Determination of InP surface state density distribution from excitation-power-dependent photoluminescence spectra using genetic algorithm-based fitting procedure

The energy distribution of the surface state density N_SS(E) on the n-type InP surface was determined from a rigorous computer analysis of the dependence of the band-to-band photoluminescence efficiency (Y_PL) versus excitation light intensity (Φ). The analysed experimental Y_PL-Φ spectrum, taken from the literature, was obtained at room temperature for the chemically polished n-InP₁₀₀ surface. The theoretical values of Y_PL were calculated using a numerical simulator, taking account of all possible bulk and surface recombination processes. We reproduced very well the experimental Y_PL-Φ dependence, which exhibited a strong increase of Y_PL. We also determined the continuous U-shaped N_SS(E) distribution from the best fit to the experimental data by applying a novel fully-computer procedure based on genetic algorithm. Furthermore, we found that the surface non-radiative recombination dominates over the bulk recombination channels (in terms of recombination rates), and governs the behaviour of radiative recombination versus Φ.