ALD strength on 5.03

To follow up on some of Dave’s work – I was curious if the ALD alumina affects the strength of pnc-Si membranes.

Punchline: 2.5 nm of Alumina seems to increase the burst pressure, while 5.0 nm reduces the burst pressure.

Methods: As part of routine QC, we had already measured robustness on 6 SepCons to be: 7.7, 8.0, 12.4, 15.0, 15.8, 17.1 for Wafer 5.03. I randomly sampled SepCons from one quadrant and pre-tested each to ensure that they would survive 8.0 psi on the burst pressure cell. I sampled until I had 16 chips. This pre-test would allow me to run a smaller sample size and have an effective ‘floor’ burst pressure. If all chips after ALD had burst pressures below 8.0 psi, we could confidently say that ALD hurt the robustness. It’s a little harder to make the conclusion the other direction. I split the 16 chips into two groups and Dave helped me (he actually ran the tool) deposit 2.5 nm and 5.0 nm of alumina based on his prior recipes.

Results: I tested burst pressure on four chips from each group. The remaining four chips will be used for separations with data shared shortly.
2.5 nm: 15.4, 20.7, 21.4, 22.9 psi
5.0 nm: 2.9, 6.5, 10.3, 14.5 psi

Conclusion: It’s clear that 2/4 chips with 5 nm alumina failed at a lower burst pressure than what they experienced during the pre-test. The average of the four was also significantly lower than the average of the control chips.
The chips with 2.5 nm of alumina had a significantly higher burst pressure than the control chips on average. However – I realized my design of experiment is biased because I would have thrown out 1 or 2 chips from the control set if I had performed a pre-test of 8.0 psi. I should have selected something like 5.0 psi. Even if you compare the four control chips with burst pressure >8.0 psi, the average is still significantly lower than the four chips with 2.5 nm of alumina. I think we can say that 2.5 nm of alumina increased burst pressure.

Dave told me about the possible tensile nature of the alumina based on some of his experiments. I’ll let him share that at the meeting – maybe this explains why a little alumina is good, but more is bad for pnc-Si.