A fused coupler basically consists of two, parallel optical fibers that have been twisted, stretched and fused together so that their cores are very close to each other. This forms a Coupling Region whose length L, determines the coupling ratio from one fiber to the other. During the manufacturing process, light is launched into an input port, P, and the output power from each output port is carefully monitored. When the desired coupling ratio is achieved, the fully automated manufacturing process is stopped. The resulting coupler is essentially one fiber with two cores that are very near to one another. This process is known as the Fused Biconical Taper (FBT) process.

The intensity profile of light traveling down a singlemode fiber is essentially Gaussian; that is, the intensity is greatest in the center and tapers off as the core/cladding interface is approached. The tail ends of the Gaussian profile extend slightly through the core and into the cladding. This tail is called the evanescent wave.

In the FBT process the cores of two identical parallel fibers are so close to one another that the evanescent wave can “leak” from one fiber core into the other core. This allows an exchange of energy, analogous to the energy exchange that takes place with two coupled pendulums. In fact, this is an excellent analogy. The amount of energy exchange is dependent upon the proximity of the two cores, d,and the length over which this exchange takes place, L . It is easy to see that if the coupling length is long enough, a complete transfer of energy can take place from one core into the other. If the length is longer still, the process will continue, shifting the energy back into the original core. By selecting the proper length, any given power transfer ratio can be realized. This is how a 50/50 or a 10/90 coupler is made.

In reality, energy transfers back and forth between the two fibers many times over the coupling region (also called the Interaction Length). The transfer rate is a function of wavelength, so if a wavelength is used that is different from the design wavelength, the energy transfer (or coupling ratio) will be different. As an example, assume that two different wavelengths are launched into the two input ports of a coupler, the interaction length (or coupling region) could be adjusted during the manufacturing process such that all of the light from both wavelengths exits at the same exit port, this is how Wavelength Division Multiplexers are made. Note that these couplers are also bi-directional, capable of splitting the two wavelengths when used in the reverse direction.