Tritium detection from fusor

[Doug Coulter]

No, we haven't proved that yet, but I think we will -- or a lot of neutrons are coming out of a fusor by magic and/or it's neglecting the other reaction pathway that makes tritium instead.
Though I'm feeling the need to finish, rather than start projects these days, some are so juicy and probably so easy that....well, here we go.

Bill scored a detector for low energy betas, a Victoreen, plastic scinitllator, preamp, insanely thin Al/Mylar window -- and we've even found pinouts for it and the preamp built in.

So this suggested itself like water going downhill.

Parts for detector.

I found a "slip in" pipe cap that could be bored a little to fit the detector face, turned off the 8 sided "head" so I could chuck it the other way in the lathe and bore to fit over the detector window, and inside the detector outer shell. My big system is already plumbed to send the exhaust outdoors, via Qest/Pex plumbing (which is fantastic stuff for its intended use as well -- freezes without breaking, FYI). So this was pretty simple to make.

It fits together like this:

Now, the total gas in my fusor for a run is a pretty tiny amount -- fraction of a cc at STP (I'd have to work it out again, but I think I remember it's about .05cc or so, it's a big tank). We generally run for quite awhile without running any gas at all out of it, and due to the way turbos work, most of the gas that would come out when we pulse the exhaust valve wouldn't be hydrogen, but heavier gasses. But at the end of a run, we take it back to hard vacuum and it all comes out. I'm assuming that letting in a little fresh D during that would flush the original tank contents out, and by arranging things right, any light gas would collect in this detector by convection, which is why I show it in the orientation I do. Obviously, it will go utterly nuts during a run from X rays and neutrons (plastic scintillator), but that's not when we'll be using it anyway.

Could be fun....we will see sometime down the road, I've got to drill a hole though the shelf in the rig for this long thing to project up through as it's going to be plumbed right at the forepump output, which is just under the shelf.

[Doug Coulter]

In the unlikely case anyone else scores one of these, or the much more likely case I forget where I put the documentation, here's the pdf for the Victoreen.

Victoreen843-22-1.pdf

Victoreen beta detector info

(158.55 KiB) Downloaded 377 times

Edit: Contains this schematic. Obviously no money spared, an LH-033 is pretty expensive for this use and much faster than needed as well. Note they even bootstrapped the input bias!

Schiz

[Bill Fain]

Hi, Looks good. I wonder if it would do any good if you filled the cup with activated charcoal ? Maybe have a u small trap on the other side with a few mm,Cm of silicon oil to contain the pressure without hurting the membrane. -bill

[Doug Coulter]

Hopefully, we won't need those things. The charcoal would quickly "fill" with other things (it's not very picky about what it absorbs), or results from previous runs -- Or so I'd guess. I am a little worried about the pulsations from a normal pumpdown from air/STP messing up the diaphragm, and adding any extra resistance would make that worse -- and the pump output would easily just blow out any oil in a small trap when it had to do actual work, like after any time the tank has been at STP -- it really blows out of there then (1Hp forepump), noisily. The way I plan to do this, any hydrogen would have to go downhill against air to get out anyway, 'we'll just have to try it! I did the plumbing to outdoors sloping down the entire way, to handle any condensation, so hydrogen won't just leap out of the system, it would have to diffuse. No point fixing what we don't know is broken yet.

This does remind me I need to put a tight mesh screen on the outdoors end to keep bugs from building nests in the pipe, though. It's that time of year, and I'd hate to anger any wasps building in there...

Lessee -- 10 liter tank (gross estimate). 10k cc. Total equivalent gas in there at 2e-2 mbar (.02 x 1/1000) would be 10 * .02, or .2 cc at STP, more or less? That pipe cap holds maybe 50 cc or so, and the light gas would float up to the detector face, or so I'd assume. It would take 10 fills and flushes to even make 2cc of output gas, probably enough to just get the T into the cup. Obviously, this will have to be glued on (silicone so it can be removed) so putting things like charcoal in there that need to be changed every time wouldn't work out operationally. And of course, the charcoal would stop the very low energy (18.6 kev) betas cold, not what we want -- that's why the detector window had to be so doggone thin and fragile. It looks very much like this detector was optimized for T (we got it from a nuclear (fission) plant) -- nothing else would need such a super thin window. (I'm guessing it's a lot thinner than a mil).

Of course, if you wanted to grind up some of that palladium you have in your "end of the world" PM stash, we could catalyze any hydrogen into very heavy water in the cup....probably only need an ounce or so (!). Just kidding

Ok, let's have a little back of envelope fun here. People who really know this math will understand how grossly oversimplified I'm doing this, but I'm making the estimates conservative by the incorrect assumptions I'll make (bear with me a minute). As Robert Heinlein said, anyone who can't get to order of magnitude quick this way doesn't deserve to call themselves an engineer. Those words and that idea have served me quite well over my career, you just gotta have perspective!

Lets assume we're making 1e6 T's per second -- not far off, we probably make more on a good run (lots more). We'll assume we catch most all of them (not conservative, but see below).
In a five minute run (300 seconds), that's 3e8 or so T's we make. T has a half life of 12.32 years, or 17740 minutes (neglecting leap years etc). So half of those will decay in one half life, making our number 1.5e8/17740 per minute for decays. Divide through: 8.455467869e+3 counts per minute if we get them all counted. Now that should be very solidly out of background!

Edit: My numbers were wrong. The real number of minutes in 12.3 years is: 6464880 (12.3x365x24x60) so dividing through again with the right number of minutes gives 2.320228682e+1 decays per minute. The question now becomes accounting for the somewhat larger number I actually did measure. This might not be rocket science, since my math that assumes constant decays for 12.3 years is clearly wrong -- decay is faster at the start of the interval (at least twice that, as 100% of the isotope is still there).


In truth, since they are all there at the beginning, the initial count rate is going to be faster yet -- the count rate per 12 years is an average (which isn't right for exponential decay, but close enough for hand grenades), higher at the start, lower at the end. But since we don't probably catch them all, and won't count them all either, it kinda still gets us into the ballpark, no?

I'd expect at least a k count/minute after a good run, given all the losses and errors. Probably more. We'll have to see what the background is on this detector to see if that's going to be good enough to declare victory or not, but I suspect it will be if my other experience with phototubes and scintillators in our background means anything. This particular scintillator is a lot smaller than the one I'm basing my guesstimate on after all, so the background shouldn't be as high as henny (about 10 times the size) or my other 1" cube (about 3 times the size). Henny gets to 6000 cpm on background, the other one about half that, and this one should be less than half that, or about 600 cpm, I'd guess (but no need to guess, this I can measure and might get that done tonight). So our signal should be well out of the noise...Cool, if true. This one's worth it.

[Doug Coulter]

Ok, finally got that horrible military circular connector cut off the thing and got it fired up. It works! It's sensitive to all of alpha, beta, gamma, with it seems a slight preference for beta, as when I tried a hot beta source, the big counts dropped off quicker with distance than the ones I assume are from gammas, well past the range of any alphas from the "hot rock.

Here's some waveforms from it. This is with a Cs-137 source, but even though the prime decay of that is beta, what this is showing are gammas -- the Cs-137 is in thick enough plastic to stop all the betas from it, so we only see the gammas from the metastable decay product. Good enough to test the thing, and nice to know.

Lots of pulses with scope in "peak detect" mode.
so you can even see them on this timescale.

Still in peak detect mode to see noise. Not much!

Beta2.gif (10.07 KiB) Viewed 7512 times

I'm guessing this will see the amount of T I predict just fine, thanks. Background (by eyeball on the scope trigger) seems to be about 1-2 per second, with some seconds no count.

I'm going to go ahead and goop it into that pipe cap so that hyper delicate window doesn't get messed up. Amazing it lasted long enough to get here, the protective cap they had on there just falls off, and did in shipping.

Since it sees gammas (no surprise for any scintillator), it will obviously go nuts during a run. What happens after one during pumpdown is what will be interesting.

[Doug Coulter]

OK, it's basically in place now, and almost ready to smoke test. Due to the layout, this was a little tricky to get installed, as you can see in the pix, but now that it's in there -- it should stay where it was put fine.

Overview -- the thing had to go through a shelf.

So I had to make a hole in the shelf between the forepump and the rest of the rig. A little tweaking and it fits. The round white thing behind is the lower neutron detector moderator.

The top sticking through the shelf

Underneath is the plumbing, since that's where the exhaust out of the "Harley V twin" forepump is. The blue pipe is the Qest stuff, on which I used hose clamps instead of the crimp rings so I can take it apart later without violence.

Bottom plumbing

You can also see the SSR I use to switch power to the forepump, based on either a manual switch or a trigger based on the turbo pump power drain -- saves power and hours on the pump.
The lead ingots are to help damp vibrations (the pump is on rubber mounts), and the shiny stuff on the underside of the shelf is DynaMat for acoustic supression -- same idea, this is the expensive stuff pro car audio guys use and it has a number of uses (I use it to damp vibrations in benchrest rifles also).

So, I'll be testing this soon, probably just with a scope at first, but the plan is to hook it through our PIC multi-geiger data AQ front end. It's getting tight down there and I'll need to add some power supply stuff as well for this.

[chrismb]

It's a mighty and bold attempt to measure the [most likely] unmeasurable, Doug. One slight hitch - with neutrons flying around, you will have to work out a scheme for working out what beta counts are from activation products in the detector itself. I'd be surprised if you had any [snowball's] chance of discriminating that over any T.

If you do get measurable T, I would expect it diffuses into the surfaces of its containment on a timescale very similar to some of the decay products you might get - meaning you may actually make a detection mistake the 'other' way [that is, actually get T, but then think it is not, because of the apparent 'decay'].

I think you are going to have to be a bit cleverer than this - some sort of hydrogen sponge in front of the detector that you can also heat up and cool down. That way, after a few hours you should get a steady reading rather than a decay curve, and heating it up should reduce the reading as the adsorbed hydrogen outgasses.

Just a few brainstorming thoughts... they may only prove to be worth what you paid for them!....

[Doug Coulter]

Well, you'd have to take my numbers, above, apart some way, and now it would be in the face of an actual measurement that seems to have succeeded.

Remember, I'm running more or less batch mode -- I'm not looking at a tiny percentage of reacted D as it flows through -- I'm looking at the entire tank contents after a run, all dumped into the detector (which would hold many-many runs worth at .05cc run). The detector T capture volume is about 50cc or so, arranged so that H floats up to the window, and heavier gases are preferentially pumped outdoors as you can see in the pix above.

I just did a run with the new stuff lashed up. Jonathan mentioned perserverence. Even though I didn't do it just right, I did in fact see (and still do) a large amount above background on the new detector -- after the run. As you say, it's noisy during one, from what I can tell it sees neutrons too, but not gammas in particular. I tested for gamma by putting a vial of U2O3 against it to see what effect it had on the background count, and it was nada. Background is roughly 1/sec with HV power (to the fusor grid) off.

After a reasonably hot (in all senses) run just now, I added some extra D to the tank, then pumped it down -- all the gas into the detector (except for one mistake that makes me think next time, probably today, will show better results). Then I pumped it down into the detector, which is now counting 4-6 cpm -- many times the background, and the pulse heights are a continuum, unlike what I see with cosmics in the background (those are all clipped).

So, I think I just did it!

My mistake? I'd left the backing pump going the entire time, which means I'd been pumping its bleed air through there too, so I only caught a little of the actual reaction products, diluted 100x or so with bleed air. As soon as things cool down, I'll go turn off the bleed and try again, the right way this time.

But I'd say that 1 cps to 4-6 cps is about 400% out of the noise, and that I can claim victory already.

I'm guessing I only caught a tiny fraction with that bleed air going through there all the time (I do have to hit the inlet and outlet buttons a couple of times during a run to keep it stable, but again -- .02-3 cc per shot there -- while the air-bleed is probably a cc every couple seconds.

I took off the neutron oven after 5 min and saw 1854 cpm off silver. I left it off as it was burning the plastic of the oven. Rough conditions 45 kv @ 20 ma. D @ 2.2 e-2 on the Pfeiffer pk-251 gage (see cal sheet on that, this isn't the true pressure for pure D).

More later -- time to pop a cork and when this all cools and the sun brings my house batteries back to full, I'll try another hot run and this time not blow what I'm trying to measure out of there with bleed air....we shall see, but given what I see right now -- my numbers above were not real far off, I should be seeing kilo counts when I do this correctly.

Yahooo! I'll get better numbers and scope pix soon. Now I'll have to add a bleed into the tank for some other gas so I can wash the T out of the detector and make it stop! I'm not going to waste that much D every time

[Doug Coulter]

A second run with more attention to pumping technique has brought the count up to 27 cps on the beta detector, with everything else off, and the tank pumped back beloe e-7 mbar.

Eyeballing the scope, these are not cosmics. The always-running geiger is nominal, about 1.5/sec, and the neutron detectors (which also see cosmic showers) are not co-incident with the beta detector -- normally a shower trips them all at once (and at full amplitude on the linear outputs). Sure is nice to have a 4 ch ghz scope...

I suppose the final proof will be when I flush out the system and get rid of whatever is in the detector to assure it resumes its normal background when I do. Soon enough! I'm planning on opening it up again soon anyway to enhance the gas inlet system a little (cap tubing post valve).

In case it tells me anything, I'm just going to leave things as they are overnight. If I pumped much hydrogen down the exhaust tubing, it may float back into the detector and increase my count.
After all, I have years to see the T activity before it self destructs. There never was any need (or desire) to make the measurement while running.

With a proportional beta detector, there never was any hope whatever of discrimination against other things -- as betas come out at energies all over the place, from near zero to the max for the isotope involved, see: neutrinos.

Edit:
After 10 minutes of flushing dry nitrogen through the system with the gas bleed wide open, the beta count is now back where I started -- about 1-2 cps.

I declare victory Hopefully future events won't cause me to need this one

[Doug Coulter]

And after another hour or two, the count rate is back up. Looks like you were right, Chris -- either I didn't flush it all the way to outdoors (12 feet of 1/2" pipe sloping down) or some sank into the plastic and then soaked back out and floated back up into the thing. Since I've got another excuse to open the tank anyway, I'll check again after pumping down from STP -- that's a lotta liters and should hopefully really flush things out.

I've not had trouble with the plastic scints or phototubes getting activated here, so far, nor the tank itself. At Bill's request I checked all that the other day (for other reasons) but it all came back cold within measurement limits.

More when I know more. Sure was fun! Maybe it even worked. Do check my calculations above to see if I made a real boner.