klayoutmatthias / sky130_xsection

KLayout XSection script for Skywater 130 process
The Unlicense
3 stars 1 forks source link

Add screenshots of cross section outputs #1

Open mithro opened 1 year ago

mithro commented 1 year ago

It would be nice to have some screenshots of the output from the cross sections and compared to the diagram at https://skywater-pdk.readthedocs.io/en/main/rules/assumptions.html#process-stack-diagram and in Tim Edward's presentation at https://isn.ucsd.edu/courses/beng207/lectures/Tim_Edwards_2021_slides.pdf

mithro commented 1 year ago

There are also some SEM images of cross sections available at https://drive.google.com/drive/folders/1g1bbKXEHyT3X1R70beO_45QzWtwWS_Ky

klayoutmatthias commented 1 year ago

I'm honestly not sure about the quality of the XSection output. Specifically in the frontend part my reasoning was that Sky130 would be a shallow trench isolation process. The stack diagram does not reflect this.

mithro commented 1 year ago

Some optical images where recently posted at https://x.com/johndmcmaster/status/1715863105782378529?s=20

mithro commented 1 year ago

I think the output would be better than random hand drawn diagrams that exist?

mithro commented 1 year ago

Some foundry provided cross section diagrams can also be found at https://skywater-pdk.readthedocs.io/en/main/rules/device-details.html#details

klayoutmatthias commented 1 year ago

The point is: the cross sections from the device details are not consistent with the stack diagram regarding the frontend part. The device details indicate shallow trench isolation for example while the stack diagram does not. My images are based on my guessing and gut feeling. The SEM images are not good enough to decide that and implants cannot be visualized by SEM anyway.

In that sense the cross sections from the script are rather virtual reality unless there is some kind of confirmation.

I'm saying that because it's very easy to draw wrong conclusions from incorrect cross section drawings. In the worst case, parasitic junctions are not considered properly and functionality is compromised or latch-up happens.

Matthias

mithro commented 1 year ago

Don't know if it helps, but a complete list of the steps in the SKY130 process can be found in the spreadsheet @ https://docs.google.com/spreadsheets/d/1PbI3IVNg93fR9Gi_hXlEDrlYtwFQuMyaD8PNEaIs3Sg/edit#gid=363070671&fvid=1378094797

mithro commented 11 months ago

FYI - I have been collecting information about the metal layers from the "Cypress Semiconductor Product Qualification Report". You can find it at spreadsheet at https://docs.google.com/spreadsheets/d/1ZB5XAwMF3zlGX8tET1-uyaFlDqHhBxabQ04PQGYwjZo/edit#gid=0 and https://docs.google.com/drawings/d/1wWBi4lQcLCZRgwqdpY8LDtjc-cHvMiD8_fxv4Ly46s4/edit

See also below; image image

mithro commented 11 months ago

So the S8 / SKY130 Process Steps spreadsheet has steps called "Shallow trench nitride etch" and "Shallow trench etch" (also shown below), so does that mean it is using shallow trench isolation?

# Step Description
1 SMAT Starting material
2 BOX Base oxidation
3 ISONIT Iso nitride deposition
4 FOM Field oxide mask
5 STINITE Shallow trench nitride etch
6 STIE Shallow trench etch
7 DNM Deep N-well mask
8 DNI Deep N+ implant
9 DNIS High V deep N-well implant strip
10 LINOX LINOX oxidation
11 FILOX Fill oxide deposition
12 CMPNIT CMP over nitride
13 NS19 Nitride strip
14 LVTNM Low Vt NOMOS mask
15 LVTNI Low Vt NOMOS implantation
16 LVTNIS Low Vt NOMOS implant strip
17 NWM N-well mask
18 NWI N-well implant
19 NWI2 NWI2 implant
20 LVTPI Low V P-channel implant
21 LVTPIS P-channel implant strip
klayoutmatthias commented 11 months ago

I'm not really an expert, but I think that is pretty much the flow of a shallow trench isolation process. The CMP with step 12 will basically planarize everything, leaving active area islands for the devices and STI regions for the space between them.

So I guess my cross section images are basically correct, but the cross section in the design manual makes less sense.

Matthias