The Pattern Problem

Note that the pn has a very strong low energy tail that can be exacerbated when using two-pixel (PATTERN <= 4) events, affecting data at energies as high 600 eV. Starting with SAS V17, ESAS has allowed the selection of either one-pixel (PATTERN == 0) events or one- and two-pixel (PATTERN <= 4) events. There are several trade-offs that the user will to need consider. If low-energy data ($E<1$ keV) data are of primary interest then only one-pixel events should be used and the analysis can be extended down to 0.3 keV. If high-energy data are of primary interest then both one- and two-pixel events should be used as excluding two-pixel events will seriously limit the number of counts accumulated at higher energies. In this case the analysis can be extended down to only 0.4 keV (or perhaps 0.45 keV) with the caveat that the spectrum should be examined for an anomalously strong low-energy tail.

An alternative is to process the observation extracting both a one-pixel event spectrum and a one- and two-pixel event spectrum. If hard and soft band images are going to be created then this doesn't require any additional effort beyond bookkeeping. Process the bands with energies less than the Al-K$\alpha$ line (1.486 keV) with pattern=0 in the call to pnspectra (PATTERN == 0 in the selection expression) and the bands including and above the Al-K$\alpha$ line with pattern=4 in the call to pnspectra (PATTERN $<=$ 4). Fit the pattern=0 and pattern=4 spectra simultaneously using an energy selection of, e.g., $0.3-2.0$ keV for the pattern=0 data and an energy selection of, e.g., $1.0-11.0$ keV for the pattern=4 data to examine the overlap region for confidence that the processing worked correctly. The image generation requires no additional effort beyond modifying the pnspectra pattern parameter.