Fluorescence and Escape Peaks


Author: Norbert S. Schulz, 2/23/99

We present analysis results for off-nominal features such as Si fluorescence and escape peak positions and efficiencies in ACIS front and back-illuminated devices. Data for front illuminated devices were obtained by the ACIS IPI team in Synchrotron measurements at Bessy as well as in house subassembly calibration measurements at MIT. The efficiencies and positions presented here were then derived from first generation eventlists obtained from the ACIS model simulator version from July 1998. For the back-illuminated devices only in house subassembly calibration data are available. Since the development of the ACIS model simulator for these devices is still under way, as a first step we present positions of off-nominal features in w134c4r (S3) and w140c4r (S1).

Introduction

The most prominent off-nominal features of the CCD spectral re-distribution function are fluorescence and escape peaks. In general we expect fluorescence emission from various elements present in CCD devices, such as oxygen, nitrogen, and silicon, however fluorescence events from elements other the silicon have negligible probabilities. Similarly L-shell fluorescence from silicon is neglected as well. However there is a significant probability, that photons with energies above the Si-K edge at 1839 eV will produce fluorescent Si K$_{\alpha}$ photons at 1740 eV. Si K$_{\beta}$ photons (1836 eV) are not likely to be generated, because the energy is too close to the K-absorption edge. The bulk of fluorescence photons comes from different areas in the device depending on energy. In front side devices energies near the K-absorption edge are is more likely to generate fluorescence photons from the polysilicon gate and gate insulator rather than from the substrate, which provides most of the photons at high energies. Once a fluorescent photon leaves the silicon substrate or interacts some distance away from the original site of the X-ray impact, an escape event is being generated. A thorough treatment of the involved physics is given by Prigozhin 1998a and 1998b and references therein. In back illuminated devices fluorescence and escape efficiencies are expected to be different because of the fundamental design differences. For incident X-ray photons with energies above the silicon K-edge we expect a fluorescent peak with a characteristic energy of 1740 eV and an escape peak at an energy that is approximately at the energy of the incident photon minus that of this characteristic energy in both device types.

For a more comprehensive study and description we refer to sections of the latest (Oct. 1998 rev2.2) ACIS calibration report concerning the physics of low energy tails and off-nominal peaks in the MIT ACIS model.

Front-illuminated Devices

For a most precise experimental assessment of fluorescence and escape efficiencies one has to illuminate the devices with a monochromatic beam at various energies. These measurements have been performed by the ACIS IPI team at the synchrotron storage ring at Bessy using a reference device (w102c3). Preliminary results of these measurements can be found in Figure 1.8 in Bautz et al. 1997. A more recent analysis was presented by Prigozhin et al. at the SPIE 1998 (Prigozhin et al 1998a). Figure 1 shows the intensities of the experimentally measured escape and fluorescence efficiencies as a function of energy. In the following we measured positions of escape and fluorescence peaks in flight device w182c4r (ACIS-S2), which we use as standard device for all flight front illuminated devices for the time being. Figure 2 shows a spectrum of a subassembly measurements using a Fe55 (Mn-k) source. ...(NOTE: at the time the Mn-k ps-figure was not yet available and the Ge-K line source spectrum is shown instead).... It shows the channel energy in ADU vs counts/ADU. Here 1 ADU is 5898/1667 = 3.65 eV. The high energy end clearly shows the two primary photopeaks at Mn-K$_{\alpha}$ and Mn-K$_{\alpha}$ at 1667 and 1850 ADU, as well as fluorescence and escape peaks as listed in table 1. Besides a broad continuum, the low energy tail and a significant Mn-L$_{\alpha}$ peak at 260 ADU there are a few other contaminants in the spectrum since there was no filter used and thus none of the applied sources in table 1, table 2, table 3, and table 4 were really monochromatic. We therefore list only the positions and position differences of the off-peaks and the primary peaks and do not attempt to measure efficiencies. Ultimately it is desired to obtain precise positions and intensities of these peaks from the ACIS model simulator. A first generation of event lists from this simulator was released in July 1998. Eventlists of 15 energies were generated containing approximately $10^6$ counts, from which 7 were far enough above the silicon K-edge to show off-nominal peaks. The event lists are "first generation", which means that the simulator is in many respects still based on idealistic assumptions, such as a monochromatic incident spectrum, no pileup effects, no read out noise, no gate structure effects on fluorescence, and no charge transfer inefficiency. Table 5 shows the fitted centroids of the primary, the escape and fluorescence peaks as well as several efficiencies for grade 0 events. Although clearly all intensity (but the escape to first order) above the silicon edge should contribute to the fluorescence efficiency. The bulk, of course, comes from the primary peak. Including the tails and shelfs (which is the right thing to do) gives significantly lower values, the inclusion of everything other than fluorescence peak and escape peak is only insignificantly lower but also gives a lower limit to the efficiencies. Table 6 shows the same for the standard grade set. Note that fluorescence and escape efficiencies show slightly different trends in both grade sets.\par However, at this stage one has to caution any conclusions. At the current state, the simulator is still under development and in the version used here an important fact was not yet included. That is, at energies near the silicon edge the bulk of the fluorescence emission does not come from the substrate, but from the gate and gate insulator. Table 5 and table 6 thus will always be updated when new generation event lists are being released. The straight lines in figure 1 show a first result of the improved simulator.

  • Figure 1: Escape and fluorescence efficiencies as measured at the synchrotron storage ring at Bessy using a reference device (w102c3), from Prigozhin et al. 1998.
  • Figure 2: The pulse height (PHA) spectrum of node 0 of device w182c4r in S2 for the Ge-K source (NOTE: this spectrum will ultimately be replace by the Mn-K line spectrum with full legend).

  • Table 1: Main and off-nominal peak positions for node 0 of w182c4r
  • Table 2: Main and off-nominal peak positions for node 1 of w182c4r
  • Table 3: Main and off-nominal peak positions for node 2 of w182c4r
  • Table 4: Main and off-nominal peak positions for node 3 of w182c4r
  • Table 5: ASCA grade 0 main and off-nominal peak positions for FI devices from first generation eventlists generated by the ACIS simulator.
  • Table 6: Standard ASCA grade set 02346 main and off-nominal peak positions for FI devices from first generation eventlists generated by the ACIS simulator.

    Back-illuminated Devices

    Because of some extended treatment during the manufacturing process back-illuminated devices show a higher affinity towards CTI effects. This became already a fact during Lincoln Lab screening of the old flexed devices, which indicated about a factor 100 higher charge loss per pixel transfer than observed in FI devices. It also appears that the two BI devices in the ACIS-S array themselves behave quite differently. The average FWHM versus energy in both devices, as specified in the ACIS IPI report, differ from 90 eV to 220 eV for w134c4r (S3) to 150 eV to 270 eV for w140c4r (S1) between 277 eV and 8029 eV respectively. In the following we therefore treat both BI CCDs as different devices.

    w134c4r in S3

    Table 7, table 8, table 9 and table 10 show the positions of main, fluorescence, and escape peaks for the 4 nodes in w134c4r derived from subassembly data. Those data are particularly non-monochromatic and contamination line emission was a problem during the analysis. Escape peak data for energies above Ni-K are therefore less reliable. The integration of efficiencies was not possible.

    w140c4r in S1

    Table 11, table 12, table 13, and table 14 show the positions of main, fluorescence, and escape peaks for the 4 nodes in w134c4r derived from subassembly data. Those data are particularly non-monochromatic and contamination line emission was a problem during the analysis. Escape peak data for energies above Ni-K are therefore less reliable. The integration of efficiencies was not possible.