Defect Etching

After planarisation of the uneven wafer front side a saw-damage removal is necessary prior to the subsequent emitter diffusion. The planarisation step as a mechanical treatment provides a high mechanical load on the wafers resulting in a heavily disturbed crystal area including micro-cracks. On the other wafer side (substrate side) an at least 25 μm thick layer of disturbed crystal structure has to be removed (Figure 7). Thus after planarisation and before solar cell processing 25 μm per side have to be etched off using an acidic isotropic etch (modified CP6 solution,

see paragraph 1.6.1.2). An anisotropic (e.g. alkaline etch) would result in a rough wafer surface due to different etching rates for each grain orientation. RGS as a multicrystalline material consists of randomly orientated grains and thus no defined surface could be obtained by using a non-isotropic etching solution.

The crystal structure shown in Figure 7 suggests that an etching depth of 25 μm per side should be sufficient to remove disturbed crystal structures on the surfaces of the RGS wafers. On the other hand on the wafers substrate side of the RGS wafers the carbon concentration is increased [24] due to the contact of the liquid silicon with the carbon based substrate plate during solidification as shown in Figure 16.

Figure 16: Distribution of the substitutional carbon concentration [Cs] in a RGS wafer.

Local FTIR measurements showed that the carbon concentration is higher at the front and the rear surface of the wafer.

The enhanced carbon concentration is a result of a supersaturation of the silicon melt with carbon and is known to produce carbon precipitates in the RGS wafers [25]

which are detrimental for the solar cell performance. Thus, to check if the applied etching depth of 25 μm per side is sufficient or if RGS solar cells would benefit from extended etching, solar cells were processed from two groups of RGS wafers. One half of the wafers were etched 25 μm per side, the other half 50 μm per side.

Figure 17: Fill factors of RGS solar cells in dependence of shunt values. The etching depth of one half of the RGS wafers was doubled (triangular symbols).

Resulting shunt values of the processed RGS solar cells are shown in Figure 17. The distribution of the shunt values shows no beneficial effect for the group with extended etching depth. The processed solar cells show for both groups nearly the same mean

values of the shunt resistances Rsh (93 Ωcm² for an etching depth of 25 μm and 90 Ωcm² for 50 μm).

As a result, an etching depth of 25 μm per side is sufficient to remove the defect-rich surface layers of planarised RGS wafers.

The absolute values of the shunt resistances of the RGS solar cells shown in Figure 17 are low compared to standard industrial multicrystalline solar cells (Rsh > 1000 Ωcm²). RGS material suffers from different shunting phenomena lowering the FFs of the solar cells. However, shunting can be avoided by means of processing as described in paragraph 1.6.7.4.

To investigate the detrimental influence of the enhanced carbon concentration on the substrate side of RGS wafers (Figure 16) on solar cell parameters, an additional experiment was performed. Two groups of RGS wafers were processed to solar cells. One group was etched after planarisation (25 μm per side) as described above.

The other group was not only planarised on the wafer front side but in addition 30 μm were removed on the wafer substrate side using the planarisation tool, followed by etching of 25 μm per side as well. As a result, 55 μm were removed from the wafer substrate side for the second group.

Figure 18: Fill factors of RGS solar cells in dependence of shunt values. The wafer substrate sides of one half of the RGS wafers were grinded prior to etching.

Due to the mechanical removal of the carbon rich layer on the substrate side of the RGS wafers the mean shunt values improved for the group with the grinded and etched substrate side (Figure 18, triangular symbols) from 74 Ωcm² to 129 Ωcm² which corresponds to an enhancement of 74%rel..

Again, the absolute shunt values are significantly lower compared to solar cell processed from standard mc material due to a RGS material specific shunting mechanism.

Figure 19: Illuminated Lock-In Thermography (iLIT) measurement of two RGS solar cells processed from neighboring wafers. The corresponding wafers were processed without (left) and with (right) a grinding of the wafer substrate side. Same scaling for both cells.

However, the possibility to improve shunt values by grinding is difficult to realise. The mechanical load during grinding induces micro-cracks limiting the shunt values and thus the FFs of the solar cells again. Figure 19 shows two iLIT [26] measurements of solar cells processed from wafers originating from the same entire RGS wafer. Thus the material quality of the two wafers is comparable (neighboring wafers). The solar cell shown on the left side in Figure 19 was processed without grinding the wafer substrate side and shows areal shunts especially in the upper right area (yellow / red areas) originating from material and process induced defects (see paragraph 1.6.7).

In contrast the solar cell shown on the right side reveals reduced areal shunting but two strong point shunts (lower left cell area) most probably originating from the additional mechanical load during grinding.