Non mechanical technology path
In advanced nanofactories electric systems will probably be used. Electric systems though can't yet be integrated into plans for nanofactories because of a lack of a set of well understood near ideal components. (For more information see: Nanosystems Section 1.3.4.b No nanoelectronic devices.)
One reason why mechanical aspects are easier to predict and estimate than electrical aspects is that: Nanomechanics is barely mechanical quantummechanics.
Applications that strongly depend on non mechanical base technologies can be fond at at the "most speculative potential applications" page.
Avoiding overestimation of capabilities
By mostly (but not exclusively!) focusing on the easier to predict mechanical aspects and not relying on difficult to predict non mechanical things or even hoping for still fully unpredictable scientific breakthroughs in these areas one finds what is at least possible. If some results from the non mechanical technology path prove useful that just means more will be possible than the things currently reliably predictable.
Molecular electronics in technology level I behave rather non digital (neither diode nor resistor behavior)
In small (or big and very cold) AP repetitive structures electrons move as Bloch-waves without being scattered. One also speaks of ballistic electron movement because in the wave picture (sharp impulse, infinitely long wave in space) the electrons move like billiard balls. Electrons can encounter problems flowing around sharp conductor bends since they do not collide with each other but only with the conductors walls. This is essentially the same effect as the limited gas conduction due to the free molecular flow in vacuum systems. At higher temperatures the unavoidable electron phonon scattering becomes stronger [Todo: check how much electrons can easier flow around bends then] Very small conductors can constrain electrons so much that the electron wave function loose all their nodes in the directions normal to the conductor surfaces (lowest mode excitation - similar to the situation in an optical single mode wave guide) This situation is called low dimensional electron gas [Todo: check wether tighter bends can be made in this case?]
A severe limit of downscaling for nanoelectronics is the fact that electrons readily tunnel through isolating layers if they become too thin. Depending on the voltage 3 to 5 nm can be the limit. [todo: add more info here] Nanoelectronics can by far compensate their lack of compactness by their higher speed compared to nanomechanic logic.
to check: Are extremely fast signals on a nanoscale conductor distinguishable from light on an ultra-thin wave-guide or is it essentially the same?
Some kinds of electronics
Restricting oneself to pure hydrocarbons like the "direct approach" motivates one can use graphene ribbons, nanotubes or other graphitic/polyaromatic structures like graphene ribbons as conductors and semiconductors and vacuum (,air) or diamond as isolator.
Note that while pyrolythic graphite is a resistive material nanotubes can conduct current between one and two orders of magnitude better than copper. Electronic properties may be heavily influenced by:
- Statically included (or dynamically applicable) high mechanical strain
- the borders of the graphitic structure - closed, hydrogen terminated, chucked between two slabs of diamond
electric contacts between parts moving relative to one another can be either made flexible for reciprocative movement or via tunneling between two combs of graphitic sheets. For low resistance the contacts need to be way bigger than the conductors [Todo:quantify]
If one allows some nonmetals one can create diamond checkerboard doped with nitrogen very similar to today's nanoelectronics.
Magnetism plays little role in nanofactories. Scaling laws make electrostatic motors and generators preferable.
Magnetism could be used for levitation of macroscopic objects like in a thermal vacuum isolation vessel (dewar).
Carbon atoms have long been thought to be completely non-magnetic. It has been found that specific radical structures can exhibit strong magnetism [Todo: verify, check which structures, note which kind of magnetism and how strong]
Spin flips in tooltips can be influenced by nearby massive atoms with high spin orbit coupling (Nanosystems 8.4.3.b Radical coupling and inter system crossing) this has some relevance for mechanosynthesis.
In a recent (when?) discovery carbon was found capable of producing rather strong localized patches of ferromagnetism. This came very unexpected since it doesn't fit well in the current models of magnetism. It's not yet clear whether a really strong macroscopic magnet can be built.
Although magnetism is not of prime interest for APM this will certainly be interesting for some specialized applications. AP motors won't need magnetism at all so magnetic carbon is not needed to end the dependency on not too abundant rare earth metals.
Some external links: Ferrocarbon EU Project | magnetic carbon | tunable magnetic properties demonstrated in carbon
High temperature superconductivity is not yet clearly understood and subject of research. AP technology will probably make this research easier.
Superconductors made from not too scarce elements like YBCO superconductors might find some use in coils for tokamak fusion reactors.
Quantum computation is obviously not a necessity for APM systems. APM systems and quantum computers may mutually boost each other though.
- It's very questionable whether pure hydrocarbon quantum computers can be built.
Nitrogen vacancy centers are currently (2014) investigated and will with APM systems be distributable to exact atomic locations.
- Searching for an optimal configuration of system components in a nanofactory relative to some chosen metric (playing a puzzle game) is essentially a search problem on which Grovers algorithm could be applied always providing a quadratic speedup. Special cases may be exponentially accellerable. While probably looking like a great artwork the downside is that those found solutions are something like "convoluted projection from some high dimensional space, or the result of an execution of some dubious program" where one can not understand how the solution was found just by looking at it and even hardly by in depth analysis.
Like heat dissipation free computation quantum computation needs reversible data processing as prerequisite.