MicroPorous Oxide Membrane Extraction
I recently popped out membranes from wafers 1064 and 1065 for our lab. I only started with about half a wafer, but these are the numbers from the extraction:
| Wafer ID | # of In-Tact Extracted Membranes : High/Low Porosity | # I broke during extraction | # already broken before extraction | # apparently under etched |
| 1065 (100W) | 147 : 81/66 | 7 | 25 | 0 |
| 1064 (150W) | 130 : 74/56 | 20 | 40 | 4 |
The extracted membranes were imaged within a few hours to determine whether they were high or low porosity. A day later, however, I observed that 2 of the previously intact 1064 membranes had ripped apart despite being stored in trenched gel boxes. None of the 1065 membranes did this. When extracting membranes from the wafer, the 1064 seemed much more prone to both membrane ripping and to wafer shattering. It was as if the 1064 was more brittle than the 1065.
I think it’s possible that something about the higher wattage process used for wafer 1064 resulted in the film being placed under a higher tensile stress. The other possibility is that the two wafers were etched differently which gave rise to the more brittle behavior in 1064 (though I don’t know why under-etching would lead to spontaneous ripping).
Assuming that cost and time requirements are roughly equivalent for both processes, I think it would be wise to use the 100W process going forward to avoid any potential tensile stress concerns.
Lower deposition power results in higher tensile stress.
There are other processing differences between the two wafers that you did not account for.
JP