Screen Printed Silver Contacting Interface in Industrial Crystalline Silicon Solar Cells

Abstract

This dissertation contributes to more insights to the fundamental understanding of the front screen printed Ag contact formation on Si n+ emitters. Screen printing Ag paste needs to be fired through the SiNX antireflexion layer, firing-compatible with the Al-BSF requirements, achieve good mechanical adhesion between the silver finger and the Si surface and contact n+ emitters with varying properties without damaging the junction. These challenging requirements need to be fulfilled by the result of the simultaneous chemical reactions between Ag, glass frit, SiNX , n+ doped Si emitter and O2 present in the atmosphere during the firing process. The final contact mainly consists of Ag crystallites distributed along the Si surface, a glass layer containing metal precipitates, and a sintered silver finger on top.To identify the key reasons behind a good ohmic contact, we metallized Si solar cell emitters with different silver pastes on textured and flat silicon surfaces and we applied a sequential selective silver-glass etching process to expose and isolate the different contact components. The surface configurations after the etching sequences were analyzed with SEM. We observed glass-free emitter areas at the tops of some Si pyramids that lead to the formation of direct contacts between the Ag crystallites grown into the Si emitter and the bulk of the silver finger. Thereafter, we focused on the understanding of the current transport mechanisms from the silicon emitter into the bulk of the silver finger. For this, liquid conductive silver after different contact etch-backs was applied and the contact resistivity was measured to determine the dominant microscopic conduction path system. We presented experimental evidence that the major current flow into the silver finger is through these Ag crystallites directly contacted to the silver finger.Hence, for screen-printed silver paste metallization, the presence of direct Ag crystallites is essential for the current conduction from the Si emitter to the silver finger and thus for low contact resistivity. For this purpose, we focused on the origin of these direct contacts. On textured surfaces we varied the Si pyramid sizes, rounded the pyramid tips to varying degrees and we also fabricated flat smooth and mc-Si surfaces. We observed that the size of the pyramids does not play an important role in the achievement of low specific contact resistivity unless the pyramid heights become smaller than the thickness of the glass layer. Contrariwise, rounding of the pyramid tips with standard heights increases specific contact resistivity significantly. Additionally, better contact resistivity on isotextured than flat surfaces were observed, even though the latter contain more Ag crystallites underneath the glass. Our observations indicate that a high density of Ag crystallites is not necessarily synonymous for a good contact, but of highest importance is how many of these Ag crystallites provide a direct connection with the silver finger. Also honeycomb textured Si features promote direct contacts at its elevated texture edges. To disregard the influence of the emitter doping concentration on the contact formation that is reportedly higher at the pyramid tips, we included surfaces without phosphorus doping in our study. Glass-free regions were also observed without P doping contained in these direct connections at the Si tips, where it is supposedly easier for Ag crystallite nucleation due to less Si atom back-bonding. Thus, from our microscopic investigations we suggested that the largest influence on the topography dependent contact resistance comes from the surface sharpness dependent glass coverage governing the amount of Ag crystallites directly connected with the silver finger bulk.Finally, we concentrated on the impact of defects originating from electrically inactive phosphorus on contact formation within silver thick film metallized silicon solar cells. For this purpose, emitters with varying sheet resistance, depth and dead layer were metallized with silver pastes from different generations. Macroscopic contact resistivity measurements were compared to the microscopic contact configurations studied by SEM. We found that the density of Si surface embedded Ag crystallites scales proportionally to the electrically inactive P, and is independent of the sheet resistance. Using the newest silver paste, the Ag crystallite density is independent of the emitter doping, but the Ag crystallite size increases as a function of the thickness of the dead layer. The presence of glass-free regions needed for the direct connection between Ag crystallites and the silver finger to achieve good quality contact depends on the paste composition and on the surface texture, and does not vary with the Si emitter properties. TEM characterization of the excess P doped Si crystal lattice shows that significant strain and Si bond weakening may play a major role for both Ag crystallite nucleation and growth. Furthermore, to study the impact of defects in general on Ag crystallite formation, mechanical defects were intentionally produced on wafers without P diffusion, and multicrystalline Si with its dislocations and grain boundaries was metallized. While sub-surface micro-cracks and dislocations promote Ag crystallite nucleation, but not their further growth, there is no Ag crystallite formation at the studied electrically inactive grain boundaries. It is suggested that tensile stress in Si triggers Ag crystallite formation while compressive stress does not. One of the aims of this thesis was to contribute to the fundamental understanding of the screen printed Ag contact formation. Requirements for achieving a good contact, like surface sharpness and glass layer coverage were discussed. For future investigations, it would be interesting to use the results of this thesis and test other contact metals in order to reach the transition towards the reduction or replacement of silver in the front screen printed paste. Finally, whether or not the Ag crystallites are in direct contact with the silver finger or in quasi direct contact (separated from it by a very thin glass layer of less than 1nm) will continue to be discussed. However, the presence of Ag crystallites is required for the current transport. In fact, for all the industrial silver pastes of different generations and manufacturers that were investigated in this thesis, we always observed Ag crystallites underneath the glass layer or directly connected with the silver finger at the Si pyramid tips for optimal firing conditions. If it was possible to avoid the Ag crystallite formation while retaining a low specific contact resistance, an ideal contact could be achieved. Because there would be no metal penetration into the emitter and therefore no contact induced recombination losses, which are currently the main efficiency limiting factor. However, the glass layer is still not sufficiently conductive to allow for current transport without Ag crystallites. Therefore for future experiments, it would be interesting to measure the conductivity of the glass layer.