I collect slide rules. You probably know a slide rule is a mechanical calculator of sorts. They usually look like a ruler (hence the name) and have a sliding part (hence the name) and by using logarithms you can multiply and divide easily by doing number line addition and subtraction (among other things).
It is easy to dismiss old technology like that out of hand as being antiquated, but mechanical computing may be making a comeback. It may seem ancient, but mechanical adding machines, cash registers, and even weapon control computers were all mechanical devices a few decades ago and there were some pretty sophisticated techniques developed to make them work. Perhaps the most sophisticated of all was Babbage’s difference engine, even though he didn’t have the technology to make one that actually functioned (the Computer History Museum did though; you should see it operating in person, but this is good too).
So why the resurgence? Nanotechnology. [Eric Drexler] (of Engines of Creation fame) proposed “rod logic” as a way to build logic circuits (and, logically then, computers) using microscopic mechanical components. You might wonder why go backwards and not try to miniaturize electronic logic elements? The problem is that when you scale down to structures only a few molecules big, semiconductors don’t work anymore. There’s been work on Josephson junctions and similar structures, but so far room temperature microscopic electronic gates remain elusive. (Also, just this month IBM announced they have had some breakthroughs with nanotube transistor contact resistance.)
However, mechanical structures scale down nicely. There are still problems, of course. To do it right, we need to be able to make perfectly meshed parts and mating surfaces. [Drexler] postulates that we’ll be able to lay down individual molecules to our whim (like 3D printing with molecules). In particular, he wanted to lay down carbon atoms in a diamond lattice. That technology isn’t exactly a practical reality yet. But we are closing in on something similar. Graphene.
Logic with Rods
There has been a lot of work about making tiny structures for rod logic. How does rod logic work? It uses very basic concepts that approach the molecular scale.
Consider the figure on the left. Let’s say the blue (long) rod is the input rod and the purple rod (it is the same size as the blue rod, but going into the page) is the output rod. If you push the purple rod into the page, it will go as far as you want because the black knobs won’t hit each other.
Now look at the figure to the right. The blue rod has been moved. Now pushing the purple rod will cause the knobs to bump together (I’m assuming the blue rod is “behind” the purple rod’s knob). You won’t be able to push the purple rod as far as before.
How does that make a logic gate? Consider an inverter. When the blue rod (the input) is pushed it, it’s knob blocks the knob on the purple rod — the purple rod will not be able to be pushed in. Blue-In, Purple-Out means the signal has been inverted. The same is true if the blue rod is out, the purple rod can the be pushed in; this inverts the signal as well. In order for this to work, you have to pull the output rod back when you are done and push it when you want to read the output–almost like a clocked logic design on an FPGA.
I won’t belabor it, but by arranging the knobs you can make all the basic logic gates. If you really want to understand the basic gates, try reading the [halfbakedmaker’s] page about rod logic or read [Drexler’s] thesis. You could–in theory–even make these diminutive logic gates in three dimensions, meaning you could pack a lot of computer into an extremely small space.
We think of mechanical devices as slow. Since these little gates are at the molecular level ([Drexler] talks about rods weighing 0.02 attograms–there are one million attograms in a picogram), they should be pretty fast compared to things in our real world experience. However, since the nucleus of an atom weighs more than an electron, moving a whole atom is a lot slower than moving an electron. So electronic molecular-scale gates are still a major research goal.
[Drexler] talks about laying down carbon atoms to make diamond. However, one of the most promising materials now is graphene which (like diamond) is carbon, but arranged in a hexagonal lattice. Consider diamonds versus graphite. Diamonds are made of carbon atoms arranged in a pyramid-like lattice. Graphite, on the other hand, is made from stacks of carbon sheets in a hexagonal lattice. Even though both are made of carbon, the arrangement of atoms changes the carbon’s properties. For example, diamonds are hard and graphite is soft. Diamonds are insulators but graphite is a fair conductor. Like graphite, graphene’s atoms are arranged in a hexagonal lattice. What distinguishes it is that rather than being made of stacked layers, graphene is a single layer just one atom thick.
Graphene, unfortunately, doesn’t have good electrical properties to act as a semiconductor (it has a zero band gap voltage). However, it has other interesting properties, including ballistic electron transport which is practically room temperature superconductivity. There is also silicene, which is a similar arrangement of silicon atoms, that may allow for nano scale electronics more easily, but that’s another post for another time.
For rod logic, though, graphene could be a great material. It is among the strongest materials known and is a good conductor of both heat and electricity. It is nearly transparent and is both stiff and elastic. The real issue is how to practically manufacture complex shapes with graphene.
Making graphene isn’t terribly hard. In fact, it was originally made using graphite and sticky tape (who knew you could win the Nobel prize using pencil lead, adhesive tape, and a microscope?). We’ve covered plenty of home lab methods to make graphene, one using a DVD burner. Researchers have even grown graphene on a semiconductor substrate, which could be a big step towards practical devices.
The biggest problem is that graphene’s magic properties come from the fact that it is essentially a two dimensional structure. Some research has worked with making transistor-like devices by doping graphene. Others use two graphene sheets stacked one on top the other. There’s also research involving alternating graphene with layers of other materials.
Making substantial mechanical structures out of graphene is a hot research topic. It is easy to make tubes (a carbon nanotube is basically graphene rolled into a tube) and a few other simple shapes. Research has included different methods of creating graphene structures, but as of today creating a practical nanoscale rod logic computer is elusive (although Lego-based gates are fairly widespread judging from YouTube).
Most products that use graphene today actually use graphene as a resin in a polymer and so the material is basically ordinary plastic imbued with a buzz word. However, there has been work at using graphene as a reinforcement in polymers that gives the polymer some of the properties of pure graphene. Oak Ridge labs has created sheets of graphene using chemical vapor deposition and made strong and conductive composite material with them.
So far, a lot of this is working in a lab and not on store shelves. Still, practical examples of graphene use include tennis rackets, LED light bulbs (apparently, the thermal properties of graphene will make the bubs last longer and cost less). and a home heating system.
The Road Ahead
It seems the big question isn’t if minuscule computers will exist, but rather will they be electrical or will they be millions of little quasi-slide rules functioning in a volume the size of a ball bearing? If the future is graphene, I am betting on mechanics. The almost superconducting nature of graphene will make it useful for applications like solar cells, but likely won’t be helpful for making electronic switches, even with doping.
Not that mechanical computing doesn’t have challenges too. At these scales, thermal noise effects are devastating and there are other barriers including just the difficulty of creating the tiny structures required. But you have to wonder: today it isn’t uncommon to see an old computers duplicated with more modern technology like an FPGA. Will we one day see Charles Babbage’s engine rebuilt at tiny scale using graphene? If it happens, I’m sure you’ll read about it on Hackaday.