Interpreting Activation Energies in Digital Image Sensors

We present a study demonstrating the dependence of apparent activation energies on signal charge in digital image sensors. The data presented in this paper are for a charge-coupled device imager, but the analysis can be applied to CMOS sensors or generally to any system that shows nonlinearity with respect to time. Activation energies for some pixels are observed to vary between values at about half the bandgap of silicon, when calculated at low signal-charge levels, to values approaching the bandgap when calculated at high signal-charge levels. As such, the traditional method of calculating activation energies using a single exposure time at varying temperatures will result in different activation energies dependent on the exposure time chosen, making it difficult to draw physically meaningful conclusions from its value. Therefore, a method of calculating activation energies using a single signal-charge level is proposed, making it easier to correlate activation energies with impurities. Further, we demonstrate how a model of a pixel with a fixed location impurity in conjunction with a moving depletion edge, due to a changing depletion region size, can lead to this behavior.