Nanoscale modelling of point and extended defects in mc-Si solar cell

Abstract

Multi-crystalline silicon (mc-Si) based photovoltaic cells are generally accepted to be cost-effective for large scale production techniques. However, it contains a relatively high concentration of point and extended defects (EDs), both of which act as recombination centres limiting the cell efficiency. EDs such as dislocations, stacking faults, grain boundaries (GBs) and voids interact with mobile point defects, such as iron which is a common contaminant in some grades of silicon. Fe is a problem since interstitial iron (Fei) diffuses rapidly, is electrically active, and is a non-radiative recombination centre. Different processing techniques are used to getter Fe to improve solar cell efficiency; internal gettering by EDs is one possibility. Fe-Si interactions are complex and not fully understood, particularly when they are associated with EDs; there is relatively little atomistic-level data for the mechanism of the segregation of iron and its complexes at EDs, partly because modelling of the EDs is challenging in terms of the system size, and partly because GB structure is a matter of debate. In this work, density functional theory was used to investigate Fe behaviour in bulk Si and at EDs. Trends in the energetics, magnetic, geometric, electronic and electrical properties of substitutional, Fe–vacancy, interstitial and interstitial-pair structures have been studied. Of particular note, the formation of Fei–pairs in bulk Si is energetically favourable, with the equilibrium spin states being sensitive to the Fe inter-nuclear separation. Overall, the third-neighbour pair structure in an antiferromagnetic form is the most stable, bound relative to two isolated Fei by 0.2 to 0.6 eV, depending upon charge state and inter-nuclear distances. Furthermore, the migration barrier for Fei- pairs is lower than that of individual interstitials, suggesting pairs may be key contributors to Fe diffusion and precipitation. Fe-defects are also modelled in intrinsic stacking faults, Σ3-(110), Σ5-(001) twist GBs and voids, and it is concluded that all forms of EDs represent binding sites for Fei. Although Fei binds relatively weakly at fully bonded GBs, it is strongly trapped by vacancies at the GB, for example and, perhaps more critically, by Fe already trapped there: the binding energy of a Fei–vacancy pair at a Σ5-(001) twist GBs is 1.6 eV, and the binding of a second Fei atom to the GB is found to be greater than the first (also by around 1.6 eV). The results showed that the most stable site for iron lies just outside the modelled void (bound by 2 eV), which may explain the experimental observations that only a single layer of iron forms at voids. Finally, quantum-chemical simulations were combined with TCAD device modelling to examine the properties of the EDs before and after segregating iron impurities, in an attempt to balance the interoperation of the advantageous and disadvantageous properties of these defects and impurities on the performance of Si solar cells. The results show that reasonable efficiency gains were obtained and the overall efficiency of the cell is improved by segregating iron from the grains to the GBs.