What we're basically doing is using thermal evaporation to lay down thin films of metal on top of each other in a high vacuum environment. We're then patterning the cell with a fiber laser to produce the cell traces and patterns needed to cell size.
Here's an example of system that does what i've just described in the context of creating CDTE cells. Note that it doesn't use a laser to create the cell divisions but rather uses a conductive ink and sand blasting.
I thoroughly analyzed their design and I have copied some aspects of their design while avoiding many of the flaws that significantly limited it's practically and efficiency.
Here's a quick summary of the design changes and flaws that I found.
For one, the design specified in the paper can barely reach high vacuum which is a requirement for producing cells of reasonable efficiency.
One major improvement I'm working on is a system that allows for automatic changing depostion powder inside that chamber instead having multiple chambers for each layer of depostion powder.
The advantage of this approach is that miniaturizes the chamber size significantly, making closer to the size of something that could fit in the back of your car then something that needs a dedicated room or floor for.
I've also looked carefully at how im going to achieve high and maybe even ultra high vacuum. And in that regard I think I've made some significant strides.
My design achieves high vacuum in three stages, the first is through a simple ventri pump and the second is a sorption pump which has been redesigned based on a old paper I found here(). The last stage uses something called a non-evaporable getter pump.
Experienced vacuum engineers might initially be baffled of choice of pumps I am using. As they are normally considered too slow for the type of operation that the chamber is being put through.
However, the downsides of the speed of these pumps can be mitigated by three measures. (1)Building a chamber that can go through bakeout(which removes contaminates and reduces pump down time) (2)Designing a chamber with metal-to metal seals and low leak rate. Lastly the obvious principal of making the chamber volume and surface as small as possible while making the size of the pumps large.
I've barely scratched the surface here but I think this should give you a rough idea of what I'm doing. I really dont think this stuff is hard as how it's made out to be.
Here some resources that really have really helped out so far.
-Building Scientific Apparatus a book that should give you a broad overview of things you need to know
-vacuum sealing techniques alexander roth
Extremely exhaustive in the amount of information of about, valves, and just general construction of the chamber. The book is really old but everything still stands and it's honestly better than most of the stuff i've found online.
This sounds cool! Building Scientific Apparatus is a truly excellent resource.
Two questions: (1) Could you get away with an inert atmosphere? I'm not familiar with the pros and cons with respect to PVD.
(2) It sounds like your vacuum setup will have a long cycle time from vent to pumpdown to operation. A load lock with a (turbomolecular?) pump adds quite a bit of expense. What's your approach to achieving high throughput?
You could, I've seen a couple of papers attempt that approach with rather poor results something like 8-10% percent. Though I'd say by the easiest approach to producing thin films cells involves basically using electroplating which achieves similar efficiency (https://onlinelibrary.wiley.com/doi/abs/10.1002/pip.417).
The cycle time is of course a highly dependent on the final design of chamber. There's no reason that Sorption and Getter Pumps combined with a ventri prestage can't preform to a degree that meets the design requirements.
However their performance is highly dependent on two things, the ability to reach bakeout quickly and the use of metal to metal seals instead of o-rings.
The actually difficult and expensive part of high vacuum engineering is figuring out how to engineer valves that can both withstand bakeout temperatures and make the tight leak free seals.
In this regard I plan to use what essential amounts to a plate valve with something called, a "powdered seal".
This valve meets the requirements for the design in every aspect with it's only downside being that it is slow to change open to close. Though this downside will not reduce the overall throughput of the system as it is designed.
Really interesting, don't know much about CDTE cell production, Just curious, do you have a small start up working on this or is it mostly theoretical at this point (or are you doing this as part of your day job at a larger company)? I did a bunch of thin/thicker films in grad school and post-doc times and now work on thicker porous films.
I'm a junior in college and hoping I'm can get a prototype out before the end of the summer.
I've worked with other PVD systems yet I have not constructed my own. The main goal of this system is to make this sort of thing accessible to as many people as possible for the lowest amount of cost.
As it stands right now I'm learning machining to actually build most of this thing from scratch as I don't have the money to get this machined from from a job shop.
