KF flange polishing

[Peter Schmelcher]

I expect others have bought used vacuum parts that were damaged, in my case it was a KF16 flange on a turbo pump. Rather than fight leaks till the end of time I decided to somehow repair the flange surface.

What finally worked was to transfer the flatness of another surface to the flange. I used sand paper face up on a glass window. This method requires a cutting fluid lubricant on the sand paper. A traditional aluminum cutting fluid is lard and kerosene, straight lard also works. I used single direction sanding; picking up the turbo pump and placing it back at the start position for each 9 inch long sanding stroke, in total about 500 strokes. The sanding grit progression was 220, 400, 800, 1600, 3000, and 6000.

A few lessons were learned along the way. Immaculately clean the flange between grits to remove all traces of lard because the lard will contain some dislodged grit particles that will contaminate the next smaller sanding grit. Changing the sanding angle slightly between grits lets you know when the previous coarser grit marks have been fully removed and its time to move onto the next finer grit. Don’t use any sand paper that is flawed for example even a slight crease and don’t sand over the paper edge, both have abnormally high sand particles that will ruin your final surface finish.

The final stage is a 10,000 grit polishing paste applied with a soft velour rag for about 15 minutes. If the mirror finish has scratches you have contamination. Try and polish for awhile longer but most likely you will need to change the sand paper and start all over. If you have the knack the whole process takes 1 to 3 hours.

Peter

[Chris Bradley]

It's a nice shiny job, for sure, but not quite sure about;

> If the mirror finish has scratches you have contamination.

I have sanded down various used flanges. As you say, sand paper laid onto a flat surface, and doesn't need to be particularly flat as rubber o rings will take up the compliance if it doesn't come out micron-perfect flat. I approach it by rotating the flange in small circles on the paper so as to leave (if any) lines generally circumferential.

The thing is, once cleaned up with solvent I think stage 3 or 4 of your photos would be just fine. I've never had a problem in the ~e-5 mbar range. Life's too short when you've got 30 flanges to get cleaned up on a system!! Actually, I rather prefer some 'honing' marks left on the surface so that a ring smeared with a teeny dab of vac oil has something to key into.

If less-than-mirror means potential contamination, then maybe more than a few chambers could do with a polishing treatment inside?

[Peter Schmelcher]

Your correct the mirror finish is overkill for an o ring. In the case of the turbo pump KF16 flange the center of mass was several inches above the sand paper somewhat unstable so an orbital sanding motion was not an option.
I intend to try a copper wire seal some day and figured the mirror surface would have the best chance. I have used inexpensive copper washers from a brake kit with VCR fittings and JIC fittings. They seal fine at roughing vacuum so I am encouraged enough to want to test copper gaskets with other flanges.

[Brian_Gage]

Here's hoping this isn't drifting too far off the subject, but I'm a rockhound and your polishing description was reminiscent of the stages for polishing rock slices or cabs. Very interesting. Congradulations.
Regards, Brian

[Dennis P Brown]

Your polishing method is classical mirror grind/polishing for telescopes.

By the way, a really great job on that flage!

I have done a different method for the old KF's - a quick face off with a lathe to get out the deep marks and then two different fine grades of paper mounted on glass to finish. Works but no where near as nice as your results.

I have used copper gaskets on systems and all these flanges used a knife edge sealing system - not sure how copper works without the knife edge - still, this was for systems in the 10-8 to 10-9 mbar so the 10-5 range might be fine.

[Jerry Biehler]

In general you don't have to go crazy. When I did the base plate for my bell jar I just smoothed out the milled surface with some scotch brite and called it good. When I got it on the system I helium leak checked it and they all were good.

I doubt you would have had a leak with the scratch on the kf-16. If it did all it takes is a rubber abrasive wheel on a dremel to get rid of the sharp corners.

[Peter Schmelcher]

Brian, Dennis, Jerry,

Thanks for the feedback. I thought about several other finishing techniques, lapidary, telescopes mirrors etc and know several that don’t work.

The photo does not show the KF16 scratch that well, it was more like a cut from a wood chisel about 0.7mm deep. The cut had a sharp bottom and one near vertical side. I imagined the o ring would end up as a suspension bridge across a valley with a torrent of air flowing underneath. I have no feeling for the vacuum you get with vacuum oil or grease on o rings, and if these will create an effective plug in a deep valley.

It is hard to know when a vacuum surface finish is good enough. VCR fittings are electro polished to a high mirror finish while other vacuum fittings look sand blasted. Well a better finish can’t hurt and just costs you some time.

My concern is a bunch of dodgy parts coming together during initial vacuum commissioning and then looking for leaks.

I stopped with sand paper but continuing down the rabbit hole the next stage is optical lapping film (30,000) and then probably some telescope mirror polishing process. All I wanted was surface finish that will seal against a dry copper gasket or copper o ring, but this is a future project.

Peter

[Richard Hull]

The vacuum seal on a KF flange relies 100% on the O ring. You were wise to polish that edge as that is where the O ring would try and seal. I would have turned out that scratch on a lathe, provided it was lathe mountable, with an ultra sharp fine tool with hemispherical edge and then polished. You would have been done a lot quicker.

Richard Hull

[Peter Schmelcher]

I did consider a milling machine. My dad used to say if you have a lathe, all problems have round solutions; if you have a milling machine, all problems have square solutions. The KF16 nipple has epoxy at the attachment point to the turbo pump body so machining operations carry some risk. Plus, solutions that the younger members can use are a good idea. This process is old school and only takes patience, something apparently I have.

My proposed metal-copper-metal seal is just because I am annoyed with high vacuum components. I know you fabricate metal all the time and have mentioned this before, but for others, if you heat up copper to dull red then cool it quickly the result is a very soft metal that is easy to deform, more like solder. After heating I didn’t have any nitric acid so I cleaned the copper with a vinegar and salt solution. The conical copper washer for the SAE flare fitting was deformed after heating/cleaning and as I mentioned seals great at roughing vacuum. I would always heat/clean to reuse the washer. SAE, AN, and JIC fittings are basically not interchangeable or for that matter intended for use with a washer but the copper washer thickness is not an issue. I think as long as a copper washer surface is not sliding during tightening it will produce a good gas tight seal.

A copper wire in place of an o ring on an ISO63 with a mirror finish, I just don’t know about yet. Perhaps two seals, viton and copper in series like the old belt with suspenders concept, would be prudent. Time will tell.

Peter

[George Schmermund]

I've had very good results using indium wire as a gasket material. A feature of this method is that the wire can be used to make any size gasket as long as the ends cross each other. Indium has excellent vacuum properties if the system doesn't need to be baked.

[Chris Bradley]

..and also wets glass, good for metal/glass seals...

[Peter Schmelcher]

George, Chris

I’ve never used indium but like the idea of the stuff. Do you think an indium sealed joint could be reopened, perhaps by heating in the oven?

I actually expect a copper wire gasket will work great. To connect the two copper wire ends together I was thinking about trying spot welding, but perhaps a scarf joint connected using indium would be a better splice. Any feeling for how strong an indium joint is in shear?

Peter

[Jerry Biehler]

Indium is very soft, I can dent it with my fingernail.

You could also use lead wire. That is what is shown in my book on deposition to seal large metal bell jars. Probably could even use thin solid core solder.

[Chris Bradley]

Peter Schmelcher wrote:
> Do you think an indium sealed joint could be reopened, perhaps by heating in the oven?
> I actually expect a copper wire gasket will work great. To connect the two copper wire ends together I was thinking about trying spot welding, but perhaps a scarf joint connected using indium would be a better splice. Any feeling for how strong an indium joint is in shear?
*Every* joint can be reopened!

You will likely find the bond to be quite mechanically solid after the metal has had time to undergo a long period of compression, a combination of 'Van der Waals' and the fact that its ductile creep behaviour will lock in tight to the surface roughness, so it may be prudent to anticipate how you would open up the joint later (if not by heat). But I would not anticipate relying on it as a mechanical connection.

After mechanical removal of the indium, post-opening a joint, you'll likely still see the grey marks left behind where not all of the indium can be removed from the surface, as it binds in tight. In fact, indium seems to stick to anything it touches, including greyed fingers after handling!

I don't think copper will do the same job. I'm not a metallurgist, but the chap at work (who is) with whom I have discussed using indium before was muttering something about the 'twinning' properties of indium metal. Not sure I understood, but as far as I took anything away from the conversation was the impression that indium is fairly unique in its ability to bond back onto itself. It's why it makes that creaking noise when you bend it, the crystal bonds are constantly being broken and then remade within the metal. Take that as 'hearsay' rather than knowledge, I don't have the background to properly understand that.

[Richard Hull]

I have extensive experience with indium and all its alloys. With proper care and scruplous cleaning, following directions carefully, indium will cold weld to indium in a manner that it can never again be separated at that joint. The bonding is at the molecular level. Little or no pressure is needed for this to occur on properly prepared surfaces.

Dirty or unprep'd indium will still stick to itself but not well. Rolls of indium wire are prep'd right out of the die in such a way that the turns will not hopelessly adhere on the roll.
Properly prepared and warmed refractory materials, glass, etc., will take a coating of indium or tin-indium alloy simply painted on its surface in the molten state with a glass brush. The result is solderable with care. Not suitable for high temp environments > 110 deg C for tin-indium (strong joint-low melting point) or 200 deg C for pure indium, (soft joint)

Pure indium can be used like chalk to draw conductive lines on paper, wood etc. Undoubtedly the softest of all metallic elements, it makes lead look like case hardened steel. So far it has shown zero toxicity to workers or users. It is a heavy metal, however, and your hands will become coated with it easily. Supposedly, some few reports exist of contact dermititus developing in regular workers, but you could say that about cheese or latex prophylactics. I handle the stuff bare.

I used indium wire on fusor III for all my seals.

It is truly a wonder metal. I put it up there with tin as a very special elemental metal.

Full details are found at Indium Corp. of America's website in the tech data section.

Small lots of ~1 oz. pure indium can be purchased from Roto-metals for about $30.00 as of this writing.

Richard Hull

[Peter Schmelcher]

You have to like a metal that creaks when you bend it, I really do need to get some. After reading the Indium Corporation application notes I learned that copper will soak up indium over time. It needs a nickel barrier layer to stop the diffusion process. Nickel plating is relatively easy and I have been contemplating nickel plating brass to reduce its vacuum out gassing.

I think plan A for the o ring scarf joint will be regular solder. Does anyone have experience with potable water lead free solders in a vacuum Sn 95%- Ag4%- Cu1%? Just guessing the other impurities will be small for drinking water service.

Peter