Three month report

From Wise Nano

NanoManufacturing Aerospace Components (NanoMAC)

A requirement of the NIAC grant is to make a report halfway through the third month.

The goal of the project is to investigate the feasibility of designing and manufacturing a wide range of aerospace products, given a source of molecularly precise nanoblocks. These nanoblocks might either be manufactured by mechanosynthesis methods outlined by Eric Drexler and Ralph Merkle, or by self-assembly methods currently under investigation by Zhenli Zhang and Sharon Glotzer http://www.engin.umich.edu/dept/che/research/glotzer/index.html.

1) What nanoscale properties of nanoblock parts are available?

• physical
• electrochemical
• electrical (conductors, insulators, and semiconductors, but there's also the piezoelectric, IPMC, thermoelectric, thermoluminescence, electromagnetic, electrostatic, electrokinetic, Peltier effect, photoconductive, photoelectric, electrochemical, and capacitive properties)

What other properties are there? or: What other ways are there to look at nanoscale properties?

Ultimately, of course, it's all in the electron wave and electromagnetic wave equations, with a few tweaks for moving the nuclei. I expect that when we eventually get into nanoscale engineering, we'll be inventing new (and probably fewer) classifications for the nanoscale phenomena that underlie all these bulk and statistical phenomena. I hope so--because the above list is incomplete; it doesn't have thermionic (or is that the same as thermoelectric?) or magnetostrictive, and I'm sure there are quite a few more.

In addition, there are properties such as thermal conductivity and thermal capacitance, not mentioning triboluminescence, mass, radiation cross section, etc.

Radiation cross section probably should be treated separately from the others. At least for the high-energy stuff that interacts probabilistically and falls off exponentially over long distances. Alpha and low-energy beta can maybe be lumped with mechanics and electrics. Neutrons are a whole another issue--actually, I guess, two issues depending on whether they're fast or thermal. But what I'd like to do, for now, is just worry about cosmic rays and solar storms. And maybe not even them--there's research that shows low-level radiation (~rem per day, IIRC) is actually, literally, good for you. It activates DNA repair mechanisms, which then clean up *all* DNA damage--and other damage mechanisms cause many orders of magnitude more damage than the radiation, so it's a net win. http://ehp.niehs.nih.gov/members/1998/Suppl-1/363-368pollycove/full.html

On the other hand, cosmic rays appear to have different damage mechanisms, and may not be so good for you. So we may still need to worry about them. On Mars they're a lot less of a worry due to the atmosphere; Mars suits probably don't need rad shielding despite the signficant level of radiation since more of it is in the good-for-you type and intensity. Of course this argument falsl apart if that paper is wrong.

Action Items: (nanoblock specification and capabilities) T- Look at apply IMPCs to the robot arm; piezo electric; dna binding T – look at logic switches/actuators of all kinds in non-diamonoid arena Look at applying Dinos Mavroidis' work

2) What human-relevant macroscale functions can be implemented with nanoblocks which have the nanoscale properties?

• structure/strength
• kinematics
• digital computation
• sensing
• actuation
• temperature control
• optics
• radiation shielding
• chemical sorting and processing (including molecular membranes)
• fluid handling and barriers
• propulsion
• communications
• mass

3) How can the nanoscale properties be engineered and constructed so as to implement the macro-scale functionality?

• chemical and/or machine-molecule fabrication
• nanoblock fabrication -nanoblock joining
• nanoblock placement (looks easy, now)
• reliability

Are we assuming atomic perfection? That means that we should mention but say that we are not testing for the types of errors mentioned at http://www.semiconfareast.com/crystaldefects.htm

(plus a few more, I'm sure).

I think we're assuming that the molecules will have exactly the desired chemical formula, or it will count as an error. For carbon lattice, that's all that needs to be said. For polymers, there are other questions like isomers and folding, and self-assembly in slightly the wrong alignment (CL is stiffer so should be easy to align).

Will we test for errors? During fabrication, maybe--whatever it takes to get the reliability high enough for a nanofactory to build another nanofactory. Post-fabrication, probably not. Self-repair is a huge pain; I'd rather just build in redundancy and detect errors at a functional level (or design automatic cutouts so they don't even have to be explicitly detected).

BTW, in addition to carbon and biopolymer, we should look at crystalline silicon as a potential nanomachine material. They've been working on it in Japan for well over a decade. Back in '94 they could pick up atoms from a Si surface and place them elsewhere, with sensory feedback, automated. This was with an STM, so I guess they couldn't build 3D Si crystal structures (not conductive). But in '03 they picked up a Si atom and replaced it using an AFM...

John Michelsen just posted on CRN's blog, saying Si machines are too squishy. I'll try to learn what this means.

Larger bearings could use graphite pads in hollows between multiple nanoblocks. Nothing says nanoblocks have to be fully filled rectilinear solids.

<insert 2d and 3d pictures)

I don't think we need ball bearings. If we have nanoscale actuators, we can "walk" or "roll" many actuators along the shaft surface to support and power it. That way, we don't need space for the balls, we don't need to assemble the balls into the race, we don't have to worry about the balls wearing down and increasing the slop in the shaft. Just put a thin carpet of actuators between the shaft and the support. And if you want an unpowered bearing, simply leave the actuators unpowered and let them coast.

The problem is, can we do arbitrarily shaped perfect curves? With an assembler, we could build a 100-billion carbon diameter bearing, but not with 200nm blocks. Even if you parameterize them, because of surface plane boundaries.

For 90 degree bends, just use crossing rods.

Gears have losses. Why wouldn't anything else?

OK, gears have frictional losses, but not kT losses. Sliding rods have frictional losses. Interlocks will have frictional and (very minor) compressive losses. You can use interlocks and drive an orthogonal set of rods; this costs you switching time, but not much friction vs. bending the rod.

Of course, a kinematic linkage could also transfer power around a corner. <insert 2D picture, and maybe 3D movie>