Since the replacement of CdS with other compounds was unsuccessful, it became important, to find out why CdS is so beneficial. All earlier attempts of explanation failed. Mostly because one tried to model the actual cell, that is polycrystalline and has a complicated morphology. The involved processes are seemingly too complex to permit a reasonable model computation that shows encouraging agreement with the experiment, i. e. mostly with the measured current-voltage characteristics.
Only recently (Boer and Ward 1967a, 1967b; Boer et al. 1969b) it was recognized that the cause of the enhancement should be researched within the CdS itself, and to do this with sufficient experimental substantiation. One therefore has to first proceed with a simple one-dimensional model, that is, by replacing the actual CdS layer with a thin CdS platelet. Here ample experiments have been performed in the 50’s and 60’s and provide the basis for an attempt to explain the behavior (for a comprehensive review see Boer et al. 1969b; Fthenakis et al. 2004). From these results it is known that CdS is an и-type semiconductor that when doped with copper becomes highly photoconductive. With an optical excitation, as it is exposed in a solar cell, the electron density is on the order of 1018 cm-3. Such photoconductivity can easily be quenched (i. e. the electron density can be reduced) by additional low energy excitation, as, e. g., by infrared light, or, what in the case of the solar cell becomes important, by an intermediate electric field. This field causes a Frenkel-Poole excitation (Noufi and Zweibel 2006) of trapped holes (that stem from the optical excitation) from Coulomb attractive hole centers into the valence band. The now set-free holes can be trapped by fast recombination centers from which they can interact with the conduction electrons and by permitting enhanced recombination, thereby reducing their density (Boer and Voss 1968a, 1968b, 1968c). These intermediate electric fields can be reached at the CdS side of the junction to the CdTe, causing in the corresponding thin CdS layer such “field-quenching” to occur (Boer and Voss 1968a, 1968b, 1968c).
It can be shown that the field quenching is one of the reasons for the cell improvement since it can reduce the electron density at the junction interface and hence its back-diffusion into the CdTe, that would represent a junction leakage. Such a leakage consequently would result in a lower solar conversion efficiency, as it is observed in the uncoated CdTe cells. However, one must now include several other important factors to more quantitatively explain the behavior.
1. The electric field at the CdS side of the junction is a function of the voltage drop that is observed at various points of the current-voltage characteristics: It is generally low (below 20 kV/cm) in forward bias and increases rapidly when approaching the open circuit voltage, and beyond in reverse bias. This means in sufficient forward bias there is no field quenching, hence electron can travel through the junction easily. However, as the field increases when Voc is approached, field quenching sets-in and substantially reduces such electron leakage (Boer 2009a, 2009b).
2. CdS in the quenched region shows a negative differential conductivity, that is, the photocurrent decreases with increasing applied voltage. When the field quenching is strong enough, a high-field domain is created in which the field is limited to approximately 50 kV/cm. This field is approximately the same in which substantial field quenching is observed. In fact, the field quenching is the cause of the negative differential conductivity that necessary to initiate the high-field domain (Boer and Voss 1968a, 1968b, 1968c). This maximum field near the CdS/CdTe interface is too low to permit electron tunneling through the junction, that otherwise would shunt the junction and reduce the solar cell efficiency.
3. The field quenching near the CdS/CdTe interface can be strong enough to turn the CdS into a ^-type layer (Boer and Ward 1967a). This causes the Fermi level in this layer to move much closer to the valence band. Since the Fermi Level at open circuit condition must remain horizontal, this means that both valence and conduction bands curved up in the field quenched region and the conduction band at the interface must disconnect from the conduction band of the CdTe, to which it was connected in forward bias (Boer 2009a, 2009b). This again limits the electron back-diffusion into the CdTe.
4. To permit such relative shift of the bands between CdS and CdTe, this requires a change of the dipole moment at the interface, which determines the band connection (or the band offset). There is experimental evidence that in copper doped CdS platelets the dipole moment that is involved in a Schottky barrier to a metal electrode can be changed by changing the photoconductivity (Boer et al. 1969b). That seems to justify the assumption that such a change of the dipole moment at an even ‘softer’ interface can occur.
With these conditions one can draw the band diagram of a CdS/CdTe solar cell close to the junction interface (Boer 2009a, 2009b) as it was shown in the previous chapter (Fig. 36.11).
As discussed before, the thin cover layer of CdS permits a substantial improvement of the conversion efficiency of the CdS/CdTe solar cell. For this it is essential that the field-quenching is initiated at a field of about 50 kV/cm. Field quenching is quite sensitive to the distance between Coulomb attractive centers that are produced
by copper doping (Boer and Dussel, 1970). If this density is too low, then the quenching is not efficient enough to reduce the electron density markedly. If, on the other hand, the density is too high, then these centers become too close and the critical field for Frenkel-Poole excitation becomes much higher. Accidentally, the saturation level of copper in CdS is about 100 ppm, that brings the distance between the copper atoms to its optimal value (Boer and Dussel 1970).
Though in some other semiconductors one can induce such field quenching and consequently negative differential conductivity, but one has to design the doping just to the right level. This is too complicated to achieve over large enough areas, causing this alternative layer to become highly inhomogeneous and its beneficial effect limited to a small percentage of the solar cell area.