Forum Springer Astron. Astrophys.
Forum Whats New Search Orders

Astron. Astrophys. 342, 839-853 (1999)

Previous Section Next Section Title Page Table of Contents

3. Results

3.1. X-ray emission morphology

The two PSPC images are pointed toward the rim of the Vela shell, and therefore they are the least affected by the pulsar emission. To properly describe and classify the X-ray emission morphology, we divided the shell in three main regions, according to the observed surface brightness [FORMULA] in the energy band 0.2-2.0 keV: Clumps: [FORMULA] cnt s-1 arcmin- 2. The clumps are the brightest regions in Fig. 1, and are irregularly distributed in the Vela SNR. They have been interpreted as tracer of inhomogeneities (e.g. Aschenbach 1993).

Diffuse emission: [FORMULA] cnt s-1 arcmin-2. This range of [FORMULA] identifies the diffuse emission probably associated with the propagation of the shock in a medium less dense than the clumps, which are embedded in it.

Background: [FORMULA] cnt s-1 arcmin- 2. This is considered background emission, although it could not be necessarily associated with the cosmic background. Aschenbach (1993) pointed out that part of this emission could be associated with the remnant itself (see also Paper I).

In Paper I two major clumps were identified in the image of RP200133, named zone A (bin [FORMULA], [FORMULA], [FORMULA] [FORMULA] in the left panel of Fig. 2) and zone B (bins [FORMULA] and [FORMULA]). In the outermost rings (6 and 7) few weaker structures are visible, like the filament in [FORMULA] (here classified as diffuse emission). For further details on the X-ray morphology of the part of the shell in RP200133, see Paper I.

The X-ray surface brightness in the image of RP500015 (Fig. 2, right panel) has a spatial distribution quite different from that of RP200133. In particular, it is possible to locate two separate "fronts", namely Front 1, the brightest of the two (classified as clump), placed nearer to the center of the remnant, and Front 2 (classified as diffuse emission), more external. Both fronts partially follow the grid curvature (in sector h, i and j, ring 1 for the Front 1 and ring 4 for the Front 2), and they show features at smaller scales. Between the two fronts, the surface brightness is low (typically classified as diffuse emission), except in sector k and j, where a bridge links Front 1 with Front 2. Like in the image of RP200133, most of the spatial bins in ring 7 include only background emission.

In the bin [FORMULA], there is a strong X-ray enhancement which we shall call Filament D (FilD). In FilD, the observed count rate is 3-5 times higher than in surrounding regions. This feature is almost aligned in the North-South direction with an extension of [FORMULA], while its dimension in the East-West direction is rather small ([FORMULA]), comparable with the dimension of the instrumental Point Spread Function: therefore, the observed spatial structure of the FilD X-ray emission is consistent with a filamentary origin. A closer inspection shows that the brightness of FilD along the North-South direction is not uniform, suggesting the presence of finer structures.

Another noticeable feature is the clump in [FORMULA], which is bounded by the edge of the FOV. The observed count rate here is 1.5 times the one measured in FilD, and [FORMULA] times the rate in the region between Front 1 and 2.

3.2. Spectral fitting results

3.2.1. General considerations

We shall present here the spectral analysis of both RP200133 and RP500015 with an optically thin plasma emission model (Raymond & Smith 1977) and two temperature components, both in CIE conditions (hereafter called RS 2T CIE model, or simply RS 2T).

In Paper I, we performed a preliminary analysis of RP200133 with the RS 2T CIE model, and we showed that, notwithstanding the goodness of the fits, there were some topics left open. In particular, we found an apparent variation of the hydrogen column density across the image. This [FORMULA] gradient is suspicious because self-absorption should be negligible and known molecular clouds lie behind the remnant itself and hence cannot contribute to the absorption. We have now explored the possibility that this results is due instead to an intrinsic spectral variation.

To understand how this effect might occur, consider the following situation: let bin A be in CIE conditions with two component of different temperatures ([FORMULA]) and same emission measures ([FORMULA]), and consider an adjacent bin B having the same two temperatures (i.e. no T gradient between A and B) but [FORMULA]. For typical temperatures found in the Vela SNR ([FORMULA] keV and [FORMULA] keV, Paper I) the first component affects the PSPC spectrum mostly in the [FORMULA] keV, and this is also true for the interstellar absorption. Because of the limitations due to photon counting statistics and/or moderate spectral resolution, the spectral variations between bin A and B could be interpreted as a fake increase of the [FORMULA] value rather than a decrease of [FORMULA].

The grid layout now adopted (the same used also in Paper II) allows us to collect a number of counts per spatial bin greater than the number previously obtained with the grid adopted in Paper I, and to test in detail if the observed [FORMULA] variations are real or not. The number of counts in the spatial bins of RP200133 have been reported in Paper II and is typically about [FORMULA] with a maximum of [FORMULA]. In RP500015 we have typically [FORMULA] counts in each bin with a maximum of [FORMULA] counts. We have performed the RS 2T fits of the spectra collected in each spatial bin of the two images treating the interstellar absorption as a fixed parameter. This emission model has 4 free parameters (two temperatures and two normalization factors) and it yields 24 degrees of freedom (PSPC SASS channels 3-30 are used, corresponding to the 0.2-2.0 keV energy range). We have fixed [FORMULA] to [FORMULA] cm-2 for both data sets, which is a value compatible with the measurement reported in Paper I and in Paper II, and it is also in agreement with the previous estimates of Kahn et al. (1985). In Fig. 3, we show, as an example, the PSPC spectrum of the spatial bin [FORMULA] ([FORMULA] counts) which contains the "Filament D" feature, and its best-fit 2T CIE model.

[FIGURE] Fig. 3. The PSPC spectrum of the spatial bin [FORMULA] with its best-fit two-temperature CIE model.

We stress that, by removing the interstellar absorption from the list of free fit parameters, the total number of free parameters is equal to the number of free parameters of the STNEI model used in Paper II. From a statistical point of view, this implies that the results here obtained with the fixed [FORMULA] RS 2T model and those obtained with the STNEI model in Paper II can be directly compared in terms of [FORMULA] goodness of the fit.

In the following, we use the symbol F to indicate the normalization factor [FORMULA] (which is usually given by the fit) and EM to indicate the emission measure [FORMULA].

3.2.2. Plasma temperatures and emission measures

In Table 2 we report the success rates of the fitting with the RS 2T CIE model in the 48 spatial bins of RP200133 and in the 56 spatial bins of RP500015, along with those results obtained in Paper II for RP200133 using the STNEI model and the ones obtained in Paper I with the RS 1T model. The table shows also the STNEI and RS 1T success rates for RP500015. The 4-parameter RS 2T model provides an acceptable [FORMULA] in most of the spatial bins (98% in RP200133 and 88% in RP500015) 1.


Table 2. Fitting success rates of RP200133 and RP500015 with various emission models
a The number of bin with acceptable [FORMULA] and the corresponding percentage.
b For these fits, [FORMULA] was fixed to [FORMULA] cm-2.

This result implies that data are consistent with a single [FORMULA] absorption value across the two images.

In Fig. 4, we report for both datasets the [FORMULA] and [FORMULA] radial profiles along each sector of the spatial grid shown in Fig. 2. The distance from the center of the remnant is reported in abscissa. For [FORMULA], we report the profiles for each sector, whereas for [FORMULA], given its larger uncertainties, we report a profile which is an averaged over all sectors, together with a plot of the expected Sedov profile derived following Landau & Lifshitz (1974). There are significant variations up to a factor of 2 between the best-fit [FORMULA] temperatures measured in the same ring but in different sectors (e.g. [FORMULA] keV and 0.08 keV in RP200133 ring 5 at [FORMULA] pc) and in different rings of the same sector ([FORMULA] and 0.11 in RP200133 ring 4 and 5 at 17-18 pc, sector b). On the other hand, from Fig. 4, it is evident that the value [FORMULA] is statistically consistent with values measured in all the spatial bins. Moreover, the ratio of the emission measures spans a range of about two orders of magnitude ([FORMULA]) and this characteristic accounts for most of the observed spectral shapes.

[FIGURE] Fig. 4a-d. [FORMULA] and [FORMULA] profiles for the two PSPC observations. We have excluded bins [FORMULA], [FORMULA], [FORMULA] and [FORMULA] because they contain less than 1000 counts. For [FORMULA], we report the profiles for each sector, while for [FORMULA] the profiles are the average of the measurements in all the sectors. We also report in the [FORMULA] plots, the expected auto-similar Sedov profiles computed assuming typical characteristic parameters of the Vela SNR.

Fig. 5, which reports the T and [FORMULA] histogram distributions, shows the clear separation between the high and low temperature component.

[FIGURE] Fig. 5a-d. [FORMULA], [FORMULA], and emission measure ratio histograms for RP200133 and RP500015 spatial bins. In the temperature histograms, dashed lines refer to the X1 component, solid lines to the X2 component.

3.2.3. X-ray emission characteristic of Filament D

The X-ray clump named Filament D (FilD) is very peculiar, because it is isolated and it has a filamentary nature. We have performed a spectral analysis of the FilD using an extraction region with a shape matching its morphology. For this analysis, we have collected the background spectrum in a region classified as "diffuse emission" (according to Sect. 3.1) adjacent to FilD. In this way, we aim to study the peculiar characteristics of the plasma in the filament. The background subtracted spectrum has a good counting statistics ([FORMULA] counts). In Table 3, we report the fitting results for FilD. Both the 1T STNEI and the 2T RS CIE model fits are acceptable. In particular, the NEI fit yields a reasonable value of the ionization time ([FORMULA] yr cm-3), while the 2T fit yields a best-fit emission measure ratio ([FORMULA]) significantly lower than the one measured in the other spatial bins. This result is due to our choice of subtracting the diffuse emission from the FilD spectrum, and indicates that there is relatively more cool plasma in the filament with respect to the surrounding regions. The [FORMULA] case (corresponding to 1T RS CIE fit) is also in the allowed emission measure range. The difference between the [FORMULA] value in the 2T RS fit and the (higher) value in the 1T NEI fit is a measure of the error associated to the use of a CIE model instead of the NEI one. The fitting results of the whole bin [FORMULA] with the 2T STNEI model is also reported in Table 3, and described in the following section.


Table 3. Best-fit parameters and derived quantities for the FilD regions
a The density was computed using Eq. 4 in Paper I and assuming a line of sight l of [FORMULA] pc, corresponding to a depth approximately equal to the N-S extension of the FilD X-ray emission. The reported uncertainty is due only to the uncertainty on F (N also scales as [FORMULA]).
b The best-fit [FORMULA] is [FORMULA], with [FORMULA] in the acceptability range.
c These results were derived from a fit to a spectrum collected in bin [FORMULA], which includes FilD and part of its diffuse emission environment, in order to increase the total number of counts in the spectrum.
d The reported value is [FORMULA]. [FORMULA] is [FORMULA].
e F2 [FORMULA] F1/50.

3.2.4. NEI effects

In Paper II, we stressed that a description of the Vela SNR X-ray emission in terms of a NEI emission model is necessary for a determination of the temperature and density in the shell. We showed the expected ionization time ([FORMULA]) of the plasma in the shell is in the range [FORMULA] yr cm-3, i.e. far from the equilibrium condition ([FORMULA] yr cm-3). On the other hand, we have pointed out that a single-temperature description of the X-ray emission yields several interpretation problems, and that the 2T nature of the plasma is de facto a growing observational evidence. There is no reason to believe that the two plasma components are both in CIE conditions; instead, they may well be in NEI conditions, but, unfortunately, a spectral analysis of the Vela SNR based on a two-temperature NEI emission model is not feasible with ROSAT PSPC data. This is clearly shown in Table 3, in which we also report the fitting results obtained with a two-component STNEI model for the spatial bin [FORMULA], in which the bright clump named "FilD" is located. Notwithstanding the very high number of counts in the spectrum ([FORMULA]), the fitting parameters are unconstrained and reliable estimates of temperature and [FORMULA] can not be derived, because of the moderate PSPC spectral resolution. For this reason, we adopt CIE emission models when investigating the 2T nature of the X-ray emission using ROSAT PSPC data.

To estimate the error introduced by the 2T CIE analysis of truly 2T NEI plasma, we have realized a simulation in which 200 2T STNEI spectra have been synthesized including Poisson noise, and then fitted with the RS 2T CIE model. The input parameters for the simulated spectra are: [FORMULA], [FORMULA] yr cm-3, [FORMULA], [FORMULA] yr cm-3. The adopted ionization times are expected for a Vela-like SNR (Paper II): note that we have assumed [FORMULA] as expected in the hypothesis that the low T component originates from a plasma about a factor 3 more dense than the environment. In Fig. 6, we report the fitting results. The distributions of the best-fit temperatures and emission measures are not centered on the input values; this is particular evident for the hotter component. By computing the centroids of the distributions for the two component, we have derived that the temperatures [FORMULA] and [FORMULA] are underestimated by [FORMULA]% and [FORMULA]%, respectively, and that the emission measures [FORMULA] and [FORMULA] are overestimated of [FORMULA]% and [FORMULA]%, respectively.

[FIGURE] Fig. 6. Results of fitting synthesized 2T STNEI spectra with [FORMULA] PSPC counts using the 2T RS CIE model. Triangles and crosses represent the X1 and X2 component, respectively. Asterisks at the intersection of the dashed lines mark input values, which are not recovered correctly in any case.

A "zero order" correction for NEI effect can be defined from these results: in fact, in order to recover the correct values of the input 2T NEI model, the best-fit parameters should be multiplied by the following factors: [FORMULA], [FORMULA], [FORMULA], and [FORMULA]. We stress that the NEI correction factor are based on a simulation with realistic but arbitrarily chosen input parameters, and that they cannot substitute a full NEI analysis which will be possible only with X-ray instrumentation with spectral resolution higher than the ROSAT PSPC.

Previous Section Next Section Title Page Table of Contents

© European Southern Observatory (ESO) 1999

Online publication: February 23, 1999