Field Map for 25x100 pixels

Dear all,
I would like to simulate 25x100 µm2 planar pixels with the design I am joining, and whose details can be found for example in page 7 of this presentation.

Indeed I am simulating in TCAD two “face-to-face” pixels as in the attached picture and I was wondering which is the correct field_mapping to be used.

Any help will be greatly appreciated!

Many thanks in advance and best regards,
Marco Bomben

basic_pair_cells_cms_25x100.pdf (245.0 KB)

Dear @mbomben ,

The correct field_mapping would be SENSOR in this case I think. This will duplicate and flip the field periodically across the sensor, starting at the lower-left corner of the bottom-left pixel. So in this case the field is not tied to the individual pixel as such, but more to the sensor.

I am not entirely sure about the placement of implants however, but I know there are more people working on those sensors using Allpix Squared, so I can imagine e.g. @annvau might know more. There was a recent-ish merge request concerning this design and the coupling matrix for capacitive transfer: Option to mirror coupling matrix every second row (or column) (!1148) · Merge requests · Allpix Squared / Allpix Squared · GitLab

Kind regards,
Håkan

Hi Marco, hi Håkan,

actually, I’m not working with these sensors myself, I just helped a colleague who needed the “mirrored” crosstalk caused by the 2-pixel design.
For the field, @jschwdt is simulating a single pixel in TCAD right now, but my colleagues don’t put the bitten implant in Allpix2 anywhere so far afaik.

Greetings from Hamburg,
Annika

1 Like

Hi @mbomben

there will be an issue if the implant spans two pixels, because then we don’t know to which pixel the charge should belong. So What I would do is to define the pixel grid as the one of the readout chip and then superimpose the field of the sensor on top of is using the SENSOR mapping just as @hwennlof has suggested.

/Simon

Dear @hwennlof , @annvau and @simonspa , thanks a lot for the feedback!
I have a couple of questions but first of all let me share an annotated version of the screenshot I had shared before:
annotated_basic_pair_cells_cms_25x100.pdf (631.3 KB)

So @simonspa I don’t think there is an implant spans two pixels but just electrodes spanning two pixel cells, if I have understood correctly your point.

Next: if I see correctly this feature will be implemented in 3.2.0, right? So I should wait for that release or is there a way to use it already now, please?

Last: how to generate the correct weighting field map for these pixels, please? Should I just create one for the bottom pixel in the figure (within a larger matrix of course) and then rely on flip_odd_rows key, please?

Many thanks again you all,
Marco Bomben

Ah, now I understand, that’s allright then!

Yes, and we just tagged 3.2 a few minutes ago! :slight_smile:

I need to think about this a bit actually, because for the weighting potential, we need a pixel-centric mapping…

/Simon

Hi @simonspa , yes, you raised a good point about weighting potential.
I gave it a thought and here is what I think:

  1. There is a 1-1 correspondence between sensor cell and roc via the underbump (https://allpix-squared-forum.web.cern.ch/uploads/short-url/g3HEPgWiRysjEu1dBo1aI8naqiE.pdf). So we can say that T(op) sensor cell corresponds to L(eft) roc and B(ottom) sensor cell corresponds to R(ight) roc.
  2. I think we should consider an induction matrix (already for calculating the weighting field) which is 3 x 3 x 2 pixels, where the ‘2’ refers to couple of T-B sensors. It is easy in TCAD to calculate the weighting potential for T and B sensors separately.
    Yes, for once I am thinking of not exploiting symmetries: I would like to have a phi_T and a phi_B weighting potential maps. The induction matrix itself does not have a well defined center and also I believe there is no 180 deg rotation symmetry in the induction matrix.
  3. For a given \vec{r}_{deposition} I would proceed to calculate the induced signal in the T-B pair of pixels whose volume contains \vec{r}_{deposition}, separately for T and B sensors, using the relative weighting potential map (phi_T for T and phi_B for B (sorry for being pedantic but I myself got confused a few times with this proposal so I want to make sure I at least explain it correctly)). Then I shift \vec{r}_{deposition} by the usual +/- 1 pitch in both directions to calculate the induced signals in all other 8 pairs of T-B sensors.
  4. Then all T sensors will induce a signal in a L roc and all B sensors will induce a signal in a R roc.

Does this seem correct from a theoretical point of view, please?

Then, for the implementation: two weighting potential maps per pixel module are needed. Unless there is a way to exploit a transformation of the map… but again, the “minimal” induction matrix I am suggesting is not symmetric and I don’t see how to build one.

Please let me know what you think.

Best regards,
Marco Bomben