DIY Magnetron Sputtering Chamber

[Sesselmann]

This project might be of interest to some of you.

As part of a larger project I need to apply a thin conductive film to some CZT crystals and for that I need a vacuum deposition system. I found a few old systems on eBay for around US $3500 that were either broken or for parts only, so I decided to make my own.

Turned out I had most of the important pieces already, vacuum pump, vacuum fittings, power supply etc.. All I needed was some aluminium and stainless steel round bar for machining, a glass tube and some silicone gaskets.

Magnetron
The magnetron is made from solid aluminium bar and has two nested Neodymium magnets, one button magnet and one ring magnet placed opposite ways north to south. Both magnets sit in the lid and are isolated from the actual vacuum chamber. A nylon bushing is used to center the button magnet inside the ring magnet.

Vacuum Pump
My vacuum pump is a standard two stage rotary vane pump - these are common and can pull a vacuum down to around 20 micron (millitorr) which is sufficient for sputtering. I didn't bother fitting a vacuum gauge mainly because I can hear when the vacuum presssure starts going below 100 millitorr - the pump starts making a flapping noise.


Power Supply
An old magnetron transformer with single diode rectifier on the centre tap and capacitor works well because it can deliver sufficient current, but it is also extremely dangerous. I don't recommend messing with these unless you know what it can do to you. For sputtering you will need around 700 to 1000 volts and a reasonable current maybe 50 to 60 mA. Running this straight from 240V would be uncontrollable, so I drive it using a Variac autotransformer - the sample was done at 50V input voltage.

Argon gas
Ordinary welding gas is fine. After evacuating the chamber fill the chamber with argoon a couple of times and let the vacuum pump do its thing, that should get rid of most of the oxygen.

Cleaning
The sample to be coated needs to be perfectly clean and dry. Degrease with soapy water dry it and do a final clean with isopropyl alcohol.

Sputtering Target
The sputtering target is a round disc that mounts to the magnetron with three small springs.
The target material will depend on the material you want to sputter, it can be aluminium, nickel, titanium, silver, copper or gold. In my case I wanted to use gold, so I bought a 10 gram gold coin and got a jeweller friend to roll it into a 50 mm x 0.5 mm round disc. A6 $200 per gram this was obviously the most expensive part of the build.

Here are some photos of the build.

Initial design was done in Fusion360 CAD - here you can see all the components. including magnets seals and sputtering target.

Design phase - Fusion360 (2)

View from below - showing the recessed flange, this flange prevents the seal from being pushed in.

Stainless steel bar

Stainless steel bar purchased for the making the base.

Machining the top

Machining the aluminium magnetron housing - which is also the lid.

Machining the top

Magnetrons can get quite hot, so I made these cooling ribs - looks a bit like a 50cc motorcycle head.

Magnetron Top open

Here is the magnetron with the top open - we can see the magnets placed inside with a nylon bushing to centre the button magnet.

Magnetron Top on desk

Here is the magnetron with cap sealed - note that the magnet chamber is vented to the outside - it does not need to be hermetic.

Bottom Chamber

This is the base, it has two fittings one is for the vacuum connection and the other is for argon gas, which is a hose from my Argon welding cylinder.

Final sputter coating setup.

here is the complete setup including my home made rectified MOT power supply. ⚠️ it should be noted that this power supply is quite dangerous, it has a single diode rectifier on the centre tap and the original microwave capacitor 😬. I drive it using a Variac (autotransformer) and run it at around 50V input. Plenty of current - enough to be scary.

Sample with serious guy..

This is the result, it is a glass lens I removed from a watchmakers eyeglass. This was coates for 5 minutes on low power - just below where plasma would first strike.