope and the high-sensitive four-feed 10.55 GHz receiver system to map the source in total power and linear polarization at an angular resolution of . The observations have been made by moving the telescope in azimuthal direction along the four feeds. The 10.55 GHz system uses the "software beam switching" technique (Morsi & Reich 1986), which relies on extremely stable total power receivers. This has been shown to be the case for the 10.55 GHz receivers used for the observations (Reich 1995). The weather effects are cancelled out by computing differences between the signals received by the individual feeds. The restoration procedure introduced by Emerson et al. (1979) has been applied to obtain an equivalent single-beam map of the observed source. A detailed description of the 10.55 GHz receiver system, its main parameters, and the method of data reduction has been given by Schmidt et al. (1993). Some parameters relevant for the current observations are summarized in Table 2. As the source size is rather large, not only complete coverages but also sections of the source have been observed at different parallactic angles. In total five complete coverages of the source have been observed. The data reduction followed the standard procedures as described by Schmidt et al. (1993), except that in this case an individual adjustment of the zero level for each observation was required, since the surroundings of G 18.95-1.1 show strong Galactic emission.

Table 2. Parameters of the 10.55 GHz observation

The result of the 10.55 GHz observations is shown in Fig. 5, where the magnetic field direction is shown assuming negligible Faraday rotation.

Fig. 5. The radio contour map of G 18.95-1.1 at . The contours of total intensity run from in steps of . The magnetic field (E-field + ) is overlaid as bars with the length proportional to the polarized intensity ().

Fürst et al. (1989) have shown that two components contribute to the radio emission of G 18.95-1.1: A large-scale diffuse component and several arc-like features. The new observations confirm the earlier results: about of the radio emission at is concentrated in the diffuse component (Fig. 6). The two components have been derived from the data by using the method of unsharp masking (Sofue & Reich 1979). In the following we use a Gaussian shaped filtering beam with a size (HPBW) of 5', which keeps the background emission at rather smooth.

Fig. 6. Grey scale map of the small scale structure of G 18.95-1.1 at . The levels start from white () in steps of . Overlaid are contours of the diffuse radio component starting at (step ).

The integrated flux density of the whole object is , which fits well onto the radio spectrum between and (Fürst et al. 1985). Comparing the data at 2.695 GHz and 4.75 GHz ((Fürst et al. 1985) with the new 10.55 GHz map we detected a small shift in Galactic latitude of 1' in the 4.75 GHz map, the reason is unknown. After correction for this shift differential spectral index plots (TT-plots, see Turtle et al. 1962) between the new data and the data have been made for individual features of G 18.95-1.1. For that purpose the map was convolved to the resolution of the map. Subsequently both maps were separated into the diffuse and small scale components using filtering beam sizes of , 5', and 10' to test the dependence of the spectral index on the filtering beam size. Including the variation due to the different filtering beam sizes we obtain the following results: For the diffuse component the flux density spectral index () is . For the central bar we get and for the prominent northern arc we get . For all the other small scale structures we obtain . The filtering beam size of produces negative structures at 10.55 GHz between the central bar and the northern arc. The large beam of 10' leaves much of the diffuse emission in the map of the small scale component. The 5' beam yields the most realistic result. We show the corresponding TT-plots in Fig. 7.

Fig. 7. TT-Plots of various structures. The two lines for each data set represent 4.75 GHz (full) and 10.55 GHz (doted) as the independent variable. Alpha is the radio flux density spectral index (). The values given in the Fi gure are the average of the two fits and for the different filtering beams (see text).

There may be a systematic error because of calibration uncertainties and the small frequency difference. However, it is confirmed that the radio spectrum of the central bar is similar to that of the the diffuse component. The radio spectral index of the prominent northern arc and probably also of the other small scale structures is slightly steeper by about .

The integrated linear polarization of G 18.95-1.1 at is about to be compared with at . For the polarization percentage we derive the following values: at the central bar, at the arc near , and near . The polarization intensity is strongly concentrated on the small-scale structure of the remnant. If we subtract the emission from the diffuse component, the polarization degree increases to about for the central bar and the arc at . The magnetic field vectors are predominantly parallel to the direction of the arcs. Only across an area of about centered on , polarisation is found exceeding the low level structures visible outside the remnant, which can, therefore, undoubtedly be attributed to the diffuse component.

© European Southern Observatory (ESO) 1997

Online publication: July 3, 1998
helpdesk@link.springer.de