The basic problem with fusion energy generation on a useful commercial scale has been publically well-known for 40 years: the fusing particles have to be confined together closely enough, and long enough to fuse. This requires a combination of high temperature and pressure. This can be either achieved by a very, very high temperature, requiring an fission bomb trigger, when it is called thermonuclear fusion; or by a very high pressure, when it is called inertial confinement. Stars use a slightly different technique: gravitational confinement.
However, all the well-known techniques make one assumption: that it is fundamentally a three-dimensional problem and the particles (Hydrogen, Helium and Lithium nuclei in engineering practice) have three spatial directions of freedom. If the nuclei could be confined to a two-dimensional plane, or, better still, a one-dimensional channel, then the effective confinement pressure and temperature can be very, very much higher. This effective temperature would be much higher than the "normal" (i.e. three-dimensional) temperature of the medium in which the events occur. The effective temperature has so little to do with this real "normal" temperature, that the fusion could even occur at room temperature.
So how might such effective temperatures be achieved ? The alignment required is ridiculous: if one fires nuclei at each other head-on, then a tiny fraction hit each other. This is done every day in nuclear accelerators such as CERN, but fusion is rarely the aim. Instead, the researchers are usually more interested in pumping in large amounts of energy to produce exotic particles, and the tiny fraction which fuse produce only a tiny fraction of the energy to keep the accelerator going. A different kind of device entirely is required to produce aligned-linear nuclear fusion.
We already know that crystal structures can have useful interactions with nuclear reaction. This is despite the fact that nuclei are 100,000 times smaller than the atoms forming the crystals. One is the Mossbauer effect effect - known for most of a century - and the other is a recent technique for triggering some forms of nuclear energy level change using X-rays of precise wavelength. Would long, perfect crystals channel nuclei accurately enough to enhance fusion ? And even if they did, how would the energy be usefully removed without destroying the crystal ?
If the crystal had to be of the order of 100,000-cubed times as long as the atomic-size channel (10 nanometres, or 10nm) to get even approximately the right level of alignment, then how long is this ? It would be approximately 1016 nanometres, which is 107 metres, or 100 kilometres. However, the likely size of a "atomic" channel for, say, a Hydrogen nucleus (a proton) could be far smaller than 10 nm. The sort of device we are talking about looks more like a long transistor than anything else; and field effects (as in field-effect transistors) could conceivably knock a few orders of magnitude off the length required. Silicon fabrication plants already make single crystals several metres long - though not without defects.
Another alternative would be to live with the length issue, but use some kind of optical fibre with atomic-sized holes. Such fibres have already been developed (though with rather larger holes) because of their benficial effects on optical communciations. This may be much closer to an engineering reality: wave packets of light (photons) could have some useful shepherding effect on the nuclei to be fused.
So on first sight, the engineering difficulties seem almost incredible, but the rewards are commeasurately vast, so the issue would seem interesting enough to pursue a little further.
The first thing to realise is that just because current fission and fusion plants are so big, does not mean that they always have to be. Most electric energy used by private individuals is actually used in tiny amounts - look at the rechargeable battery industry. So a generator that just got warm and only produced a few Watts, but did so essentially forever, would have a large number of applications. So the average heat output need not be enough to destroy the device, but what about the radiation damage ?
If two nuclei beams are aligned towards each other so that most of them hit each other, then the resulting debris comes out sideways. Typically, two nuclei merge to form one nucleus and a couple of other light particles. Conservation of momentum means that the nucleus must come out at almost exactly 90 degrees. Nuclei that miss each other presumably would have the same effect as nuclei that are ineffectively channelled: they would be absorbed all along the duct (fibre or crystal) and cause displacement damage but no fusion and no radiation. At the "hot zone", any crystal structure would be destroyed fairly quickly. If it operated at a high enough real temperature, this might anneal out, but there would probably be lots of defects: bubbles of Helium at least. Thus at first sight, an interaction in free space, perhaps just a fraction of a micron, would be appropriate. The emitted particles would then miss the channel guides and be absorbed in some blanket which would heat up - and also in all probability become unfortunately radioactive.
The precise alignment of the fusion product, a flat, circular, razor-thin fan, could mean that the geometrical design could ensure that the nuclei produced are absorbed inventively in an element which only produces very short-lived isotopes. Thus radiation may only be produced during operation and die down entirely when it is turned off. This is part of the aim of existing Tokomak fusion devices, but in those the nuclei go everywhere and hit everything, including all the supporting steelwork. It is the radioactive steel which produces much of the practical operational problems.
The ideas included here are not patented and are published on the Internet as an open disclosure to prevent such a fundamental idea being patented in future. However, to make it really work would, I am sure, require tens of thousands of additional patentable inventions.