CCD quantum efficiency.

Having all the absorption coefficients in hand it is possible to calculate the quantum efficiency of the CCD taking into account transmission of all the layers in the gate structure of the frontside CCD. In this model we accounted for the overlaps between the gates of the CCD, and assumed triangular edges of the channel stop oxide layer (which is a good approximation for the so-called ``bird's beak'' shape of those edges). Thicknesses of the layers were taken from the results of Scanning Electron Microscope (SEM) measurement of a sibling device cross section (see Fig 4.35). In Fig. 4.60 is shown a plot of quantum efficiency as a function of energy for the frontside illuminated device.

  

Figure 4.60: Quantum efficiency of the frontside illuminated CCD. Dashed line is the result of the QE measurement at SX700 line at BESSY.

The dashed line in the Figure represents results of the quantum efficiency measurement of the frontside CCD at the SX700 beamline at BESSY. No fitting of the model to the data has been made for this plot. The deviation of the experimental points from the model at higher energies can be explained by the known second order light penetration of the SX700 monochromator at higher energies. In the analysis of the SX700 results the second order light counts are excluded in the CCD, while the reference photodiode cannot discriminate between orders. This results in the underestimation of the QE values for the CCD. It should also be mentioned that the above model accounts only for the transmission losses in all the layers (in great detail, though) and does not deal with the redistribution of counts into the low energy tail. This should not be a very significant factor for the frontside illuminated CCDs.

A similar plot for the backside illuminated device is shown in Fig. 4.61. Triangular points show the results of experimental measurements of the quantum efficiency of the backside illuminated device at the SX700 beamline at BESSY. The plot extends to very low energies around the silicon L edges, although the device loses its ability for spectral resolution below 250 eV. It still can be used for imaging, and its QE can be high enough, as can be seen from the Figure.

  

Figure 4.61: Quantum efficiency of the backside illuminated CCD. Triangles mark experimental results from SX700 at BESSY.

The best agreement of experimental data with the transmission model was reached assuming a surface thickness of oxide layer of 0.055 microns, whereas the nominal thickness specified by Lincoln Lab is 0.035 microns. Besides that, we had to multiply the calculated QE by a factor of 0.96 to obtain an agreement with the experiment. This is expected, since for the backside devices a noticeable fraction of events is lost to the low energy tail. A more sophisticated model is necessary to account for this effect. As in the case with the frontside devices, a discrepancy at higher energies is attributed to second order light contamination.