A quantum switch routes photons one by one

Just as the internet relies on routers to ensure that packets of data reach their intended recipients, quantum information networks will need switches that can direct the flow of quantum bits between nodes. In networks linked by photons—say, communication networks that use photons to transmit quantum encrypted data or parallel computing schemes that use them to link individual quantum processors—switches must manipulate photons’ trajectories without destroying the information they carry. (See the article by Michael Raymer and Kartik Srinivasan, Physics Today, November 2012, page 32 .) A new device developed by Barak Dayan and coworkers at the Weizmann Institute of Science in Rehovot, Israel, may do the job. ¹

In the researchers’ scheme, illustrated in the figure, a rubidium atom coupled with a microsphere optical resonator directs the traffic of photons along an optical fiber connecting two input and two output ports: The switch routes incoming photons to one of the two outputs depending on the atom’s internal state. Because the atom and photons interact coherently, the switch should be able to guide photons without destroying the superposition states that encode their quantum information.

Within the past year, other groups have designed devices that similarly exploit atom–photon interactions to switch a photon’s direction or phase. ² But whereas those schemes require auxiliary control fields, Dayan and company’s switch is controlled exclusively with fiber-guided photons and is therefore compatible with scalable quantum network architectures.

Left from right

At the heart of the switch is a silica microsphere resonator that traps light in so-called whispering gallery modes (WGMs), in which photons circulate around the resonator’s equator guided by total internal reflection at the walls. When a WGM resonator is brought to within a few hundred nanometers of an optical fiber, the electromagnetic fields of the two can couple, allowing photons to jump back and forth between them. As viewed from above in the figure, photons propagating right to left couple with the clockwise-propagating mode, and those traveling left to right couple with the counterclockwise mode. Once inside the resonator, a photon may couple with a nearby atom: An atom close enough to interact with the resonator’s evanescent field may absorb and reemit the photon into the cavity. Eventually the photon jumps back to the fiber and exits in the direction corresponding to its propagation mode in the resonator.

One challenge for Dayan and his coworkers was getting the atom to distinguish between clockwise- and counterclockwise-propagating photons. Until recently the prevailing wisdom was that atoms make no such distinction. Models of atom–photon interactions treated the photons’ electromagnetic fields as lying transverse to their propagation direction, as is the case for photons in free space. The resulting electric field has the same orientation for both counterpropagating modes.

In waveguides and resonators, however, confinement effects can force part of a photon’s electromagnetic field into the propagation direction. In a WGM resonator, trapped photons can exist in TM modes, for which the electric field has a longitudinal component but the magnetic field is strictly transverse, and TE modes, for which the opposite is true. The TE modes’ electric-field orientation is independent of the propagation direction. But in the TM modes, the electric field’s transverse and longitudinal components oscillate out of phase, which yields a rotating electric field resembling that of circularly polarized light. For photons traveling clockwise around the resonator, the electric field rotates one way; for those traveling counterclockwise it rotates the other.

Last year a group led by Arno Rauschenbeutel (Vienna University of Technology) worked out the details of how atoms interact with those modes. ³ They showed that the TM modes, due to their spin angular momentum, could change the orientation of an atom’s spin; because the modes’ spin and propagation direction are linked, they effectively give the atom a means to distinguish left from right.

That finding was fortuitous for Dayan, who had been looking to build a switch based on toggling an atom’s angular momentum states. “We had been working on a complicated scheme to inject circularly polarized photons into the resonator,” recalls Dayan. “Then we learned that we didn’t need to—the resonator creates them nearly perfectly on its own.”

Flipping the switch

In Dayan and company’s scheme, photons propagating left to right become approximately right-hand circularly polarized (σ⁺) inside the resonator, and those traveling right to left become approximately left-hand circularly polarized (σ⁻). Those photons interact with a rubidium atom in a three-level configuration, illustrated in the inset, in which two degenerate ground states having magnetic quantum numbers −1 and +1 share an excited state.

An atom in the ∣−1〉 state that absorbs a σ⁺ photon will reradiate it as a σ⁻ photon and decay to the ∣+1〉 state. That the atom doesn’t instead emit a σ⁺ while decaying back to the ∣−1〉 state is due to a peculiarity of quantum cavity electrodynamics first uncovered a few years ago. ⁴ Technically, the atom radiates into the σ⁺ and σ⁻ polarization states with equal probability. But because the emitted σ⁺ photon has a π phase shift and is indistinguishable from the incident σ⁺ photon, those two destructively interfere, and only the σ⁻ photon survives.

In the ∣+1〉 state, the atom is transparent to σ⁺ photons. By symmetry, the ∣+1〉 state reflects, and the ∣−1〉 state transmits, σ⁻ photons. “The key,” says Dayan, “is that the atom ‘remembers’ when it reflected a photon—it changes its state.” That means photons can toggle the switch: A control photon injected at input 1 switches the atom to the ∣+1〉 state, and ensures that a subsequent photon, the target, will be routed to output 2. Likewise, a control photon injected at input 2 steers the target to output 1.

To integrate their switch into large-scale networks, Dayan and his colleagues will first need to devise a better atom–resonator coupling. At present they use an approach that Dayan helped develop as a postdoc with Jeff Kimble and Kerry Vahala at Caltech: They trap a cloud of laser-cooled atoms just above the silica resonator and then release it. Occasionally a free-falling atom passes through a Goldilocks zone about 100 nm from the resonator surface—close enough to interact with the resonator’s evanescent field but not so close as to be pulled onto the resonator’s surface by van der Waals forces. The atoms spend just microseconds in the evanescent field, but that’s plenty of time to complete an experiment.

Still, a less ephemeral coupling might be established by holding the atom in an optical trap—an approach that’s been successfully used to couple atoms with optical fibers. Also, Dayan notes that real atoms can be substituted with artificial ones such as quantum dots, which can more easily be pinned into place. “It’s not the atom that’s important,” he says. “It’s the scheme.”

The researchers also look to improve the switch’s fidelity. At present, the switch correctly transmits photons about 90% of the time but correctly reflects them only 65% of the time. In theory, those numbers should be closer to 100% and 90%, respectively. (The reflection operation’s inefficiency is mainly because the polarization of the resonator’s field isn’t perfectly circular.) Assuming they can get close to those theoretical limits, Dayan and company should then be able to tackle more sophisticated quantum experiments, such as using the switches as quantum memories or quantum logic gates.

Rauschenbeutel thinks those are reachable goals. “There are some technical considerations they’ll have to work out. For instance, one could imagine using a control photon in a superposition of TM and TE states to place the atomic switch in a superposition of toggled and untoggled states, but then you’d need to carefully orchestrate the ensuing switching sequence to avoid destroying that superposition. It will take some thought, but I’m pretty sure it’s doable.”