In this study, synchrotron-based x-ray absorption microspectroscopy (mu-XAS) is applied to identifying the chemical states of copper-rich clusters within a variety of silicon materials, including as-grown cast multicrystalline silicon solar cell material with high oxygen concentration and other silicon materials with varying degrees of oxygen concentration and copper contamination pathways. In all samples, copper silicide (Cu3Si) is the only phase of copper identified. It is noted from thermodynamic considerations that unlike certain metal species, copper tends to form a silicide and not an oxidized compound because of the strong silicon-oxygen bonding energy; consequently the likelihood of encountering an oxidized copper particle in silicon is small, in agreement with experimental data. In light of these results, the effectiveness of aluminum gettering for the removal of copper from bulk silicon is quantified via x-ray fluorescence microscopy (mu-XRF), and a segregation coefficient is determined from experimental data to be at least (1-2)'103. Additionally, mu-XAS data directly demonstrates that the segregation mechanism of Cu in Al is the higher solubility of Cu in the liquid phase. In light of these results, possible limitations for the complete removal of Cu from bulk mc-Si are discussed.

Synchrotron-based microprobe techniques have been applied to study the distribution, size, chemical state, and recombination activity of Fe clusters in two types of mc-Si materials: block cast mc-Si, and AstroPower Silicon Film(TM) sheet material. In sheet material, high concentrations of metals were found at recombination-active, micron-sized intragranular clusters consisting of micron and sub-micron sized particles. In addition, Fe nanoparticles were located in densities of ~;2'107 cm-2 along recombination-active grain boundaries. In cast mc-Si, two types of particles were identified at grain boundaries: (1) micron-sized oxidized Fe particles accompanied by other metals (Cr, Mn, Ca, Ti), and (2) a higher number of sub-micron FeSi2 precipitates that exhibited a preferred orientation along the crystal growth direction. In both materials, it is believed that the larger Fe clusters are inclusions of foreign particles, from which Fe dissolves in the melt to form the smaller FeSi2 nanoprecipitates, which by virtue of their more homogeneous distribution are deemed more dangerous to solar cell device performance. Based on this understanding, strategies proposed to reduce the impact of Fe on mc-Si electrical properties include gettering, passivation, and limiting the dissolution of foreign Fe-rich particles in the melt.

Synchrotron-based microprobe techniques were used to obtain precise and systematic information about the size distribution, spatial distribution, shape, electrical activity, and chemical states of iron-rich impurity clusters in multicrystalline silicon materials used for cost-effective solar cells. These experimentally observed properties of iron-rich clusters allow one to derive conclusions about the origins of iron contamination, the mechanisms for incorporating large amounts of Fe into mc-Si, quantitative information about the distribution of Fe in mc-Si and the impacts of such contamination on solar cell performance. Two distinct groups of iron-rich clusters have been identified in both materials: (a) the occasional large (diameter greater than or equal to 1 mu-m) particles, either oxidized and/or present with multiple other metal species reminiscent of stainless steels or ceramics, which are believed to originate from a foreign source such as the growth surfaces, production equipment, or feedstock, and (b) the more numerous, homogeneously distributed, and smaller iron silicide precipitates (dia. less than or equal to 800 nm, often < 100 nm), originating from a variety of possible formation mechanisms involving atomically dissolved iron in the melt or in the crystal. It was found that iron silicide nanoprecipitates account for bulk Fe concentrations as high as 1014-15 cm-3 and can have a large negative impact on device performance because of their homogeneous distribution along structural defects. The large (dia. greater than or equal to 1 mu-m) particles, while containing elevated amounts of metals, are low in spatial density and thus deemed to have a low direct impact on device performance, although they may have a large indirect impact via the dissolution of Fe, thus assisting the formation of iron silicide nanoprecipitates. These results demonstrate that it is not necessarily the total Fe content that limits mc-Si device performance, but the distribution of Fe within the material.