Apparently, fast annealing at high temperature (400 degrees C) is detrimental to the wafer quality (wafer 80).
Four 6 in. wafers have been annealed at 150 degrees C for 2 weeks (one wafer at LBNL and three at Honeywell). The following is the combined summary of changes in die classification for these wafers:
good -> fair : 0 die
good -> bad : 5
fair -> good : 3
fair -> bad : 5
bad -> good : 3
bad -> fair : 4
No change : 522
The changes in the classification do not seem to be directly related to the problem of bad vias. For example, from 5 chips which went from good to bad, 2 failed analog logic tests (bad cell IDs), 2 got several bad channels, and 1 exhibited too large DVDD curent.
Annealing at 150 degrees C for 2 weeks didn't change anything on the 4 in. wafer used as a control (wafer 148).