Eye Candy My Lab

[johnf]

So you guys can get a handle of what I do for a day job.
The pic following (I hope) shows the lab I spend most of my time in.
bottom of pic shows the 4 beam lines coming from the single ended 3MV accelerator left most beam line = PIXIE and proton microprobe where we can scan the proton beam over a sample to get elemental map, beam spot 10um
next beam line is associated with an ion implanter and 1kW E-Beam annealer-- this allows us to depth profile with RBS while doing implantations and subsequent annealings.
The third beam line is for NRA, RBS and external proton beam (through thin Be window) for samples incompatible with vacuum
the forth beam line is under construction for high resolution RBS we hope to get depth profiling down to 0.1nm or better] Vacuum in sample chamber for this is 2 x 10 to the minus 9 millibar..

Top of pic left to right dual ion implanter so two ion species can be implanted at once
next argon ion sputter system
old SEM rejigged with ultramodern image capture outputs 10meg tiff files

[Doug Coulter]

Extremely high drool quotient there! The only thing that would make that better is if it were here and we could just play, rather than doing whatever makes the boss money.
Well, some things are both I'm sure.
Vain hope, I know, but....just wow!

I note you have some square feet of table top showing there. I've never mastered that one -- every horizontal surface here has junk on it. Guess I need to get a guy with a large whip
to enforce cleaning up and finishing projects before beginning the next. There does seem to be an optimal number of projects to have in progress, so you can work on the one
that will make the most progress at any given time, but I blew past that limit long ago....passing from "creative chaos" to simple chaos. Going to have to do something about that.

[fusordoug]

John,
I'm noticing you have some pretty serious magnets there, that look like they are post-acceleration. Are you using these for energy filters/selectors? That's my guess from the picture.

Which brings up an interesting topic -- why does everyone seem to do it post-acceleration? I can surely think of some applications where doing it (as much as you can) pre-acceleration would result in some big efficiency gains. Does recombination or some other effect make that idea useless?

My thought is that for a simple beam on target device (like a borehole tube) you might make some great gains with the usual lousy ion source (eg puts out lots of diatomic ions) by keeping them out of the main accelerator in the first place, thus not wasting the energy to accelerate them and the heat on target that implies. Some work needs lots of 9's accuracy, but this would greatly benefit on the very first one you got.

[johnf]

Doug
to get single mass unit selection it has to be done after the mass is accelerated f=mxa comes into this deal.
We have a table of mass vs acceleration to set the magnetic field for each magnet that gets you to around +- 2 mass units depending on the remnance of the magnet then you maximise beam on target making fine adjustments.

i'll post the super simple calc tomorrow
these magnets go to around 1.5 tesla linear with a 400mm bend radius and weigh 2 tons with a pole gap of 3". Power supplies are xantrex 6kW units.
magnets designed by us but made in Auckland NZ by Buckley engineering. They make 90% of all mass selection magnets for the semiconductor industry. I was in the factory a few years ago when they were flat out making all the magnets for the synchnotron in Australia.
The boss up there has a super yacht "maximus" and the day we were there was launching day kanting keel and all that.
Doug you would want to work for them, the factory knocks off around midday on the friday and all the mud track midget racers come out of the woodwork and are got ready for a weekends fun. Their big 5 axis mill did the keel bulb for maximus forget how many tons of lead it was but the keel bulb is accurite to a couple of thou.

Edit in the pic of the lab just right of centre is a box covered with a greenish tablecloth. This box contains the latest magnet for our high resolution RBS detector. Another orange beast
There were lots of implanter bits being built while we were there as well for AMAT if I remember right, huge blocks of tooling plate aluminium 3 feet cubed being turned into high vacuum chambers on a line of huge CNC mills.

[Doug Coulter]

I bet I would love working for them, if I could get into the habit of working again in a "real job". I've been freelance since about the middle '70's. Moved it from Washington DC to here when my jazz band toured here and we all said -- what in heck are we doing living in DC when there's Floyd? Both nicer people and well, you've seen what the area looks like.

I think I didn't phrase my question right, as I think this can work. Let's suppose I want to use a crummy ion source -- in this case because it's going to have to float off ground some +60kv or so -- don't want to mess with running my good mostly-monatomic microwave ECR source up there (which would require an isolation transformer of 100w+ good for 60kv isolation -- and it's too large anyway). So I would use a penning-type source, or even an E beam one, which makes mostly diatomic D ions. I don't need that much current in the resulting beam, but I do only want D+ out of this, not that and D2+ -- to a first approximation.

One has mass 2, the other 4 (a two to one ratio). To get them to a speed where you can separate them decently you first accelerate them to say, 1000v, then put them through a magnetic separator, then do the main acceleration from +60 or so kV to -60kv or so. This is a reasonable supposition because it is something I plan to do. I only need ~microamp currents of D+ for this, but don't want the have to put in the power to accelerate the 90%+ of D2+ ions, or take that heat on my target, that your usual crummy ion source produces. Would that not be a workable plan? It adds some mechanical complexity to be sure, but that seems manageable. I know the "filtering" wouldn't get to the "purity" many other applications need -- but I don't need it perfect here -- just decent.

So I would need to select a 2::1 mass difference -- seems pretty easy, with ions at 1kv energy so the energy spread isn't that huge (maybe ~100v or 10% of the beam energy). Since I'm just designing this, I can make that 1kv some other number to get reduced percentage beam energy spread if required. And if I lose some of the desired ones, again, that's not a big deal in the design.

What I'm planning is a sort of beam on target device, with split supplies for the main acceleration, and an ion/pre acceleration source running off small batteries floating up at +60kv or thereabouts. If a few of the wrong kind of ions leak through -- it's not a really big deal, that first "9" of "purity" I get makes it good enough for this and a second 9 fantastic -- just saves a ton of wasted energy in the main acceleration, and a lot of needless damage to the target, which in this case might be a D compound (see my smart-target theorizing elsewhere here) that won't take much heat without decomposing (which itself would be off ground at -60 kV so it's hard to keep cool easily with other than forced air or convection in oil).

I did see your formula, and thank you for one in human-usable units -- it implies I think that I could pull this off, and not even need a big magnet to do it -- a small permanent magnet would get me there. I have a decent hall-effect magnetometer to tweak designs with for that. I've found that necessary as many PM makers are a bit, errrrrm optimistic about their magnet specs and calculations of reluctance in the pole pieces can be a bit off depending on dumb things like exact shapes and the iron alloy. At any rate, I wouldn't need the big voltages to concept-test this in a prototype.

old borehole tube design from Phillips

In other words, this design more or less with a mass selection before the main acceleration gap. Most of the rest is already built.

[johnf]

Doug
if you want to mass select before acceleration ie the extraction potential say 2kV
and you are going to bend it 90 degrees with r=43cm as in our beasts then using the formula you would get 212 gauss needed in the magnet
begger all field. the next problem with this is that anything magnetic in your lab will affect the beam. Also sending the low energy beam around a radius like 43cm is not going to give you much at the output due to beam spreading. This is why we mass select after acceleration were the beam is going much faster having less time to spread. Now if you redo the calc for 30keV then you need 823 gauss to swing mass 2 around r=43.
Maybe a velocity filter would be better on the pre acceleration side ie a Wein filter

[Doug Coulter]

Right, but I want this small and 8k gauss is trivial to get with permanent magnets....so that's kind of what I'm thinking about.
Would something on that scale not make the bend radius suitably small so beam spread not so bad an issue? In this case I don't mind
wasting some ions at that point anyway -- I've not invested a lot of energy in them yet, which is the point of doing it pre acceleration for me.
Couldn't use a big electromagnet right at the HV potential anyway -- I'd then have to insulate from that, and power it besides.

But yes, another "filter" design is fine too -- I'm all ears. I've kind of been looking over the older mass spec literature for ideas, with the
thought that what used to take tons of copper and steel now fits in my pocket where magnets are concerned.

The target for this is something that will fit onto my smaller pfeiffer pump station, or even be sealed of, and sans power supplies be
pretty dinky -- smaller than a watermelon. With beam on target devices like what I have in mind, you don't even want good beam focus,
as that makes the energy density on the target too high -- so even a really crummy power supply (couple of auto spark coils) could work
in the end thing. Of course, I would start with the nice regulated bench supplies, but it's a thought.

I've already been working on making zirconium deuteride for a gas supply/regulator/resevoir as shown in the pic above.