(Image by John Burch of Lizard Fire Studios.)

Some nanotechnologists, such as Eric Drexler, believe molecular nanotechnology (MNT) and nanofactories are physically feasible. Others, such as George Whitesides, are skeptical. The UK’s “nano champion”, Richard Jones, lists six challenges for molecular nanotechnology in a blog post from two years ago:

1. Stability of nanoclusters and surface reconstruction. Surfaces have a tendency to “reconstruct” - seek out stable equilibria in ways not necessarily predicted by molecular dynamics simulations.

2. Thermal noise, Brownian motion and tolerance. Atoms on the nanoscale may be too wobbly to build complex machines out of. Drexler addressed this, but not in thorough detail.

3. Friction and energy dissipation. Surface area becomes much larger as machinery scales down, and high functional densities will give rise to high power densities in molecular machine systems. The friction and heat may be so intense that molecular machine systems cannot be reliably constructed.

4. Design for a motor. Richard is skeptical that the electrostatic motor as described in Drexler’s Nanosystems would actually work. More detail needs to be fleshed out and supported by experimental testing.

5. The eutactic environment and the feed-through problem. For MNT systems to work, they would need to operate in ultra-high vacuum. But, interacting with the outside, they’d be exposed to a very atomically messy environment. Valves and pumps need to be around 100% efficient to exclude foreign molecules.

6. Implementation path. How do we get there from here? If “soft” nanotechnology is all that works, how do we transition from there to hard?

These are all valid arguments, but some are a bit more interesting than others. To estimate them roughly in order of declining importance based on my own opinion, I’d list them as 3, 1, 2, 4, 5, and 6.

For 3 I definitely recommend taking a look at the full text as written by Dr. Jones. He anticipates a major issue in MNT machines will be energy leakage from the driving modes of a smaller machine to the larger vibrational modes of the structure it is embedded within. Jones writes, “MNT systems will have very large internal areas, and as they are envisaged as operating at very high power densities; thus even rather low values of friction may in practise compromise the operations of the devices by generating high levels of local heating which in turn will make any chemical stability issues (see challenge 1) much more serious.” To address this, power densities can merely be kept lower than the theoretical maximum - scaling laws would still allow the construction of MNT systems with much higher throughput and product customization performance than conventional factories.

1 has to do with surface stability of nanostructures. Part of the argument is that more careful quantum chemistry techniques should be used where mere molecular dynamics simulations are being used today. I’d don’t know much about the details of this issue so I won’t comment. More research is definitely needed.

2 is the thermal noise, Brownian motion, etc. As Jones mentions in his blog post, Drexler laid out a framework in Nanosystems to calculate the impact of thermal noise, which was used to estimate positional uncertainty at the tip of a molecular positioner. The uncertainty was found to be less than an atomic diameter, which is promising, but Jones would like to see simulations with more complex structures where both the positioners and their foundations are subjected to thermal noise and Brownian motion. Making serious progress on this will likely require hundreds or thousands of molecular engineer hobbyists, using programs like Nanorex’s NanoEngineer-1 to try out a wide range of possible designs and see what works. For a look at the work of someone already playing with a beta release of NanoEngineer, see the Machine Phase blog. The projects on this blog help you get a visual idea of the challenges in designing molecular machine components.

4 is the design for a motor. Drexler’s electrostatic motor needs to be thoroughly simulated with quantum chemistry techniques, and eventually, of course, experimentally tested. However, even if it doesn’t work out, some other power source will surely be devised. Even MNT motors using ATP as an energy currency would probably be able to achieve power densities far superior to today’s best manufacturing machinery, so I don’t think that the availability of a nanoscale electrostatic motor is a showstopping issue.

5, the concern with maintaining the molecular integrity (ultra-high-vacuum) of the MNT workspace, seems like one of the weakest challenges to me. If filters only work at 99% efficiency, say, then they can merely be daisy chained until the desired purities are achieved. Because these are nanoscale filters, they’d barely take up much space in comparison to the functional machinery. Also, there is no need for MNT machinery to interact directly with the chaotic external environment. Nanobots would be pretty poor at locomotion anyway. What we want is molecular manufacturing systems to build microbots stable in a variety of external environments. Most commonly-discussed MNT applications: nanoscale implants, utility fog, diamondoid products, etc., do not depend on autonomous nanoscale assemblers operating in messy surroundings. The delicate molecular machinery can be kept safe in an ultra-high vacuum, shielded by multiple layers of containment and filtering systems. Airlock-type systems can be used to extrude the product without permitting dust inside the factory.

6 is the least bothersome of all. If the potentially huge payoff of developing molecular machine systems becomes obvious to more people, then we will be able to afford to try out a very large number of different implementation paths. Technologies for viewing and manipulating the nanoscale are growing ever more accurate and inexpensive, so the right tools will be there, we just have to embed ourselves in the engineering challenges and see what works. Granted, a major implementation route holdup could delay progress for a decade or two, but much longer than that seems implausible - the number of possible routes is so large that it seems very likely one of them will work.

A more recent list of challenges can be found on the Nanofactory Collaboration site.