If a thickness of a semiconductor is smaller than the penetration depth of the electron beam, e.g. in silicon on insulator (SOI) structures, only a small portion of incident electrons energy , which is lost in a superficial silicon layer separated by the oxide from the substrate, contributes to the electron beam induced current (EBIC). Because the energy loss distribution of primary beam is not uniform and varies with beam energy, it is not straightforward to predict the optimum conditions for using this technique. Moreover, the energy losses in an ohmic or Schottky contact complicate this prediction. None of the existing theories, which are based on an assumption of a point-like region of electron beam generation, can be used satisfactorily on SOI structures. We have used a Monte Carlo technique which provide a simulation of the electron beam interactions with thin multilayer structures. The EBIC current was calculated using a simple one dimensional geometry, i.e. depletion layer separating electron- hole pairs spreads out to infinity in x- and y-direction. A point-type generation function with location being an actual location of an incident electron energy loss event has been assumed. A collection efficiency of electron-hole pairs was assumed to be 100% for carriers generated within the depletion layer, and inversely proportional to the exponential function of depth with the effective diffusion length as a parameter outside this layer. A series of simulations were performed for various thicknesses of superficial silicon layer. The geometries used for simulations were chosen to match the "real" samples used in the experimental part of this work. The theoretical data presented in Fig. 1 show how significandy the gain decreases with a decrease in superficial layer thickness in comparison with bulk material. Moreover, there is an optimum beam energy at which the gain reaches its maximum value for particular silicon thickness.
Study of Low Voltage Prebreakdown Sites in Multicrystalline Si Based Cells by the LBIC, EL, and EDS Methods