Tritium detection from fusor

[chrismb]

Despite your envelope calc above, I still maintain my surprise that you get something detectable. But spectrometry indicating T betas would be a clincher, which it looks like you might have there.

The other way that I'd be fully persuaded is (where I was getting to) that you take your sealed chamber of exhaust and then heat it, cool it, heat it, cool it, etc., on a defined time cycle - a half hour, or whatever is the shortest 'soak' time - then show that the beta count changes on the same time cycle. I'd see no other explanation at all then, other than a radioactive gas adsorbing and desorbing.

Nice work Doug. There's little that can be more convincing than experimental measures that address skeptical comments!

[Doug Coulter]

Well, my envelope calc above has a boo boo, which I edited (left the original, but added the correction in orange, see above). Now it all makes a lot more sense. I'd run 20 minutes for the first run, and about the same for the second. Assuming I really caught all the gas on that second one (I did try harder to get it right) then the numbers actually agree pretty closely! Or, great for such a simple estimate.

In a way, that bit of accidental self-deception made this a more-honest experiment, I was expecting lots more (see the orignal math) but I reported what I did see....and a measurement beat a guess once again.

I will continue this series of tests and get some really good numeric measurements, rather than eyeballing the scope counts in 5 second sweeps -- with a little signal conditioning I should be able to push this signal into our multi-geiger (tm) counter/data aq. It seems obvious now that I should have made provision to flush the system better without having to take air through the system and pumps. I will add a fitting between pump outlet and test chamber so I can push shop air through there and get rid of last run's T and baseline every time.

It appears that running the fusor at record levels (twice the normal power input) for long times (4 times a normal run time) is bad for some of the pieces in there, I melted some things inside it seems, and will have to get in there and replace a quartz sleeve (or swap it end for end) on the feedthrough, if it didn't weld to the pyrex, in which case I have to make two more pieces. No big deal, and for sure that will flush the system, but if I can finde a T and a plug in my plumbing junk, I'll use that time to also add a flush port for shop compressed air to clean things out.

Late last night, what appeared to happen is that any T I'd pushed out too far (due to bleed air from the backing pump) didn't quite make it outdoors, and during a wait it simply found its way back up to the chamber, being lighter than the air it was mixed with, and the system being designed to slope down all the way to outdoors. So after the flush that reduced the counts, they went back up again. Hopefully I don't have issues with the stuff soaking into the plastic, or worse yet, the detector. I'll know soon enough, but it seems wise to go ahead and fix the issues I know about (messed up FT and need for flush port and signal conditioning so I can get real counts rather than eyeballs) and try some more times.

I can't do a beta spectrum very easily, as "there is no such thing" in reality. Betas have a continuum of energy levels, the rest of the fixed Q of the decay coming out as a neutrino in a variable ratio. So if I hooked a MCA up to this detector (possible, we are evaluating one now for purchase) all I would get is the classic "Square box" from zero energy up to the limit for T, which is 18 kev or so, pretty low. Here's the wiki on beta decay and why I can't do a spectrum that means anything much.
This is a special detector head to even see energies that low, the window over the scintillator plastic (NE-102) is only a few microns thick to allow such low energy particles (electrons) through at all.

Tritium link at wikipedia. claims range in air to zero energy is only 6mm (for the few full-energy betas)...this is pretty "weak tea" as radiation goes.

But hey, I found the mistake and now it all lines up pretty close. I'd expect to count something less than half the decays because half or more would simply emit a beta away from the detector face -- it's all adding up pretty well (it kind of surprises me -- usually my back of envelope will only get me to the right order magnitude -- and this seems right to about 1 digit significance, a little better than that).

So, now I feel a little better myself. Nothing more to do but repeat a few times and get better numbers (or not) to see if this is viable for sure, but even after the correction, yes, it looks possible, and not even all that hard. There could be various ways to optimize T detection as well. In a proper box (light tight) you could eliminate the window, and arrange the scintillator to capture light from the full 4 pi angle of emissions too. This particular detector was designed for a fission plant (I guess they worry about low energy betas in those too -- T in the cooling water?) and isn't perfect for "best possible sensitivity" -- wouldn't need to be for what it was designed for - safety. We are no way into a range where the amount of T we can make is a safety issue!

Edit:

Thanks Alex for sending me a mail with the correct math. Maybe you should join, we evidently use some guys who understand the calendar

[johnf]

Doug there is a quantum 8 on ebay for 250 including NaI detector

I've also got one of these all fairly commomn TTL and a uP

but I have no manual so if you are looking at it a pdf of the manual would be good

[Doug Coulter]

John, we're looking at this one:
http://seintl.com/products/ursa_II.html
And I've got one in hand, a demo/factory mod sample. It seems to work at least as well as any of my NaI heads. The guy who wrote the software says that's "terrible" and I can believe it, since my current heads are all specialized for something else other than doing nice looking spectra. We've ordered a 4" Harshaw from someone on ebay, and I'll be posting about the thing once I get spectra that don't look so nasty.

Still wouldn't do squiddly on a beta detector for the reasons given in links above -- betas don't strictly have a line spectra, there's this neutrino thing going on there...

[chrismb]

The Saint-Gobain literature says that the plastic scint material Geo is selling at the moment (PVT 408) is a good choice for beta detection, in particular - but I guess the weedy beta from tritium would be a tough call to detect with scintillation material?

[Doug Coulter]

The issue with low energy betas is that they're low energy (big surprise). From any beta emitter, the energy distribution goes from the peak number all the way to zero -- those pesky neutrinos carry away the rest. This gives you issues with any window in front of a scintillator - they have to get though with enough energy left to make some light (a few eV?). Now, if you can contrive to have no window -- you're fat. Since the betas are also slowed or stopped by air (or anything with mass) then you can't have them coming from far away either. If you were going to design some sort of ideal T detector, you'd have a way to pump the T into a light tight box, and no window between it, and some scintillator of maximum surface area, since air at STP stops even the top energy ones in 6 mm. So I'd imagine having, say, a bunch of thin plates of scintillator, all joined at one edge to some sort of light conductor that would in turn bring the light to a phototube face. Say you had a perfect saw that made really smooth (optically smooth) cuts. You could take a block of scintillator and slit it a bunch of times, so the T in question could be pushed through the slits and always be near some scintillator that could bring light to the tube for detection. Now, since there is no such thing as a saw for plastic that leaves optical surfaces behind -- they don't do that. I'd suppose if the kerf was wide enough, you could get in there and polish the sides of the cut -- I might try that someday. But polishing these scint plastics is a real chore even when the surfaces are easy to get at. From what I've been able to find out, almost any scintillator sees betas just fine -- other considerations are more important, like matching light wavelength to the tube, how well they conduct their own light into the tube, and the other usual stuff. ZnS::Ag in other words, would work fine -- but it's almost opaque to its own light...and has such a high index of refraction, it won't let light out into most other light guides -- it just bounces around inside the ZnS.

The one I have is more or less your basic scintillator, thin plastic (no need for thick -- the betas stop easily in a short distance of it) with an insanely thin window of aluminized mylar over it for light tightness - barely thick enough to be opaque (and probably not in a strong light -- it's much thinner than a space blanket and I can see some light through those). It came from the old stock of a fission plant, where it was evidently intended to be used in safety measurements (this one didn't seem to have ever been used, was just a backup, in the original packing). I presume it was made to detect low energy betas, as it's labeled as such, and that window is hyper thin - and therefore not something that is rugged, something you'd only do if you had no other choice in the design. I would bet a no-window one would be much better, but harder to make -- and you'd have to tackle making the gas inlet/outlet light tight enough for a phototube - no small task, but one possible of accomplishment with enough work and careful thinking. In other words, a lot of work.

You'd think that for example, the pump outlet, with its threaded fitting, is a pipe into total darkness, and keeping light out of that end of the cup would be trivial with metal fittings. Not really, light is fairly snaky and will go down around the fitting threads....unless some really opaque thread goop is used -- opaque for the wavelengths the tube can see. Many tubes see near-IR, to which carbon black is transparent, for just one nasty example - TiO2 "white" is more opaque there! But whatever binder you use (as in spray paint) is probably transparent and will conduct light between the grains... Detectors tend not to be just "cobble something together"-class efforts, in my experience.

Possible, sure, but this is also one reason I like phototubes where separate anode connection is possible -- so you can see dark current without the dynode divider chain current and noise. If it basically isn't zero (See Charles Wenzel's work on what constitutes "zero") -- you've got a light leak, or a noisy tube; I use a 1 meg resistor and a millivolt/div scope to check, DC coupled - if the baseline is basically right at zero volts, with 30-100hz pulses on that -- you're there.

I like to be able to see single photons, and most modern tubes will show those if there's no light leak -- you can actually count them (virtual photons from accidental electron emission due to thermal agitation in the cathode) in a lot of cases - the dark noise tends to correspond to some number between 30 and 100 photons/second and you can see them as individual pulses with the right gear (after the 10e6 gain of most tubes, that is). That's the noise floor. A single beta might make from 1 to thousands of photons in the scintillator, and I'd guess where you'd declare a beta for sure would be in the hundreds of photons, nicely out of the noise floor -- you can set a threshold and lose a few of the lower energy ones and still get pretty good statistical accuracy since the beta "spectrum" isn't really time varying for a given sample, other than its decay. Not an issue with a "long" half life like with T -- I'm not going to wait long enough to see that decay curve!

If you see DC current -- that's either leakage (which will be noisy) or enough photons to look smooth -- and one or a burst of a few more will have a harder time getting out of that noise floor. In this case, the DC is noise itself -- just rectified and smoothed by statistics.

[johnf]

Doug
what you need is a channeltron or a multichannel plate inside the fusor that you power up after a run.Basically like a photomultiplier with out the envelope

[Doug Coulter]

Wow!
Duh. I even have a couple channeltrons, but I've not taken them out of vacuum pack - I hear my habits of tank-opening often give them short lives (and I had something else in mind for them). I wonder the tradeoff there -- detector sensitive area vs how "spread out" the T is in fusor conditions after a run? It's pretty tenuous in there, usually 2e-2 mbar on my gage, more or less. The calibration curves are dodgy at that point, but it's maybe reading double the real pressure. I didn't even think of trying to measure it directly, but I suppose those betas have really long range though almost nothing. One might figure out a way to herd any such low energy electrons to the detector, though...

Interesting


I'd really like to find some multichannel plate stuff for entirely other things, though -- very fast camera for example (telescope and earthly physics both).

[JonathanH13]

Fantastic result Doug - if it is repeatable then I think that certainly overturns some long-held wisdom around measuring fusor generated Tritium.

Does your Deuterium bottle specify how much Tritium is in the Deuterium gas, from the supplier (I'm presuming it is in the uCi/litre range)?

Would a channeltron detector operate at fusor pressures?

Also thinking it would be nice if you could calibrate your new detector against a known Tritium source…

[Doug Coulter]

Yes, I'm on cloud 9 -- especially since the math now agrees pretty well. I doubt there's any T in the D, but they don't say (the main contaminant is supposedly plain H). I did notice that a flush-out with D does take the counts back down, but is kind of wasteful, so I'll institute a better flush scheme real soon.

I think the "long held wisdom" is held by people with one heck of a lot less experience handling small (minute, really) gas samples, mixed with some sour grapes on none of them owning a mass spec either. 100% of them run their fusors in flow-through mode, and in their case, yeah -- no way you'd be able to detect it as a tiny percentage of that huge flow -- they are probably running hundreds of times the gas through I am. Run 20-40 minutes in the same gas batch, at 4-5 million neutrons/second output, and I have a better sample to work with than they will anytime soon.

I still think we might see it on a mass spec too with refined technique, but it'd be a lot harder. That radioactive tracer sure is helpful! But of course, you'd know that already from what you've been doing for a living. I might make a try on the second fusor once I really tune up that mass spec -- it has some pretty fine resolution down there, so we should be able to resolve less than an amu (much less) and see TT, DT, HT and so on all with that little offset from plain H's mass due to binding energy. One nice thing about being able to suck the data into a PC is I can really bring some super signal processing to bear on it that just wasn't even in the imagination of the guys who built that thing n the first place. Then, it was "impossible". Now, not so much, and when it comes to DSP, I'm the man - at this nuclear stuff I'm a relative newbie, well read, and with some experience, but not in the same class as what I did digging signals out of noise for a living for decades.

At the moment I'm off on another metrology quest - a 5" NaI head just arrived (wow, it's like 10 lbs or more of just the xtal)....and a MCA that works right (for a change)....I just need to slap together a new 2kv power supply so I can get it out of the noise, and maybe a preamp. Just gotta love it when a plan comes together .

Maybe we can enter the impromptu "see who can activate the entire periodic table" contest at some point, with proof. But that's a side issue for those of us looking for gain out of fusion.

A channeltron would probably be ruined very quickly at fusor pressures, if you put it under voltage. The "stuff" is an oxide, which would be reduced by the H in there when hit hard.
But it seems this is actually good enough, so no problems. I'm saving the channeltrons for the "tabletop ring collider" experiment, which runs at channeltron-friendly conditions.
It certainly is an intriguing idea, though.