定义：一种光纤装置，用来将光从一根或几根输入光纤中耦合到一根或多根光纤中，或者从自由空间耦合进光纤中。

光纤耦合器是一种基本的光纤光学器件。需要注意的是，光纤耦合器有两种不同的含义： 

• 可以是一个包含一根或多根输入光纤和一根或多根输出光纤的光纤器件。一根输入光纤中的光可能在一根或者多根输出光纤中出现，其中功率分布与波长和偏振有关。 
• 也可以是一个将光从自由空间耦合到光纤中的装置。 

这里指的光纤耦合器是第一种类型，将光在光纤之间耦合。可以采用不同的方法制作耦合器： 

 图1:2*2光纤耦合器。
 

• 可以将两个或更多光纤受热拉成锥形然后熔接，这样它们的纤芯在几厘米长度上紧密接触。也可以采用保偏光纤，得到保偏耦合器或分离器。 
• 有些耦合器采用边抛光光纤，提供与光纤纤芯的接触。 
• 有的利用平面光波导，包含一些分支波导，采用光纤耦合输入和输出。 
• 耦合器也可以由体光学材料制作，例如以微透镜和分束器的形式，也可以耦合进光纤中。 

图2给出了以上描述第一种类型光纤耦合器中光束传播的数值模拟。其中，光在两光纤纤芯中振荡，最终大部分光强保留在初始光纤中。但是，对于另一个波长的光来说，耦合过程可能差别很大。因此，这种耦合器只能工作在有限的光学带宽中。它可以用作二色性耦合器或合束器，用来将两个波长组分分离或合在一起（例如光纤放大器中的泵浦光和信号光）。 

图2：光纤耦合器的振幅分布，由数值模拟光束传播过程得到的。 
光纤耦合器通常都是定向耦合器，即进入入射端口的光不会再返回输入端口中。还有一个特征量是回波损耗，表征反射光相对于入射光的大小。 

光纤合束器的限制因素 
耦合损耗
如果采用的光纤都是单模的（即给定波长情况下，单位偏振方向上只传播一个模式），耦合器的性能会受到一些物理因素的限制。尤其是，将两个或更多具有相同频率的入射光耦合进单偏振的输出光纤时不可避免的存在很大的附加损耗，当然如果精准的调节和稳定入射光束的相位是可以消除损耗的。即两入射光束合束时需要完全相干。 
但是，这一限制在不同入射波长时就不存在了，有的耦合器可以将两个不同波长的光合在一起耦合到一个输出光纤中，而不会产生很大损耗。这种二色性耦合器通常用在光纤放大器中将信号光和泵浦光合在一起。两个输入光的插入损耗都很小（例如，远小于1 dB）。还有对波长敏感的耦合器，可以用在波分复用通信系统中用作复用器，将几个具有不同波长的入射信号结合在一起，或者分离开。 
多模光纤合束器可以将两个不相干的光束合在一起，也不会产生耦合损耗。但是，这会使光亮度减小。 
带宽
大多数的耦合器都工作在有限波长范围内（带宽有限），因为耦合强度与波长有关（通常还与偏振有关）。这是耦合长度一定的耦合器的典型性质。熔融耦合器的典型带宽为几十纳米。上面已经提到了，它可以用作二色性耦合器或者合束器。有时也被称为WDM耦合器（参阅波分复用）。 

典型应用 
光纤耦合器的一些典型应用包括： 

• 在有线电视系统中，从一个发射器中得到的强的信号光被送进光纤分离器，它将光分配到很大数目的不同用户的输出光纤中。 
• 光纤耦合器可以用在光纤干涉仪中，例如光学相干断层扫描（OCT）。 
• 在光纤激光器的谐振腔中，二色性光纤耦合器可用来注入泵浦光，再采用另一个耦合器作为输出耦合器。尤其是在光纤环形激光器中通常采用这种技术，不需要在谐振腔端口注入光。 
• 在光纤放大器和激光器中，二色性耦合器通常用来注入泵浦光或者消除信号输入中的剩余泵浦光。 
• 在高功率光纤激光器和放大器中，多模光纤耦合器用来将几个激光二极管的输出结合在一起，然后输入到活性光纤（双包层光纤）的内包层中。

Definition: fiber devices for coupling light from one or several input fibers to one or several output fibers, or from free space into a fiber

Fiber couplers belong to the basic components of many fiber-optic setups. Note that the term fiber coupler is used with two different meanings:

• It can be an optical fiber device with one or more input fibers and one or several output fibers. Light from an input fiber can appear at one or more outputs, with the power distribution potentially depending on the wavelength and polarization.
• It can also be a device for coupling (launching) light from free space into a fiber; see the article on fiber launch systems.

This article treats fiber couplers of the first type, coupling light from fibers to fibers. Such couplers can be fabricated in different ways:

Figure 1: A 2-by-2 fiber coupler.

• Two or more fibers can be thermally tapered and fused so that their cores come into intimate contact over some length of a few centimeters, for example. Such fused couplers can also be made with polarization-maintaining fibers, leading to polarization-maintaining couplers (PM couplers) or splitters.
• Some couplers use side-polished fibers, providing access to the fiber core.
• There are fiber-optic pump combiners and pump–signal combiners, which usually work with multimode pump fibers.
• There are planar lightwave circuits, containing things like branching waveguides, with fibers coupled to the inputs and outputs.
• Couplers can also be made from bulk optics, for example in the form of microlenses and beam splitters, which can be coupled to fibers (“fiber pig-tailed”).

One may omit one of the input ports of a 2-by-2 fiber coupler, obtaining a Y coupler, also called T coupler. It may also be called a tap coupler, particularly if only a small fraction of power is obtained at one output and used e.g. for power monitoring. Couplers with many inputs or outputs are called star couplers; they may be used, e.g., as fiber-optic splitters, e.g. for distributing cable-TV signals.

Figure 2 shows a numerical beam propagation simulation for a fiber coupler based of the first type as explained above. Here, the light distribution oscillates between the two fiber cores, and finally the larger part of the power remains in the original (upper) fiber. For light with other wavelengths, however, the coupling can be very different. Therefore, such couplers work only in a limited optical bandwidth. They can be used as dichroic couplers or beam combiners, for example for separating or combining two wavelength components (such as pump and signal light in a fiber amplifier).

Figure 2: Amplitude distribution in a fiber coupler, obtained with a numerical simulation of beam propagation, done with the software RP Fiber Power.

Fiber couplers are usually directional couplers, which means that essentially no optical power sent into some input port can go back into one of the input ports. There is often a specification of return loss, which indicates how much weaker the back-reflected light is, compared with the input, and is usually quite large (many tens of decibels).

Limitations for Fiber Combiners

Coupling Loss

In many cases, all fibers involved are single-mode, i.e., they support only a single mode per polarization direction for a given wavelength. There are then certain physical restrictions on the performance of the coupler. In particular, it is not possible to combine two or more inputs of the same optical frequency into a single-polarization output without significant excess losses, except if the optical phases of the input beams are precisely adjusted and stabilized. That means that the two inputs to be combined would have to be mutually coherent.

However, such a restriction does not occur for different input wavelengths: there are couplers which can combine two inputs at different wavelengths into one output without exhibiting significant losses. Such dichroic couplers are used in fiber amplifiers to combine the signal input and the pump wave. Their insertion loss may be very small (e.g. far below 1 dB) for both inputs. Other wavelength-sensitive couplers are used as multiplexers (WDM couplers) in wavelength division multiplexing (WDM) telecom systems to combine several input channels with different wavelengths, or to separate channels.

Multimode fiber combiners allow the powers of two mutually incoherent beams to be combined without a power loss. However, this will cause some loss of brightness.

Bandwidth

Most types of couplers work only in a limited range of wavelength (a limited bandwidth), since the coupling strength is wavelength-dependent (and often also polarization-dependent). This is a typical property of those couplers where the coupling occurs over a certain length. Typical bandwidths of fused couplers are a few tens of nanometers. As mentioned above, they can be used as dichroic couplers or beam combiners. They are sometimes also called WDM couplers (→ wavelength division multiplexing).

Typical Applications

Some typical applications of fiber couplers are:

• In a cable TV system, the powerful signal from one transmitter is sent in to a fiber-optic splitter, which distributes the power over a large number of output fibers for different customers.
• Fiber couplers can be used in fiber interferometers, for example for optical coherence tomography (OCT). Specially designed broadband couplers are often required for such purposes.
• Within the resonator of a fiber laser, a dichroic fiber coupler can be used to inject pump light, and another fiber coupler can be used as the output coupler. This technique is used particularly in fiber ring lasers, having no resonator ends where light could be injected.
• In fiber amplifiers and lasers, dichroic couplers are often used for injecting pump light or eliminating residual pump light from the signal output.
• In high-power fiber lasers and amplifiers, multimode fiber couplers are often used for combining the radiation of several laser diodes and sending them into inner cladding of the active fiber (a double-clad fiber).

Bibliography