Wednesday, June 01, 2016

TG Barrier and Low Light Linearity

A paper from Xi’an Microelectronics Technology Institute, China, "Measurement of charge transfer potential barrier in pinned photodiode CMOS image sensors" by Cao Chen, Zhang Bing, Wang Junfeng, and Wu Longsheng explores connection between TG barrier height, image lag and low light non-linearity:


  1. I wonder who reviews papers for J. Semiconductors? There are some serious questions I would have for the authors regarding their interpretation of measured results. For example, the "jump" at an exposure time of 0ms is attributed to partition "noise". I wonder where those electrons come from? And is it noise or just offset? How is capacitive coupling (opposite direction) manifested? Is the PPD SW truly empty at the start of each exposure cycle? How does one know? Why does thermionic emission start when the relative barrier gets small - isn't always in effect? It could be that the work is ok and just the interpretation of the data and device has some weak areas, but good reviewers should catch these things and get the paper tidied up. Anyway, haven't we all observed this initial flat zone, attributed it to charging lag, and estimated the lag (charge) from the width of the initial flat foot? In the end, of course, the flat zones need to be eliminated otherwise the PPD is not working properly. Might as well make a 3T device. Still, in course of fixing the PPD implants or operation, the estimation is useful. I suppose if you adjusted TG-high to a lower value, you should also see the same phenomena and further validate this model.
    The last comment, which is just wry humor, is that I notice this Chinese paper has 4T pixel charge transferring from right to left, but if the paper is written in English, shouldn't the charge transfer go from left to right?

    1. There is indeed a misleading A-A' path which is not coherent between Fig. 1-2 and 3 (TCAD plots).
      I agree with the questions about the 350e- that pop up instantly. If those electrons were left below the TG (charge partition hypothesis), they would end up in the FD anyway when TG goes off unless there is a worse barrier next to the FD :-)


All comments are moderated to avoid spam.