Impact on device performance and monitoring of a low dose of tungsten contamination by Dark Current Spectroscopy

The dark current in CMOS Image Sensors induced by deliberate contamination with tungsten ion implantation is studied with the Dark Current Spectroscopy (DCS) technique. We obtain quantized dark current peaks associated to the donor level of W in silicon. Accounting for rigorously Schockley-Read-Hall formalism and Poisson distribution of metal atoms, the technique allows to check the generation rate and the fingerprint of this deep level. We use this information for the study of a very low level of accidental contamination that produced an increase in the white pixel number of the sensors. This contamination is then identified as coming from W clusters having an average number of 30 atoms and impacting a limited number of pixels. A hypothesis on the origin of this contamination is proposed and confirmed by snapshots of the pixels in the dark environment.