Scatterings image

In the SUNY-led team’s setup, a waveguide engineered to have highly asymmetric internal reflections allows a high-power signal beam (left) to be interferometrically turned on and off by a control beam (right) operating at only one-third the signal beam’s power. [Image: State University of New York, Buffalo]

The ability to control one light source using another one commonly requires a control beam that consumes as much energy as, or even more than, the beam being controlled—not an especially efficient setup. A team of researchers in the United States has now devised an approach to such all-optical switching that involves a control beam running at only one-third the power of the signal beam (Phys. Rev. Lett., doi: 10.1103/PhysRevLett.117.193901).

What’s more, the scientists behind the work believe that the system can be further refined to the point where the control beam might require one-ten-thousandth the power of the signal. That’s a level that could make all-optical switching a lot more practical for energy-efficient computer chips, optical communications and other applications.

Nonlinear power penalty

Light-light switching commonly involves nonlinear effects in optical materials, in which an intense laser field from a control beam reversibly modifies an optical property of the material (such as its refractive index), enabling switching of the weaker signal beam. That’s fine in principle, but the requirement of an intense control or pumping laser field imposes a power penalty on such systems that makes them hard to channel into practical applications.

To clear that power hurdle, the U.S. research team—led by OSA Member Liang Feng of the State University of New York (SUNY), Buffalo, along with other scientists at SUNY, the California Institute of Technology and the City University of New York—turned to a familiar linear process: destructive interference. The researchers reasoned that a switching scheme that used a tunable control beam to turn the signal beam on and off directly, depending on whether the two beams were in or out of phase, could open the possibility of much lower power consumption than is achievable using approaches that require nonlinear light-matter interactions.

Metamaterial magic

Previous efforts have used destructive interference as the basis for all-optical switching—but those attempts have required that the control beam and signal beam have equal power. To cut the control-beam power requirement, the SUNY-led team leveraged the practical magic of metamaterials.

The team’s setup begins with an 800-nm-wide silicon waveguide that can serve as a channel for an infrared laser signal. The scientists then fitted the waveguide with a delicately calibrated set of nanophotonic elements, including asymmetric notches that introduced variations in the channel’s refractive index, and off-center diamond-shaped germanium-chromium pads atop the waveguide that acted as partial light absorbers.

The net effect is to introduce a highly asymmetric reflectivity within the waveguide that enables the amplitude of the reflected light from the lower-power control beam (as a standing wave within the waveguide) to match, and potentially cancel out, the amplitude of the beam being transmitted through the waveguide. (More technically, the system exploited the physics of so-called non-Hermitian metamaterials, which for certain conditions of the scattering matrix include “exceptional points” that allow asymmetric resonances.)

Efficient switch

To test the setup, the team shone an infrared laser into the waveguide from the left, and a control beam at one-third of the power from the right. They then adjusted the phase of the control beam relative to the signal beam, to see if they could implement a workable optical switch. As expected, the researchers found that the interference between the signal beam and the reflected part of the control beam could cut the signal transmitted through the waveguide by some 60 db—roughly a million-fold reduction.

The scientists also write that the control-beam power can, in principle, “be further engineered to be much weaker than the signal power,” through additional tweaking of the geometry and metamaterial elements and the selective integration of nonlinear elements to nudge the system closer to specific exceptional points. The result, the researchers believe, could be a promising platform for “low-power interferometric light-light switching for the next generation of optical devices and networks.”