Membrane anneal study
Last week, we reported that it was possible to anneal membranes after they had been etched. Below is a summary of results from membranes that were annealed over a range of temps and under different conditions.
| w600 (20 nm oxide) | w510X (40 nm oxide) | |
| amorph. anneal 700 C | no defects | no defects |
| amorph. anneal 750 C | no defects | no defects* |
| amorph. anneal 800 C | no defects | no defects |
| anneal w/ oxide 950 C | no defects | no defects |
| anneal w/ oxide 1000 C | defects | defects |
| anneal w/ oxide 1050 C | defects | defects |
* This sample should have flattened after the anneal (see 20 nm oxide sample annealed at 750 C). I believe there was still oxide remaining on this sample (hence the wrinkles).
“Amorph. anneal” indicates that the protective oxide was stripped prior to RTP treatment thus exposing the amorphous silicon. “Anneal w/ oxide” denotes membranes that were annealed with the protective oxide intact and subsequently stripped with BOE. All anneals were done with the susceptor for 60 s with a ramp rate of 10 C/s.
We can see that annealing membranes with protective oxide above 1000 C creates defects concentrated along the edges of the window. These are not circular “pinholes” but rather “tears” in the silicon.
It’s not surprising that the defects are concentrated along the edges of the membrane since the stress/strain is greatest at the interface. It’s also interesting that when the silicon film is unrestrained (no protective oxide), defects do not occur.
My next group of experiments will focus on a two step anneal: ramp to a intermediate temperature ~ 500 C, hold for 30 s, then ramp to the set point, hold for 60 s. The idea is to minimize the strain.
There’s also the question of what the nanostructure looks like for these conditions.
Dave,
Have you had a chance to TEM some of these recent samples? I’m interested to know when (temp.) we see pore formation, and if pores form when the oxide is stripped off the a-Si before the anneal.
I like the idea of double annealing. If it doesn’t work I have a few more ideas maybe we could discuss.
This is a huge parameter space when you consider the timing aspects of multiple anneals, oxide stripping, gas type and flow, temperature profiles, etc. I would suggest working with a partner who can do flow rate and burst pressure in real time after these membranes are made. We cannot TEM everything, but if we throw out anything with visible pinholes or no flow or <2PSI burst strength, we can limit our imaging to to those materials that show promise. Once we identify some promising recipes, then we can explore the local parameter space with TEM. There must be a set of conditions that will work, since these first few attempts are so close…