Tales from the Lab

I didn't post for a week, and the reason for that was just plain laziness and low motivation on my part. Additionally, I spent a very frustrating week in the lab, due to stupidity on my part.

It all started last Friday. I had been puzzled the last few days about the lack of self-bias in my plasma. An asymmetric capacitively coupled plasma (asymmetric is just a fancy way of saying the RF electrode and the mass electrode are not the same size) develops a self-bias - the RF electrode charges up a bit. But I was seeing an absurdly low self-bias - millivolts instead of tens of volts. The reason for this behavior turned out to be two things: I had changed my matchbox from an L to a π matching network, and the RF generator was not DC decoupled.

A π-network has not capacitor between generator and plasma, hence it couples also in DC. Generators are normally decoupled, safe the one I am using. Those two things combined meant that there was a DC short to mass - no self-bias could develop.

The solution to that problem is easy - just install a decoupling capacitor somewhere between generator and plasma. I did that (and a couple of other things, and that would come back to haunt me!), and could not match anymore. So I fiddled around a bit with different sized coils in my matchbox. Nothing. So I dug out a Network Analyzer, with whom I proceeded to, well analyze my network. I calculated the size of the new coil, made it, put it into my matchbox - no matching.

Phillipe (of our group) finally told me that the readings were strange for a reactor, and after a couple of hours of fiddling around, we found out that the Network Analyzer was broken. Fortunately, there was still an old, analog one kicking around. I dug it out of some dusty cupbord, tried it, and couldn't even calibrate it.

As a last (well, next to last, but I really didn't want to do a gazillion Voltage-Current measurements) I got an Impedance-meter, which finally worked. So I measured the impedance of PADEX, designed a new coil and -SUCCESS! Or so I thought. I was able to match as long as there was no plasma. As soon as it ignited I had about 50% reflected power, which is a Bad Thing ^TM.

So in despair I went back to the old configuration. Then I started changing things bit by bit. Shorter cables, remove mass electrode, remove screening from RF electrode, add decoupling capacitor... This is what I should have done in the first place - never change too much of your system at once! Anyway, I was ready to jump in and modify my matchbox at any step, but I didn't have to. Matching can be hard to find, especially with π-networks, but by slowly changing the system I was able to keep up with the changing characteristics. In the end, all the hassle with network analyzers and whatnot was for nothing. I feel dumb now.