In an earlier post I discussed how laser light is brought into waveguides on a chip. Now let’s look at some of the reasons we want to have light guided on chips in the first place. One device that shows the advantage of integrated optics is the ring resonator. It is a very compact device which allows precise discrimination of wavelengths.
Just like optical fibers, waveguides on chips have the advantage of being able to support data signals at different wavelengths. While an electrical wire has a set voltage and carries a single signal, optical waveguides can support many signals on top of each other, often with little interference between them. But sometimes you want to filter out one of the wavelengths and leave the others on their way. This can be done with a ring resonator.
When two waveguides are placed near each other, but not touching, light from one couples into the other. This is somewhat analogous to quantum tunneling. In a ring resonator, a waveguide is bent into a circular loop and placed alongside a straight waveguide. At the location where the straight waveguide comes close to the ring, the light couples from one to the other. But over reasonable time scales, the light has a chance to go around the ring and reach the coupling region again, so it has the chance to couple back to the straight waveguide. So there is an interference effect between the waves of light on the waveguide with the light from the ring.
The main feature of the ring is its circumference. If the path light travels on the ring corresponds to a set fraction of multiple wavelengths, then the light at that wavelength from the ring interferes with light on the straight waveguide destructively. That one wavelength, which was present before reaching the ring, is no longer propagating on the waveguide. But other wavelengths don’t divide into the circumference the same way, and so they don’t see the same destructive interference. They go through. The net result is that the ring filtered out a single wavelength and allowed the others through. This filtering can be extremely sharply defined, so even if one wavelength channel is at 1550 nm and the other is at 1550.1 nm, it’s possible to filter one and not the other.
BRIAN STERN – PHD STUDENT, CORNELL UNIVERSITY
Brian is currently researching ways to improve integrated photonics systems and developing new on-chip technologies.