Could be a blown-up pin which was shorted or connected to an disadvantageous voltage-level. I would try to check (with a program) for dead or connected pins if the visual inspection of the board doesn't show any hint. I do not believe in chipworms or very small chipmunks. (;-)
That is interesting. We had some chips failed during protytping with 24V accidentally applied to the chip and the center part discolored. Never have a hole in the chip so far, :-)
Not sure if this works for todys plastic package material, but we used to use a
standard propane torch, take a part out to the parking lot, and torch it, plastic would
turn to ash, and we could examine die and leadframe for issues.
The principle way CMOS can develop high power shorts, localized heating, is the
parasitic SCR structure hung on I/O pins. If that gets triggered a short exists in that
region, between Vdd and Vss, usually blows out the bond wire on that pin. Or the
Vdd bond wire. Vss bond usually multiply bonded. Triggering mechanism is excess
current into a pin that has been driven below ground.
I'm afraid that fabricating Crème brûlée out of an integrated circuit could destroy some of the structures you'd like to investigate. Have a look at http://en.wikipedia.org/wiki/File:Creme_brulee.jpg. I remember to have red that "they" use a mixture of concentrated sulfuric acid with 5% chrome (VI) oxide. This will remove all organic material (including fingers) and should not be handled by amateurs or outside of a laboratory. I presume that there are a few companies offering a service for that.
Dana, Bob et al,
Do you think the missing decoupling capacitors could in any way be related to SCR latchup and subsequent 'venting' of the chip. Of course by 'venting' I really mean explosion/destruction. :-)
Short answer is yes.
Input excursions beyond Vdd or < Vss will do it, it only takes typically
a few 100 nS to a couple of uS to trigger the SCR, latchup.
Sorry for the delay in replying!
Out of idle curiosity I wondered if any of the hardware team at Cypress could identify roughly which part of the chip that was which had vented - it's probably one of the VCC bond wires, or above an SCR pair I'd guess.
Most semiconductor operations have a Failure Analysis department that
can "de-lid" parts and establish what happened. Try contacting Cypress's
department on this.