“WHY cannot we write the entire 24 volumes of the Encyclopaedia Britannica on the head of a pin?” When the fabled physicist Richard Feynman posed the question during a famous speech in December 1959, he was not looking for an easy-to-carry version of the illustrated reference guide. He was drawing attention to the problem of manipulating things on a vanishingly small scale.
As he warmed to his theme, Feynman dared to imagine that we might one day build with the building blocks from which all the known matter in the universe is made. “I am not afraid to consider the final question as to whether, ultimately – in the great future – we can arrange the atoms the way we want; the very atoms, all the way down!”
Advertisement
In this vision of atomic Legoland, we could build all manner of wonder stuff. We could make silicon’s successor, a material that would allow us to keep stuffing ever more computing power into tiny devices. We could come up with a substance that would beef up our puny solar cells or supercharge the ultimate battery, so we could store all that clean energy. We might even trigger chemical reactions that are impossible today (see “Elusive chemistry”).
The trouble with that vision is that atoms are ridiculously minuscule, so much so that more than 10 trillion iron atoms would fit on to the head of a steel pin. And yet, in the bowels of giant brushed-steel contraptions that bring to mind steampunk machines, we have begun to nudge individual atoms around by the thousand, and with astonishing precision. Now we just have to figure out where to put them.
For most of human history, we made do with the stuff nature gives us. Then we found ways to enhance it, taking metals and adding a dash of other elements to create alloys like steel – materials that have given us everything from cutlery and the kitchen sink to the jet engine. We have even managed to engineer materials that can control the flow of electrons, creating the microchips that power your smartphone and laptop.
But for the most part, we are still limited by what we can dig out of the ground – and that’s holding us back. No matter how craftily we combine the available ingredients, we can’t seem to crack the recipe for inexpensive thermoelectric materials to scavenge waste heat, for example. Commercial solar panels still max out at 20 per cent efficiency. Magnets for electric car motors rely on elements whose supply is anything but reliable. And batteries, as anyone who has watched their phone die at a crucial moment will know, have plenty of room for improvement.
To create stuff with whatever properties we happen to desire, we need to build novel materials from scratch – and that means building with atoms, as Feynman dreamed of almost 60 years ago.
Grasping the atom
He was still around to see the beginnings of the “great future” when, in the early 1980s, Heinrich Rohrer and Gerd Binnig at the IBM Zurich Research Laboratory in Switzerland invented a powerful new kind of microscope. It took advantage of a bizarre phenomenon of quantum mechanics called tunnelling, in which particles do things they shouldn’t be able to, according to classical physics. Rohrer and Binning spotted that when you place a metal stylus an atom’s width away from a sample and apply a voltage, electrons tunnel across the gap. This creates a tunnelling current that, crucially, varies exponentially with the size of the gap. By reading the current while slowly scanning the stylus tip across the sample, you can map its surface, atom by atom.
Scanning tunnelling microscopy meant we could see atoms for the first time, opening our eyes to the world of the very small. But seeing was just the start.
It soon became clear that the microscopes could pick up atoms and move them around. Lower the tip close enough and short-range electrostatic forces called van der Waals forces grab the atom beneath, which means you can drag it along by moving the tip across the surface. Lift the tip away and the atom stops in its new position.
By 1989, a team at IBM Research Almaden in California were nudging 35 xenon atoms into place on a nickel surface to create the world’s smallest logo. It was an impressive demonstration, but still a long way from manipulating many thousands of atoms, which is what we need to do to conjure up new materials.
That remained something of a pipe dream until last year, when at the Technical University of Delft in the Netherlands and his colleagues went one better – or 59,965 better, to be precise. Having noticed that chlorine ions were easy to push around on a copper surface, they wrote an algorithm to automate the scanning and shuffling of 60,000 of them. The result was that stores data by positioning individual chlorine atoms into patterns to encode each bit. If scaled up to 1 square centimetre, their 1 kilobyte rewritable chip would hold about 10 terabytes of data, orders of magnitude better than the best comparable chips around today.
More importantly, Otte’s achievement is a proof of principle, one that suggests we are on the verge of being able to manipulate atoms according to our hearts’ desires. “The notion we can move atoms around now means we can basically start to design materials the way we want them, rather than being limited to the materials nature gave us,” says Otte.
“We are on the verge of manipulating atoms according to our hearts’ desires”
He is aware of the size of the task. One of the biggest problems is that you’d have to assemble about 1023 atoms to make a single gram of material. Using a microscope to painstakingly drag each one into place, layer by layer, would take forever. So while we figure out how to move atoms more efficiently, we should be concentrating our efforts on applications where moving a few atoms has a big impact, says Otte.
Take the chips inside your phone. Each has billions of transistors that function like a tap, turning the flow of electricity on and off. Transistors are now so small they are prone to leak even when off, which wastes power, generates heat, and in turn stop the chips running at maximum speed. If you could only modify the handful of atoms that make each transistor function, stopping the leaking, you would boost the performance of the whole device. Indeed, doing so will probably be crucial if we want to maintain Moore’s law of shrinking transistor size.
For this particular application, scanning tunnelling microscopy may not be our best bet, because it can only manipulate surface atoms. That’s not a problem when building a material from scratch, but you can’t reach inside an existing slab of semiconductor to shuffle its atoms about. A new technique, scanning transmission electron microscopy, might do the trick. This fires an electron beam as fine as a single atom through a material to image its internal crystal structure – but it so happens that the beam sometimes nudges single atoms about. Several teams are now looking to exploit this effect to assemble materials with atomic precision.
Even if they pull it off, there is a more daunting problem: we don’t have a clue where we should put the atoms we move. If you’re aiming to create materials with particular properties, you can’t just randomly arrange atoms and hope for the best. Nor is there time for trial and error. Instead, you need a way to simulate new materials, which is exactly what is striving to do at the Duke University Center for Materials Genomics in Durham, North Carolina.
Curtarolo has developed a rapid-fire method for testing the chemical stability and physical properties of predicted atomic arrangements, quickly assessing combination after combination without stepping into a chemistry lab. The approach has already met with some success. This year, his collaborators made two of the materials that Curtarolo’s computers had flagged as potential magnets – and showed . It was the first time computer modelling had predicted magnetism in a new material. Other researchers are using the approach to hunt down better battery materials and home in on the best recipe for solar cells.
But even Curtarolo is not yet able to run simulations at the scale of individual atoms. His new magnets were mixtures of three chemical elements that arrange their atoms in a regular, repeating pattern – a much more manageable problem for a computer to tackle. Predicting the properties of a material atom by atom requires ridiculous amounts of computing power, and right now we simply don’t have it.
So is Feynman’s dream is destined to remain just that? Not quite. Because it turns out that the very atom-nudging contraptions we might use to make our miracle materials could come to the rescue, helping us build a device that can do the intensive simulations required to find the right atomic arrangements.
“Predicting the properties of a material atom by atom requires ridiculous amounts of computing power, and right now we simply don’t have it”
What we need is a quantum computer – a device that exploits the weird properties of quantum mechanics to achieve the sort of processing power classical computers can only dream of. The concept is simple enough. In a regular computer, a transistor is in one of two binary positions, either on or off. But a quantum property like the spin of a quantum bit, or qubit, can be up, down, or in a superposition of both. Harness that behaviour to perform calculations, and you have a device that can consider many possible solutions to a problem at once. Link two qubits together, and the system can be in four states simultaneously. Link three, and you get eight states. The processing power grows exponentially.
Link 300, say, and you’d have a system more powerful than all the world’s computers combined. For certain problems, not least simulating how large numbers of atoms interact to generate a material with particular properties, that power would be a game changer. “A quantum computer would be a quantum leap of the kind of simulations we could do,” says Curtarolo.
The problem is that we don’t yet have a practical quantum computer, although it’s not for want of trying. Google is developing a device based on aluminium circuits cooled until they become superconducting, while Microsoft wants to use “topological qubits” – elusive quasiparticles conjured on 2D surfaces that would maintain the required quantum state without too much fuss.
and her team at the University of New South Wales in Australia, on the other hand, are banking on the power of atom-nudging. They are using scanning electron microscopes to carefully place single phosphorus atoms on to silicon, with each phosphorus atom forming one qubit.
The reason they prefer this approach is that the computer-chip industry is already familiar with silicon. “We believe it will be much easier to manufacture a full-scale processor chip than for any of the leading designs, which rely on more exotic technologies,” says team member . They might just be right, but first they must prove they can do it.
“When we proposed it back in 2000, a lot of people thought it wasn’t possible to control the world at that level – the technology just didn’t exist,” says Simmons. So they had to invent it. The first problem was a sticky one. Phosphorus and silicon form such a strong bond that you can’t use a scanning tunnelling microscope to nudge phosphorus atoms across a silicon surface. Instead, the team found a workaround: they coated the silicon with a non-stick hydrogen surface, then used the microscope to pluck off just the hydrogen atoms covering the sites where they wanted the phosphorus atom to go. It worked perfectly.
Having overcome several more engineering problems, in 2012 Simmons and her colleagues demonstrated a functioning transistor made from a single phosphorus atom in a silicon circuit. They followed it up three years later by demonstrating a two-qubit system, consisting of two phosphorus atoms connected to form a logic gate. Now they have funds to build a 10-qubit device within five years.
“The solution is a device capable of churning out recipes for miracle materials”
Ultimately, they want to produce a quantum computer with a whopping 1024 qubits. As Otte’s nanoscale memory chip has shown, with a fair wind it’s possible to shuffle tens of thousands of individual atoms, so Simmonds should be OK. Having mastered manipulating individual atoms, her challenge now is to design the surrounding circuitry to form a fully functioning quantum device.
But even once we have quantum computers churning out recipes for miracle materials, for Otte, Curtarolo and anyone who wants to build from the atom up, one last challenge will remain: scaling up. A decent-sounding strategy would be to use scanning microscopes with multiple tips working in parallel. But in practice, the very sensitivity that allows the tips to manipulate single atoms makes them too vulnerable to vibrations. “There are four-probe machines, but that’s just four times as hard to use as a one-probe,” says Otte. “Scaling that up to 1000 times is just incredibly difficult.”
Then again, scanning microscopes were never developed with industrial processing in mind. “I think it just requires other people with a different background and philosophy to pick this up and to start to think about it,” Otte says. “I think it should be possible to come up with new ideas.”
Elusive chemistry
A drop of this, 3 grams of that; give it a stir and watch it bubble. Until now, chemistry has been akin to cookery, a matter of seeking the right recipe for the substance we were after.
and his colleagues at IBM Research Zurich are helping to pioneer a more surgical approach. Using a microscope that can nudge atoms around, Gross is triggering reactions by removing single atoms from molecules.
Scanning tunnelling microscopes work at very low temperature, which means we can use them to create and probe highly reactive molecules that would otherwise remain mysterious, says Gross. That could help us master the catalytic production of polymers, drugs and other molecules we rely on every day. More importantly, perhaps, “we can start to build elusive molecules that cannot be made by any other means,” says Gross.
He is already doing it. For 60-odd years, chemists had tried and failed to make an exotic molecule called triangulene, a flat cluster of carbon atoms resembling a tiny triangular flake of graphene. This year, Gross and his team finally (see picture, above), using their scanning tunnelling microscope to selectively rip a couple of hydrogen atoms from a precursor molecule.
Triangulene has no immediate use, even though it boasts some interesting electronic properties. But the feat of its creation could have big implications for chemistry.
“After so many decades of rigorous research, the chance of finding new important reactions are quite limited,” says , an organic chemist at the University of Santiago de Compostela in Spain, who has been collaborating with the IBM team. With this atomic surgery, we appear to have started discovering them again.
This article appeared in print under the headline “The anything factory”
Article amended on 23 August 2017
We corrected the number of iron atoms on the head of a pin
