Neutral atoms are trapped and imaged in a three-dimensional optical lattice

The essence of quantum computing is simple to state: If you could entangle N two-level qubits, you’d have access not just to 2N combinations of binary states but also to their myriad quantum superpositions. Running on just 50 or so qubits, a quantum computer could crack integer factorization and other problems beyond the reach of the most powerful classical computer.

Physicists have already made single qubits from a range of ingredients, including Josephson junctions, quantum dots, and trapped ions. Other, more exotic recipes, like those based on topological states of fractional quantum Hall systems and superconducting strontium ruthenate, await their qubit debut.

As in juggling, the more objects you have to coordinate, the more precise and delicate your control has to be. Curiously, the record for juggling balls, 10, is about the same for entangling qubits. Three years ago, Rainer Blatt of the University of Innsbruck in Austria and his collaborators entangled eight trapped ions. Breaking that record is difficult. Vibrational motion, which serves to entangle the ions, develops more collective modes as more ions are added. When a new mode appears just above the ground state, an ion can easily hop out of its qubit state. Ways around the problem are being pursued by several groups.

Because trapped atoms interact far more weakly than trapped ions, they’re far less susceptible to such collective disruptions; they’re also harder to entangle. Although Immanuel Bloch of the University of Mainz in Germany and others have trapped and entangled atoms in collective states, no one has yet matched the ion trappers and created qubits from small groups of individually addressable atoms.

The current lack of qubits hasn’t deterred David Weiss of the Pennsylvania State University from building a system in which many individual atoms could one day be initialized, entangled, and read out. In a recent experiment, he, postdoc Karl Nelson, and graduate student Xiao Li set up a three-dimensional optical lattice, spread 250 or so cesium atoms among its nodes, and imaged the atoms. ¹

The accompanying figure shows that atoms in the focal plane are clearly resolved. (The background blur comes from atoms trapped in out-of-focus sites.) The figure also shows the central atoms remained in their sites for the 3 seconds between exposures. That duration is more than enough for quantum computation. At 5 µm, the lattice spacing is both large enough for light to address individual atoms and small enough to sustain the entanglement schemes that theorists have devised.

The Penn State optical lattice is formed by three orthogonal pairs of nearly copropagating lasers whose interference creates a 3D standing wave. The lasers’ 845.5-nm wavelength is far from any transition of the Cs atom. Being off resonance keeps the atoms trapped but requires a second set of lasers for cooling them.

At the start of an experimental run, 3000 or so Cs atoms are loaded into a magneto-optical trap. The lattice lasers are turned on, which corrals about six atoms into each lattice site. Turning on the cooling lasers promotes two two-body processes: molecule formation and energetic collisions, both of which eject pairs of atoms from the lattice.

After 5 ms of cooling, sites that start off with an even number of atoms end up empty, while sites that start off with an odd number of atoms end up with one trapped atom. The pattern of occupied sites is random, as the figure shows.

To image the atoms, the Penn State team uses the same set of lasers that provides the cooling. Light scatters off the trapped atoms and passes through an objective lens whose 2.8 µm depth of field brings only one lattice plane into focus. A second lens, acting like the eyepiece of a microscope, produces an image on a CCD camera. A piezoelectric actuator moves the objective lens back and forth to image successive planes.

Weiss’s lattice meets some requirements for quantum computation. Other requirements await further work. The random distribution of atoms is not ideal, but Weiss has proposed schemes to fill the empty sites. The 3D geometry provides each atom with more entanglement partners than 1D and 2D geometries, but at the cost of making it more difficult to address individual sites without perturbing other sites. For that reason, other groups, among them Philippe Grangier’s at the Université de Paris-Sud, Dieter Meschede’s at the University of Bonn, and Mark Saffman’s at the University of Wisconsin, favor lower dimensions.

There are several schemes for entangling trapped atoms. Lasers could excite adjacent atoms into Rydberg states, whose strong dipole–dipole interactions extend over the atoms’ micron-scale separation. In another scheme, the optical lattice could be shrunk to induce phase-dependent collisions. Atoms could also be entangled in high-finesse optical cavities. Although none of the schemes has yielded qubits, the atom trappers are optimistic. After all, algorithms for quantum computation were written years before the first qubit ever appeared.