Materials that Learn

Suppose that we have a suspension of long electrically-conductive particles (such as metal filings or buckytubes) suspended in an insulating liquid resin. If we then try to force the liquid to conduct electricity in a particular direction, the particles will tend to self-organise to make that outcome achievable more efficiently.

A physicist will say that what actually happens is that when we apply a high voltage across the material in an attempt to force it to conduct, the particles become charge-polarised, and line up "lengthwise" in the electric field ... then the oppositely-charged "heads" and "tails" of adjacent particles tend to link up, and pretty soon you have lines of conducting threads running through the material linking the two electrical contact points. If your insulating resin's electrical resistance breaks down over small distances (above a given threshold voltage between a pair of particles), and if the sides of your particles can be persuaded to repel each other, to prevent the formation of additional conductive paths at right angles to your applied voltage, then, if you allow the resin to set, you should have a new type of material whose electrical conductivity depends on direction.

In itself, this doesn't sound particularly interesting: after all, we can already produce a solid directionally-conducting block by mechanically glueing or fusing a stack of insulated wires together and then machining the block to the desired shape. The advantage of using self-organising materials is that we can use them to build conduction patterns into films or coatings, or to build more exotic structures into solid blocks. You might want to tailor the electrical response of the paint on an aircraft or satellite to produce certain effects when it's hit by an incoming EM wave (say, to deflect radar or focus an incoming signal), or you might want to produce solid waveguides or field guides for electrical engineering, without laboriously building them from layers of laminated conductors, or winding them as coils.

The idea of self-organising materials isn't new. We use the idea dynamically with liquid crystal displays, and we've recently spent a lot of R&D money coming up with a "freezable" counterpart to LCDs, "electronic paper" (as used in the Amazon Kindle). But the idea of being able to "print" field structures into or onto materials, in a way that automatically self-corrects for any structural defects or variations in the material, is rather interesting. You could use superconducting grains to build exotic superconducting structures in two or three dimensions, or you could use a resin that's conductive when liquid, and freeze it from one end while the applied field is varied, to grow field structures that would be impossible to achieve by other means. You could even try coupling the process with 3D printer technology to produce independent conduction-alignment of each point within a structure to produce extremely ornate conductor structures. Then we have the interesting idea of field holography: if we create a complex external field around a device, and "freeze" the critical regions into superconducting blocks, then when those blocks are milled and reassembled, will Nature tend to recreate the original field by "joining the dots" between the separated blocks? What if we have a containment field with complex external field junctions that tend to destabilise under load – if we could freeze that field junction topology into a set of surrounding superconducting blocks, would they tend to stabilise the field?

We might be able to use expensive hardware to set up, say, a toroidal containment field, place a container of "smart resin" in the field, and "freeze" the external EM image of the device into the external block. If this was a useful component to have, we'd have a method of mass-producing them for use as as field guides or field stabilising devices.
With a number of interconnected connected and energised surrounding blocks, and the original device removed and replaced with a container of "smart resin", you might also be able to use the process in reverse, to recreate a rough electromagnetic approximation of the internal structure of the original device (crudely analogous to the old stereotype process originally used by printers to preserve and recreate the shape of blocks of moveable type).

Admittedly most of the potential applications for this sort of process don't exist yet. We're not mass-producing cage-confinement fusion reactors, and the LHC's magnets don't need miniaturisation. Fusion-powered vehicles are still some way off. But it's nice to know that there are still some fabrication tricks that we might be able to use that don't require laborious hand-tooling and impossible levels of molecular-level precision.