I was looking at the debugging plots, found in the TransientPropagation module, and I noticed a big difference in the induced charge between holes and electrons with and without gain (I send you the plots).
The reason is that the electrons from impact ionisation are generated close to the implant, hence they drift for about a micron or even less and don’t contribute too much to the signal, if the weighting potential is about linear. The additional holes, though, drift through the whole sensor depth and thus contribute with a large signal.
We have added a draft for an alternative impact ionisation algorithm:
Would you be interested in testing it? In short, what you’d need to add to the TransientPropagation would be
multiplication_depth = 10 # that's more a guess, can be increased
multiplication_probability_based = true
thanks a lot for your suggestion!
we’re testing your new Impact Ionization algorithm and we can confirm that we obtain some better result. In the following plot you can find 3 average waveforms with no gain (green), Van Overstraeten-De Man Model (blue) and Okuto-Crowell Model (red), with the additional parameters multiplication_depth and multiplication_probability_based, as you suggested to use.
from the TCAD simulation we expected a maximum current value around 6uA (with Van Overstraeten-De Man model) and we would like to tune our Monte Carlo simulation to have an output as similar as possible to the one from the TCAD. We can also discuss about this offline,
thanks a lot for all your work!!