The copper process
- Thread starter nsaspook
- Start date
https://www.eetimes.com/ibm-combines-low-k-material-copper-in-130-nm-process/
IBM combines low-k material, copper in 130-nm process
IBM combines low-k material, copper in 130-nm process
I remember well the headaches caused by switching to copper.
Copper metal required the introduction of minimum and maximum metal density and maximum width requirements that we had never had with aluminum. Density checks weren't compatible with the way any project did layout and I had to find a way to do design rule checks at different densities at different levels of construction so layout designers didn't waste time trying to fix density errors that they couldn't possibly fix, while also preventing major problems from being found later in the layout construction (and therefore closer to tapeout).
It did give us a one process generation pass on reliability verification.
I can't remember if we also introduced dualVt at 130nm, or if we did it at 180nm. Rule development was so late in the design flow that I needed to write a program to synthesize that mask. The program was so complex that we only used that methodology for one process generation.
My projects were always the first to go to high volume manufacturing on our new process, so it wasn't unusual for us to get hit with surprises on each new process.
Copper metal required the introduction of minimum and maximum metal density and maximum width requirements that we had never had with aluminum. Density checks weren't compatible with the way any project did layout and I had to find a way to do design rule checks at different densities at different levels of construction so layout designers didn't waste time trying to fix density errors that they couldn't possibly fix, while also preventing major problems from being found later in the layout construction (and therefore closer to tapeout).
It did give us a one process generation pass on reliability verification.
I can't remember if we also introduced dualVt at 130nm, or if we did it at 180nm. Rule development was so late in the design flow that I needed to write a program to synthesize that mask. The program was so complex that we only used that methodology for one process generation.
My projects were always the first to go to high volume manufacturing on our new process, so it wasn't unusual for us to get hit with surprises on each new process.
