There is a type of controlled-nuclear-fusion reactor that already exists. It was invented by Philo T. Farnsworth, who also invented the basic electronics behind television. It is known as the Farnsworth Fusor, and it is simple enough to build (and small enough! -- fits on a tabletop!) that at least
two have been entered into high school science fair competitions.
However, like the massive machines that have been in research-and-development by various governments for fifty-odd years, this fusion reactor produces substantially less energy than it takes to run it (but it can take as little as a hundred watts to run it...). There are some plans afoot to see if it can be scaled up to the "break-even" point, but I don't know anything else about the status of those plans. What I shall do here is start by assuming that the Farnsworth Fusor CAN be scaled up to become a useful power plant. (According to the first link, Farnsworth claimed to have achieved self-sustaining reactions, which is the next step beyond break-even.) And there is an aspect to its method of operation that lets me bring up a notion that I've not seen elsewhere, regarding the device. That's why I'm writing this, of course!
The Farnsworth Fusor uses "electrostatic confinement" of the electrically charged particles in its reaction chamber. Other fusion reactors are designed around "magnetic confinement" or "inertial confinement", while stars, of course, use "gravitational confinement". The original Fusor was fairly complex, but a modified version also works and is simpler to describe. It consists of a spherical chamber, evacuated, which contains a wire-form spherical grid, perhaps a third the diameter of the overall outer sphere. See the second link. Both the interior grid and the outer shell are given electric charges (opposite charges, and a few tens of thousands of volts -- heh, CRT/television voltages!). Control circuitry will modify those voltages as necessary.
Fusable atoms are ionized and injected into the sphere. The voltage gradient between the outer shell and inner grid cause the ions to accelerate -- but after they pass through the center of the sphere, the ions decelerate. Those electric fields make the ions come to a full stop before they hit the outer wall of the sphere, and then start accelerating again, back toward the center of the sphere. So, with a significant number of ions present, all moving at speed in the central-most volume of the sphere, collisions and even fusions can and do occur. Those that miss are turned around to try again.
Scaling this up merely has the goal of increasing the rate at which fusions occur. To some extent, the overall average density of ions within a sphere can be increased, but mostly the sphere itself will have to be enlarged. Possibly an alternate design, a long cylinder with hemispherical endcaps, would assist in the goal, too.
The main difference in approach, between this fusion reactor and others, is the fact that "temperature" is not the measurement that matters. Here all we want is enough ion velocity to overcome mutual repulsion -- and this is easily achieved via acceleration within the electrostatic field-gradient inside the sphere, for practically all the ions. Over in a magnetic-confinement reactor, they want to sustain a hundred million degrees in a power-production reactor, so that a sufficient FRACTION of the plasma will be moving fast enough.
It is the fact that ion velocity can be rather precisely controlled in the Farnsworth Fusor that leads me to this Idea. Consider two deuterium nuclei: Each has a single electric charge that they use to repel each other, and a certain minimal velocity X is needed to overcome it. Two helium nuclei each have two electric charges, so they repel each other with four times the force as two deuterium nuclei -- and consequently higher ion velocities are needed to overcome that repulsion. Or, consider one deuterium encountering one helium-3 (a VERY desirable reaction in fusion research, since the result produces no neutrons that can make stuff radioactive): Fusing these will take higher ion speeds than fusing two deuteriums, and lower ion speeds that fusing two heliums. (I should mention that fusing two heliums doesn't work very well; helium-4 is such a stable nucleus that beryllium-8, the fusion result, is unstable and breaks back down to two heliums in a hundred-millionth of a second.)
OK, what I am going to suggest is that we set the voltages in our large-scale Farnsworth Fusors to fuse hydrogens such as deuterium only -- and then we feed ORDINARY hydrogen, not deuterium, into the reactors! Here is what will then happen: Ordinary hydrogen (it's name is "protium") has a nucleus that consists of a single proton. If two of them are fused together, one of the protons will turn into a neutron, and the result will be a deuterium nucleus. Since the Idea here is that we feed this reactor with only protium, those produced deuteriums will be far outnumbered, and will very rarely fuse with each other. IF they do, then about 50% of the time, the result will be radioactive tritium-hydrogen and a loose protium-hydrogen -- and about 50% of the time the result will be a helium-3 nucleus and a loose neutron. Remember, loose neutrons cause other things to become radioactive. However, since the fusion of two deuteriums in this reactor will be comparitively rare, the rate at which the reactor becomes radioactive will be very very low. MORE, part of this Idea is to actively attempt to sort out the deuteriums, and keep their numbers low! (Such sorting, by mass of nucleus, will also catch the tritium and helium-3 nuclei. Note that if radioactive tritium is confined in storage, it eventually -- a 12-year half-life -- becomes non-radioactive by turning into helium-3.)
Meanwhile, because the deuteriums ARE hugely outnumbered by protiums, a more likely scenario is that deuteriums and protiums will fuse. This is a reaction that is seldom talked about, so it probably happens somewhat rarely, but it can indeed happen. When it does, the result will be helium-3. Note that since I specified that we set the voltages in this reactor to prevent hydrogen from being able to fuse with helium, we should always be able to extract the helium during the sorting described above.
So, what do we accomplish, by implementing this Idea?
1. We gain a nuclear fusion powerplant reactor that is fueled with ordinary hydrogen, the commonest stuff in the Universe and 600 times as common as deuterium (--on Earth! Protium is even more common than deuterium in outer space). YES, I know that this reactor will have to be several times as large as originally hinted in terms of scaling up the Farnsworth Fusor, because the energy produced by protium reactions is rather less than the energy produced by other reactions. I think the availability of fuel will make it worthwhile.
2. We gain a factory for making deuterium and helium-3, which are the most-desired fuels for OTHER types of nuclear fusion reactors! (When fused perfectly, no neutrons are produced!) I suspect that such a factory -- which is also a power plant, remember -- will yield those two items rather more cheaply than trying to sort one atom out of 600 to obtain deuterium, or mine the Moon to obtain helium-3. They are the waste products of this power plant!
In other words, implementing this Idea means HELPING the goal of making those other fusion reactors work, by providing them with reliable fuel sources. Also, those other reactors may be physically smaller than our scaled-up protium-burning Farnsworth Fusor, and so can be built in a wider variety of locations. Some might even be "portable".