Quiescent Particle Background

Table 6: Instrumental Lines.
Line Energy (keV) Instrument
Mg K$\alpha$ 1.253 MOS & pn
Al K$\alpha$ 1.486 MOS & pn
Al K$\beta$ 1.557 MOS only?
Si K$\alpha$ 1.739 MOS
Au M$\alpha$ 2.123 MOS & pn
??? 4.535 MOS
Cr K$\alpha$ 5.405+5.415 MOS & pn
Cr K$\beta$ 5.947 MOS & pn
Fe K$\alpha$ 6.391+6.404 MOS & pn
Fe K$\beta$ 7.058 MOS & pn
Ni K$\alpha$ 7.460+7.478 MOS & pn
Cu K$\alpha$ 8.028+8.048 pn
Zn K$\alpha$ 8.615+8.638 MOS & pn
Cu K$\beta$ 8.905 pn
Au L$\alpha$ 9.628 pn
Au L$\alpha$ 9.713 MOS only?
Au L $\beta$ 11.442+11.584 MOS & pn

The quiescent particle background (QPB) spectra created in the previous section is primarily a continuum spectrum. There are a number of lines due to emission by materials in the vicinity of the detector, as can be seen in Figures 12 and 13. Many of these lines are included in the particle background spectra created in the previous section. However, there are a number of emission lines that are too strong to be well modeled in this manner, and these lines were excised from the quiescent particle background spectra. It should also be remembered that the background spectra are created from collections of corner data and filter-wheel-closed data that have accumulated over the entire mission. The line spread function has varied significantly over the mission and thus the background spectra derived from either corner data or the filter wheel closed (FWC) data have a different line width than the actual observation data. This is not a problem for weak lines, but it is a substantial problem for the strong lines. Further, there are slight variations in the gain which are often not noticeable except around the very strongest lines.

Since the Al K$\alpha$ and Si K$\alpha$ lines for the MOS and the Al K$\alpha$ and Cu K$\alpha$ lines for the pn are excised from the QPB spectra, these lines must be fit explicitly to the data. For the pn it may be more reasonable to simply exclude the Cu line region from the analysis, although this may shorten the lever arm for fitting the SPF component.

For the MOS, fitting the instrumental lines is done easily by the addition of two Gaussians in the spectral model with energies of $E\sim1.49$ keV and $E\sim1.75$ keV with zero width. The intensities of these two lines vary over the individual detectors but they seem to be in relatively good agreement between the two MOS instruments. We note, however, that since the lines are strong, minor errors in the line redistribution function can produce significant residuals, particularly at energies just below that of the line.

The pn, unfortunately, requires six lines at $E\sim1.49$ keV, $E\sim7.47$ keV, $E\sim7.06$ keV, $E\sim8.03$ keV, $E\sim8.62$ keV, and $E\sim8.90$ keV. The normalization of the pn Si line should be fixed at zero.

We note that for the typical exposure, only the brightest of the lines in Table 6 need be fit. However, for the QPB spectra available from the corner data demonstrate the existence of several more lines. Fitting the QPB spectra by themselves is difficult given that the intrinsic line spread function has increased with time, but it is possible, as shown in Figure 18. As we move to more statistically correct ways of fitting the data, such fitting may become important; currently this is shown merely as a demonstration of the number and location of instrumental lines in the QPB spectra.

Figure 18: Fits and residuals to the mean MOS2 (left) and pn (right) spectra. The spectra have not been binned or smoothed. We show the complete decomposition of the background spectrum (black) into its components. Green: power law components for the bulk of the continuum. Red: low amplitude features in the continuum fitted with Gaussians. It is not clear what some of these features are. Solid Purple: emission lines fitted with zero-width Gaussians and no instrumental line spread function.
\includegraphics[width=8.0cm]{mos_fit.eps} \includegraphics[width=8.0cm]{pn_fit.eps}