Disk Jet Diffusion Pump

[George Schmermund]

I've had a request to explain the particulars of my claim to have come up with a novel diffusion pump design. As suggested in the previous post, DP's have been around and evolving for a century. That's a lot of time, and as the prime movers for achieving high vacuum for so long one could easily come to the conclusion that every possible configuration and design attribute has been tried. To confirm that this new idea is novel and not obvious to those skilled in the art, I went back through 100 years of patents and all the literature that was readily available.

Modern oil DP's have two modes of pumping. One is downward momentum transfer and the other is true diffusion into the oil vapor stream. I'll explain these two traits in very simple terms and the nit-pickers and naysayers will have to bicker in their own circles. The fact that what I'm about to relate works and can be demonstrated is all that concerns me.

The first vapor pumps designed all used mercury as the motive vapor. This method is purely momentum transfer. Mercury doesn't dissolve gases. If it did, the McLeod gauge wouldn't be very useful.

When low vapor pressure oils came into use these early pumps were only marginally modified to accept the new fluids. Oils improved over time and the DPs were refined to the point where remarkably high levels of vacuum became routine. The primary engineering consideration was (and still is) based solely on the momentum transfer idea. The gasses absorbed in the oil were ignored or, sometimes, seen as a less than desirable trait.

Over the years I've noticed that whenever a diffusion pumped system was turned on from a cold start, there would be a marked increase in the foreline pressure as the boiler came up to temperature. The pressure would increase until the boiler was quite hot, but below working temperature. Then the pressure would drop significantly and soon after that the DP would start pumping and the foreline pressure would again rise as the system above the DP was evacuated. When these systems (including the DP's) were vented back to atmospheric pressure with dry nitrogen, the cold start-ups produced even higher initial foreline pressure. The highest of all was when new pump oil was put into the boiler. Everyone who has worked with a DP has had this experience. No one, as far as I know, has ever attempted to exploit this remarkable property of modern DP oils to produce high vacuum.

Having been cursed all my life with an excess of curiosity, the decision was made to build a DP using ALMOST exclusively gas absorption as a transport mechanism. The results so far demonstrate that the a low 10^-6 Torr vacuum can easily be achieved with DC 704 using a jet assembly where the jets terminate at 90° to the wall. Were talkin' 0° inclination.
The only true and exclusive momentum transfer is where the ejector faces into the foreline.

As can be seen in the photos, the parts are few and easy to make. This pump is a "no brain-er", once you get past the invention of it. I have some tweaks to add that will get me down well into the 10^-7 Torr range with just an optically dense baffle.

[John Futter]

George
Yes I have seen the pressure rise at switch on many times and put it down to water condensing in the oil as it cooled.
well done.
I wonder if this is the underlying reason why diffstaks work better than diff pumps --the bulbous region giving a chance of adsorption due to increased distance to wall ????
diffstaks also get into the high 10 to the minus 8 region without ln2 baffles just a water cooled one above the stack.

I see no chance of boredom at your house

[Doug Browning]

Interesting to see some experimenting with the "old" technologies. The big test will be to see how well it performs for pumping speed versus Watts consumed. The best diffusion pumps are getting over 2 Liter/second per Watt. The worst (VMF-11 comes to mind) are around 0.1 L/sec Watt. A thermal wall isolator barrier could get one up to around 5 Liter/sec per Watt.

Since you are into experimenting on this stuff, let me throw in a few old ideas of mine for high vacuum pumping, not necessarily practical, but worth some mental exercise. Maybe leading to some new ideas.

The gas to be pumped is constantly bouncing off the pipe walls, so if some means could be found to implement directional momentum transfer at the wall surface, it could push the gas along. A transverse traveling acoustic wave (ultrasonic for short wavelength) along the pipe wall might do this, but the amplitudes to get substantial wall material displacement to form a wave shape are prohibitive, at least for metal walls. Some easily flexed material might get somewhere. Maybe mylar film (a phased electrostatic speaker along the wall, or better yet, a transverse stack of closely spaced mylar films with coated metal film electrodes to form phased traveling bend waves, from electrostatics, down their length).

In a higher pressure regime, like the roughing range, the non-linearity of acoustic waves thru the pumped gas could be put to use. For a single stage, a funnel would be snugly placed in the center of the surrounding pipe, with only the small center tip hole facing downstream for gas flow. Acoustic vibrations in the gas would be generated at the large funnel inlet side by a pulsating electric gas discharge thru the gas. The high pressure peaks would push compressed gas down thru the funnel and out the tip (an exponential shape, like an acoustic horn, would facilitate wave compression). The acoustic valleys would be much less efficient at pulling gas backwards thru the funnel, since the sound wave would be rectified by the inability to go lower than zero pressure, plus the acoustic mis-match for a broad wave approaching the back tip side of the funnel. The acoustic valleys nevertheless are still needed to allow new inlet gas into the chamber inlet side for the subsequent cycle. Multiple stages could be stacked for higher compression. A diaphram pump would still be necessary to reach atmospheric.

An alternate version would be to just use the ionic wind from a sharp charged point facing a large counter charged ring. Gas ions formed at the point are accelerated toward the ring. Multiple stages could be combined for improved compression. Traveling high voltage electric waves down a series of rings/sleeves (like a phased linear accelerator) could make repeated use of the same ions (with a synchronously pulsed sharp point ion generator) for better efficiency and compression. This CAN be made to operate up into the atmospheric range. MIT has recently shown this to be something like 50 times more efficient for propulsion than a propeller. Not very all-weather though.

Aerogel foam. You just place this stuff in the vacuum and let the gas diffuse into it. Then mechanically move it to the exhaust region and squeeze the gas out. Recycle. Repeat. Maybe call it the "vacuum sponge". Could use some kind of continuous mode of operation, like a Van de Graff belt, with a set of pinch rollers in one spot to squeeze it like an old washing machine wringer.

[Andrew Seltzman]

Very interesting George, though I do believe it's still pumping by momentum transfer. Although the plates that deflect the oil vapor are perpendicular to the outer wall, there is no symmetric plate under the holes that the oil vapor escapes the central column from. As a result the oil vapor will expand downward as well, giving it net momentum towards the fore line, and allowing it to pump via momentum transfer.

[George Schmermund]

John - Thanks for understanding the experiment.

Don - It's time to return to your home planet. Come back when you have some results from your experiments.

Andrew - Thanks for giving the pump some thought. I'm not discounting your idea about how the vapor might expands when it leaves the jet orifices. but If what you're postulating was a significant part of the pumping action in this new pump, then a disk jet should work as well in a mercury vapor pump. It won't.

Also, keep in mind that I purposely put ALMOST in caps. This pump is a proof of concept prototype.

[Doug Browning]

Gas diffusion into the oil vapor would just mean the gas molecules have achieved thermal equilibrium with the oil vapor, hence momentum transfer completed. Unless you are thinking the oil is still in tiny liquid droplets capable of absorbing or adsorbing gas. CVC went to some extra measure to superheat the oil vapor to prevent that.

MIT: http://inhabitat.com/mit-developing-ion ... t-engines/
(55 x better efficiency, so long for turbo pumps)

I would take up your recommendation, except all this energy related madness is long ago elegantly solved there. Much more interesting to watch the clueless going in circles on this planet while it enters final meltdown.

You can lead a horse to water, but you can not make it drink the water. I can not save you.
John Heywood (c.1497-1580) Jeremiah 32.1-15 Luke 16.19-31

Sayonara

[Jerry Biehler]

MIT: http://inhabitat.com/mit-developing-ion ... t-engines/
(55 x better efficiency, so long for turbo pumps)

So long for turbo pumps? An ion thruster won't even work at the pressures you use turbos for.

[Doug Browning]

I didn't say it would, if you read the above more carefully:

"In a higher pressure regime, like the roughing range, ....."
"An alternate version would be to just use the ionic wind"

But a HYBRID version, with electrified oil vapor up front might just work. So might an electrostatic traveling wave speaker up front. The vapor approach goes directly to the gas, versus the traveling wave approach lets the gas come to the surfaces. (the Turbo IS a traveling wave approach by the way)

The issue is how much energy is wasted in putting momentum into vapor that never hits any gas for the vapor approaches. Or, acoustic energy that gets dissipated in the walls, without transfer to the gas, for the traveling wave approach. Ultimately, you only want to put momentum into the pumped gas. The turbo is quite good at that, but its downside is fragility and bearing wear (and cost). (the Jet engine of course falls down on efficiency due to all the thermodynamically wasted HOT gas being exhausted)

What would really get somewhere useful, would be if the gas could be ionized on contact with the walls or passing through the electron cloud near an (efficient) filament. Seeing as there are thin film tunneling electron emitters (TiO2 for example) of superb efficiency now, which can readily be made into planar surfaces, I would say this is solvable. (MIT is probably already on it for their thruster purposes)

George:
Addressing your venture to make cheap vacuum pumps for students. These thin film emitter surfaces can be made by simple gel electroplating on metal under atmospheric conditions.

[George Schmermund]

Gosh! This thread sure got driven off course and into a ditch quickly!

[Monroe Lee King Jr]

This goes to show how simple a DP is and how easy it is to make one. In fact I bet you could even make an upside down "tree" and it would still work. I don't see any indication of absorption? Where did you figure that? Way to go however on making your own. Moisture is the effect your seeing on start up.

Monroe

[George Schmermund]

What a novel concept, an upside down tree as a diffusion pump! Yes, you are correct that moisture retained in the roots would be an issue at start up.

[Dennis P Brown]

I really like your desire to do this project and your design and execution appear near perfect. I have one concern - any substance, but DP oils included, do have an extremely strong relation between gas absorbed and temperature. Unfortunately, it isn't what your design needs. That is, as temperature rises, gas absorbed drops rapidly (very non-linear.) So, cold fluids absorb huge amounts of gas - that is why/how absorption pumps work and some cryo-pumps, as well (those with a carbon matrix assembly on the finger) - so, as any liquid is heated, gas solubility goes down rapidly and the dissolved gas is released; this is what happens as you heat up a diffusion pump - see your description of what happens to the chamber pressure. Still, hope something else occurs to make your system perform and you get a good vacuum. Best of luck!

[Doug Browning]

"In fact I bet you could even make an upside down "tree" and it would still work."

"What a novel concept, an upside down tree as a diffusion pump!"

Someone has already tried that, guess it didn't work too well:
http://www.ebay.com/itm/Edwards-Speediv ... 566b1ca677

[Monroe Lee King Jr]

Point I was making is diffusion pumps seem to for some reason be confused with some other device that is complicated to make. Because the diffusion pump is a very very simple device. Nobody seems to want to hear it as this device they talk about seems to have some "special" properties that make it function and are some kind of a secret. You can make a diffusion pump from some stainless steel exhaust pipe reducers and tubing. You can also use DOT 5 silicone brake fluid as your pumping medium. (once you run it for a while to boil of any impurities).

The disk working is no surprise to me at all because it is a simple principal the pump works on. Of course it's not a good idea to have the "tree" vents shoot upward instead of down as I was suggesting but it would probably still function that way (albeit not very well). It's hard to screw this principal up.

Over the last 100 years of diffusion pumps they did hone the process and ones available work about as good as engineering will allow. But there's lots of room to play when building one. Will you reach the ultimate vacuum of the best engineered pumps? NO, but you can do pretty darn good just eyeballing a pump design you like.

Will this disk pump work? Sure I see no reason why not. Will it attain better ultimate vacuum than the 100 year old engineering efforts, NO. Is it novel? I'm sure it's been done already and discarded forgotten.

Might be a good design to allow others to build there own from inexpensive materials though. I find that highly useful.

If you can't afford a $250 diffusion pump and oil off ebay they might as well cost $10,000 will that stop you? Doesn't have too all I'm saying.

I think it's a great interesting way to make your own and I appreciate the open sharing of the design that's really cool.

But trying to proclaim it novel or especially worthy? I don't think so. Just wrong tact IMO. But hey! It's a free country last time I checked and razor blades are cheap.

Monroe

[Doug Browning]

Well, conventional diffusion pump technology that I have seen ("Foundations of Vacuum Science and Technology" by J.M. Lafferty or "Handbook of Vacuum Science and Technology" by Dorothy Hoffman) depends on momentum transfer from the moving oil vapor to the entrained gas molecules. Similarly for turbo pumps, momentum transfer from moving vanes.

That's not to say that one couldn't make a pump that works by absorption in a fluid. But as you said, the temperature profile would most likely need reversing.

As to diffusion pumps being a completely mature technology, I beg to differ. These things were designed in an era when energy was cheap. It is well known (see above texts) that 3/4 of the heat input from the heater just goes up through the casing to the cooled section. Totally wasted. There is a patent ( #5137429) for a thermal insulating collar in the casing to prevent that, but the manufacturers could not be bothered (the user pays the electric bill), with the possible exception of the Edwards TVA50/60.
Electrostatic means (like an electrostatic cleaner for example) could be employed to prevent oil backstreaming. But cooled baffles do reasonably well (but typically reducing pumping speed by 1/2). No doubt other refinements could be found using more recent technologies. Ferro-fluids..... Magnetic nano-dust.... electrostatic "paint" sprayers....

[Monroe Lee King Jr]

Perhaps your right about some of the items you mentioned but how would that apply to this particular design? I don't see where it is addressing any of these possibilities. I don't see absorption at work here. The other possibilities you mention are interesting but this is a basic design issue here not one of the more novel approaches you mention.

But I am out of time here again I have other projects that are demanding my attention. I have not forgotten fusor.net and I hope to get back to the stuff I was working on here soon. I will get around to it I'm pretty sure of that. I don't leave things unfinished forever.

But have fun you guy's!

Monroe

[Doug Browning]

Since George was redesigning the diffusion pump, I though I would just throw some ideas out. Not saying they are developed or tested.

One comment might be in order about the diff. pumps available on Eb_y. If you look at the sold items, they are rarely any of the expensive items listed. And Eb_y has gone to the trouble now of obscuring the actual price paid on "make an offer" type sales.

I always try 1/2 the listed price 1st, see if that goes. I got a VHS-4 a little while back that was in quite good condition for $50. I wasn't even looking for one, but it sat there for a week with no offers. Couldn't resist a bargain I guess. Now I have to modify it with a thermal collar (two flanges and a Viton O ring) to get the 1450 Watts down to 400 so I can plug it in without blowing a fuse.

[George Schmermund]

I guess I'd have to say that I'm in accord with the posters here that make claims that a diffusion pump is easy to design and build. Their suggestions that a working pump can be constructed and preform well with only the proper assembly of such things as spray guns, car mufflers, stove pipes, inverted tree stumps, and the like make the job look simple. They're all brilliant ideas to talk about..... so long as you don't have to really understand or build anything. As I've pointed out before, talk is cheap and you'll find plenty of that here.

[Monroe Lee King Jr]

I guess you have not seen my stainless steel tail pipe diffusion pump? And I guess it's not obvious that the "tree" I'm speaking of is the diverter baffles in any standard DP. The tree in this case being disk.

[Richard Hull]

For me, any discussion of energy efficiency in a diff pump is a rather bizarre thing. Again, if it took 2000 watts to run even a small 2" diff pump and it got me to 10e-6 torr, I'd suffer the massive 14 cents/hour electrical bill. Efficiency is sweet if you are in the "another penny or two will break the bank" game, but you are blowing about 1000 watts in the fusor, pumps and instrumentation to get 1 microwatt of actual goodie out of it. Saving even 100 watts at the diff pump, still leaves COP down a very deep well.

As always, the power company and I have a sacred trust....... They supply me with whatever juice I need to get a real job done and I send them money in return. I ran the fusor for many hours on Friday and Saturday of the HEAS event and all that pleasure with all the pumps and HV supply rockin' and rollin' probably cost under $1.50 for both days.

Efficiency! Gimme a break.

Ricahrd Hull