WINDOW-ABSORBER STRUCTURE WITH ABSORBER INTERFACE RECOMBINATION
Earlier in this chapter we expressed concern about the possible impact of localized states at and around a HJ's metallurgical junction on device performance （mechanism 8 in Fig. 5.5）. Having developed a good feel
FIGURE 5.22 Some short-circuit simulation results. （a） Electron conventional current density components and total conventional current density and （b） the hole conventional current density components and total conventional current density.
FIGURE 5.23 Some simulation results for the maximum power point for the cell of Figure 5.20. （a） Band bending at the maximum power point with the electric field given by the derivative of the local vacuum level; （b） Plot of the bulk recombination R（x） as a function of position at the maximum power point.
for heterojunction behavior without such defects, we are ready now to add them and examine their impact using numerical analysis. We do this using the cell shown in TE in Figure 5.24. It is the same device as that of Figure 5.20, except now it has defects at the HJ interface. As seen from the second row of Table 5.3, the defects have been added to a 10-nm interface layer in the absorber at the HJ. Taking the number density of atoms in the absorber to be about 1021/cm3, it can be seen from Table 5.3 that the defect density used is such that 1 in every 100 atoms in the 10-nm interface layer is a defect （impurity, interdiffused species, vacancy, etc） in this HJ example. The defect states have been chosen to be effective, from a loss point of view, S-R-H interface recombination sites. This can be noted from their properties listed in Table 5.3, which exhibit the telltale signs of efficient S-R-H recombination paths; i.e., relatively large capture cross-sections with both cross-sections relatively close in magnitude （making them attractive to both carriers） and an energy location for these states away from the band edges. The defects have been taken to be acceptor-like, so the capture cross-section for holes is higher due to coulombic attraction, as discussed in Chapter 2. In the numerical analysis undertaken for this example, it has also been assumed that the interface recombination path has electrons at the absorber conduction band edge （absorber EC or LUMo） recombining with holes at the absorber valence band edge （absorber EV or HoMo）； hence, the band gap involved is that of the absorber. However, there are many more holes at the valence band edge （window EV or HoMo） of the window. Consequently, due to intermixing of the materials at theinterface, tunneling, etc., it is very possible that the recombination path could be from the absorber LUMo to the window HoMo. This would result in an effective band gap, a quantity mentioned earlier, for this recombination process that would equal the window Homo - absorber LUMo difference.