Nanotechnology is great. New materials for advanced displays, molecular recognition for more selective catalysis, or new drug delivery pathways through dendrimers and nanoparticles. And there is the cool property of gold nanoparticles being red rather than yellow (with a size dependence on colour) due to the differences in light interactions at macro and microscopic scales.
Except few want to hear about that stuff. Make no mistake, nanotechnology is very real science – but it would be called “chemistry”, “biotechnology” or “materials science” if it wasn’t for the imagination-capturing concept of nanoscopic robots presented in decades of science fiction books and television.
The dream is simple: build a small enough intelligent robot and it will be able to manipulate not just pieces of plastic and metal as they do in a production line for cars, but atoms and molecules themselves. The possibilities of this are endless. Any molecule can be made with ease and without thought for a retrosynthesis process. Any part of the body can be repaired with just a computer program. Or they can just go wild and eat the entire Earth, converting its entire mass to a lifeless grey goo.
We’re all gonna die.
Trouble is, taken literally this pipe dream is far from realistic. Not far from realistic in the sense that such abilities are more 500 years away than 50 years away, but far from realistic in the sense of “chemistry doesn’t work that way” and “physics doesn’t work that way”. It’s therefore unsurprising that such robotic nanotech is popular with people like Eric Drexler and Ray Kurtzweil – both singularity-obsessed engineers with no formal experience or training in how molecules work in real life. It’s not that these people lack the intelligence to fathom this (far from it) it’s just that they seem to have little first-hand appreciation for the real difficulties their ideas have to overcome. In a way, they’re like theologians discussing, with great depth and clarity, how many angels can dance on the head of a pin without much thought to whether angels even dance in the first place.
Far from nanoscopic problems
Nanotechnology is said to be a blend between biochemistry and modern engineering. This is a fair description, but overly simple in many ways. Nanoscopic robots have to cope with conditions that have absolutely no similarity to the macroscopic world we experience around us because of the change in scale. A bacteria can propel itself through water with a measly flagellum, while a ocean liner cannot, simply because at the bacteria’s scale, moving through water is more akin to burrowing through wet sand than swimming through easily parted waves. In organometallic chemistry, parts of molecules fall on and off all the time, often permanently, in direct competition with the solvent they’re in. While our own arms don’t necessarily come on and off so rapidly because we have many more bonds holding them together. More often than not, this sort of thing leads to a massive solvent-dependence on the reaction – what works in dichloromethane doesn’t necessarily work in water. Solvent molecules need to move out of the way before any reactivity can happen, and this can hinder any chemical reaction far more than you might expect if the molecule was on its own. Macroscopic machines don’t experience this; air simply moves out of the way with ease. If you want to think of a nanobot as something like a car assembly plant, imagine filling the entire production line from floor to ceiling with sticky gravel and you have a better idea of the environment such bots would work in.
No, you don’t get it. You are still in a pretend world where atoms go where you want because your computer program directs them to go there.
Next is the physical work they have to do. Chemical bonds take a lot of energy to break. This is why you can happily have oxygen and hydrogen floating around with each other without a problem until you heat it up with a flame. It’s why we’re constantly developing new catalysts and new processes to boost efficiency. Without this, few chemical reactions are straightforward. A nanobot would have to be able to overcome this by being able to stabilise all the reactive intermediates that simply don’t want to be in that exposed, half-built, state and will tear at anything that comes close in order to become stable. If thermodynamics says that involves tearing the nanobot apart, so be it; giving it clever programming won’t stop nature doing that. The only thing stopping oxygen eating away the steel in a conventional machine is its relatively huge size – at the nanometre scale, reactivity is much faster, and a metal nanobot would be oxidised and denatured fairly quickly.
But let’s finally wind down with some of the practicalities of how a magic nanobot is even supposed to know what it’s doing. For a single cell, its operation is chemical. Certain stimuli happen here, the cell releases a certain chemical there. It all works. For proteins and enzymes, these work as very specific catalysts that facilitate chemical reactions. And these can be very specific. While many chemical or transition metal catalysts may react with, say, “any alcohol”, an enzyme can pin its reactivity down to a specific kind of alcohol molecule. This is one of the reasons that propanol and methanol are more poisonous to us humans than ethanol (which we can drink in much larger quantities before killing ourselves). Our bodies can tell the difference even though, chemically speaking, they’re remarkably similar chemical compounds. Loading a program onto something so small is also an issue. What can store the information and how does it process it? A chemical will just move according to the chemical environment it’s in, getting some bonds to rearrange at that scale is infinitely more complicated than just sending a pulse of electricity down a macroscopic copper wire.
A nanobot would have to play at being a combination between Maxwell’s Demon (to be able to identify chemical compounds by some magic process) and the Incredible Hulk (to have the energy to pull molecules apart without being destroyed itself) in order to perform its supposed function of universal molecule builder. This is a shame, because real nanotechnology can work some wonders, but not miracles.
