Alas, it was finally concluded that the maximum field strength WB-4 could produce running continuously and cooled, about 3 kilogauss, was not good enough, and it was not going to be possible to push any harder with cooled magnets of the sizes possible at that facility. The scaling law said the effectiveness of the machine should scale as B4R3 (magnetic field strength to the 4th power, times radius cubed), and at the higher field that was needed, even pumping ice-water under pressure into the coil was going to produce steam in short order. That would stop the water flow. At that point, it became obvious that making serious fusion with the little machines required very short-pulsed operation of the magnets. There was no question that cooled magnets would work with larger machines, with radii on the order of 1.5 meters, but those were not going to fit in the vacuum chambers available, much less operate with the limited power available.
They even contacted the best superconducting magnet maker in the world, to see what it would take to build a superconducting magrid. It turned out, for the little machines with a diameter of under half a meter, it just was not practical. The structure of the magnet required would have an inner superconducting core soaked in liquid helium, a vacuum jacket around that, a jacket of liquid nitrogen encasing that, and another vacuum jacket around that, all with a well-thought-out structure to minimize thermal leaks while maintaining strength against the enormous mutual repulsive forces of the magnets. At a larger scale, he saw no problems, but he could not do it as small as was needed.
Simultaneously with the tests of WB-3 and WB-4, two other machines were run with a configuration we called MPG. The simplest to describe is MPG-1. This machine was formed from a length of copper tubing bent so that it formed a single-turn magnetic structure approximating the WB-3 size and form, a truncated cube (a cube with the corners cut off). MPG-1 was limited to fairly low magnetic field strength, but that was partly offset by the fact that the magrid it formed didn't have much area to start with. Furthermore, the conductor was round and so the magnetic field it produced circled it cleanly. And finally, the conductor was spaced so that it never touched itself, and the result was that it had no funny cusps.
And darned if that simple piece of hardware store tubing didn't manage to make a little fusion!
It turned out that the electron losses and the mysterious generation of hydrogen gas were of the same cause. In an ultra-high-vacuum chamber, it barely takes a trace of gas to raise the pressure by a factor of a thousand or more. Electrons bombarding the magrid case corners and funny cusps were not only being lost, they were digging out hydrogen buried in the metal. That hydrogen diluted the fusion fuel, sometimes so greatly that they couldn't produce fusion even when they had a good potential well depth. And the abrupt increase in neutral gas flooded the area between the magrid and the chamber walls, and produced Paschen discharge, effectively shorting out the power supply driving the electron acceleration to the magrid.
Dr. Bussard had resisted building pulsed machines, knowing that what we really wanted were machines that could run continuously, or at least for seconds at a time, but finally gave in when the limits of magnet cooling at the available size was apparent. As a result, WB-4 finally produced neutrons at a rate of about a million per second when run at higher fields for very short times, in a pulsed mode. But it was still plagued by excessive electron losses and all of the problems that caused.
* * * *
And Finally, The Solution!
At last, he realized that they had to build a machine that had the right structure, even if the size and budget requirements meant it must be uncooled and intended only for pulsed operation right from the start. Time was running out. The stubborn loss mechanisms had dragged the program out longer than intended, and the source of funding was about to be shut off. They finally realized the right way to build a magrid, but had to do it in haste. The result was a design called WB-6, and it was one gorgeous magrid. One day that thing should be set up with spotlights on it at the entrance to a fusion museum. It is that pretty, and it is that important. The magnet cases are circular-cross-section toroids. Instead of touching, they are spaced apart by a few electron gyroradii. Electrons don't actually follow magnetic field lines, they spiral around them, at a radius determined by the field strength and their kinetic energy. By spacing the coils apart to clear this spiral, the funny cusp is eliminated, and the electrons slip past the grid instead of impacting it. This was, at last, a magrid of the proper form. Figure 5 is a photograph of the finished magrid before installation, and figure 6 illustrates the critical difference between this machine and the earlier magrids.
* * * *
Figure 5. WB-6, the gorgeous piece of hardware that finally did the trick.
* * * *
Pretty though it was, it had limitations. The coils were wound from plain varnish-insulated magnet wire, with no cooling mechanism. Like WB-2 and WB-3, the wire was going to get very hot, very fast. The tests would have to be quite short. Also, realize that the configuration of coils on any Polywell produces mutual repulsion. The coils experience high forces as they press against their containers attempting to get away from each other, and the individual windings also tend to mutually repel. The WB-6 magrid, built from wire meant for more ordinary applications, was destined for abuse. Both of the previous magnet-wire machines had reached end of life due to coil blowouts, and this one was going to be hammered even harder.
Another limitation was apparent. Scaling information generated by the other machines revealed that the drive power requirements for this device exceeded the power available to the little light industrial bay that housed the lab. The previous experiments had been bumping up against this limit all along. Typically, we had to use huge battery banks to operate the magnets, as we didn't have the power to run the magnets and high voltage supplies simultaneously. But WB-6 was going to require more electron beam power than the building could provide. It would have to be driven from a large capacitor bank. The capacitor bank could only deliver current for a very short time, with no current regulation in case of Paschen breakdown.
* * * *
Figure 6. The flawed design of WB2, 3, and 4 was replaced by the good design of WB-6, shown on the right. This greatly reduced electron losses, and allowed the machine to validate the concept.
* * * *
One final problem limited this machine. With so much effort needing to go to perfecting the basic magrid electron trapping performance, carefully metered ion production was beyond the scope of the program. A practical fusion machine is going to need something akin to a carburetor to produce ions, with no neutrals, in the right quantity and location. Fuel to WB-6 was metered by what amounts to an eye-dropper, a “puff-gas” system, which delivered a pulse of neutral deuterium to the inside of the magrid. Ion formation in this device was not produced in as well-controlled a manner as might be desired, although it did work well enough for the purposes at hand. A byproduct of this method of introducing fuel was that Paschen discharge tended to occur between the magrid and the outside walls about 0.5 to 2 milliseconds after the pulse, the time it took any un-ionized gas to migrate to the area outside the magrid. Had this particular puff-gas system been capable of turning off after the initial pulse, this could have been avoided, but the machine had to be built with what was already on-hand. Once the puff was started, gas would continue to flow into the machine until the small reservoir was exhausted.
Following a number of preliminary non-fusion runs at reduced power to characterize the machine's electron trapping properties, four tests of this machine were run in early November 2005 in an attempt to produce fusion. The drive voltage was approximately 12.5 kV in most of the runs, with a resulting expected potential well depth of about 10 kV. The tests were actually run several days after the lab had officially closed, and produced a few neutron counts each. One additional test threw caution to the wind, hammering the device harder than ever before, and that ended the predictably short career of WB-6. This magrid, too, satisfied G. Harry Stine's axiom
. The last test pushed it too far, and blew a coil. And so they sadly closed down the lab, before even having a chance to analyze the data.
A month later, the data were finally analyzed, and they discovered that the relatively few neutron counts produced, corrected for the counting efficiency and the geometry of the test (13,000 neutrons per count), and the fact that there are two fusions for every neutron in the D-D reaction, showed that nearly a hundred thousand fusions were produced in about a quarter of a millisecond! That's a rate of roughly half a billion fusions per second. The low count meant that the number could not be stated with precision, but it was certainly statistically significant enough to establish the order of magnitude. And the burst of neutrons repeated on each of the four tests, right where the potential well was right for fusion.
One of the things that amazed me when I learned it was that WB-6 actually ran at a considerably lower magnetic field strength than WB-4, and yet still outperformed the more robust machine by orders of magnitude. The improvement was not due to boosting the magnetic field strength. The improvement was due to the subtly improved geometry. The other thing that amazed me was the low drive voltage.
By comparison, a Hirsch-Farnsworth fusor running straight deuterium at ten kilovolts produces neutrons, but at a level so low as to be barely above background. My own fusor (presented at the PhilCon and LepreCon science fiction conventions circa 1998-1999), which I ran with EMC2's neutron counters, produced around 3000 fusions per second at 18 kV, and I was hard pressed to count anything above background at 10-13 kV.
Robert Hirsch reportedly managed about a billion fusions per second once[5], but that required pushing his fusor to 150 kV. And for a rate that high, he used deuterium and tritium, a much easier fuel mix for producing fusion. In spite of the relatively few counts, crude deuterium metering, unregulated pulsed power, and small size, WB-6 was making fusions like crazy.
[Footnote 5: “Inertial-Electrostatic Confinement of Ionized Fusion Gases,” Robert L. Hirsch, Journal of Applied Physics, v. 38, no. 11, October 1967.]
So what does that prove? Did WB-6 reach breakeven? No. But it demonstrated that, once the configuration of the machine was correct, it does, indeed, produce fusion at a rate in line with the models. The data said the machine was, at last, working properly, around three orders of magnitude better than WB-4. The fundamental problem was fixed.
And the models say that the reactor output will scale as B4R3. Power gain scales as B4R. If the magnetic field can be made stronger in proportion to machine radius, that would mean output increases as the 7th power of radius, and power output with the 5th power of radius. If this is even vaguely close to being correct, then at some modestly larger scale, this type of reactor is virtually assured to produce net power. And Dr. Bussard says the results of the tests of WB-6 put that point at a radius of about 1.5 meters for a deuterium machine, and 2 meters for p-B11.
At a radius of 1.5 meters, cooled copper magnets of the required field strength and capable of continuous operation are practical. It is also at about that size that superconducting magnets become practical for building magrids, and superconducting magnets make possible much stronger magnetic fields, which the model says tremendously improve performance. While one could imagine spending some years sneaking up on a net power version of this technology, it is apparent that very little is to be gained by doing so. Machines much smaller than 1.5 meters will still have to be run for very short durations, not the continuous operation needed for a working technology. And the size described for net power is not an extraordinary effort. Dr. Bussard estimates about $150 million for a D-D machine, and $200 million for a p-B11 machine. Now, we all know that estimates are estimates, and going over-budget is a long-standing tradition in technical projects, so let's just suppose for a minute that Dr. Bussard is wrong by a factor of five. That would put a p-B11 demo reactor at a billion dollars, to demonstrate a technology to save the world. And let's say he is low by a whopping and exceptionally unlikely factor of 50. That would put the program at $10 billion. That would still be a fraction of what has been spent over the last few decades on mainstream fusion research.
The problem is not that too much money has been spent on fusion efforts to date. The world spends something like five trillion dollars a year on energy, and an R&D effort of a couple of billion dollars a year on a better alternative, over a decade or so, is hardly an unreasonable expenditure if the probability of success is high. If anything, fusion is under-funded. So under-funded, in fact, that existing programs jealously guard their budgets, and the result is some very ugly politics that make it difficult for competing ideas to get a fair chance.
I admit that I think the present tokamak program is a dead end, and Dr. Bussard does not make a secret of the fact he also believes it. The majority of the public probably thinks the same thing, after decades of promises that always seem to be 30 years or more away from bearing fruit. But, in fact, tokamak research has developed or proven most of the technologies needed to build electrodynamic fusion machines, and has even explored tapping some fusion power using direct conversion from charged particles. Remember, Dr. Bussard's concept springs from the realization that a tokamak would easily confine electrons. The proposed reactors would be a small fraction of the physical size of the ITER tokamak, and the technology to build them should be far less challenging. Cost of machines like this tends to scale with size, and comparing the size of Dr. Bussard's proposed machines to the cost and size of ITER suggests that his cost estimates should be pretty good. Nothing entirely new and mysterious needs to be done, we just need to decide to put the pieces together and do it. At this point, the challenge is engineering.
And this little fusion program knows where to go. The intent is clearly not to sneak up on fusion a little at a time over many decades, the intent is to target a breakeven demonstrator on the first shot, which could be built in a decade, or even far less if it had the right commitment. When that reactor is built, we'll quickly know if it is a success, a very near miss that needs one more attempt, or hopeless.
* * * *
What Is This Pie-In-The-Sky p-Whatever?
Let's examine p-B11 fusion fuel. What is it, and how realistic is it?
Most fusion efforts expect to burn a mix of deuterium and tritium. The reaction produces most of its energy from neutron emissions, which tend to limit life of the reactor and render the whole structure radioactive. This might be worthwhile if nothing better can be found, compared to the adverse environmental problems of burning fossil fuels, but it is hardly ideal. It is the fusion fuel you usually hear about because it is, by far, the easiest reaction to produce, and it is the only fuel system with any chance at all of making net power from tokamaks, laser fusion, and the like. One look at the reaction cross-section versus initiating energy graphs will show you why they don't consider p-B11 for tokamaks. The “temperature” it would take to trigger the p-B11 reaction in a system operating on Maxwellian heat is vastly higher than for D-T. Expressed in Kelvins, the “temperature” required would be nearly 6 billion degrees!
Nevertheless, the p-B11 reaction is, and has been for decades, on the “short list” of fusion fuels. Why, if it is so much harder to burn? For one thing, it is remarkably clean. The reaction results in three alpha particles, which, recombined with electrons, are plain old helium. Breathe the waste product of this reactor, and the worst that happens is you talk like a duck for a few seconds. The reaction produces almost no neutrons. Natural boron is 80% B11, it is abundant, and is somewhat toxic. This reaction turns a toxin into an inert gas.
The way the energy comes off has always been attractive. Alpha particles have a charge of +2. The first particle carries 43% of the reaction energy, and comes off at 3.76 million electron volts. The other two alphas come off at around 2.46 million electron volts each (skewed somewhat by the velocity of the intermediate particle). If you wanted to make an alpha with 3.76 MeV of energy, you would knock both electrons off a helium atom, and accelerate it with an electr
ic field of 3.76/2 = 1.88 million volts. To get that energy back, simply decelerate that alpha against a 1.88 million volt field, let it kiss gently into a metal plate as it comes to a stop, and it will produce two electrons of current at that voltage. This has been done on a small scale using radioisotopes, and it is very simple to do. Since virtually all of the energy from this reaction comes off as alphas, and since their energies are relatively close together, it should be possible to devise a method of doing the same thing with the products of the p-B11 reaction. The principles are straightforward, although one can bet the engineering will not be trivial. But the benefits of doing it this way are enormous. Even the klutziest approach, setting the decelerating potential at 2.46/2 = 1.23 million volts, would presumably recover something like 85% of the energy. Considering that any nuclear reactor that generates its power as heat will wind up running steam turbines that waste 2/3 of the energy, this is a stunning technology. The environmental benefit of avoiding all that waste heat, the economic benefits of avoiding large cooling towers, and implications for lightweight space propulsion systems all make this efficiency highly desirable.
If this works, it has got to be the greenest technology to come along since photosynthesis.
But can an electrodynamic fusion reactor burn boron? If the models are right, yes, and surprisingly easily. Boron has five electrons. Knock them all off, and the nucleus has a charge of +5. That means an electrostatic or electrodynamic acceleration system will work five times as hard on that nucleus as it would on a proton of charge +1. The net result is that one only needs a potential well depth of something like 100-150 kilovolts. So build the machine a little larger and run it at higher voltage, and it should be straightforward.
Analog SFF, January-February 2008 Page 12