Best method for digitizing existing mask

[alanrocelec]

Hi, I'm researching the best methods for digitizing an existing mask. The mask shop we use indicates they cannot "blow-back" anything smaller than 2um and I'm dealing with .9um technology. I have been playing with stitching and converting microscope images however somthing tells me theres a better way. Any experts on the subject?

 

[IRstuff]

??? Isn't your existing mask made from a digital database?

TTFN

FAQ731-376

 

[alanrocelec]

I can't go into too much detail. Suffice to say we are licensed by the original manufacturer to reproduce the device. They have provided masks however the design files have been lost.

 

[IRstuff]

OK, been there...

Have you tried any other mask shop?

Barring that, your only other choice might be to find someone with a laser profilometer and have them scan and stitch something together. This could be a problem if the die size is large, though.

Are you wanting simply to make modifications to a single, existing layer? If so, you might be able to get your mask vendor to do a 1x copy of the mask to a new mask, then, before photodevelopment, add in the changes directly to the photoresist.

TTFN

FAQ731-376

 

[alanrocelec]

We need to reproduce the entire database to migrate the parts to a new process. I am going to demo a confocal lsm next week that should be able to automatically generate multiple images for stitching with sufficient resolution.

 

[IRstuff]

OK, I'm still a bit hazy here. Are you looking to do a die shrink? Otherwise, you could still run your old masks on a new process.

If you're talking die shrink, then you'd need to not only digitize, but convert to CAD format, since the scan format is probably not going to be the same as the raster format for the mask generator.

The other issue might be any built-in process compensations, which you'd need to undo or redo for the new process.

That's all assuming that the die shrink is drastic. If it's only 10-20%, people used to be able to get away with a straight photoshrink of their existing masks, and living with any issues with mask compensations or smaller bonding pads.

TTFN

FAQ731-376

 

[alanrocelec]

The old masks aren't compatible with the target fab photo process equipment. Also the stuff we've tried to shotgun into a new fab in the past unaltered was enough of a pain to tweak that we've learned to tweak the design to the new target process design rules up front, hence the need for design files. We have been experimenting with software for converting image (BMP, JPG, TIF etc) files to semicon design (gdsii) files with good luck. We just haven't generated an image for a whole layer for the device in question yet. I know mask shops can "blowback" (enlarge) and digitize masks I'm just not very familiar with the methodology. furthermore the shop we use indicated they can't resolve less than 2um structures anyway.