MURRAY HOLLAND dreams of the day when we can switch on a very different kind of computer. Forget pressing a button and waiting for your hard drive to whirr into action. If Holland’s idea works out, your computer’s circuits will materialise in front of you, made from little more than beams of light and a puff of gas. Your new machine would be more versatile, too. Flick a switch and the shimmering electronics will reconfigure a component from, say, data storage into an extra processor. In fact, you would be able to transform its parts into any electronic circuitry you like.
Welcome to the strange, shape-shifting world of atomtronics, where light beams are used to generate and control a current that is not a flow of electrons but a flow of atoms. Holland and his team at the Joint Institute for Laboratory Astrophysics (JILA) in Boulder, Colorado, reckon they can use a current of atoms to build anything from batteries to amplifiers and even transistors, which could eventually become the building blocks of “atomtronic” computers. Admittedly, you might have a long wait. “Atomtronics won’t happen tomorrow,” says Holland. “As an idea though, it is really taking off.”
How so? For more than a decade, researchers have been using light to trap atoms, creating artificial crystals that are much like the real thing but on a larger scale. The atoms in these crystals mirror the behaviour of electrons in solid matter, and the prospect of being able to model real materials this way is proving irresistible. Not only are these collections of atoms large enough to see, physicists can control the interactions between atoms, switching a material from insulator to metal, say. At last they can explore phenomena that are difficult or impossible to study in real materials.
Advertisement
The idea of using light to trap atoms was suggested more than 30 years ago by Soviet physicist Vladilen Letokhov, and was developed in the US by Arthur Ashkin at Bell Telephone Labs in New Jersey. They proposed creating an interference pattern in a light beam – a pattern of bright and dark stripes – and using it to ensnare atoms. Place an atom in a laser beam, and the beam’s intense electric field induces a charge imbalance on the atom which, in turn, interacts with the light’s electric field. Their calculations showed that the effect would be to move the atom into the dark region in the interference pattern and hold it there.
Things get really interesting when you add more light beams. Create a criss-cross grid of laser beams, with a dark patch of interference where each pair of beams cross, and you have a chessboard array known as an optical lattice. Atoms sprinkled into the lattice fall into the dark “wells”, where they remain trapped like marbles in an egg box.
That’s the theory, anyway. In practice, atoms at room temperature whizz around at hundreds of metres per second and have far too much kinetic energy to be trapped by the minute forces generated in an optical lattice, so to make the idea work they have to be cooled. In 1992, Philippe Verkerk and his colleagues at the École Normale Supérieure in Paris, France, used a combination of laser and magnetic fields to remove energy from a cloud of caesium atoms, slowing them down to less than walking pace – the equivalent to cooling them to within a few millionths of a degree above absolute zero. At these temperatures, Verkerk’s caesium atoms fell into the optical lattice’s wells and stayed there, resembling beads on a string.
Measurements confirmed the material’s similarities to a real crystal – though the caesium atoms in the traps were some 10,000 times further apart than those in the real thing. Today, research groups all over the world are studying ultracold atoms held in optical lattices. “Of course, you don’t capture all the complexity,” says physicist Immanuel Bloch at the University of Mainz in Germany. “We are trying to understand the interactions that matter at the simplest level.”
It might seem strange that optical lattices have so much in common with real materials. The reason lies in quantum mechanics. The properties of any real crystal depend on its energy landscape: the potential energy maxima are hills, while the minima are the valleys where atoms nestle. The ability of electrons to flow through a crystal, by moving between atoms in adjoining valleys, determines whether a crystal is a metal, a semiconductor or an insulator.
Classical physics struggles to explain how electrons do this. In the 1960s, British theorist John Hubbard developed a quantum mechanical model to describe this motion. He suggested that since electrons are quantum particles, they can “tunnel” through the hills, from one atom to its neighbour. Hubbard showed that the probability of this happening depends on the depth of the energy valley, the distance between neighbouring valleys and interactions between electrons on each atom.
In 1998, a team of theorists led by Dieter Jaksch, then at the University of Innsbruck in Austria, suggested that Hubbard’s model should also work for atoms in an optical lattice (Physical Review Letters, vol 81, p 3108). Here, atoms would tunnel between valleys corresponding to the dark regions in the laser interference pattern, creating the atom equivalent of an electron current flowing through a metal-like conductor.
Jaksch and his colleagues even predicted that physicists would be able to create an optical lattice that behaved like a type of insulator known as a Mott insulator. The secret was the choice of atoms – they must repel each other when they collide – and to make sure the valleys are so deep that it is very difficult for the atoms to tunnel through. In theory, the atom flow should stop altogether.
Transformers
In 2002, Ted Hänsch’s group at the Max Planck Institute for Quantum Optics in Garching, Germany, demonstrated this in spectacular style. Electrically neutral atoms can repel or attract each other, so Hänsch needed a way of tuning the force between them. Magnetic fields are perfect for this. The team made a 3D crystal using ultracold rubidium atoms bathed in a magnetic field and the light from six orthogonal laser beams. Cranking up the intensity of the laser made the wells deeper and stopped the atoms tunnelling, exactly as Jaksch’s group had predicted. At the turn of a dial, the researchers had morphed a material made of atoms and light beams from a metal into an insulator – a material with entirely different characteristics.
“At the turn of a dial they morphed the material from a metal into an insulator”
Back at JILA, this set Holland wondering just how versatile the morphing optical lattices could be. If you could turn a metal into an insulator, he asked himself, could you also make a semiconductor? Perhaps you could even turn these atom semiconductors into electrical components – an idea he dubbed atomtronics. “When we started we didn’t know how far we could push the analogy between electronics and atomtronics,” Holland says. “It’s not at all obvious that you could dream up a diode junction.” For a start, there is the fact that atoms are so much heavier than electrons – over 150,000 times more massive in the case of rubidium. Unlike electrons, atoms are electrically neutral and their quantum mechanical spin is different. Can atomtronics be any match for electronics?
Holland’s team looked to the Hubbard model for answers. These were not easily found. Theorists can write down equations describing one atom in a lattice. Add another atom, and you have an interaction between the two atoms as well as between the atoms and the lattice. “Add more atoms and you soon need a supercomputer,” says Holland. “The equations get very hard very quickly.” He started with the simplest electric circuit possible: a wire joining the two ends of a battery. Connecting a wire to a conventional battery starts a chemical reaction that produces electrons, which travel along the wire from the battery’s negative terminal to its positive terminal.
In Holland’s scheme, the atomtronic equivalent of the negative terminal is a dense cloud of ultracold atoms trapped by laser beams. The positive end of the battery uses identical beams, but without the atom cloud. The wire between them is an optical lattice containing plenty of empty wells (see Diagram). The atomtronic current starts flowing when repulsive interactions among the atoms in the cloud become too great for some atoms. Their response is to escape by tunnelling into the optical lattice “wire”. From there, they tunnel their way along the length of the wire towards the empty trap. How fast the atoms hop, Holland’s team inferred by comparison with the Hubbard model, depends on the depth of the wells in the optical lattice.
With a simple circuit in the bag, Holland’s team next turned to the atomtronic equivalent of a diode. In conventional electronics, diodes are the simplest semiconductor devices you can build. Add a few extra atoms to a silicon crystal and you turn it from an insulator into a semiconductor. Toss in phosphorus or arsenic, whose atoms carry one more electron than silicon and you have an “n-type” semiconductor that conducts electrons. Add atoms with one less electron, such as boron or gallium, and you create a p-type semiconductor whose crystal structure is missing electrons. Known as holes, these absent electrons in effect move through the crystal and also conduct electricity.
By marrying n-type and p-type semiconductors together, you make a p-n diode junction that allows current to flow in one direction only. To get any measurable current, you need to connect the negative terminal of the battery to the n-type semiconductor and the positive terminal to the p-type. This arrangement causes electrons to flow from the n-type silicon across the p-n junction. Swap the battery around and the current stops.
Last year, Holland’s team reported that they believe might be the easiest to make (Physical Review A, vol 75, p 023615). One way to do this is to create an optical lattice as usual and then expose one half of it to laser light with a slightly different frequency. The effect is to raise the energy of this part of the lattice and the atoms it contains. This step in the energy landscape is what acts as the p-n junction in a diode. Holland’s calculations show that if you connect an atomtronic battery to the device, the atoms flow in one direction only. Ultracold atoms can tunnel easily through the high-energy half of the lattice. When they reach the junction, they simply drop down to the low-energy region and carry on tunnelling until they reach the other end of the battery. However, this is not true for atoms starting in the low-energy half of the lattice. When they reach the p-n junction, they do not have sufficient energy to make the leap. The atoms soon fill up the low-energy half of the lattice and turn it into a Mott insulator, blocking the flow of current.
Holland thinks there is no limit to what physicists could make using lasers and atoms. His group has designed an atomtronic transistor by joining two p-n diodes together, mirroring conventional transistors made from p-n-p semiconductor junctions. As the building block of all kinds of digital logic gates, an atomtronic transistor is the key to memory and even microprocessor chips. This raises the question: could you build an entire computer from light beams and atoms? “In principle, it should work,” Holland says.
Could they ever replace silicon chips? Probably not. Atoms are much heavier than electrons so atomtronic devices will always run more slowly than ones driven by electrons. But the research might still speed up your PC. Computer power has roughly doubled every two years as manufacturers pack twice the number of transistors into integrated circuits. This requires ever smaller transistors. Chip maker Intel has now launched a processor made from transistors just 45 nanometres wide and plans to reduce this to 16 nanometres by 2013. The laws of quantum mechanics could stymie progress beyond this, though. Intel has already found that electrons can tunnel through transistors’ insulating layers, threatening to create a short-circuit. While novel materials could help solve this, other quantum effects could still wreak havoc.
Holland believes atomtronics might help tackle these problems. “We need to understand these effects if we want our electronics to keep working,” he says. “Atoms offer exquisite control. You can use them to understand the physics and feed that back to the electronics.”
Quantum logic
Beyond that, atomtronics might help the development of quantum computers. In theory, these computers would manipulate huge amounts of data stored in quantum bits, or qubits. The main difficulty is that you need to preserve the delicate superposition that qubits exist in. But in 1999 Gavin Brennan and colleagues at the University of New Mexico in Albuquerque suggested that neutral atoms in an optical lattice might offer an excellent route to stable qubits. Since neutral atoms in a lattice are protected from outside disturbances, atomtronics might offer a route to a more robust quantum computer that uses light to start a calculation and to read the answers.
“Atomtronics might even help the development of quantum computers”
So far no one has made an atomtronic diode or transistor, let alone an atomtronic quantum computer. In fact, no one has even created optical lattices with the shapes suggested by Holland. “All these experiments are very hard,” says Dana Anderson, an experimental physicist who works with Holland at JILA.
Anderson is taking a different approach to atomtronics, trapping the ultracold atoms with magnetic fields generated by wires on a chip rather than with laser beams. Design such a chip carefully, as Anderson and several groups around the world have, and you can cool a small cloud of ultracold atoms, trap it and then transport it around the circuit. He believes this could eventually lead to atomtronics circuits comprising several transistors packed together. “I think that it’s harder to do with optics, but I don’t want to underestimate the ability of my colleagues.”
He is right to respect their abilities. By playing around with laser beams, researchers have already made triangular lattices, as well as lattices that trap atoms spinning in certain directions, and lattices with wells of alternating depth. Bloch believes it’s only a matter of time before someone sculpts the optical lattices needed for a diode. This could even happen in the next year, he says.
Eventually, it could be possible to morph the atomtronic circuits from one type into another. Markus Greiner and his group at Harvard University are attempting to shape artificial crystals by shining laser light through a hologram onto a cloud of ultracold atoms. Similar techniques have been used by David Grier’s group at New York University to simultaneously manipulate hundreds of objects in the micrometre scale using a single laser beam. Grier’s holograms are computer-generated and created by shining laser light through an LCD screen, which splits the light into any number of beams and moves the objects such as cells at will. By adjusting the pattern on the screen, the researchers can create new holograms. One day, perhaps, this same idea will work with atomtronics.
To adapt the method to atoms that are thousands of times smaller is “very demanding”, says Bloch. If they pull it off, the technique could eventually mean you could watch an atomtronic diode transform into an atomtronic transistor and back again at the flick of a switch. Holland’s dream would finally be on the way to reality.
Quantum World – Learn more about a weird world in our comprehensive special report.
