Only recently, methods for quality control of multicrystalline silicon wafers have been published, which allow the efficiency of solar cells to be predicted precisely from photoluminescence (PL) images taken in the as-cut state. In this letter it is shown that oxygen precipitates, present in standard Czochralski silicon wafers, can cause efficiency losses of more than 4% (absolute) within an industrial solar cell proc- ess. These efficiency losses correlate with ring-like defect structures of reduced intensity in the PL image. In comparison with QSSPC-based lifetime measurements, we introduce a PL-based method of quality control which allows the critical wafers to be identified and sorted out reliably at an early state of production and thus increases yield and average efficiency of production lines

No abstract

Light-induced degradation (LID) due to boron-oxygen complex formation seriously diminishes the minority carrier lifetime of p-type Czochralski-grown (Cz) wafers. Depending linearly on the boron concentration NA in uncompensated silicon, the boron-oxygen defect density was suggested to depend on the net doping concentration p0 = NA - ND in compensated p-type samples, containing similar amounts of boron and phosphorus [D. Macdonald, F. Rougieux, A. Cuevas, etal., Journal of Applied Physics 105, 093704 (2009)]. However, this dependency contradicts observations of LID in compensated n-type silicon wafers [T. Schutz-Kuchly, J. Veirman, S. Dubois, etal., Applied Physics Letters 96, 1 (2010)], which are confirmed in this study by investigating the boron-oxygen complex formation on a large variety of compensated p- and n-type samples. In spite of their high boron content, compensated n-type samples may show a less pronounced LID than p-type samples containing less boron. Our ex periments indicate that in compensated silicon, the defect concentration is only a function of the compensation ratio RC = (NA ND)/(NA - ND)

Knowledge of the carrier mobility in silicon is of utmost importance for photovoltaic applications, as it directly influences the diffusion length and thereby the cell efficiency. Moreover, its value is needed for a correct quantitative evaluation of a variety of lifetime measurements. However, models that describe the carrier mobility in silicon are based on theoretical calculations or fits to experimental data in monocrystalline silicon. Multicrystalline (mc) silicon features crystal defects such as dislocations and grain boundaries, with the latter possibly leading to potential barriers through the trapping of charge carriers and thereby influencing the mobility, as shown, for example, by Maruska [Appl. Phys. Lett. 36, 381 (1980)]. To quantify the mobilities in multicrystalline silicon, we performed Hall measurements in p-type mc-Si samples of various resistivities and different crystal structures and compared the data to majority carrier Hall mobilities in p-type mo nocrystalline floatzone (FZ) silicon. For lack of a model that provides reliable values of the Hall mobility in silicon, an empirical fit similar to existing models for conductivity mobilities is proposed based on Hall measurements of monocrystalline p-type FZ silicon. By comparing the measured Hall mobilities obtained from mc silicon with the corresponding Hall mobilities in monocrystalline silicon of the same resistivity, we found that the mobility reduction due to the presence of crystal defects in mc-Si ranges between 0 and 5 only. Mobility decreases of up to 30 as reported by Peter [Proceedings of the 23rd European Photovoltaic Solar Energy Conference, Valencia, Spain, 1-5 September 2008], or even of a factor of 2 to 3 as detected by Palais [Mater. Sci. Eng. B 102, 184 (2003)], in multicrystalline silicon were not observed