I have to preface my comments about what I think of as "the ageing problem" of the MEMS with the rider that I am only guessing as to the nature of the problem and its causes. Even an educated guess is still a guess, and mine may not even be that. It is certainly not an informed guess. I simply picked up on comments made in the virtual AGM and sent a query and comment in on-line. It was said (by Yuval Cohen?) that MEMS were testing OK when new but that some then failed when retested after a day or so. My comment was not responded to at the time and further questioning later has also had no response. I suspected that a thin silicon membrane with thin flexors that were made of composite layers could be subject to oxidation of silicon metal (semi-metal) to silicon dioxide, and that this could affect the mechanical properties of the structure. I do not believe this is a new or unexpected phenomenon in the semiconductor industry, but when these get used as rapidly moving parts such issues may become more significant. However I would also expect that this would be a relatively easily solvable problem.
I would assume that the chip's film protector will be made from, or incorporate, a layer of fluoropolymer (e.g. such as Teflon) to keep out dust, water, smoke particles, and oil mist etc. Silicone polymers would not do all of these duties as well as FP. Ideally it would also be electrically conductive so it does not accumulate a static charge. But again I am only taking a slightly educated guess on this.
-Claydon
The goal would be to make the most rugged, most highly performing product, as inexpensively and reliably as possible. Lessons from each wafer production run are clearly being worked into later production cycles to optimise the process. It is apparent that they are working closely with TowerJazz and that the Fab is very much accommodating their need to get to a production ready process on a shoestring budget. It is likely that upgraded versions of the first production run would also be made to second and subsequent runs (1.0 -> 1.x). I would be very surprised if no ways were found to improve or tweak the original product to make it even better over the years following its introduction (1.x -> 2.0 etc).
I also read this as alluding to the fact that the original device was to be the principal and only product offered by AP. But that was before the partial or half structure device was found to have potentially commercialisable properties. That could easily become a second product offered.
The other potential applications for the device and for other useful applications built upon its structure and function could easily prove very useful products for future development.
It is early days still.
NB: My comments are pure speculation and are neither technical nor investment advice.
(20min delay)