The difference between the rear passivation layers consists of the doping type and the doping density of the respective a-Si1-xCx films. For the phosphorous doped films, the different behavior depending on the contacting scheme can be explained when referring to the concept of floating junction (FJ) passivation (section 2.5.2). Practical problems arise from parasitic shunts in the FJ as also observed for p-type cells with SiNx rear surface passivation [27]. There exist basically two leakage paths that can occur in the presented solar cell approach. (i) The first one is the direct contacting between the floating emitter formed by the a-Si1-xCx (n) film and the base at the alu-minium point contacts. The latter must also be taken into account in the case of p+ regions directly neighboring n+ regions (which might be the case in the LFC approach) because this configuration allows for significant tunneling currents. (ii) Other leakage paths can exist due to pinholes in the SiOx layer which separates the a-Si1-xCx films from the rear aluminium. Again a shunt between the floating junction and the base is established and the result is a drop in cell performance.
Independently of its origin, the shunting of the FJ can be characterized in terms of a shunt resistance RFJ,sh. At low voltages, most of the current flows through the shunt (shunt-dominated region) and therefore effectively enhances the rear side recombina-tion, whereas at higher voltages the main current path is the diode itself (diode-dominated region). Hence the magnitude of RFJ,sh determines the transition from shunt- to diode-dominated region (transition region). In the case of a very low RFJ,sh, the
Fig. 6-4: IQE measurements of solar cells with plasma etched rear openings (left) and LFC approach (right) from 700-1200 nm. Best cells of the respective categories are depicted.
Silicon solar cells with a-Si1-xCx rear side schemes 101
junction behaves as if it were a surface with an infinite recombination velocity (SRV) since all carriers diffusing to the rear junction are collected by RFJ,sh and are not re-turned to the front junction. Due to the voltage dependent SRV in the case of a shunted FJ, this effect can be observed as a kink in the dark-IV curves or, equivalently, by considering the local ideality factor mloc. The latter is defined by the 1-diode equation
⎟⎟
that can be written as
( ) ( )
V (neglecting the -1 term for increased voltages). J0 is the dark saturation current density, q is the elementary charge and J and V refer to the current density and the voltage of the solar cell, respectively. Eq. (6-2) shows that mloc is proportional to the inverse slope of the log(J)-V curve. A kink in the latter is therefore accentuated as a small bump in the mloc-V representation.The mloc-V curves derived from dark-IV measurements for a set of cells with rear plasma openings are displayed in Fig. 6-5 left. The shape of the dark-IV curve and hence of the mloc curve at low voltages is dominated by the current flow through the shunt resistance of the front junction Rsh,F. At high voltages, the series resistance RS
dominates the dark IV-curve and the curve flattens (mloc increases). A pronounced bump in the mloc-V curve at 0.45 V of the cell-type n15/plasma evidences the transition region of a typical, shunted high resistive FJ. No bump is visible for cells with intrinsic
Fig. 6-5: Local ideality factors mloc as a function of voltage. Left: mloc derived from dark-IV measurements. Right: mloc derived from Suns-Voc measurements.
102 Silicon solar cells with a-Si1-xCx rear side schemes
passivation. A smaller RFJ,sh results in a dark-IV kink occurring at higher voltages because poorer RFJ,sh require the collection of a larger amount of minority carriers before switching to the diode-dominated case. Therefore, if RFJ,sh is very small, the flattening of the IV curve due to the series resistance of the cell and the kink may occur in the same voltage regime. This problem can be avoided by deriving the mloc-V data from Suns-Voc measurements since this technique basically quantifies the dark-IV characteristics of a solar cell in absence of series resistance (chapter 3.6). One limita-tion of this approach is that the evalualimita-tion of mloc can only occur up to the Voc of the considered cell. This implicates that the mloc information for cells with poor perform-ance are rather restricted. Fig. 6-5 right displays the mloc-V curves for an intrinsic/LFC and a n15/LFC cell derived form Suns-Voc. While mloc is rather constant in the intrin-sic case, a bump appears at around 0.62 V for the cell passivated by a lowly n-doped a-Si1-xCx layer, pointing to the shunting of a low resistive FJ. No comparable feature occurs in the mloc-V curves of cells with highly n-doped a-Si1-xCx films (n85/plasma and n85/LFC), which may be explained by a shift of the bump to even higher voltages not observable by the Suns-Voc data.
Owing to the presented evidences, it seems reasonable to address the severe per-formance loss of the a-Si1-xCx(n) passivated cells to shunts in the rear FJ (low RFJ,sh).
However, as already mentioned, there is a considerable scatter in the cell data. This may point to an additional deteriorating effect caused by statistically distributed pin-holes in the SiOx layer. It can be argued that the difference between the two rear con-tacting approaches is due to the undercutting of the SiOx layer and to the at least partial prevention of a shunt between the a-Si1-xCx(n) film and the silicon base by evaporated
Fig. 6-6: Schema of a cross section through a plasma etched opening. Possibly the shunt resistance of the rear diode (floating junction) can be increased by undercutting the SiOx layer enabled through the isotropic character of the used etch gas.
Silicon solar cells with a-Si1-xCx rear side schemes 103
aluminium (Fig. 6-6). Both current leakage paths may be present, however in the case of the highly doped films the shunts originating from pinholes probably dominate the cell performance due to a considerable reduction of the RFJ,sh because of an increased conductivity of the a-Si1-xCx(n) film.
An equivalent “shunting-effect” of the HLJ in the case of the boron doped a-Si1-xCx film does not exist. As could be expected from the ongoing argumentation, the p20/LFC results show no contacting related decrease in cell performance.