Characterization of Structured Doping Areas Diffused from Doped CVD Layers on Crystalline Si Substrates

Abstract

Due to the special cell design, BC-BJ solar cells require adapted characterization methods, in particular to investigate the junction of the highly doped regions on the cell's rear. The focus of this Master Thesis was the development and implementation of a method to characterize single pn-junctions, which are located on the back of a BC-BJ solar cell. To achieve this objective, an simple process flow was developed to produce suiting samples with a single illuminated pn-junction and two TLM fields. For this purpose, the development of a precise alignment method was essential, which allows alignment with an accuracy of ~20 micrometer. All local processing steps are aligned directly or indirectly with the initially produced pn-junction. This applies to the contact openings, the metallization, and even the illumination mask, which is positioned on the front of the sample during the illuminated J-V curve measurement.Simulations conducted with Sentaurus Synopsis show, that the characterization method can distinguish samples with and without intrinsic gap region between the highly doped regions on the back of the cell, if this region is not defect free. A difference in the fill factor of 1.9 % between samples with and without the intrinsic gap is expected for the 'zebra structure', without consideration of tunnel recombination, which would further increase this difference. Furthermore, a comparison was made between the simulations programs Sentaurus Synopsis and Quokka, and differences were presented. In the parameter range of a highly efficient IBC solar cells, there is only a small error for the J-V characteristic quantities. In a more detailed investigation with the FELA, differences in the resistance loss at the BSF contact were found, which indicate that this loss channel is differently implemented into the two simulation programs. Due to the general consistency of the simulation results, Quokka was used as a simulation program to illustrate the influence of certain experimental parameters in the analysis.The difficulties encountered in the preparation of samples with a single pn-junction and the results obtained, were presented in the context of the individual process steps. In particular, the local removal of dielectric layers was discussed. This layer removal is relevant for the creation of the pn-junction, which is formed in a single diffusion process from borosilicate glass and POCl3 gas, but also for the opening of the contact surfaces. The local removal of dielectric layers by means of photolithography and laser ablation (ns and ps pulsed lasers) was carried out, analyzed, and comparisons were drawn between the two methods. The focus was clearly on laser ablation due to the industrial relevance of this method. In addition, J-V measurements show that the applied photolithography process does not provide performance advantages over laser ablation. Consequences of effects such as laser doping (with the ns pulsed laser), surface roughening in overlap areas, and shading CVD material remains were observed, characterized and tested for relevance in the sample production process. As a result, a process could be presented for the production of pn-junctions with and without the intrinsic region.Furthermore, the contacts were characterized by TLM measurements. Specific contact resistances of 9-24 mOhmcm2 were measured for the deposited Ti/Pd/Ag and Al/Ag contacts. The contact formation takes place directly during the deposition or during a short thermal annealing process. The specific contact resistance values are independent of the selected method of contact opening and the deposited metal. Shunt problems were also analyzed and corrected.Further relevant findings were obtained with regard to surface passivation. With thermal anneal, a high passivation quality could be achieved with Al2O3 and a-Si passivation. For this purpose, the passivation layers were coated with SiNx and thermally annealed at a temperature between 300 and 420 °C. A recombination parameter J_0 below 30 fA/cm2 was obtained for a 140 Ohm/sq p-type emitter passivated with Al2O3 or a-Si. The J-V measurements also underline the importance of adequate surface passivation. The passivation quality was tested for the rear (BSF, laser treated surface and emitter) and the front surface (texture). The absorption of a-Si was also analyzed because of its relevance for the passivation of the front surface with a-Si.The processing difficulties seem to be overcome at this time, which is why further sample processing with the presented manufacturing process should lead to a result with high characteristic quantities. In particular the increased lifetime, due to the postponed texturing step, and the non-existent laser doping, due to the use of the ps laser on the entire area that has to be ablated, are an improvement to the last processed samples. With these processed samples, the comparison between samples with and without the intrinsic gap can then be drawn. A clear distinction between these samples would be a milestone for the characterization method and would provide the proof-of-concept. Further investigations, with applied reverse bias voltage or with varied rear surface passivation, could provide additional information on the properties of the pn-junction.In addition, the concept can be extended to more than one pn-junction per sample. Sufficient information on the contact formation on samples has already been collected in the course of this work, which is why the TLM structure is no longer required on all samples. Adding pn-junctions on a sample has the advantage that a considerably higher number of measurable pn-junctions can be produced. In addition, conclusions about possible processing problems can be classified more easily. Alternating p- and n-doped regions on the entire sample's rear side also represent an approximation to the structure of an IBC solar cell.Moreover, an examination of the pn-junction using KPFM would be useful. Following the proposal in this work, the pn-junction could be polished at a small angle to the surface so as to obtain a depth resolution of the pn-junction with KPFM measurement. Accordingly, one obtains insight into the doping profile as a resolution in depth.