An Examination of Three Common Assumptions Used to Simulate Recombination in Heavily Doped Silicon

A well-designed silicon solar cell entails a complicated optimisation of its heavily doped surfaces. There exist many analytical and numerical models to assist this optimisation, and each involves a foray of assumptions. In this paper, we examine three of the common assumptions: (i) quasi-neutrality, (ii) the use of effective recombination parameters as a boundary condition at the surface, and (iii) the use of Boltzman rather than the more complicated but more precise Fermi–Dirac statistics. We examine their validity and the computational benefits of their inclusion. We find that (i) the quasi-neutrality assumption is valid over a wide range of conditions, enabling a fast and relatively simple simulation of emitter recombination, (ii) the effective surface recombination velocity depends on doping under most conditions and can give the misleading impression that the density of interface defects depend strongly on surface concentration, and (iii) at carrier concentrations where Boltzmann statistics are invalid (above 10 cm, and especially above 10 cm), it is more difficult to mitigate the error by employing an effective model that combines degeneracy with band-gap narrowing, than to include the more general Fermi–Dirac statistics themselves.