Discrete lines (Pollock) and Shock models - 15 October 2006

The E0102 spectrum as observed with the Chandra HETG (Flanagan et al. 2004) and the XMM-Newton RGS (Rasmussen et al. 2001) clearly show the bright lines from H-like and He-like ions of elements O, Ne, Mg, Si and H-like C. Some hints of Fe ion lines were indicated in the RGS spectrum. It should be noted that neither of these papers undertook to create a model to fit the complete spectrum of E0102.

For calibration purposes a continuum-plus-lines model has been created by Andy Pollock based on re-analysis of RGS data; that model expressed in XSPEC format (.xcm) and converted to ISIS format (which uses the XSPEC model library) as well (.par) is given here in these files:

• Pollock_E0102_RGS.2005-05-09.model.xcm
• pollock_model_sf15.par
  This par file was created using the ISIS script pollock_model.i. Note that the "sf15" here means that the line sigmas are a factor of 0.15 times the Pollock model value allowing the addition or not of further bluring.

An early approximation to the global E0102 spectrum was made using wabs times a simple shock model - the model and it's parameters are given by the ISIS .par file:

• e0102_simple_shock.par

  Because it is sometimes tricky for fitting algorithms to minimize complex (and table-lookup) models like vnpshock, the temp and tau parameters were randomly selected and then the norm and abundances fit. After thousands of instances the best overall temp-taus were then selected. Fe was then added and other elements adjusted.

Plots of the Pollock model and the simple shock model (calculated using the XSPEC NEIVERS of 1.1) are given here:

e0102_pollock_shock_ICWG .pdf , .ps

The three plots on this page are made in roughly similar format to the plots on pages 6,7,8 in Paul Plucinky's (2nd) presentation to the "ICWG" in June 2006, "E0102 as a Standard Candle for X-Ray Astronomy". Specifically, the Pollock model is shown in Red and the Shock model in Green. To emphasise where Fe lines are in the shock model, the shock model without Fe included is shown as the Blue line. The y-axis shows counts per bin and the same ARF and RMF have been applied to both models. (For reference the ARF and RMF are appropriate to the LETG as it was during an SNR 1987A observation. This has the property of good low-E response and a resolution that is not too good but not too blurry either - similar to the velocity blur Pollock's model includes...)

e0102_pollock_shock_hires .pdf , .ps

The three plots on this page are shown at the model resolution - narrow lines. The Pollock model (Red) has its lines 0.15 times narrower than the xcm model specifies whereas the Shock model (Blue) has bin-narrow lines - this helps to distinquish them. The Grey lines are from the Fe in the shock model and clearly show the "forest" these ions produce.
The middle panel is zoomed in a bit and plotted in wavelength.
The bottom panel shows the low-Energy end in the same color coding as above - note the Blue (from Silicon, I believe) lines in the shock model at ~ 0.37 and 0.40 keV.
In addition, this bottom panel includes, for reference, Carbon (orange) and Nitrogen (green) lines.

Comparing the two models

"Why compare with some random shock model?"
The idea is that this shock model is based on a range of temp and ionization times and so gives a good idea of all the lines that might be present - it can also be recomputed using improved line databasess and shock model physics. Comparing the shock model with the Pollock model then gives us some guidance for what to look for in the real data, e.g., to decide if Fe is present, etc. That will be the subject of further work.

There are similarities and some clear differences between these two models:

Similarities:

• There is clear agreement on most of the bright lines and their locations; the flux agreement is good as well (given that the simple shock was meant to be an approximation at the 20% level and has only 9 adjusted parameters in addition to the norm and N_H.)
• Looking at the 0.5-0.8 keV and 1.4-1.5 keV ranges, the continuum level for the two models is very similar, including a similar edge at ~ 0.53 keV.

Fe emission:

Fe emission/lines are (claimed to be) not explicitly included in the Pollock model, however comparing these spectra it is clear that the main two places where Fe lines show up in the shock model is also where there are "extra" lines in the Pollock model. So, the Pollock model has the degrees of freedom to "cover" for the Fe emission lines. Specifically the regions: 0.69 to 0.75 keV and 0.80 to 0.89 keV contain 5 lines each in the Pollock model (red), whereas in the shock model these same regions show 4 and 7 peaks (green) --- of these, 3 and 4 peaks, respectively, are due to Fe emission (not seen in blue).

• The presence of Fe in the shock model also shows up as a peak-like bit of extra flux at ~ 0.965 keV and an enhanced "continuum level" in the 1.08 to 1.3 keV range. It is interesting that the Pollock model continuum in the 0.9 to 1.3 keV range (red) is at about the level of this "pseudo-continuum" due to the many (blurred) Fe lines (green):
  
  In the Pollock model the first redge component at 0.87 keV adds the equivalent of the Fe pseudo-continuum; this edge does show up at a lower level in the shock model, and is seen more clearly in the high-resolution plots:
  

Other, easily corrected, differences:

• The Shock model does not have a line at ~ 1.15 keV that is in the Pollock model - this looks to be the Ne IX "Lyman delta" 5-1 transition and probably should be in the shock model (perhaps the line is not in the XSPEC NEIVERS=1.1 line list? )
• The Pollock model probably needs to add lines that show up at ~ 1.58 and 1.65 keV (from 3-1, 4-1 transitions of Mg XI/XII?)
• The N_H(in 10^22) value used in the wabs component is different for the two models: 0.0536 for Pollock and 0.08 for the simple shock. For both models, however, N_H is likely degenerate with other parameters --- the agreement of the resulting continuum level, e.g., around 0.55 keV, is more relevant to calibration issues.
• The models diverge at the energy extremes: below 0.45 keV and above 1.5 keV. This is not surprising since most of the data-overlap which constrains the models is in the 0.5 to 1.5 keV range. Note that at higher energies, the Pollock model only has a brems and redge components with kT = 0.36 keV, whereas the shock model has a range of kT's up to 0.6 keV --- better at including the Hughes et al. (2000, Table 2) estimated blastwave NEI-kT of ~ 0.48 keV and the Flanagan et al. (2004, Table 6) Ne emission temp of kT ~ 0.58 keV.

Velocity broadening of the lines:

• The Pollock model includes line width values, sigmas, to take into account the high velocities that are present in E0102. This aspect of the spectrum is not part of the simple shock model given here and is better examined using a spatial-spectral model of E0102 that takes into account the various components (ejecta, blastwave) and their motions. Such a model, similar to the E0102_Jun05 modeling, will be used in future work in this series.