It's come to our (well, everyone's) attention that there needs to be some standard way to measure neutron outputs between labs, and within a lab to see the results of tuning changes.
For the first, I've been an advocate of silver activation, as it is impossible to fool, and integrates over a decently long run, going to Indium for very hot or longer runs. There's going to be a little variability there depending on neutron oven build and placement, as well as counting the result, for which the solution is our Standard Counter - these are all guaranteed to count the same, and be particularly suited for activation counts as they are more beta sensitive than most counters.
However, activations aren't the end-all, be-all, though it's a great start. Silver has some metastable states so short lived that it is more sensitive to the neutron output at the end of a run than the rest, In takes a lot of neutrons to activate well out of background, and neither give real-time feedback to the operator as he tunes things. While we have a much better chance at getting to absolute numbers with activations (within say 20% or so), for tuning we really don't need that. For that matter, as long as everyone is using the same neutron counters, we don't need absolute numbers there either.
While my own lab is blessed with just about every type of neutron detector there is, from BTI's to 3He tubes, both kinds of boron tubes etc, I find that especially for newbs around EMI (which would include almost everyone who hasn't made a life study in it), EMI induced false counts in those tend to create significant errors, and people (even I sometimes) just wind up tuning for max EMI, not max neutrons out of their fusors. I've been using cross-checks with activations to make sure of that one, as well as monitoring the raw signal on scope and audio, but that's more work than many will likely do. So we need an affordable, stable, EMI resistant, neutron counter.
In recent tests with a pretty standard cheap photomultiplier, the Hornyak detector I got from Eljen has come though with flying colors in this regard. Stable, repeatable, pretty immune to EMI because the signal is pretty good and large, and slow, so it's easy to get the right counts in the presence of noise. Since the store bought ones are expensive ($150 for just the button) and this isn't rocket surgery, I propose to start making the button heads. They might not all be precisely the same sensitivity (same would be true of the bought ones) but they'll be stable. I've made one before that worked, despite having the "wrong" phosphor for it. Well, now I have the right stuff, so why not make up a few of improved design? This thread will document doing just that.
The plan is pretty simple. I will cut up a bunch of 2" sq pieces of 1/16" thick plexiglass for light guiding. I will stack these up with phosphor/epoxy slurry between them, and harden this under pressure in a jig, so we will have alternating layers of ZnS:Ag and clear plex light guide. This is required as ZnS:Ag has such a large index of refraction that there's nothing you can cast it in that will result in the unit being transparent to its own light output. A standard Hornyak tries to get around this by interleaving rings of phosphor/resin with just resin in a bullseye pattern.
That's hard to make, and it doesn't permit too many layers for a given phototube face size, here desired to be 1" (so the tube is not expensive or rare -- those are common from old PET scanners and easy to find). I plan to make a 2" by 2" by 1" thick one, then dice it up into 1x1x1" pieces to get 4 made per run on the glue jig. After polishing the plex edges, I'll coat the non phototube facing ones with something reflective, like space blanket material. This actually works better than white stuff, or metallizing the plastic, as if there's a tiny air gap, you get total internal reflection in the plexiglass for the angles that works for, wheras if things are touching, they can suck light out and only return it with some losses. (This is well known in detector lore, I'm just repeating it here for the non experts). Plex and most casting resins or epoxies have an index of about 1.5. ZnS:Ag is in the 4's, which is why it always looks white when cast into things, and Hornyak worked hard to find the right mix for best performance. Later, someone else came along with the interspersed light guide idea, and I'm doing my own version of that here.
Assuming this works out, I'll be able to provide neutron detector heads for the standard counter to complete the kit for fusor radiation monitoring, which is the goal. I expect the main issue will be some variations in sensitivity, as the interlayer slurry thickness will not be super easy to control, but we'll see how that works out in practice vs theory.
So, I've already cut up some plex, now I'll make a jig to hold them in alignment and under pressure while whatever I use as the adhesive gets hard. I plan to go for a fairly heavy concentration of phosphor in those layers, since the rest is clear anyway, and I'd never have a working thing with that concentration low enough that much light would get through the dispersion inside that layer anyway - it will have to get out into the plexiglass and go to the tube from there no matter what.
One big advantage of a Hornyak - it only sees fast neutrons. This means no moderator is required, and unlike say a 3He tube, the counting rate won't vary as a big bag of water moderator (eg me) moves around the lab during a run and changes the neutron flux for slow neutrons. We are also looking at other neutron-recoil type detectors, but this one is known to work, so it will be the standard the others are judged by. You can in theory just use a light gas - H2 or He, in a proportional tube and count that, but again you're getting into low signal output there, though you save having to buy a phototube. Those will get their own threads at any rate so we can do honest side by side comparisons with them all with the same source (fusor) running during our testing.
Even with unpolished sides and the brown paper on there, light gets through this stack just fine.