Some moments you just never forget. A few weeks later I was sitting at my computer when my business phone rang. I answered eagerly (that thing didn't ring much), and the voice on the other end called himself R.W. Bussard, and wondered if Tom Ligon was available. “I certainly am,” I said, “but I just have to know if this is the Robert W. Bussard, as in interstellar ramjets?”
“I guess I'll never live that down,” he admitted, and then he invited me in for an interview.
We hit it off smashingly, and I instantly recognized that his scheme was a superior version of something I had dreamed up while studying health physics, but never pursued. I went to work for him as soon as he got a little dribble of funding to start back up. I felt like I'd just gotten the opportunity of a thousand lifetimes. If you can only afford one lab rat for your project, I'm the jack-of-all-technical-trades you want, and the company could only afford one lab rat at that time. I worked on the project for about five and a half years. By that time, the project had gotten more funding, and moved out to San Diego. I had agreed to move out there for a year or two to help him get the bigger lab going.
So, what was it like to work at EMC2? The small, hand-selected technical staff was composed of amazingly talented people, highly motivated by a desire to make fusion, a true dream team. The lab ran on physics and passion. Dr. Bussard is awesome. His career stems from a very early and intense desire to make spaceflight practical, and, with R. D. DeLauer, he literally “wrote the book” on nuclear rocket propulsion. His first degree was as an engineer, and only after developing the first working fission rocket engine did he head off to Princeton to earn his Ph.D. in physics. And, while he is a first-rate physicist, he still has the heart of an engineer and inventor. It was always great fun to watch him at a blackboard, NRL Plasma Formulary in one hand, chalk in the other, working problems faster than I could follow them with a calculator.
I should mention a few of the others in this effort. Dr. Nicholas A. Krall is one of the best theoretical plasma physicists ever, and has collaborated on this project from the start. Lorin Jameson is a computer whiz and physicist that put the math of Bussard and Krall into functioning computer programs like EIXL, to analyze the data from the experimental runs and predict performance of larger systems. Later on, Mike Wray, Mike Skillicorn, Ray Hulsman, and Noli Casama were the ones who made the machine that finally worked. And none of this would have happened without EMC2 president, Dolly Gray, the only person on Earth who can make R. W. Bussard do what he needs to do when he doesn't want to.
Two and a half years after the move to San Diego, he had hired enough talent that I started to feel like excess baggage, so we parted company and I came back to Virginia and found another amazing job. But we stayed in touch, I remain a true believer in the project, and wonder if I'll ever again have a chance to literally save the world. Maybe this article will help.
* * * *
Basic Principles
Let's review the idea of Inertial Electrostatic Fusion. A little earlier I described a device called a Hirsch-Farnsworth fusor. This remarkably simple device consists of an outer vacuum container (typically spherical, but any shape will work) and a spherical inner grid (Figure 1). This device is the descendent of a spherical vacuum tube designed by Langmuir and Blodgette in 1924[3]. If you pump the chamber down to a vacuum, backfill with a trace of deuterium gas, and apply high negative voltage to the inner grid, at the right combination of pressure and voltage, a glow discharge will light off. A glow discharge, also called a Paschen discharge or Paschen arc, is what occurs in a neon sign. Near the low-pressure end of this glow discharge region, with voltages of ten kilovolts or higher, a distinct bright spot can be seen in the center of the fusor's inner grid, and deuterium-deuterium fusion starts to occur. Deuterium ions (deuterons, the + marks) formed near the outer grid or chamber walls are attracted to the inner grid by the high negative voltage. They accelerate toward the grid, which is very open, and most pass on to the center, where density rises rapidly and the chance for fusion goes up with it. That's just how simple fusion is. The fusion reaction is driven by particle velocity, not heat. You don't need to “heat” the fuel by applying a hundred million degree Kelvin torch the way tokamaks or laser fusion approaches run. Plain old high voltage acceleration works fine. In fact, because it is not random, but instead both directed and monoenergetic, it works a lot better than Maxwellian heat.
[Footnote 3: Irving Langmuir and Katharine B. Blodgett, “Currents Limited by Space Charge Between Concentric Spheres,” Phys. Rev., 23, pps 49-59, 1924.]
* * * *
Figure 1. A Hirsch-Farnsworth “fusor” ion accelerator.
* * * *
Alas, the simple machine shown in Figure 1 is flawed. Not every deuteron that shoots into the center of the device manages to collide sufficiently head-on with another deuteron to produce fusion. Some hit and bounce off, and most just miss. In fact, only a few ions produce fusion on any pass. But, if you could build the machine so that the ions conserve energy and can make many passes thru the machine, that wouldn't matter. Eventually they would fuse. But those darned grids are the problem. It just isn't practical to make them more than about 98% transparent, and the usual figures are more like 90-95%. On every pass, 2% or more of the deuterons will hit a grid wire and be lost, and that's too many to make a breakeven reactor, by a very wide margin.
So that brings us to Figure 2. The Elmore-Tuck-Watson machine[4] is the reverse of a Hirsch-Farnsworth machine. The inner grid is positively charged instead of negatively charged, so it attracts electrons instead of ions. Electrons pass thru the inner grid and converge on the center, pass out the other side, then come back for another pass. The result, at sufficiently high current and voltage, is a very dense region of negative charge in the center of the machine. This is really what you want instead of the negatively charged grid in the Hirsch-Farnsworth machine. If you generate deuterons just inside the inner grid of an Elmore-Tuck-Watson machine, they'll oscillate happily thru that cloud of electrons for a very long time. The electrons and ions are at such high energy that they essentially can't recombine to any significant degree, so, in principle, the ions might make enough passes thru that central region to produce meaningful fusion.
[Footnote 4: “On the Inertial-Electrostatic Confinement of a Plasma,” William C. Elmore, James L. Tuck, Kenneth M. Watson, The Physics of Fluids, v. 2, no. 3, May-June 1959.]
* * * *
Figure 2. An Elmore-Tuck-Watson electron accelerator with ion accelerating center potential well.
* * * *
Ah, but what about the electrons? Of course, the machine still has grids, and they still will have about the same limits of transparency, and about 2% or more of the electrons will be lost on every pass. That loss kills this machine as a power reactor.
* * * *
Dr. Bussard once designed a small tokamak called the “Riggatron” (described as “World's Simplest"). Although he maintains to this day that it would have worked, he was unable to secure sufficient funds to actually build it. Perhaps that was just as well, for it got Dr. Bussard thinking about how to get around the problems plaguing tokamaks.
Tokamaks work by employing intense magnetic fields around a toroidal vacuum vessel. The idea is that ions will spiral around the resulting “lines” of magnetic force that the magnets produce that parallel the inside surface of the torus. Deuterium and tritium ions are thousands of times as massive as electrons, and it takes a really intense magnetic field to make them stay on a line. If you run their density and energy up enough to make fusion, they tend to hop from line to line with each collision until they hit the wall and are lost. The bottom line is that that's why we've been messing with tokamaks for all these decades and we're still not using them to light our homes. The most optimistic estimates say they may be working by 2040, but a more realistic estimate might be post-2100. The Electric Power Research Institute fears they'll never make power economically due to high capital costs and short life.
* *
* *
The New and Improved Elmore-Tuck-Watson Machine
As Dr. Bussard thought about this, he had the thought that it was a shame ions are so much more massive than electrons, because a tokamak would be able to confine electrons at high density far more easily than it would ions of fusion fuel. And then he thought about Hirsch and Farnsworth and the idea that Elmore, Tuck, and Watson had, and a little light went on in his head. He began to wonder if an Elmore-Tuck-Watson machine might actually work if the accelerating grid could be magnetically insulated. And so was born the notion of building a “quasi-spherical” device into which one could inject high-energy electrons. He realized if he took certain geometries, including an equilateral pyramid, a cube, or a dodecahedron, and placed a circular solenoid electromagnet on each face, each pointing with the same pole inward, a suitable electron containment might be achieved. All of these devices are described as PolywellTM designs.
Please understand that this is not a thermonuclear approach to fusion. It does not confine a Maxwellianized plasma in order to produce fusion. The “confinement” principle for the fuel ions here is purely due to their attraction to the electric field produced by the electrons. What that gives us is an ideal form of the Farnsworth fusor, with no grids, within the confines of the magnetic grid. The fusion is more straightforward because the fuel ions converge to a region of high density with the same kinetic energy, sufficient to trigger fusion, rather than depending on some tiny tail of a Boltzmann distribution of energies. Ions not producing fusion have an excellent chance of circulating in and out of this potential well for many passes until they do fuse. The magnets are there strictly for the electrons.
The underlying principle driving the concept is the electron potential well. In order to achieve the desired well depths for D-D fusion, excess electron densities on the order of a million electrons per cubic centimeter are required. It is very important to realize that this is not a multiplier. You don't need a million times more electrons than ions in the center of the machine, you only need a million more electrons per cc than ions. If there are 1.000000 x 1012 ions per cubic centimeter, then 1.000001 x 1012 electrons are sufficient to maintain the well. That suggested a certain robustness of the machine to allow manipulation of ion densities to achieve fusion conditions. The plasma does not need to be overwhelmingly negative; it can, in fact, be almost neutral, and the higher the ion density, the closer the plasma is to neutral.
Another important principle is that the potential well is not some static thing. The electrons forming it are in constant, and very vigorous, motion. They pass in and out of the well continuously, as do the ions. While the inner grid of a fusor might classify its driving force as “electrostatic,” Dr. Bussard's concept (and for that matter the Elmore-Tuck-Watson machine) is more properly considered “electrodynamic.”
* * * *
Figure 3. Magrid operation and formation of the diamagnetic trap “ball."
* * * *
Several devices have been built according to one version of this scheme, in which the magnets are mounted on the outside surfaces of a vacuum chamber of the shape described. The first was HEPS, a very large pulsed machine built in the 1980s. I personally assembled and ran PXL1, a miniature of HEPS that could be run for many seconds at a time. Finally WB-5, a scaled-up and improved version of PXL-1, was built and run. None of these were successful fusion machines, but all were quite capable of trapping a lot of electrons, and provided important insights into the trapping mechanism. I won't go into them here because they were not found to be the right approach for making fusion, but the interested reader will find references at the end of the article if they wish to explore these further. Their fundamental flaw is that they cannot recirculate electrons around the magnets.
The machines that can produce useful fusion use magnets that operate in the geometries described, located inside the vacuum chamber, and covered with metal shells that tightly conform to the magnets. The faces and corners are open. These are magnetic grids, or magrids, and they are charged to a high positive voltage. They serve the exact same function as the inner electron-accelerating grid of the Elmore-Tuck-Watson machine, except that magnetic fields are used to keep the electrons from being able to actually hit the grid. But the magnetic field also turned out to have an additional effect.
When I arrived at EMC2, WB-1 and WB-2 had already been built and tested. The designation WB describes the shape of the magnetic field inside a Polywell as it “inflates” with large populations of trapped electrons. The resulting field looks just like the plastic toy Wiffle BallTM on the computer models, the holes corresponding to cusps going thru the magnets’ central holes and corners. The toy-ball phenomenon tends to make the population of electrons, and consequently ions, much higher inside the magrid than outside. A high trapping factor is very helpful in achieving fusion conditions inside the magrid without having excessive densities of charged species outside the magrid. Ball formation can be visualized by imagining that the magnetic fields’ graceful convex hyperbolic arches penetrating into the volume within the magrid are made of foam rubber. High electron populations act more or less like a balloon, and push back the fields, producing a nearly spherical volume, and squeeze the cusp holes to a very small effective diameter. Figure 3 illustrates the nominal magrid field condition and what happens when it operates with a large population of energetic electrons.
* * * *
WB-1 was made with ceramic donut magnets, the kind used to make audio speakers. These were crudely encased in stainless steel shells, assembled into a cube, and used as a magrid. The best thing about this device was that it was really cheap and simple. The machine did show some electron trapping, and made some pretty glows, but suffered high electron losses due to the fact that magnets of this type have lines of flux going into their faces. This property will affect all permanent magnets and all iron core magnets. The first thing that pops into everybody's mind when they start trying to understand a magrid is that permanent or iron-core magnets might work, but all such devices have cusp-lines going into faces, and are unusable. But this machine would be almost as easy to build as a fusor, and I'll be disappointed if I don't see them start to show up at science fairs.
WB-2, on the other hand, was built of six copper-wire electromagnets, on square cross-section spools welded together at the corners to make a square box. Although small, WB-2 was the right general idea, and it turned out to be a vigorous electron trap. I did a large number of test runs with it, as we tried to increase both drive voltage and magnetic field strength. We even tried it with deuterium gas to see if it could be coaxed to make a little fusion. Per G. Harry Stine's axiom that the tests are not over until the prototype is destroyed, WB-2 blew a coil when we pushed it to about 4.5 kilovolts and a couple of kilogauss. It was a brave little machine, but it was too small for fusion (proving that at least that much of the mathematical models were correct).
Next we built WB-3, which was simply WB-2 scaled up to double the magnet diameters. While it did show signs of trapping electrons aggressively, and lit off all the spectacular effects seen in WB-2, it just never seemed to want to “clean up.” Every time we cranked up the voltage past about half the potential needed to make deuterium fusion, it seemed to generate huge quantities of hydrogen gas, swamped out the potential well, and lit off the whole interior of the machine with a bright glow. WB-3 did serve as a good platform for testing several new instruments and ionization methods. And, had we run it with deuterium-tritium, it very likely would have made measurable fusion. WB-3 was capable of running at about half the voltage needed for D-D fusion for a fairly long time before the big glow set in, and is one of the reasons I'm sure these machines are capable of running steady-state.
I left the company just as the finishing touches were being put on WB-4, a very nicely put-together machine about 50% larger than WB-3, made with water-cooled coils, and with elegantly fabricated magnet shells that were sealed up so that outgassing from the insulated copper coils was no
longer a problem. They had high hopes for WB-4, and they were, in fact, able to coax some fusion from it, although at far lower levels than hoped for. It shared some of the problems with WB-2 and WB-3. It lost more electrons than it was supposed to, and tended to generate excessive gas when run hard. That excessive gas, in turn, often triggered a Paschen discharge between the magrid and the outer Faraday cage and chamber walls, that same bright discharge the earlier machines tended to generate. A Paschen discharge is what makes most amateur-built fusors run, but in a magrid machine, it shorts out the high voltage supply driving the electrons, so it must be avoided.
* * * *
Figure 4. WB-4 in operation.
* * * *
At several points in the development of these machines, Dr. Bussard described to me a problem he suspected might be plaguing them. WB-2, -3, and -4 were all assembled from six magnets whose cases were welded at four spots on each case to form a rigid cube. Each point where they touched was something he called a “funny cusp.” In principle, we knew that magnetic field lines penetrated these points, much like the lines that entered the faces on the solid-state magnets of WB-1. However, the hope was that, since this phenomenon really was just a short line, and since lines have no area, it really was not important. Also, with two magnets touching, those points had the highest magnetic field strengths in the machine, which we hoped would make them more prone to act as magnetic mirrors, which also insulate. But what these three machines had been whispering in our ears was that the funny cusp might not be so trivial after all. And the other thing these machines had in common was that their magnets were made on square cross-section spools. The magnetic fields produced by the magnets were not square, though, so that meant lines of magnetic force tended to cut across the corners, making another loss path.
Analog SFF, January-February 2008 Page 11
