Polarization in Fiber Optics

A beam of light can be thought of as being composed of two orthogonal electrical vector field components that vary in amplitude and frequency. Polarized light occurs when these two components differ in phase or amplitude. Polarization in optical fiber has been extensively studied and a variety of methods are available to either minimize or exploit the phenomenon. In this tutorial, basic principles and technical background are introduced to help explain how the polarization in fiber optics works.

Polarization Manifestation in Optical Fibers

Birefringence in Optical Fibers

Birefringence is a term used to describe a phenomenon that occurs in certain types of materials, in which light is split into two different paths. This phenomenon occurs because these materials have different indices of refraction, depending on the polarization direction of light.

Birefringence is also observed in an optical fiber, due to the slight asymmetry in the fiber core cross-section along the length and due to external stresses applied on the fiber such as bending. In general, the stress-induced birefringence dominates the geometry-induced one.

Polarization Maintaining (PM) Fiber

A specialty fiber called the Polarization Maintaining (PM) Fiber intentionally creates consistent birefringence pattern along its length, prohibiting coupling between the two orthogonal polarization directions. In any design, the geometry of the fiber and the materials used create a large amount of stress in one direction, and thus create high birefringence compared to that generated by the random birefringence. There are a number of designs available commercially, using various stress inducing architectures, such as Panda and Bow Tie PM Fibers available with various cut-off wavelengths.

Poincare Sphere

The Poincare sphere is one of the conventional ways of describing the polarization and changes in polarization of a propagating electromagnetic wave. It provides a convenient way of predicting how any given retarder will change the polarization form. Any given polarization state corresponds to a unique point on the sphere. The two poles of the sphere represent left and right-hand circularly polarized light. Points on the equator indicate linear polarizations. All other points on the sphere represent elliptical polarization states. An arbitrarily chosen point H on the equator designates horizontal linear polarization, and the diametrically opposite point V designates vertical linear polarization

Poincare sphere representation of polarization states
Figure 1. Poincare sphere representation of polarization states.

Measurable Polarization Properties


Degree of Polarization (DOP) is defined as

DOP = Ipol / (Ipol + Iunp)

Where Ipol and Iunp are the intensity of polarized light and unpolarized light, respectively.

When DOP = 0, light is said to be unpolarized, and when DOP = 1, it is totally polarized. Intermediate cases correspond to partially polarized light


Polarization Extinction Ratio (PER) is the ratio of the minimum polarized power and the maximum polarized power, expressed in dB. Any polarization component will specify this value as a specification.


Polarization Dependent Loss (PDL) is the maximum (peak to peak) variation in insertion loss as the input polarization varies over all its states, expressed in dB.


Polarization Mode Dispersion (PMD) is actually another form of material dispersion. Single-mode fiber supports a mode, which in fact consists of two orthogonal polarization modes. Ideally, the core of an optical fiber is perfectly circular. However, in reality, the core is not perfectly circular, and mechanical stresses such as bending, introduce birefringence in the fiber. This causes one of the orthogonal polarization modes to travel faster than the other, hence causing dispersion of the optical pulse.

The maximum difference in the mode propagation times due to this dispersion is called Differential Group Delay (DGD), whose unit is typically given in picoseconds. Because of its dynamic properties, PMD does not have a single, fixed value for a given section of fiber, but has a distribution of DGD values over time. The probability of a DGD with a certain value at any particular time follows the Maxwellian distribution shown in Figure 2. As an approximation, the maximum instantaneous DGD is about 3.2 times the average DGD of a fiber

Maxwellian distribution of DGD
Figure 2. Maxwellian distribution of DGD.

Controlling Polarization in Optical Fiber

Methods of Controlling Polarization

Controlling the polarization state in optical fiber is similar to the free space control using waveplates via phase changes in the two orthogonal states of polarization. In general, three configurations are commonly used.

In the first configuration, a Half-Wave Plate (HWP) is sandwiched between two Quarter-Wave Plates (QWP) and the retardation plates are free to rotate around the optical beam with respect to each other. The first QWP converts any arbitrary input polarization into a linear polarization. The HWP then rotates the linear polarization to a desired angle so that the second QWP can translate the linear polarization to any desired polarization state.

An all-fiber controller based on this mechanism can be constructed, with several desirable properties such as the low insertion loss and cost, as shown in Figure 3. In this device, three fiber coils replace the three free-space retardation plates. Coiling the fiber induces stress, producing birefringence inversely proportional to the square of the coils’ diameters. Adjusting the diameters and number of turns can create any desired fiber wave plate. Because bending the fiber generally induces insertion loss, the fiber coils must remain relatively large.

Figure 3: Polarization control using multiple coiled fiber

The second approach is based on the Babinet-Soleil Compensator. An all-fiber polarization controller based on this technique is shown in Figure 4. The device comprises a fiber squeezer that rotates around the optical fiber. Applying a pressure to the fiber produces a linear birefringence, effectively creating a fiber wave plate whose retardation varies with the pressure. Simple squeeze-and-turn operations can generate any desired polarization state from any arbitrary input polarization

Figure 4: Polarization control using Babinet-Soleil compensator principle.

Polarization controllers also can be made with multiple free-space wave plates oriented 45° from each other. An all-fiber device based on the same operation principle would reduce the insertion loss and cost. The retardation of each wave plate components varies with the pressure of each fiber squeezer. The challenge is making the device reliable, compact and cost-effective.

Piezoelectric actuators drive the fiber squeezers for high speed. Because it is an all-fiber device, it has no back reflection and has extremely low insertion loss and polarization-dependent loss. All new 25xxP Series Polarization Control instruments employ the fiber squeeze technique.

Figure 5: Polarization control by squeezing fiber from various directions.


Newport would like to acknowledge General Photonics for their contributions to this tutorial.