> I have trouble getting a clear understanding of what the "parts" are
> in nanotech. Most workers in the field will explain that they aren't
> literally planning on putting things together atom by atom. Rather,
> there will be molecular building blocks and complex work sequences
> which can assemble a wide variety of useful structures. So I think
> it is misleading to think of nanotech as having only a few "parts".
Eric is pretty clear in Nanosystems, that the feedstock for diamondoid materials would be small molecule hydrocarbons, such as acetone or alcohols. You could probably use methane/ethane/propane, as well. Ralph Merkle has done the most work trying to define assembler chemistries. See: "A proposed "metabolism" for a hydrocarbon assembler": http://nano.xerox.com/nanotech/hydroCarbonMetabolism.html
Ultimately, nanotechnology & biotechnology are reduced to a very small number of "parts". A few dozen elements make up almost all organisms and 5 elements form the bulk of the material (H, O, C, N, S). Even if you go up to the next level: proteins, carbohydrates, lipids; most of the activity revolves around 30-40 major molecules and their assembly precursors or breakdown products. Where biochemistry gets really complicated is when you start talking about enzymes that manage the reactions. Still, it looks like you can produce a minimal self-replicating bacteria "nanomachine" with something in the neighborhood of ~300 genes. These bacteria have to live in a "rich" environment (inside other cells) where many of the building blocks can be sucked up from the surroundings.
The earliest namomachines will probably be built out of diamondoid materials so only a couple of types feedstock molecules will probably be required. Eric's designs for pumps & gears only have 8 different elements in them, so I would expect that on the order of a dozen feedstock "molecules" could build most nanomachines. It isn't clear at this point whether different source molecules would be required for "displacing" an atom from the source onto the target (depending on the adjoining atom types, valence states, bonding strengths, etc.). It is distinctly possible that this might be required.
The general approaches I've seen for bottom-up assembly *do* envision assembling structures atom-by-atom, this is the only way to get the atomic structure exact. This *is* the way it is done in biology with the possible exception that enzyme systems don't break things down any more than is necessary. I suspect that "mature" nanoassembly, would take approaches that are similar (e.g. force the addition of a 10 carbon atoms from decane onto a diamond crystal in one shot, rather than adding them one by one from methane). This just takes a more complex jig to hold the molecule and apply any energy needed to push through the transition state.
> I gather that there will be a large number of building block molecules
> and modification enzymes (which act as tools) which will have to be
> brought into play in the proper sequence in order to construct materials.
> Where these building blocks and tools come from is not clear, but I
> gather that conventional, bulk chemistry would be involved.
The feedstock molecules would generally be assembled by bulk chemistry, Enzymes might play a role in feedstock assembly because they could be more energy efficient. I doubt you will see "enzymes" (i.e. amino-acid based proteins) used in the actual assembly of hard-nanotech (diamond/sapphire). Enzymes are designed to function in water while hard-nanotech assembly is typically described as being done in a vacuum or inert atmosphere.
> A nanotech
> assembler will have to be part of a relatively large and complicated
> system which produces the wide variety of feed stocks that it needs to
> perform its tasks.
I would agree that the assembly of macro-sized objects from nanoassembly will be a complex systems engineering task. However, the assemblers for nanostructures themselves should not be very large (I think Eric's nanoassembler design is a few million atoms. Small compared to some of the designs for nano-cells that Robert Freitas has done. Nano-cytes run 1-10 billion atoms). The feed stocks, as mentioned above, should be fairly limited and they can probably be generated "in-line" from simple sources (e.g. from Air (C/N), water (H) and dirt (Al/Si/S) if enough energy is available.
You might want to keep in mind that we do primitive nanoassembly today in DNA and Protein synthesizers. A DNA synthesizer requires 6-8 bottles of reagents and a Protein synthesizer requires 25-30 bottles. Since we don't have the tools to mainipulate individual molecules in biotechnology this assembly typically occurs at the micro-mole scale (~10^17 molecules). Interestingly enough one of the major barriers to working at smaller scales is keeping the water containing the sample from evaporating.
Robert