Some locations still have M5-M4 vias, while others do not. Need to check the progress more often.
And so, another attempt at the same. This is certainly overdone as metal-4 seems to have been attacked in some places, but a marked improvement overall.
Even more interesting, while oxide interference disappeared, some other colours persisted. Silicon nitride?
Easy to confirm how many metal layers we have here.
Going back to my practice chip, Infineon M7690, I tried the alumina suspension on it too. Certainly, the results are different compared to diamond slurry. Confirmed it goes *fast*. The right shot was made after about 4x60s on low speed, and the most of passivation and top oxide + metal is gone, only nitride lining remains in a couple spots. The left image is for reference; wherever pad metal was eaten away, there are pits at or below metal-X.
Another interesting artefact is the pattern on nitride lining. I wonder how this came to be and whether this is something produced by my setup or a manufacturing quirk.
The last image seems to show a variation in thickness caused by the previous layer, which is interesting because this should have been removed by planarization.
The same in oil immersion. All the interference colours gone, and there's slightly more contrast.
Full-res map: http://infosecdj.net:8086/map/nxp/mp091a0/infosecdj_mz_nikpa60x_s2/
11/n
After a bit more work. This is imaged using a dry objective.
Planarity is shit but no visible scratches, which is amazing.
Materials used: Allied 0.3u alumina suspension, Ted Pella Met-X cloth
10/n
One digit is gone from the maskset revisions. :-)
9/n
mole99
