open-source-silicon.dev

image.png

First attempt at lithography simulation of the 1RW bitcell array poly using <https://github.com/vincentlv/DimmiLitho|DimmiLitho> (with a bunch of hacks/fixes).

Right now, this is NOT using the OPC but instead using its own ILT-generated OPC.

Set the NA to 0.69 which seems reasonable for 130nm from what I've seen. Not sure about the other parameters for the resist, lens, etc.

Left: Mask
Mid: Exposure intensity
Right: ILT Mask

The LI layer gets a bit more complicated:

This is poly and LI with the OPC from Skywater:

It's hard to tell, but there is some difference in the exposure:

<@U0175T39732>:  This is very cool.  What are you doing with the software hacks and fixes?  Is the project active?  Can you do a pull request?

<@U016EM8L91B> I'm planning to. 

1) They have a hacky gds parser that ignores the database units and scale in gds. 
2) it assumes that the pattern is tiled or with a percentage of width/height boundary. I added an option to use a bounding box option but then realized the array has orientation changes as well so changed to a small array.
3) it needs to use add/drop mask layers with boolean operations. I manually did the add layers and this one didn't have drop, but the diff layer has both add and drop layers.

4) the intensity doesn't account for the resist and so on. They use those models for the ILT but not the "simulation". So I need to add that too