c-Si solar cells based on laser-processed dielectric films

Abstract

This paper shows an innovative and low temperature fabrication technology for crystalline silicon (c-Si) solar cells where the highly-doped regions are punctually defined through laser processed dielectric films. Phosphorus-doped silicon carbide stacks (SiCx(n)) and aluminium oxide/silicon carbide (Al2O3/SiCx) stacks are used for the creation of n+ and p+ regions respectively. These films provide excellent surface passivation on both n- and p-type substrates with effective surface recombination velocity below 20 cm/s. Moreover, a wide laser parameter window for laser processing them leads to low recombination highly-doped regions that show emitter saturation current densities of 21 and 113 fA/cm2 for n+ and p+ emitters respectively. All this is combined in the DopLa (Doped by Laser) cell structure whose fabrication process can be reduced to wafer cleaning, film depositions, laser processing and metallization. As a proof of concept, 1x1 cm2 solar cells were finished on both p- and n-type substrates with promising results. The analysis of loss mechanisms shows that optical losses and technological issues in the translation of surface passivation from the solar cell precursors to the final device are limiting efficiency. Both are not inherent to DopLa structure and suggest room for improvement in these devices. On the other hand, the crucial role played by ohmic losses in the proposed structure is identified. These ohmic losses arise from the fact that both base contacts and emitter regions are defined in point-like patterns. Firstly, base contacts are defined in a 1 mm pitch square matrix introducing additional series resistance that can be reduced by using low resistivity substrates. Secondly, the emitter consists of laser processed local diffusions with an inversion layer emitter in-between characterized by very high sheet resistances. The ohmic losses introduced by this induced emitter are closely linked to the fixed charge density located at the dielectric/c-Si interface. As a result, we conclude that low resistivity n-type substrates fit better to DopLa cell concept because emitters are based on Al2O3/SiCx stacks which have higher fixed charge density, i.e. inversion layer emitters with lower sheet resistances, than their SiCx(n) counterparts.