Metallic impurities in and on polysilicon

Metallic impurities, particularly metallic impurities interacting with crystal defects detrimentally limit the lifetime of charge carriers in silicon solar cells (Breitenstein et al., 2010; Bronsveld et al., 2010; Buonassisi et al., 2005; Ceccaroli and Lohne, 2006; Hashim et al., 2007; Kwapil et al., 2009; Seibt et al., 2009b; Seifert et al., 2009). For metallic contamination, the instrumental neutron activation analysis (INAA) provides a very reliable, however, tedious, method, metal results can be expected with detection limits down to 10_ 16g/g silicon (Huber et al., 1993; Verheijke, 1992). After optimized preparation and g-spectrometry, SEMI PV10 describes the INAA procedure for analytes with atomic numbers beyond silicon (SEMI PV10-1110, n. d.). Arsenic, antimony, and phosphorous intensify the background, therefore cannot lie higher than 1013 atoms/cm3 and 1016 atoms/cm3, respectively.

Using glow discharge mass spectrometry (GDMS; Verheijke, 1992; Hockett, 2008; Putyera et al., 2010; Wang et al., 2009; Di Sabatino et al., 2011) or ultrasonic sampling electrothermal vaporization inductively coupled mass spectrometry (USS-ETV-ICP-MS; Hsiao et al., 2011), polysilicon can be analyzed without dissolution. GDMS provides a good overview on metallic and nonmetallic contamination but requires advanced skills to sort out mass interferences. The reliability of quantification of GDMS depends on the homogeneity of elemental distribution, analytical instrument, and the calibration, the relative reproducibility can vary 5-12% (Di Sabatino et al., 2011). USS-ETV – ICP-MS’ detection limits lies in the region oflower ng/g (Hsiao et al., 2011).

After dissolution ofthe polysilicon sample, metallic impurities can be de­termined by ICP-MS (Masafumi et al., n. d.; Nikkei M., 1996) like metallic impurities in monocrystalline wafers (Fabry et al., 2006; Shabani and Shigeru, 2002). For reporting results and the method detection limit (SEMI C10-0305, n. d.), matrix interferences must be sorted out in order to avoid erroneous identification of analytes (Shabani et al., 2003). Interlaboratory testing leads to better understanding of sampling, sample preparation, and instrumental analysis (Hang Chui et al., 2005). However, the experimenter must be aware of its statistical limitations due the low level of concentration (Horwitz, 1982).

The precision of trace-analytical results fundamentally depends on the sample size: Si can be determined in silicon from a reasonable size of sample 99.1 ± 0.2%, but for Ca 0.05 ± 0.01% 100-fold of that size would be needed (Nygaard, 1996). In homogeneity of the samples, particularly the inhomo­geneous particle size distribution contributed most strongly to the uncer­tainties in the internally repeated results as well as in the interlaboratory analyses (Kennedy and Buseth, 1996; Siu Hang Chui et al., 2004). Facing the complexity of feedstock effects on solar cells, the industry has requested standardized specification for solar polysilicon for a long time (Nyhus, 2010).

Updated: June 30, 2015 — 7:21 pm