The Goblet of Fire

Plasma, that fourth state of matter:

-“When an ionised gaseous substance becomes highly electrically conductive to the point that long-range electric and magnetic fields dominate its behaviour.”
(wikipedia)

As you know, plasma is a quite hot stuff. Usually it’s density is so low that thermal transfer quickly reaches an equilibrium much lower than the melting point of materials. However, in a sputtering machine, you do concentrate plasma in a small area, so, continuous operation requires a bit more than just having a passive aluminium heatsink.

And yes, you can just powermelt the target:

An aluminium plate with concentrated plasma heats quickly, as I have found out. After medium sputtering runs (20/25 minutes, NO Argon), the magnets are so hot you can barely hold them in your hand. That must be around 40/50ºC, wich obviously can only get worse. Normal (cheap) neodimium magnets have a curie point of just 100ºC, so it’s better not to risk damaging them. And just because fuck yeah, I want to liquid cool it to allow extended operating time.

Let’s have some water channels cut into the plasma trap plate:

At first I was thinking about machining it by hand in stainless, as having a rotary table makes it a fairly easy task. However, later on, I realized I could simply make it in aluminium wich is a better heat conductor AND my 3020CNC can machine it, if you are patient. I’ll still have to tap all the holes by hand, but that’s a minor nuisance.

Having a laser cutter helps immensely with reducing waste (milling acrylic is a pain in the ass)

Some might say so many screws are overkill, and might be true, however, given the diameters involved, and that I am using viton O-rings wich are not super flexible either, I have just preferred going all the way. The acrylic is 7mm to ensure minimum deflection.

Yeah, I should have made the utside border a tad larger. (The inside diameter has to fit the magnets, so there’s not much room there.)

The screws are stainless steel, so I don’t think they will affect the shape of the magnetic trap.


Test fit on top of a non machined plate (with recess for the magnet)

Since I received my new 3D printer before machining the plate, I printed a 15mm section to test fit the magnets and general visual guidance, plus screw check.

The magnet slot had a bit too much room left (2mm) so it was reduced 1,25mm in Fusion. (to 40,75mm). The screw lenght allowance was correct.

Overall fit was good, so the minor changes where validated and saved.

Machining!

Spot drilling.
I was too lazy to actually make a separate program for this, so I used the same drilling program, but added a +6.5mm offset so it would only peck 1.5mm with the 90º spot drill, I know, extremely ineficient, but it would have taken longer to start the computer and actually make the program.

After drilling, the o-ring recesses where cut.

Quick seal check, just in case y messed up, or the o-rings where wrong.

As an afterthought, I could ave made the slots 0,5mm deeper, so the acrylic would fully compress the rings and sit flat against the plate, but well, as I said, an afterthought.

The magnet slot was cut first:

Then the water channels:


I know the pic is shaky, but I was busy, you know…

Little misshap with mach 3 going first to Z-Ø instad of X/Y-Ø, and a not-so-exact Z height setup. Fortunatey, the cut was ony 0,15 deep, so it doesn’t affect the overall performance, just my inner pride. (I tested a newish 2-flute endmill instead of the oldish 4-flute I was using, but it underperformed, so I changed back, and messed with the paper thickness Z height procedure. This part is not critical in the Z axis, so I wasn’t particularly worried.

Anyhow, everyone hail the awesome cooling plate:

I accept the fact that there will be some blow-by between inlet and outlet, but most of the water should still flow along the path.

Magnets!

The CNC part took 6.5h to machine in a cheap (ballscrew) 3020CNC router, with the following setup:

  • 2mm carbide 90º spot drill, 2,5mm carbide drill.
  • 8mm depth, 0,25 pecking depth. Lube between holes.
  • 2mm (R1) carbide ball endmill.
  • 4mm-4flute carbide endmill, flat, center cutting.
  • 0.5mm depth of cut up to 5mm depth (6mm for magnet), 1.25mm stepover. 75mm/min plunge. 150mm/min feed.
  • LUBE as if there was no tomorrow. Also make sure all the router bolts are tight.

These kind of machines are not particularly rigid, so this is the absolute limit of what they can do. If you are patient, tough, you will be able to squeeze it to the very last drop.

Using a locating setup, one could just machine the outside border and flip the part to machine the lip, however, since I have a lathe, I just prefer to do it there and avoid the hassle.

First, bandsaw the corners.

Chuck the plate from the inside…

…and slowly carve the diameters away.

After polishing all the important surfaces:

Oh! Also some custom fittings should be made, amirite?

So gorgeous.

In the heat exchanging department I would like to employ some passive cooling like this:

I know I could try to hook it up to one of my beer refrigeration units, however I pretty much prefer not to, because the vacuum chamber is at potential, and having liquid wiring around is…well, not my thing. Also it will make for a standalone unit (not that it is going to go anywhere, but hey, what if I want to sputter in the living room?) Should that not suffice, I can always connect it to one of the machines later, or make a heat exchanger or something.

In any case, the particulars of the radiator are irrelevant, any liquid computer cooling setup should work (to some extent, if you put a small one). So, pick your poison, as some say.

See ya!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s