A new design

I tried rotating the probe above the lower electrode to position it directly above the cloud, but now with the greater distance between probe and cloud I was unable to see any movement of the cloud at all. If we were able to generate a pulse of more than 210V we would probably be able to induce a strong wave in the cloud, and maybe even a shock, but it would be hard to find a stronger waveform generator or put together a system for pulsing our standard power supply in the time I have left.

So, I'm done with the probe for now. I decided on a design for the nozzle that will let me control its height without opening the vacuum chamber. I made a little mock-up of it for my mentor:

It's pretty simple: the plate hangs by a few pieces of wire from a rod, which can be rotated to wind the plate up or down. The rod sticks through two openings of the chamber and is sealed by a Wilson seal that allows the rod to rotate while remaining sealed. In the actual design, the wire will be replaced by nylon line (fishing line), the metal tube replaced by a plastic rod (so as to be electrically isolated), and the plate will be a 1mm-thick steel disc. My mentor was skeptical of the design, saying that it would be better to just start with a simple construction that would support the plate from the base of the chamber, but I still feel that I would spend too much time adjusting it and moving the plate after each test if I needed to open the chamber each time.

There are a few problems with positioning the rod, since the top openings that I'd wanted to use are too close to parts of the chamber to attach Wilson seals to them (I'll add a picture tomorrow). My plan now is to put it in level with the bottom electrode, since the cloud should always be below that level anyway. We ordered all of the materials, which should come tomorrow or the day after, and then we'll have to get them machined. Grooves will need to be cut into the rod for the fishing line to spool into, the disc will need to be cut out of the sheet of metal we bought, and holes will need to be drilled into the sheet for the line and nozzle. We should be able to put it into the chamber by the end of the week, though hopefully sooner.

In other news, my primary computer (at the lab) died today. First all of it's USB devices (mouse and keyboard) stopped working, and when we tried to reboot it, it gave us a blue screen of death. We've been unable to start it since; it doesn't even respond to the power button. My secondary computer is still functional (though it's not very comfortable to work at), and all of my videos and images were backed up on Dropbox, but most of my recent MATLAB code is only on that computer. We should (hopefully) be able to recover it even if the computer is dead, but it's good that I'm not going to be analyzing any data for a while until we get the new setup running.

The end of one chapter, the beginning of another

It looks like I've reached a dead-end in my search for shocks with the method I've been using. I've tried a number of things with the waveform generator, but other problems are arising, along with lots of unanswered questions. At the maximum wave we can generate, the cloud barely budges, its movement hardly detectable when the cloud is filled with dust acoustic waves. The movement is also only visible by the movement of the edges of the cloud; its not large enough for us to be able to detect any wave on top of the acoustic ones. Furthermore, the cloud moves radially away from the probe. This would seem to be the expected behavior, if only it had acted that way before. When I was first generating waves in the cloud from pulses to the probe (before the leak was fixed), the waves (which were then large) traveled vertically downwards; neither I nor my mentor had any idea why they traveled that way regardless of the probe's position. However, since the acoustic waves we generate travel in the same direction, we had hoped to be able to use the pulse to compress the wave and produce the shock. Now that the pulse travels radially, we can't do that, since a radial pulse (given our geometry) travels mostly perpendicular to the direction of the acoustic waves. Now that I think about it, it might be possible to twist the probe around to position it directly above the cloud, making the pulse travel downwards through the cloud; still, the pulse's small amplitude means that it will be even less visible among the dust acoustic waves.

 Another mystery (though somewhat less interesting and relevant) is the question of what determines the repulsion of the cloud from the probe. When I first began working here, the probe was grounded (unintentionally) by an oscilloscope which was meant to measure the pulse that was generated. In this case the probe strongly repelled the cloud, likely due to becoming negatively charged when bombarded by the plasma's electrons. When we disconnected the ground, the probe no longer repelled the cloud to any significant degree, and the cloud hardly reacted to the probe's movement. Now, it acts the same as it did before when grounded, but repels the cloud a small amount when not grounded. Moving the probe into the cloud now results in a circular space around the probe with no dust. My mentor suggested that the probe, when floating (not grounded), would attract the dust as it acquired a voltage from the electric field, but that produces more questions than answers. As my mentor said, I could spend the rest of my time here just studying the factors controlling that repulsion.

 So, the "combine acoustic waves with pulses" idea seems to have led us nowhere. Another method we could use to produce a shock is placing a nozzle in the path of the dust acoustic wave; this would compress the wave as it passes through, creating a shock behind the nozzle. Other experiments have successfully produced shocks in this manner, though with a different setup that ours. My job now is to figure out how to make and position such a nozzle inside our vacuum chamber. Ideally, I want a small opening in a disk (or between plates) that could be positioned in the center of our cloud and later adjusted. My mentor suggested attaching it to supports connected to the grounding plate or chamber base, but it would also have to be attached in a way that would allow fine adjustments when needed. Another problem with this kind of setup is that it would be impossible to modify without opening the vacuum chamber, and so we'd be limited to doing one experiment a day when we want to change the probe. Initially placing it in the cloud could also take several days, as it would probably be a matter of trial and error, with each trial separated by a full day of pumping down the vacuum. Furthermore, we'd be unable to change any parameters of the experiment (electric field, pressure), since those changes would move the position of the cloud and misalign (?) it with the nozzle.

 The best solution, of course, would be to make it adjustable from outside without breaking the vacuum seal. Though my mentor said that such a solution would probably involve more time spent engineering it than would be saved on the experiment, I have a few ideas I'm thinking about to make it work. One of these is to suspend the disk from above using some string or non-metallic wire, which could allow me to pull the wire in or let it out from above through a vacuum seal. That has its share of problems, too, but that's what I'm thinking about now.

An unproductive day

I spent most of today fiddling with the probe and plasma parameters, but I wasn't able to get a non-arcing wave today at all, unfortunately. I did move the grounding plate down a bit, but I needed some materials from certain people all of whom left early, so I wasn't able to move it down as far as I wanted (I needed a new nylon screw to cut to a shorter length and thread with a die, but ended up just moving the plate down half an inch on the screw already in the chamber). By the slow speed at which the vacuum is going down now I suspect I may have introduced a small leak in the chamber, though it could just be outgassing from water vapor as before. I doubt that the short distance I moved the plate will have much of an effect on improving the arcing problem, so I'll probably have to bring the chamber up to atmospheric pressure again on Monday to switch to a new screw. Since I have nothing else to talk about, I'll mention another problem that's stumped me (and my mentor). The setup is mostly symmetric: our electrodes are circular and share a common axis, the chamber is cylindrical (with also the same axis), and the grounding plate, which used to be circular, is now a rectangle that stretches the width of the chamber. The probe is exactly in the center. The laser itself should have no effect. So why, then, do we regularly get a smaller, secondary cloud on the left side of the chamber (near the laser), and not on the other side? We thought it might be a slight tilt of the electrodes, but twisting them didn't drastically change the location of this second cloud. The only other thing I can think of is the place where the electrodes come through the chamber wall, as this is off-center, but those pieces are fairly small and insulated. It's not a question of any particular importance, but I thought it was an interesting quirk.

Unstable plasma and other thoughts

I think I know why I'm having problems with my waves now that the probe is back in. The probe seems to be pushing the cloud downward, and the large grounding plate I installed is increasing the cloud's size and thus lowering its lower edge. Since the cloud is now much closer to the grounding plate, arcing between the plate and cloud is more common. The arcs are also smaller and more frequent; normally arcs are brightly visible and occur every few seconds, but now they're on the order of 4 per second, and too weak to see. It's pretty frustrating to record what seems to be a good wave, only to see when I play it back that it is unstable. Since the arcing frequency is fairly consistent, my mentor recommended simply removing that frequency (ie 4 Hz) from the resulting wave; I'll try it, but I think that the instability has farther reaching effects that will influence the data. The solution will probably be to lower the grounding plate; I'll do that tomorrow afternoon so the vacuum can stabilize over the weekend.

On the topic of instability, I filmed a pretty odd effect that occurred fairly regularly at certain conditions. A strong plasma glow would appear suddenly on one side of the bottom electrode, pulling the cloud towards it and stabilizing it. Then it would disappear, and the cloud would return to its previous position and resume oscillating.

I'm not sure what's causing it, and neither is my mentor; she says that it may be a result of dirt on the lower electrode, but from what I've seen, that results in a steady, concentrated glow around that area, rather than a wide glow that appears and disappears like in this scenario.

I've also found the problem I was having with the probe. The square wave I was generating was apparently only 1 millisecond long, when I'd assumed it was 10 milliseconds long. Anyway, that's another variable I will need to play around with tomorrow before I move the plate. I haven't yet tried other waveforms, so I'll need to do that too.

I spent far more time than I should have this afternoon battling a persistent bug in my wave analysis code and a separate one in my GUI, but managed to fix (more or less) both of them. Since most of my time before lunch was taken up by a seminar and an evacuation drill, I didn't spend as much time working with the dust waves as I had hoped.

My clouds are getting frustratingly faint, too, and I can't see the dust in the wave valleys at all. This isn't really anything new, but it means I need to spend more time trying to get bright and dense clouds. Near the beginning of my internship I talked to someone who was in the process of building an image intensifier, and who said that they should be done by the end of the summer; I should find out whether any progress was made on that and whether I might be able to borrow it at some point (although if I remember correctly he offered to let us use it on the condition that we helped him build it...). Unless we do end up getting one of those devices, I'm still ultimately going to be limited by my laser power. Another option is to adjust the cloud density; dense clouds are brighter, but the density is mostly fixed by the particle characteristics. I probably won't be switching to a different dust material, but I could filter the dust to get a uniform particle size. That would probably change the characteristics of the cloud significantly, though I'm not exactly sure which ones and in what direction.

No shocks yet

I've been mostly fiddling with different parameters today. My acoustic waves aren't as nicely defined today as they were before I installed the probe, but they are still decent. They are a lot more turbulent than before, which makes it harder to isolate things like shocks. They are also exhibiting some odd "bouncing" behavior, where the back edge of a wavefront would occasionally change direction for a brief period of time, and I have no idea what would be causing it. The most likely explanation is probably that there are multiple dust acoustic waves travelling through the cloud at a slightly different frequencies, causing the irregularity. I haven't gotten a shock yet (or not one that I've noticed, at any rate); I'll need to play with the waveforms I'm generating on the probe some more and see what comes out of it. So far I've been using a basic square wave, but I may get better results with a ramp. I also haven't tried changing the length of the pulse, or fiddling with any other parameters relating to it. There's seems to be some sort of problem with the generator at the moment; hopefully I can get that sorted out fairly quickly. It's probably just something I set incorrectly.

I've also mostly finished up my GUI for now. It's not very pretty, but it gets the job done.

(Click for full image)

Probe reinstalled

I got the probe tested with the leak-checker, and it seemed in good shape. I installed it back into our chamber, brought down the pressure, and tested it again; everything went well. The pressure will stabilize overnight, and I'll need to bake the plasma for a few hours tomorrow to get rid of anything that may have gotten in when I opened the chamber up. Unlike the previous times, though, I left the nitrogen flowing when I opened it to keep a steady stream of gas leaving the chamber whenever the chamber was not sealed. This should have significantly reduced the amount of junk that managed to get in while I was swapping in the probe and greasing the O-rings.

Since I have had consistent dust acoustic waves in the chamber since we fixed the leak, I should be able to generate a pulse on top of that wave as soon as the plasma settles tomorrow. Hopefully I can get something interesting!

Besides probe-related stuff, I spent most of today writing a graphical user interface for my MATLAB code that analyzes the videos I take. Not too exciting, but it means that I'll hopefully be able to stop mucking about at the command line and visualize the waves better.

Fixing the probe

I took the probe apart today, and sealed up the wire with epoxy. The epoxy needs 24 hours to set, so we'll bring it up to have it leak-checked tomorrow morning.

Very nice acoustic waves

With the nice vacuum I got, I was able to form some really large and stable clouds, and was able to pretty quickly produce some consistent dust acoustic waves! Here are some pictures and video of the inside of the chamber:

This is what the inside of the chamber looks like, illuminated by the green laser. The primary electric field is generated by the two ring electrodes, while the dust (silica) sits on the grounded plate below. The plasma charges the dust on the plate, and it is lifted up by the electric field to form the cone-shaped cloud. You can see a small dust acoustic wave in this picture by the light and dark bands of high and low density within the cloud. You can also see the faint purple plasma near upper electrode and in a few spots on the lower one.

By bringing up the pressure and increasing the voltage, we can clearly see the plasma around the top electrode and the grounded plate. The purple spots on the lower electrode are places where the plasma is burning off something.

A spontaneous, stable, dust acoustic wave in the dust cloud. This video was shot by the Phantom camera at 100 fps.

This short video shows a few seconds of the plasma arcing to the sides of the chamber. By significantly raising the voltage, we can get the plasma to arc like this; arcing burns off contaminants in the chamber and gives us a more stable plasma when we decrease the voltage again. Doing this also charges up many dust particles and pulls them up into the field; the longer we arc the plasma, the larger the dust cloud that settles when we bring down the voltage.

Pictures

Some photos of my setup and that of my labmate:

View of the lab

This is my labmate's experiment. His goal is to study aerosols; the final construction will be a long tube at a near-vacuum. A pressure gradient will suck gas (nitrogen) from one end of the tube to the other. Dust will be dropped into one end and pulled through a very small opening, letting him study its dispersion and other behavior. Eventually plasma would be added at some point, but I'm not exactly sure when or why.

Everybody who comes into the lab can't help but make a joke about the beer keg being used to contain and measure out gas. 

My workspace after some massive cleanup. The large box encloses the laser that I use to illuminate my dust clouds. The two boxes on top are the laser's power supply and the display for the pressure gauge. To the right is the incredibly expensive Phantom Miro camera that I use to record the dust clouds. It can record at over 100,000 frames per second, but the faster you shoot, the less light you get in each frame; our laser strength limits us to about 150 fps, and I usually shoot at 100.

These are the valves controlling flow into and out of the chamber. The thick hose on the right and left leads into the chamber, and the thin hose in the center leads to the vacuum pump. 
The black and white valve in the middle regulates the flow of argon, while the green valve in the top-left controls nitrogen. Argon is what the plasma is made of, while nitrogen is just used as a clean, cheap, and harmless gas for bringing the pressure in the chamber up to atmospheric pressure when I need to open it. If I were to just open the chamber, the air that would be sucked in would disperse the dust inside, in addition to bringing with it water vapor and other junk.

The back of the chamber. The plastic tubes on the right cover the two primary electrodes that we use to form the electric field, and the green plug connects the pressure gauge. The large T-shaped piece on the left side of the chamber is designed to disperse the laser once it hits the other side of the chamber, so we don't get any reflections or unwanted light from that side. The window on top (just like the primary window on the other side) has a piece of polycarbonate attached to it; this is designed to stop glass from flying everywhere if we were to accidentally overpressurize the chamber and it were to explode.
The probe that we use to produce pulses is usually attached to the middle opening, but I took it out to fix the leak in it.

This cabinet houses a server that connects to the Phantom camera (top), a waveform-generator that I use to send pulses to the probe (middle), and the power supplies for the two primary electrodes (bottom).

My desk, with a pile of CCD cameras and lenses that I was swapping out.

Small update

With the leak fixed, we were able to get the chamber all the way down to 0.0000007 atmospheres, more than 200 times lower than what we had before (a measly 0.00013 atmospheres).

Not a leak, actually

Good news: what I though was a leak wasn't actually, so I'm in good shape. Andrew Zwicker (the director of the lab and my mentor's boss) came by to take a look, along with a visiting physicist who works in exactly my field of dusty plasma, and they pretty much laughed at my attempts to find a leak, since the speed at which I was depressurizing the chamber now was better than great. The reason I was seeing a pressure rise when I turned off the pump, and the reason it got worse after I had opened up the main window, was because of outgassing: water vapor had condensed on the walls, and was slowly becoming water vapor again and increasing the pressure. When I opened up the window and kept it open for about an hour, water vapor had had lots of time to get into the chamber, and that's what I was seeing now. I'd read about outgassing when researching vacuum leaks, but didn't realize that water vapor would have such a large effect. It's called a "virtual leak"; it looks like the chamber is leaking, but it's really just a matter of gas forming on the inside.

So I don't have a leak, and I have a good vacuum, but I don't have a working probe. Yet.

Leak sealed... or not?

We found and sealed the vacuum leak, but it wasn't in the place we thought it was. In the probe I mentioned, there is a wire that leads into the chamber to create the pulses. The wire is sealed on the outside with the probe casing, but apparently it was not sealed completely at the place where it touches the probe's electrode. That meant that air was leaking through the wire, underneath the insulation. It was sucked through about 10 feet of this wire, all the way until that wire connected with a different wire and the insulation changed. That's why it was so hard to find; the reading on our helium detector wouldn't change until the helium had traveled all the way down this wire, and even then, the place where the air was sucked in was pretty far from our main chamber.

Anyway, my temporary solution was to remove the probe entirely and seal up that opening. We'll probably need to redesign the probe a bit to fix the issue. In the meantime, we are getting low pressures, but I'm not sure what kind of dust clouds we'll be able to produce.

Meanwhile, we have a new leak. When I first sealed the chamber after removing the probe, I was getting very good results, but then I opened up the chamber to move some stuff around and clean the window, and since then I've had this leak. I'm pretty sure it's at the window seal, but I'm having trouble getting rid of it.

First post: vacuum leaks, dust cloud waves, and more

Hello everyone!

I have some interesting developments going on with my project at the Princeton Plasma Physics Lab I'm interning at, so I figured I would start this blog to share my progress. So here's what I've been doing over the past few days, and what I've been trying to do over the past 3 weeks. My summary of what I'm studying is farther down the page.

Overview of the lab. I work in the right half of the room, and my labmate, another intern my age on this program, works in the left half. At the back of the room on the left you can see the compressed gas cylinders (black is nitrogen, green is argon) that we both use.

Recent News

As I've told some of you, last Friday I pressurized the vacuum chamber and opened it up for the first time to replace the grounded dust-holder. It was a small disk, but I replaced it with a large aluminum plate so we could levitate more dust at a time and get bigger clouds. I sealed up the chamber and it depressurized over the weekend, but on Monday I still wasn't getting the results I wanted (I am trying to generate "dust acoustic" waves; more on that below). So I talked to my mentor about decreasing the pressure in the chamber (since other similar experiments used lower pressures), and she told me that we actually had a leak that was preventing me from getting a lower-pressure vacuum. 

Fixing a leak is often a pretty difficult task, not only because vacuum-tight seals can be difficult to achieve, but because it's pretty hard to find where the leak is in the first place. I could easily confirm that the leak was there; as soon as I disconnected the vacuum pump from the chamber, the pressure would immediately go up. To find the leak, I spent all of Tuesday and a few hours today pouring increasingly-large amounts of ethanol over every connection and valve of the chamber. The theory behind it is this: if you pour ethanol over a leak, the ethanol will get sucked into the chamber, and the ethanol vapor will cause the pressure to increase. The problem is that there is a large number of places where plates are screwed together where a leak could occur. Besides the primary cover of the chamber, there are three glass viewing ports, three electrical connections (for the electrodes and probe), the tube connecting the pressure gauge to the chamber, the hose leading to the vacuum chamber, a number of valves for pumping in different gases, etc. Each electrical connection also had several insulating elements and other connections. The hose itself could have a hole in it, too. In short, there were a lot of places for leaks. The good thing about ethanol, though, is that it evaporates very quickly, so pouring it all over my work area is not a problem.

Eventually, I found what I thought was the leak using the ethanol method (though it took more than a day). I wasn't sure I had found it, though, since the pressure difference wasn't all that large. So I got in touch with a specialist on vacuum leaks, and he brought down his huge, hi-tech mobile leak detector and a cylinder of helium. This machine is a very interesting device. It is attached to where the vacuum pump normally connects, and it has its own vacuum pump which sucks down whatever is in the chamber. The gas that enters the machine goes through a mass spectrometer, a device that can determine the gas's composition. Essentially, the atoms of gas are given an electric charge, and then shot past an electromagnetic field; how far the atoms are deflected by the field tells you the mass of the particles, which tells you what element you are looking at. Anyway, this particular machine is tuned to detect the presence of helium. To find the leak, we'd spray sections of the chamber with helium from the cylinder, and when we hit a leak, the helium would be sucked into the vacuum and down into the machine, where a pretty graph showed us that the helium it is detecting has drastically increased. We still had to go connection-by-connection, and cover individual valves with aluminum to stop the helium from dispersing everywhere, but it gave us a good confirmation that I had correctly found the leak with the ethanol, and that there weren't any others.

To make a long story short, I pressurized the chamber again (spraying dust all over the inside because I accidentally did it too fast), changed the rubber O-ring that we thought was the problem, depressurized the chamber again, and discovered that the leak was still present. Hopefully we can seal it up for good tomorrow, but at least we know where it is.

What am I trying to do, anyway?

The ultimate goal of my project (or rather, my mentor's project) is to induce a shock in our levitated dust cloud and study its properties. So, what is a shock? A shock is basically a special kind of complicated wave. We have a number of different types of waves we are looking at, and each wave has a characteristic driving and restoring force: something speeds the particles up, and something else slows them down. 

1. A basic, linear wave caused by a changing electric field. When we send a pulse in voltage to the probe in our vacuum chamber, the particles move in response to the new field from the probe (the driving force), but are slowed down by the global electric field that we use to levitate the dust and keep it stationary.
2. A "dust acoustic" wave is a special wave unique to dust in plasma. Plasma is, essentially, a gas made of positively-charged ions (atoms missing electrons) and very hot, fast, free electrons flying around. Both the ions and electrons exert pressures on our cloud of dust. Dust acoustic waves are driven by both of these pressures, and are restored by the inertia of the dust itself, since the dust is much heavier than the other particles. These waves are slower than the speeds of the electrons and ions.
3. A "dust-ion acoustic" wave is similar to a "dust acoustic wave", but they are driven by just the pressure from the electrons, and is restored by both the dust's inertia and the inertia of the ions. This wave is slower than the electrons, but faster than the ions.
4. A shock is a much more complicated and interesting wave; one way to create it is by combining a dust acoustic or dust-ion acoustic wave with a wave generated by an electric pulse. Shocks travel faster than the speed of sound in the cloud, unlike the other waves I mentioned. (Dust acoustic and dust-ion acoustic waves are both essentially sound waves, hence the name "acoustic"). This means we have all of the interesting effects of supersonic waves; for instance, dust behind the wavefront will be moving at a supersonic speed, but the dust in front of the wave will have no information about the wave, and it won't move until it is hit by the wave (just like the sonic boom from an supersonic airplane).

So far, I've done a good deal of analysis of the first type of wave from the electric pulse. They are pretty easy to create; anytime I have a large dust cloud, I can produce that kind of wave by sending a pulse to the probe. I managed to get a steady dust acoustic wave going, and recorded it, but was unable to reproduce it again, so I only have one datapoint for it so far. Acoustic waves are much harder to produce, since they happen spontaneously only at certain conditions. Right now, I'm hoping that the lower pressure and lack of air-leaks should make it much easier to generate acoustic waves. Once I get these waves going, I'll start sending pulses to the probe while the waves are active to try to make a shock.

That's where I am right now! My own project is very interesting, and all of the resources and machinery that make a plasma physics lab work are incredibly cool. 

P.S. Today I got to watch people dropping boxes from the top of a 100-foot firetruck ladder into a pile of inflated garbage bags. Each box had a camera inside and some sort of experiment; the fall allowed for a couple seconds of microgravity conditions inside the box. I don't know much about the science they were doing, but firetrucks are always pretty cool.