Generation–recombination noise modelling in semiconductor devices through population or approximate equivalent current density fluctuations

Abstract Generation–recombination noise modelling of semiconductor devices through the impedance field method customarily exploits, as microscopic noise sources, approximate equivalent current density fluctuations derived in a homogeneous system; moreover, such an equivalent approach often makes use of a monopolar (majority carrier) model only. Taking, as a reference model, an accurate numerical implementation of the Langevin approach wherein electron and hole density fluctuations are applied to a bipolar drift-diffusion physics-based model, we show that the approximate equivalent approach is inaccurate, even in a uniformly doped sample, unless the effect of the ohmic boundary conditions is negligible. Moreover, the monopolar equivalent model introduces further inaccuracies whenever minority carrier fluctuations are not negligible; this is shown to occur not only for direct processes, but also for trap-assisted transitions involving deep levels in a p - or n -doped semiconductor. The equivalent models are finally shown to match the exact approach only if the carrier lifetimes and fluctuation spectra are corrected empirically.