-
The response of a Si based sensor looks very much like the Landau curve: This distribution is a distribution of signal, which is proportional to the number of electron hole pairs, which again is proportional (says H. Bichsel) to deposited energy in the sensitive part (the so called EPI layer) of the sensor. So the MC calculation misses the smooth start (left part) of the curve. Literature tells me there is just a simple conversion from deposited energy to # of electron hole pairs (1 eh per 3.7 eV). Mr Tessier already gave some thoughts to this subject in #826, which stressed the peaky form of the MC calculation even more. Thank you Mr Tessier ! The peak is due to electrons traveling past the (thin) sensitive layer without leaving a trace ... because all their interactions are below cutoffs. Is there still a model of these energy transitions so we can fill the part between 0 and 4 keV ? And get rid of the ringing? Thanks again. |
Beta Was this translation helpful? Give feedback.
Replies: 3 comments 10 replies
-
The top curve is the response from a silicon detector, but I am not sure what it represents exactly (voltage distribution?). There is a lot going on between the energy deposition events and the production of the signal: bias, drift, recombination, gain, time resolution and so on. The Monte Carlo curve represents the distribution of energy deposited for each incident electron, so it is closer to what would be measured if you could detect a single electron at a time in the sensor; state of the art time resolution seems to be ~100 ps, corresponding to incident beam currents in the nanoamps range. It would be interesting to understand how this signal is convoluted into the detector response. To some extent, you can investigate this convolution with EGSnrc by scoring the energy of multiple histories bunched together. Or else more efficiently by just sampling the one-electron distribution to create bunches "by hand". Either way, the main sharp peak will get smoothed out. Keep in mind that the first "ringing" after the sharp peak is due to the |
Beta Was this translation helpful? Give feedback.
-
To clarify, from the original post:
The first peak, or rather the initial flat region is not an artifcat. It corresponds to physical reality: no electron can physically cross the layer and loose less that the corresponding CSDA energy loss, where you find the first peak (the fact that the distribution is not 0 below the peak is due to relatively rare backscatter events). The arttifact from the 1 keV |
Beta Was this translation helpful? Give feedback.
-
For those who share my (too low) patch abilities , here are the 2 files needed for the 'bunch' option: Just put them instead of egs_phd.cpp and egs_phd.h. spectrum bunch = 5 before This will score the average of 5 histories instead of each history in the spectruim file. |
Beta Was this translation helpful? Give feedback.
The top curve is the response from a silicon detector, but I am not sure what it represents exactly (voltage distribution?). There is a lot going on between the energy deposition events and the production of the signal: bias, drift, recombination, gain, time resolution and so on.
The Monte Carlo curve represents the distribution of energy deposited for each incident electron, so it is closer to what would be measured if you could detect a single electron at a time in the sensor; state of the art time resolution seems to be ~100 ps, corresponding to incident beam currents in the nanoamps range. It would be interesting to understand how this signal is convoluted into the detector response. …