Design of gem-gum on-chip factories
raw notes:
- level throughput balancing
- layer and stage ratios
- (folding)
- level0 splitup in hot and cold section
- power dissipation bottleneck - dynamic drag & breaking losses
- threading by
- 3D fractal speedup
- cooling & isolation
Although diamondoid mechanosynthesis works at room temperature cryogenic cooling will probably be employed just because it seems rather easy to do (see: "Diamondoid heat pump system") and error rates can be shrunken by many orders of magnitude.
Depending on whether general purpose mechanosynthetic fabricators or mill style fabricators (serial chain of tools with no spaces) are used predominantly more or less layers and channels for threading parts by are needed [?].
- slow speed of assembly level 0: The mechanisms to assemble parts normally can be potentially smaller than the parts being assembled. Since The mechsanisms to assemble minimal sized DMEs need themselves to be DMEs The density of actual building sites of atomic size is rather low consequently mechanisms at the bottommost layers are quite a bit slower than the ones above and need to be included in greater numbers.
Component router systems
For the transport of unfinished product parts of different sizes from lower to higher assembly levels nanofactories may use routing structures.
The routing structures can either have separate or merged multiplexing and de-multiplexing steps where the former provides redundancy of rails. (Nanosystems Fig 14.7.)
There are two in some respects similar yet in other respects very different steps where this can occur.
- when diamondoid molecular elements (DMEs) are transported from assembly level I to II
- when microcomponents are transported from assembly level II to III
For all the optional steps in convergent assembly (assembly level IV) the lower stages should be programmable/steerable enough that no further shuffling is required. (Depending on the programmability the lower stages may too be simplified.)
Since direct control of those systems would clog the IO bottleneck hirachical heterogenous nanomechanical computing system must be integrated in parallel (one layer might suffice). Temporary storage facilities for microcomponents are optional and may be more useful as seperate macroscopic entity.
[Todo: explain free space designs, analyze parallelism]
External links
Articles from E. Drexlers Blog:
- Complete molecular manufacturing systems will have many subsystems, designed to meet many constraints
- Physical scaling laws enable small machines to be highly productive
External references
- Nanosystems chapter 14
