Traditional high data rate communication realized by electric components – such as A/D converters, encoders, modulators, amplifiers, processors and similar integrated circuits – faces a challenge as further increase of frequency would result in cooling problems due to the generated amount of thermal energy in the form of Joule heating. An alternative approach is light-based information transfer in guided structures, such as optical fibers. However, optical fibers occupy relatively large spaces due to the limited minimum size of bends allowed by the small refractive index difference between the wave guiding core and the outer cladding.
The following figure illustrates a typical optical fiber structure. The core material, having a larger refractive index compared to the cladding material, allows for guiding light due to total internal reflection. The buffer layer is usually present for mechanical isolation and protection purposes and holds a bundle of fibers together.
Increasing refractive index difference between core and cladding allows shrinking device sizes substantially. One approach is to have silicon and silicon dioxide structures, that are usually referred to as silicon photonics or silicon on insulator structures, as small as a few millimeters in size.
These devices can serve various functions, such as transporting signals using waveguides, splitting by directional couplers, mixing, multiplexing or filtering, usually in a frequency dependent manner. The wavelength of interest is either 1310 or 1550 nm to agree with existing fiber optics solutions, that operate at either of these wavelengths. These wavelengths from the near-infrared range of the spectrum provide low attenuation in silica materials and therefore are favored for long-range optical communication links.
The light (electromagnetic wave) is well confined within these structures as the following plot of the electric field illustrates.
However, the boundaries of the core are not as perfectly closed for electromagnetic radiation as for a metal waveguide, and some small leakage occurs, allowing for an interesting phenomenon: evanescent wave coupling. Let’s say one wants to design a simple optical splitter: when a wave propagates in a waveguide near another waveguide, due to the leaking fields a second wave is excited in the neighboring waveguide. The following plots illustrate a device example with the electric fields within the two waveguides, where the original signal is launched at the top waveguide from the right and coupled to the bottom waveguide.
The coupling length (where the two waveguides are in close proximity) can be adjusted to allow for splitting the incoming signal equally to the two downstream waveguides for both used infrared frequencies, 1310 and 1550 nm.
Fine tuning the coupling distance between waveguide cores and the coupling length also allows for frequency-selective operation, as illustrated below. For the 1310 nm wavelength, the geometrical parameters couple most of the signal to the bottom waveguide, however, for the 1550 nm wavelength, the signal travels through the top waveguide.
A complete transmission plot over various wavelengths can also be obtained to analyze the performance better when carrying out a time domain simulation. A single simulation gives results for a whole broad frequency band:
In this blog post a brief overview of evanescent coupling was provided and we saw how simple design changes such as coupling length can have a large impact on overall device function and performance. This also shows that silicon photonics devices cannot have arbitrary small distance between waveguides, where coupling is to be avoided. In these cases, evanescent coupling is an unwanted phenomenon.
Parametric and material sweeps to allow efficient silicon photonics single device design and complex system evaluation are possible using OnScale, either when designing for or against evanescent coupling.