Help with which Transfer modules to use

Dear Allpix-Squared Experts,

Hello. We are trying to simulate the response of planar sensors connected to RD53A readout in the presence of fluence of 4e15 neq/cm/cm. We turn on trapping in our allpix-squared configurations and we use a TCAD Electric and weighting fields in our simulations. We wanted to take your expert opinion on which propagation and transfer modules to use. Is [TransientPropagation] followed by [PulseTransfer] or [TransientPropagation] and [InducedTransfer] more accurate? The FAQ and FEP slides from the last Allpix-squared workshop suggest using [PulseTransfer] with [Transient Propagation]. The slides also suggest using [InducedTransfer] when you anticipate a large fraction of particles getting trapped.

A follow-up question in this regard. We also wanted to study the response of our detector in the presence of a linear electric field and a fluence of 4e15. We again turn on trapping and use TCAD weighting fields in our Allpix-squared configuration file. We use the [GenericPropagation] module and tried to see the effects of using [InducedTranfer] and [PulseTransfer] modules. The mean value of the pixel charge distribution in the above-mentioned cases is quite different and we are having a bit of difficulty with which transfer modules to use here. It’d be great if you could help us with this or point us in some direction to understand these differences in the distribution we see.

PS: I am not able to attach my plots on the forum. When I try to do this, I get a message saying that new users cannot add attachments on the forum :face_with_diagonal_mouth:

Here’s the link to my cerbox. You should be able to see the distribution here: CERNBox

CC: @mbomben

Hi @knakkali

thanks for your post! My suggestion would be to use PulseTransfer. The difference between the two modules is the following:

  • PulseTransfer takes the current pulse induced by each individual charge carrier (group) and combines them at the electrodes, essentially forming the signal you would have measured at each of them during the motion. This of course requires a time-resolved propagation simulation, which you have with TransientPropagation. In case you would use this module in conjunction with GenericPropagation, it would take the arrival time of each charge carrier (group) at the electrode and build a “pseudo pulse” from them - an approximation that works okay if the weighting potential is limited to the volume close to the electrode.
  • InducedTransfer is meant as a module for non-time resolved simulations performed with GenericPropagation. The only difference to SimpleTransfer is that it in addition to counting collected charge carriers which have reached the electrode, it adds contributions from charge carrier that only moved a fraction of the way through the sensor and e.g. got stuck because of trapping or recombined with the lattice. Here, we’re calculating the total induced current by looking at the initial and the final positions and their respective weighting potentials. The result of InducedTransfer will be a total charge, not a time-resolved pulse.

Concerning your follow-up question, form the information you provided it is hard to see where the difference might come from. One thing I’d like to know is why you are using a linear electric field here if the TCAD field seems to be available. When doing a transient simulation with a e-field and weighting potential that do not fit together, you will very quickly get unphysical results because the drift path does not match the weighting potential gradient, leading to odd induced currents mostly around the electrodes. Is there a reason for doing so?


PS: not sure what the attachment thing is, I’ll dive in to see if there is a setting for this…

Hello Simon,

Thank you so much for the very nice and clear explanation. We will proceed with our simulations using [Transient Propagation] and [PulseTransfer] modules as you suggested.

Regarding your question on why we are using a linear electric field even when TCAD simulations were available, this was just to explore and study one effect at a time. We understand now that this would in fact result in some unphysical results. Our end goal indeed is to study the behaviour of our detectors in irradiated electric fields.

[CC: @mbomben ]

Thanks a lot.

Hi @knakkali

if you want to do things step by step I would start with a non-timeresolved simulation using only the electric field and GenericPropagation. Then adding the weighting potential and moving to TransientPropagation is the next step in complexity, both computational and in the physics effects