No abstract

Variable-area diodes are routinely used to separate surface from bulk effects in the characterization ofHgCdTe photodiodes. In this work, the traditional models are reviewed and clarified by comparison to the results of numerical models of diffusion and other effects. It is shown that the characteristic lengths associated with lateral collection of both thermal and photo-generated carriers are not necessarily equal to each other and also are not equal to the minority carrier diffusion length. The effective surface recombination velocity in the traditional model for surface generation-recombination current is shown to have an alternative explanation as the effect of a surface inversion channel. Finally, the relation of the traditional models to the effects of point defects is examined. The average effects of defects with bulk or surface distributions are shown to contribute either zero or positive slope, respectively, in a plot of the inverse of the zero-bias resistance-area product vs the perimeter to area ratio. A negative slope arises from defects whose conductance increases faster than linearly with the diode radius. Distributions of diode dynamic resistance data from over 100 HgCdTe test arrays were analyzed to determine how the probability of a diode being defective varies with diode size. Plots of the diode conductance data ordered by increasing conductance indicate that the onset of defects occurs for a percentage of diodes of a given size which increases approximately linearly with the diode perimeter. This suggests that these defects have a perimeter rather than bulk origin.