FAQ - Flow Control: Why, and how, is deuterium flowed during

[Chris Bradley]

** Why do we need flow control?

Flowing of deuterium during fusing operation is required even for the best amateur fusors. This is because creating a perfectly sealed system with no real or virtual leaks is well beyond amateur means. There are also risks from sputtering mechanisms, for which it is desirable to maintain continual evacuation to avoid a build up of contaminations of heavier species (when non-deuterium nucleii are in the chamber then the probability of fusing collisions is significantly reduced). Therefore, continual pumping is required which, in turn, requires continual admission of deuterium into the chamber.

So the objective is to, first, reduce real and virtual leaks to the minimum that is practically possible, and then allow a given, minimum, amount of deuterium to flow through the device whilst still pumping the chamber sufficiently. This will then maintain the required pressures and gas purity.


____________________________________________________________________________



** What base pressures do you need, and what flow rates of deuterium?

Let's set a target of 99% purity for the gas flow through the device, and assume a desired operating pressure of 10^-2 mbar. (Let's assume the source gas is 100% pure in the first instance to make the maths easy.)

So to get to 10^-2 mbar we'd have needed initial conditions of 10^-4 mbar before adding in our gas.

We also need to look at the flow rate. Let's say we had a turbo pump rated at 60l/s, and that the base pressure achieved was 10^-4 mbar with this pump. With additional pipework and a flow-rate valve between it and the chamber, it'll be more like 30l/s. We can calculate the required flow rate of deuterium from the base pressure and pumping rate: 10^-4mbar is one ten millionth of an atmosphere, so 30l/s is 30,000cc/10,000,000 per second = 0.003cc of atmospheric gas per second. Generally we discuss flow rates of gas in sccm (standard cc per minute) so this equates to 0.18 sccm.

So our 30l/s pumping achieving 10^-4 mbar is pumping 0.18sccm of the ambient gasses from real and virtual leaks. Now we also need 99 parts to 1 for the flow rate, as well as to the base pressure. This means in this example we'd need to be flowing 18sccm of deuterium to keep the purity up to 99%.


____________________________________________________________________________



** What base pressure should I expect can be practically achieved in my first build attempt?

The example above would be a poorly performing chamber. A base pressure of 10^-4 with a turbo or diffusion pump is pretty poor, and your 0.18sccm leaks are excessive if all you have is a small chamber with a few metallic and ceramic components. Although you can still go ahead with a 99% purity by flowing deuterium fast, at 18sccm, it likely means there is a leak in the chamber which is probably easily fixed. Either that, or you've accidentally dropped a bit of a cheese sandwich, or something else undesirable, into the chamber!!

Clean it out, seal it up again, and keep pumping it for a few hours will rid the chamber of most out-gassing sources like fingerprints and solvents, and should get you to 10^-5 mbar easily enough, which seems a typical value for first reports of systems put together competently, and this is perfectly adequate to achieve a 99% purity we're typically seeking whilst flowing deuterium in the order of around an sccm or two.

One can make a fine art from honing a vacuum system until the base pressure is in the 10^-7 or -8 range, but in reality once you are at 10^-5 and flowing 1 sccm of deuterium, a 50 litre bottle will last you almost a 1,000 hours of continuous operation. It is your call as to how to balance your money on improving your vacuum system, or spend money flowing more deuterium.

After operating for a while, your base pressure will likely drop quite naturally to around 10^-6 as the chamber is further rid of outgassing sources (providing you have no real leaks!). You can then either continue to flow deuterium at 1 sccm which will serve to further improve the purity of the gases in the chamber, or you can eek out what you have further with a lower flow rate so that you can run your fusor *continuously*, day and night, for more than a whole year! I'd tend to recommend you run at 1sccm and benefit from the improved purity, as you're unlikely to use up a whole 50l bottle in 30 years if you run it for a half-hour a week at 1sccm!


____________________________________________________________________________



** Isn't the deuterium flow rate required affected by throttling the pump?

The pump needs to be throttled back in some way (e.g. by a gate valve, or slowing a turbo) so that the pressure rises to the desired operating pressure (~10^-2 mbar)

The calculation above (e.g., showing 1.8sccm minimum for a system with a base pressure of 10^-5mbar) is based on the calculated virtual and real leaks in the chamber. These will be similar even if the chamber pressure is higher due to the throttling. (There will be some reduction in outgassing rates between 10^-4 and 10^-2 mbar, but it is unlikely to be significantly different for the purposes of this discussion.)

So if you hit 10^-5 mbar with 30l/s, then you'll need to flow 1.8sccm deuterium at whatever operating pressure your throttling lifts the chamber pressure to, to maintain 99% gas purity.


____________________________________________________________________________



** How do we flow gas at just 1 sccm?

This flow rate is a very small quantity and regular mass flow controllers usually operate a few orders of magnitude higher than this.

So instead what we need to do is introduce a fixed and very limiting restriction into the flow of gas so that if we have a few atmospheres on one side of it (from a gas regulator sourcing from a bottle) and the vacuum on the other, then it will flow just this ~1 sccm range.

The idea then is that we can control the flow rate through this restriction by controlling that differential pressure, instead of adjusting the restriction itself. Some here have tried to use very small flowing needle valves to regulate the whole flow in one go. This is almost certainly doomed from a practical point of view because the sort of restriction we need to achieve 1 sccm with one atmosphere across it is something in the order of microns of gap size. There are such adjustable needle valves in existence and maybe you'll get lucky with scrounging, or can afford one, but they are expensive and, ultimately, unnecessary.


____________________________________________________________________________



** What flow restrictors do we use?

Two particular types have been generally used here: a length of capillary tube, or a laser drilled orifice. The capillary tube can be cut to a certain length to achieve the desired flow rate. The laser drilled orifice cannot be adjusted but gives the opportunity to create a system with a fast response time. Both are used here, with good effect.

I am, here below, introducing a third type of flow restrictor which uses a blunt dispensing needle. (I'm fairly sure this is the best value way of constructing a flow restrictor I've seen!)


____________________________________________________________________________



** How do we use a flow restrictor, and what is the 'response time'?

The restrictor is put in between your gas regulator and your chamber. So on one side of the restrictor is the vacuum of your chamber, and on the other is the pressure you have dialled up on your regulator. I would recommend [just my opinion] you aim to design your system so that 1 to 2 atmospheres differential gives you around 1 sccm flow. Then if you increase the pressure your flow rate will increase, and if you reduce it then the flow rate drops. The pressure regulator maintains, approximately, a regulated pressure (as per its function) so in theory the flow rate will stay the same because the vacuum on the other side can vary but to all practical effect it does not alter the pressure drop across the restrictor.

Now let's say you are running your system and have 2 atm across the restriction and are flowing 1 sccm. There is a finite space between your regulator and this restriction. If you want to increase the flow rate, you simply increase the pressure in that space. Say to 3 atm to achieve 2 sccm (practically, the pressure drop versus flow rate does not appear to be linear – I've yet to figure out the interrelation between differential pressure and flow rate!). Now, it is easy to increase the pressure as you have a big tank of gas, so the flow rate will go up straight away. However, if you wanted to drop the flow rate to 0.5sccm (let's say it corresponds with 1 atm differential) then you have to wait a while for the gas in the space between the regulator and restrictor to go down.

If that space is, say, 20cc then you'll have to wait around 10 minutes before the flow rate drops off. So you can see that if you reduce the volume of that space in the deign of your system then you can drop the flow rate more quickly. A capillary tends to be a little more awkward in this respect because firstly you will have to create an interconnection for the tube to go into at the pressure end, and then the capillary will have its own internal volume along its length. Whereas the drilled orifice can be constructed with a very small volume (e.g. if it is attached directly to the output of an additional control valve) and it has no internal volume of its own, so that can be made to have a very quick response time.


____________________________________________________________________________



** What's the 'budget' idea?!

For my epicyclotron project I have investigated a new means; to use flow control needles, which are cheap and readily available. These are used in many industrial and medical areas, e.g. for medicine dispensing, and paint-shops for paint flow control, so can be acquired very cheaply. Look up on ebay using the terms 'dispensing' or 'dispenser' and 'tip', 'needle' or 'syringe'.

But as they are, the orifices of dispensing needles are too large for this application, so my 'trick' is that I've simply threaded a wire through the needle, which gives a fine and tunable annulus. By selecting your wire diameter and needle ID, you can get some very fine orifice sizes (indeed, much smaller than required for ~1 sccm flows).

After shopping and experimentation, I have found that a good way to achieve the flow rate I recommend above is to use a 1” long needle with a 0.21mm ID, and thread a 0.18mm wire through it. You will have to work with what you find, so if you can only get shorter needles maybe the wire needs to be thicker/orifice smaller, etc. Also to bear in mind is the manufacturing tolerances involved – if the needle is 0.215 instead of 0.205 then you might find the flow rate is double the expected rate (because we are using a subtractive process to control overall orifice size).

[I'll mention also that this might possibly be a means to consider adjusting laser orifices too – they are expensive and it may be cheaper and better to buy an oversized orifice, then fine-tune it by inserting wires of the right diameter through it.]

Below is how I did this. Firstly, I used NW10 connections – these have the same flange size as NW16, but the bore is 10mm, so this means there is a relatively large flange area (we don't want the rubber bulging under the pressure, so it needs to span a narrow flange). Then I cut out two circular pieces from 3mm flat nitrile to cover the flange area, with some excess around it. I then made a hole through the centre of both with a sewing needle. It is worth noting that the rubber seals itself back up again, so if you were to try to use this as your 'orifice' then nothing gets through! This self-sealing is important for the next stage because we want these pieces to seal themselves around the needle and not leak through (so use a good, clean and thin sewing needle!).

Now I flexed one piece so the hole opened up on the convex side of the rubber and I held the end of the needle against the hole. Then I flexed the rubber the other way and the needle pops through the hole. Do that twice for each piece (you need around 5mm thickness to replace the usual NW gasket), then I cut off the excess so that both pieces will fit together in the flange.

Finally, thread the wire through the needle.

Then simply clamp it up between two NW10 flanges somewhere along the gas flow path. Job done!

...

[Dennis P Brown]

Very impressive and informative post. The flow control is simple but ingenious.

[Chris Bradley]

Thanks Dennis. Blimey, you're a quick reader! I hope it is not too long, but there is detail worth discussing on the finer points of flow control, I think, to help avoid [potentially expensive] snags in buying parts and setting up flow control for the first time.

[Ryan Atkinson]

I'm new to high vacuum control. What is "virtual leak"? Thank you.

-Ryan Atkinson

[Jim Kovalchick]

A virtual leak is a slow release of gas from a pocket within the vacuum envelope. A common example is a threaded joint where gas slowly makes its way into the chamber empty space by passing from within the joint through the threads. This looks like a low leak. For ultra high vacuum applications screws are drilled through so there is a free communication as vacuum is first established.

[Chris Bradley]

Or from some lump of something that is not vacuum compatible (a lump of grease, or a cheese sandwich), or from surface adsorbed gases, often water vapour (very sticky stuff!). This is why 'baking out' a chamber helps because it heats up all the surfaces so adsorbed stuff gets kicked off and sucked out.

[Ryan Atkinson]

Interesting. Thank you.

-Ryan Atkinson

[Mike Flander]

Chris,
Thank you for this post. Much appreciated.