Cell structure and mask design

In the previous chapters and sections the processes to manufacture a high-efficiency multicrystalline silicon solar cell were characterised. These were in detail:

• Gettering to increase the material quality (chapter 3)

• Texturing (chapter 4)

• Emitter diffusion adjusted to textured surface (chapter 5)

• Highly reflective rear surface passivation without degradation of the bulk (chapter 3.2.4 and section 6.2.2)

• Laser-fired contacts for the rear (section 6.2.4).

The resulting cell structure for multicrystalline silicon developed in this thesis is shown in Fig. 6.10. In comparison to the cell design developed for high-efficiency monocrystalline silicon solar cells (see section 2.4) the oxidations necessary to build the cell structure were reduced to the minimum. The masking oxides for the local boron diffusion on the rear and the local emitter diffusion under the contacts on the front were omitted and the masking oxide for the front surface texturing process was replaced by a mask made of photoresist.

A photoresist-guided electroplating of the front contacts was not applied since the increased process complexity would require more wafer handling. This increases the risk of breaking the thin multicrystalline wafers.

Fig. 6.10: High-efficiency cell structure for multicrystalline silicon. For the sake of clarity the double layer antireflection-coating on the front is not shown.

The main features of the developed cell structure are (from front to rear):

• Evaporated contacts. The front metal contact is built from an evaporated stack system of titanium, palladium and silver (Ti/Pd/Ag). This thin layer has a low contact resistivity and for a high grid conductivity the metallisation lines are thickened by electroplating of silver.

• Double layer antireflection coating. The use of a double layer coating of TiOx/MgF2 significantly reduces the primary reflection losses in comparison to

a single layer by adjusting layer thickness and refractive indices to the whole solar spectrum.

• Thin silicon oxide for surface passivation. A thin silicon oxide of about 15 nm thickness is thermally grown on top of the emitter. There was no aluminium anneal performed due to the increased process complexity and concerns about the long-term stability when exposed to UV-light [110].

• Plasma-textured front surface. The surface consists of deep cones which were etched using a lithographically defined mask of photoresist. The structured surface increases the possibility for incident light to be coupled into the cell.

Furthermore the tilted surface leads to a refracted ray path and increases the distance of light coupled into the cell to reach the rear surface. This improves the absorption probability.

• Thin base of multicrystalline silicon. Although a thick wafer increases the absorption probability for long wavelength photons, multicrystalline silicon can benefit from a reduced cell thickness. The ratio of diffusion length/cell thickness is increased and the distance which generated carriers need to diffuse to the p/n-junction is decreased.

• Silicon oxide passivated rear surface. The rear surface is excellently passivated with an Al-nealed oxide of about 100 nm thickness. This reduces the rear surface recombination velocity and increases the minority carrier diffusion length. The excellent optical and electrical properties compensate the loss in carrier generation caused by the reduced cell thickness.

• Aluminium evaporated onto the silicon oxide. An aluminium layer of 1-2 µm thickness evaporated on silicon oxide works as a nearly perfect mirror. Long wavelength photons reaching the rear surface are reflected and pass through the whole wafer a second time. Together with the textured front this acts as a very good light-trapping scheme and enables several internal reflections.

• High dopant concentration under the rear contacts. A local silicon/aluminium alloy (p⁺) underneath the rear contacts minimises the contact resistance and decreases recombination at the contacts. This is done by the laser-fired contact process and leaves the passivating oxide around the contacts unaffected [101].

A large number of solar cells which are electrically separated from each other are desirable since with such cells the material quality can be studied in detail on small areas (see chapter 7). Furthermore many process parameters can be tested on one wafer of 100 cm² size. This was done for the front metallisation by varying the

number of grid fingers (11 and 13) and the contact width (3 and 5 µm) for 36 solar cells of 1 cm² area. Six cells of 4 cm² with 25 grid fingers and 5 µm contact opening were defined. All cells were electrically isolated from each other by masking the emitter diffusion around the cells on the front. To enable the exact determination of the cell parameters, the cells were confined by a photo-lithographically defined evaporated aluminium mask between the cells. Every cell was measured individually with all neighbouring cells being shaded during the measurement. A picture of a completely processed wafer is shown in Fig. 6.11.

Due to the effective front surface texturing, the multicrystalline structure is hardly visible on the cells but clearly detectable on the aluminium area mask.

Fig. 6.11: Picture of 42 small solar cells on one wafer of 10⋅10 cm². Every small solar cell can be measured independently and the area is defined by an evaporated shadow mask. The metallisation including the busbar is completely inside the defined aperture area. The structures in the middle of the picture (left and right edge) are metallisation test structures and symbols used for alignment of the lithographic masks.