Date of Award
Bachelor of Science Honours
Faculty of Computing, Health and Science
Dr Stephen Hinckley
This research has made a comparative investigated of crosstalk in backside illuminated and frontside illuminated single junction photodiode and vertical double junction photodiode CMOS compatible pixels, using a commercial 2D device simulation package, SEMICAD DEVICE (1994). Comparison of pixel total, electron and hole quantum efficiency response and Absorption Volume data is undertaken. This is so that the underlying carrier drift-diffusion dynamics responsible for optimal pixel response resolution may be qualitatively understood, allowing prediction of even more optimal photodiode pixel configurations. The effect of varying the double junction and single junction photodiode pixel's geometry on response resolution is considered. Only for the former is the effect of doping, biasing and introducing highly doped pixel boundary trenches on response resolution undertaken. Additionally, for the former pixel, the effect of introducing a guard-ring electrode on its electrical response resolution is investigated. For single junction photodiode pixels, the boundary trench isolation, a highly doped recombination boundary trench placed either side of each pixel's well, showed considerably less pixel response resolution and hence more crosstalk than using the guard-ring electrode configuration. However the boundary-trench-isolation-pixel's response was an improvement on the unguarded single junction photodiode pixel's response. The outer junction of the double junction photodiode pixel acts in the same way as the guard-ring electrode for the single junction photodiode pixel, by suppressing the pixel response away from the pixel centre. However for pixels with similar geometry of outer well and substrate to the single junction photodiode guarded pixel, the outer junction "guard" improves the pixel response resolution more than the electrode guard does. However for shallow pixels their response resolutions are not significantly different, in that response outside the "well" (outer well in double junction photodiode pixels) is insignificant. The response resolution is more flexibly varied inside this "well" for the double junction than for the guarded single junction photodiode pixel. Generally the frontside illuminated photodiode pixels have better response resolution and hence crosstalk suppression than the same pixel backside illuminated. This is due primarily to their greater depletion region absorption volume proportion. This results from the closer proximity of their photogenerated carrier-envelope to their pixel's depletion region. However as frontside and backside illuminated pixel absorption volume proportions converge, their response resolution becomes less distinguishable. The predictive advantage of pixel absorption volume data for optimal pixel response resolution is evident. Such data can help to narrow the selection of possible optimal pixel configurations. However simulation is still the necessary final arbiter without the more costly fabricated-device testing option available.
Jansz, P. V. (2003). Device Structure Effects on Electrical Crosstalk in Backwall Illuminated CMOS Compatible Photodiode Arrays. Retrieved from http://ro.ecu.edu.au/theses_hons/355