>> Background History of Optical Circulator
An optical circulator is a multi-port (minimum three ports) nonreciprocal passive component.
The function of an optical circulator is similar to that of a microwave circulator—to transmit a lightwave from one port to the next sequential port with a maximum intensity, but at the same time to block any light transmission from one port to the previous port. Optical circulators are based on the nonreciprocal polarization rotation of the Faraday effect.
Starting from the 1990s optical circulators has become one of the indispensable elements in advanced optical communication systems, especially WDM systems. The applications of the optical circulator expanded within the telecommunications industry (together with erbium-doped fiber amplifiers and fiber Bragg gratings), but also expanded into the medical and imaging fields.
>> Background Technology
Since optical circulators are based on several components, including Faraday rotator, birefringent crystal, waveplate, and beam displacer, we will have to explain these technologies before jumping into the detail of circulator.
1. Faraday Effect
The Faraday effect is a magneto-optic effect discovered by Michael Faraday in 1845. It is a phenomenon in which the polarization plane of an electromagnetic (light) wave is rotated in a material under a magnetic field applied parallel to the propagation direction of the lightwave. A unique feature of the Faraday effect is that the direction of the rotation is independent of the propagation direction of the light, that is, the rotation is nonreciprocal. The angle of the rotation θ is a function of the type of Faraday material, the magnetic field strength, and the length of the Faraday material, and can be expressed as
θ = VBL
where V is the Verdet constant of a Faraday material, B the magnetic field strength parallel to the propagation direction of the lightwave, and L the length of the Faraday material.
The Verdet constant is a measure of the strength of the Faraday effect in a particular material, and a large Verdet constant indicates that the material has a strong Faraday effect. The Verdet constant normally varies with wavelength and temperature. Therefore, an optical circulator is typically only functional within a specific wavelength band and its performance typically varies with temperature. Depending on the operating wavelength range, different Faraday materials are used in the optical circulator.
Rare-earth-doped glasses and garnet crystals are the common Faraday materials used in optical circulators for optical communication applications due to their large Verdet constant at 1310 nm and 1550 nm wavelength windows. Yttrium Iron Garnet and Bismuth-substituted Iron Garnets are the most common materials.
The Verdet constant of the BIG is typically more than 5 times larger the YIG, so a compact device can be made using the BIG crystals. All these materials usually need an external magnet to be functional as a Faraday rotator. Recently, however, a pre-magnetized garnet (also call latching garnet) crystal has been developed that eliminates the use of an external magnet, providing further potential benefit in reducing overall size.
Faraday rotators in optical circulators are mostly used under a saturated magnetic field, and the rotation angle increases almost linearly with the thickness of the rotator in a given wavelength (typically 40 nm) range. The temperature and wavelength dependence of the Faraday rotation angle of the typical BIG crystals at wavelength of 1550 nm is 0.04-0.07 deg/°C and 0.04-0.06 deg/nm, respectively.
2. Light Propagation in Birefringent Crystals
Another common material used in the construction of optical circulators is the birefringent crystal. Birefringent crystals used in optical circulators are typically anisotropic uniaxial crystals (having two refractive indices with one optical axis). In an anisotropic medium, the phase velocity of the light depends on the direction of the propagation in the medium and the polarization state of the light. Therefore, depending on the polarization state of the light beam and the relative orientation of the crystal, the polarization of the beam can be changed or the beam can be split into two beams with orthogonal polarization states.
The refractive index ellipsoid for a uniaxial crystal is shown in the above figure. When the direction of the propagation is along the z-axis (optic axis), the intersection of the plane through the origin and normal to the propagation direction So is a circle; therefore, the refractive index is a constant and independent of the polarization of the light. When the direction of the propagation S forms an angle θ with the optic axis, the intersection of the plane through the origin and normal to S becomes an ellipse. In this case, for the light with the polarization direction perpendicular to the plane defined by the optic axis and S, the refractive index, is called the ordinary refractive index no, is given by the radius ro and independent of the angle θ. This light is called ordinary ray and it propagates in the birefringent material as if in an isotropic medium and follows the Snell’s law at the boundary.
On the other hand, for light with the polarization direction along the plane defined by the optic axis and S, the refractive index is determined by the radius re and varies with the angle θ. This light is called the extraordinary ray and the corresponding refractive index is called the extraordinary refractive index ne. In this case ne is a function of θ and can be expressed as
The ne varies from no to ne depending on the direction of propagation. A birefringent crystal with no < ne is called a positive crystal, and one with no > ne is called a negative crystal.
Therefore, the function of a birefringent crystal depends on its optic axis orientation (crystal cutting) and the direction of the propagation of a light. Birefringent crystals commonly used in optical circulators are quartz, rutile, calcite, and YVO4.
One of the applications of the birefringent crystal is the waveplate (also called retardation plate). A waveplate can be made by cutting a birefringent crystal to a particular orientation such that the optic axis of the crystal is in the incident plane and is parallel to the crystal boundary (zx-plane in the second figure). When a plane wave is perpendicularly incident onto the incident plane (zx-plane), the refractive index for the polarization component parallel to the x-axis equals no and that parallel to the z-axis equals ne.
Therefore, when a linearly polarized light with the polarization direction parallel to the z- or x-axis is incident to the waveplate, the light beam experiences no effect of the waveplate except for the propagation time delay due to the refractive index. However, when the polarization direction of the incident light is at an angle to the optic axis, the components parallel to the x- and z-axes travel at difference velocities due to the refractive index difference. Therefore, after passing through the waveplate, a phase difference exists between these two components, and the resulting polarization of the output beam depends on the phase difference. The phase difference can be expressed as
where δ is the wavelength of the light, Δn the refractive index difference between the ordinary and extraordinary refractive indices, and t the thickness of the crystal.
When the thickness of the crystal is selected such that the phase difference equals to m • (π/2) (quarter of the wave), the waveplate is called a quarter-waveplate, and similarly the phase difference in a half-waveplate is m • π (where m is called the order of the waveplate, and is an integer and odd number).
The quarter-waveplate is best known for converting a linearly polarized light into a circularly polarized light or vice versa, when a light beam is passed through the quarter-waveplate with the polarization direction at 45° to the optic axis (see the following figure). The half-waveplate is used most frequently to rotate the polarization direction of a linearly polarized light.
When a linearly polarized light beam is launched into a half-waveplate with an angle θ against the optic axis of the waveplate, the polarization direction of the output beam is rotated and the rotation angle equals to 2θ (see the above figure (b)). Crystal quartz is widely used for making waveplates, due to its small birefringence.
4. Beam Displacer
Another commonly used form of the birefringent crystal is the beam displacer, which is used to split an incoming beam into two beams with orthogonal polarization states, the intensity of each beam dependent on the polarization direction of the incoming beam.
The birefringent crystal-based beam displacer is made by cutting a birefringent crystal in a specific orientation such that the optic axis of the crystal is in a plane parallel to the propagation direction and having an angle α to the propagation direction (see the above figure). The separation d between the two output beams depends on the thickness of the crystal and the angle between the optic axis and the propagation direction, and can be expressed as
where t is the thickness of the crystal.
The optic axis angle to yield a maximum separation is given as
Rutile, calcite, and YVO4 are common birefringent materials for the beam displacer due to their large birefringence (Δn of more than 0.2 at 1550 nm wavelength). For rutile crystal, the ne and no at a wavelength of 1550 nm are 2.709 and 2.453, respectively, resulting in the αmax of 47.8°. For YV04 crystal, the ne and no at a wavelength of 1550nm are 2.149 and 1.945, respectively. Calcite is rarely used in optical circulators due to its softness and instability in a damp heat environment.
>> How Optical Circulator Works
Optical circulators can be divided into two categories.
- polarization-dependent optical circulator, which is only functional for a light with a particular polarization state. The polarization-dependent circulators are only used in limited applications such as free-space communications between satellites, and optical sensing.
- polarization-independent optical circulator, which is functional independent of the polarization state of a light. It is known that the state of polarization of a light is not maintained and varies during the propagation in a standard optical fiber due to the birefringence caused by the imperfection of the fiber. Therefore, the majority of optical circulators used in fiber optic communication systems are designed for polarization-independent operation.
Optical circulators can be divided into two groups based on their functionality.
- Full circulator, in which light passes through all ports in a complete circle (i.e., light from the last port is transmitted back to the first port). In the case of a full three-port circulator, light passes through from port 1 to port 2, port 2 to port 3, and port 3 back to port 1.
- Quasi-circulator, in which light passes through all ports sequentially but light from the last port is lost and cannot be transmitted back to the first port. In a quasi-three-port circulator, light passes through from port 1 to port 2 and port 2 to port 3, but any light from port 3 is lost and cannot be propagated back to port 1. In most applications only a quasi-circulator is required.
The operation of optical circulators is based on two main principles.
- Polarization splitting and recombining together with nonreciprocal polarization rotation.
- Asymmetric field conversion with nonreciprocal phase shift.
We will explain both designs in detail.
1. Nonreciprocal Polarization Rotation-based Circulators
A. Early Development
Dielectric coatings-based polarization beam splitters were used to construct optical circulators in the early stage of circulator development. A schematic diagram of a 4-port circulator is shown in the figure above, where two dielectric coating-based polarization beam splitter cubes were used to split the incoming beam into two beams with orthogonal polarization.
In operation, a light beam launched into port 1 is split into two beams by the polarization beam splitter that transmits the light with horizontal polarization (along the y-axis) and reflects the light with vertical polarization (along the x-axis). The two beams are then passed through a half-waveplate and a Faraday rotator. The optic axis of the half-waveplate is arranged at 22.5° to the x-axis so that the vertically polarized light is rotated by +45°. The thickness of the Faraday rotator is selected for providing 45°-polarization rotation and the rotation direction is selected to be counter-clockwise when light propagates along the z-axis direction.
Therefore, the polarization of the two beams is unchanged after passing through the half-waveplate and Faraday rotator because the polarization rotation introduced by the half-waveplate (+45°) is cancelled by that of the Faraday rotator (-45°). The two beams are recombined by the second polarization splitter and coupled into port 2.
Similarly, when a light beam is launched into port 2, it is split into two beams with orthogonal polarization by the second polarization beam splitter. Due to the non-reciprocal rotation of the Faraday rotator, in this direction the polarization rotations introduced by both the half-waveplate and Faraday rotator are in the same direction, resulting in a total rotation of 90°. Therefore, the two beams are combined by the first polarization splitter in a direction orthogonal to port 1 and coupled into port 3. The operation from port 3 to port 4 is the same as that from port 1 to port 2.
B. Current Development
However, the isolation of this type of optical circulator was relatively low due to limited extinction ratio (around 20 dB) of the polarization beam splitters. Various designs using birefringent crystals have been proposed to increase the isolation by utilizing the high extinction ratio property of the crystal.
One of the designs is shown in the figure below (a), where birefringent beam displacers are used for splitting and combining of the orthogonally polarized light beams. As shown in the figure blow (b), where each circle indicates the beam position and the arrow inside the circle indicates the polarization direction of the beam.
A light beam launched into port 1 is split into two beams with orthogonal polarization states along the y-axis. Two half-waveplates, one (upper) with its optic axis oriented at 22.5° and the other (lower) at -22.5°, are used to rotate the two beams so that their polarization direction becomes the same. The Faraday rotator rotates the polarization of both beams 45° counter-clockwise, and the two beams are vertically polarized (along the y-axis) and passed through the second birefringent crystal without any spatial position change because the polarization directions of the two beams match the ordinary ray direction of the crystal. After passing through another Faraday rotator and half-waveplate set, the two beams are recombined by the third birefringent crystal, which is identical to the first one.
Similarly, a light beam launched into port 2 is split into two beams and passed through the half-waveplate set and the Faraday rotator. Due to the nonreciprocal rotation of the Faraday rotator, the two beams become horizontally polarized (along the x-axis), and are therefore spatially shifted along the x-axis by the second birefringent crystal because they match the extraordinary ray direction of the crystal. The two beams are recombined by the first crystal at a location different from port 1 after passing through the Faraday rotator and half-waveplates. The distance between port 1 and port 3 is determined by the length of the second birefringent crystal.
The use of the birefringent crystals generally results in an increase in size and cost of the circulator due to the cost of crystal fabrication. Extensive development efforts have been concentrated on improvement of various designs. Due to the performance advantages, currently all commercially deployed optical circulators are based on the use of birefringent crystals.
2. Asymmetric Field Conversion-based Circulators
An optical circulator can be constructed using two-beam interference with nonreciprocal phase shifting without the need for polarization beam splitting. One example of this kind of optical circulator is shown in the figure below, where a four-port circulator is constructed using two power splitters and nonreciprocal phase shifters.
In operation, a light beam launched into port 1 is split into two beams with equal intensity by the first power splitter. The two beams are then passed through two sets of phase-shifting elements (half-waveplate and Faraday rotator) that are selected such that they provide no phase shift between the two beams in one direction, but in the reverse direction a phase shift of π is introduced between the two beams. Therefore, from port 1 to port 2 the two beams are in phase and will be constructively recombined by the second splitter and coupled into port 2.
Similarly, a light beam launched into port 2 is split into two beams by the second splitter, and passed through the phase shifter set. Because a phase shift of π is introduced this time, the two beams are out of phase and no longer will be coupled into port 1 but will be coupled into port 3 due to the out-of-phase relation between port 1 and port 3.
The structure of this type of circulator is very simple and potentially could lead to lower cost. However, because phase information is used for the circulator function, control of the phase in each element and control of path length difference between beams are very critical for the performance.
Currently, circulators based on this principle have only been investigated in waveguide devices and no commercial products are available due to manufacturing challenges and performance disadvantages.
>> Newer Optical Circulator Designs to Reduce the Use of Materials and Size
Cost and stability have been the main limiting factors in expanding the applications of optical circulators. Recently, several designs have been developed in an effort to reduce the cost and realize high reliability. In the design shown in the second figure above, the circulator is used in a collimated beam and each port is collimated using a lens; therefore, relatively large size elements have to be used in order to construct the design due to the beam size. In recent designs, efforts have been concentrated on reducing the use of materials and size.
1. Circulator Design Using Diverging Beam
A compact low-cost circulator design has been proposed, placing optical elements in a diverging beam instead of in a collimating beam to reduce the overall use of expensive materials.
As shown in the figure below, in this design, all optical elements are placed in a diverging beam between the input/output ports and lenses. Two identical groups of elements are placed near the focal point of the lens, resulting in reduced size and manufacturing complexity. Each group of elements consists of two birefringent crystals, one Faraday rotator with 45° rotation angle, and two half-waveplates with their optic axes oriented in opposite directions (22.5° and -22.5°).
In operation, a light beam from port 1 is split into two orthogonally polarized beams in the y-axis by the first birefringent crystal. The two half-waveplates and the Faraday rotator are arranged such that after passing through the rotators the polarization directions of the two beams are the same and match the ordinary ray direction of the second birefringent crystal. Therefore, the two beams pass through the second birefringent crystal without any displacement. Two lenses are used for providing a one-to-one imaging system. Because the second group of the element is the same as the first one, the two beams are recombined and launched into port 2.
Similarly, a light beam launched into port 2 is split and passed through to the rotators. Due to the nonreciprocal rotation of the Faraday rotator, the polarization directions of the two beams are rotated matching the extraordinary ray direction of the second birefringent crystal. Therefore, the two beams are shifted a certain amount along the x-axis and shifted again the same amount by the second birefringent crystal in the other group. If the sum of beam shifting by the two birefringent crystals is designed such that it is the same as the distance between the first and third ports, the two beams will be recombined and coupled into port 3.
Because port 1 and port 3 share a single lens and the beam shifting is done at the diverging beam, the required beam shifting in this case is very small and typically equal to the fiber diameter of 125 μm. On the other hand, the required beam shifting in the design shown previously is determined by the diameter of a lens due to the use of collimated beams and is typically in the order of millimeters.
To further reduce the required thickness of the birefringent crystal, mode-field diameter of the input and output fiber is expanded to reduce the divergence angle of the beam. With this compact design, a circulator with a size of 5.5 mm in diameter and less than 60 mm in length has been developed, as shown in the figure below, compared to a typical size of over 25 mm in cross-section and over 90 mm in length for the design shown in previous figure.
2. Circulator Design Using Beam Deflection
A compact circulator using collimated beam deflection is also proposed and demonstrated. In the design, polarization-dependent angle deflection is used instead of the polarization-dependent position shift. As shown in the figure below, a single lens is used to collimate the light for both port 1 and 3 and all elements of the circulator are positioned in the collimated beam. The main difference is that a Wollaston prism is used in place of a birefringent beam displacer and a single lens is used for collimating two beams.
In operation, a light beam launched into port 1 is collimated and split into two beams with orthogonal polarization by the first birefringent crystal. The polarization directions of the two beams are rotated by the half-waveplates and Faraday rotator so that they become the same. Because port 1 is off-axis of the lens, the resulting collimated beam from the lens forms an angle θ to the propagation axis. This angle is corrected by the Wollaston prism and the two beams are propagated straight to the second Faraday rotator (solid lines). After passing through the half-waveplates and being recombined by the third birefringent crystal, the combined beam is focused by the second lens into port 2.
Similarly, light launched into port 2 is collimated and split into two beams with their polarization direction rotated. Due to the nonreciprocal rotation of the Faraday rotator, the two beams from port 2 are deflected to a direction opposite to the angle θ by the Wollaston prism (dotted lines). Therefore, after passing through the polarization rotators and the first birefringent crystal, the combined beam is focused by the first lens to a position different from that of port 1. The required deflecting angle of the Wollaston prism can be determined by the position distance between port 1 and port 3 and the focal length of the lens.
This design reduces the size of materials considerably. However, because the beam splitting and recombining is still performed in the collimated beam, it still requires relatively long crystals.
3. Reflective Optical Circulators
As shown in design examples described in this section, most optical circulators have a symmetric structure in terms of element materials and their relative positions. Therefore, a proposed design concept using imaging folding to redirect the light beam and reuse the common elements has advantages in reducing the overall device size and cost. A schematic diagram of one of the compact reflective circulator designs is shown in the figure below, where a single lens and a mirror are used to couple lights between all ports that are at the same side of the circulator. In this design, all elements are passed through twice to reduce the element account to half while maintaining the same performance as a conventional circulator.
In operation, a light beam launched into port 1 is split into two beams by the first birefringent crystal, and passed through the second crystal without any lateral position change, because the rotation angles of the polarization rotators (+45° or -45° rotation) are designed such that the polarization directions of the two beams match the ordinary ray direction of the second birefringent crystal. After being collimated by the lens and reflected by the mirror, the two beams are passed through the same elements again except for half-waveplates and recombined into port 2.
Similarly, a light beam launched into port 2 is split into two beams with orthogonal polarization directions. After passing through the polarization rotators, the polarization directions of both beams are aligned with the extraordinary ray direction of the second birefringent crystal due to the nonreciprocal rotation of the Faraday rotator, and the physical locations of the two beams are shifted after passing through the crystal. The two beams receive the location shift again after being reflected by the mirror and passed through the crystal. Therefore, after the proper polarization rotation the two beams are recombined at a location different from port 1 and will be coupled into port 3 if the distance between port 1 and port 3 matches two times the beam shift introduced by the second birefringent crystal. Multi-port circulators can be made by adding more ports into the design.
With the reflective design, the size and required optical elements can be significantly reduced, resulting in overall cost savings.
There are many variations in the circulator design, however, all nonreciprocal polarization rotation-based designs share a common structure with a minimum of three functional elements; polarization splitting and recombining elements, nonreciprocal polarization rotation elements, and polarization-dependent beam steering (angular or positional) elements.
>> Applications of Optical Circulators
Optical circulators were originally used in telecommunication systems for increasing transmission capacity of existing networks. By using optical circulators in a bi-directional transmission system, the transmission capacity of the network can be easily doubled without the need for deploying additional fibers, which has become increasingly expensive.
However, with the rapid advancement in optical communication technologies and the readily availability of low-cost and high-performance circulators, the applications of optical circulators have drastically expanded into not only the telecommunication industries but also the sensing and imaging fields. Optical circulators have become an especially important element in advanced optical networks such as DWDM networks.
In the traditional bi-directional optical communication system, a 50/50 (3 dB) coupler, which splits a light beam into two beams with equal intensity, was used to couple the transmitters and receivers as shown in the following figure (a). However, there are two main problems with this kind of structure. One is the need for an optical isolator in the transmitters to prevent light crosstalk between the transmitters, and the other is the high insertion loss associated with the use of the 50/50 coupler, because two couplers have to be used and each has a minimum loss of 3 dB, which results in a minimum 6 dB reduction of the link budget from the system.
The use of an optical circulator can solve both of the problems by providing the isolation function as well as a loss of less than 3 dB as shown in the following figure (b).
Optical circulators are powerful devices for extracting optical signals from a reflective device. Therefore, optical circulators are often used in conjunction with the fiber Bragg gratings that are typically reflective devices. Together with fiber Bragg gratings, optical circulators have become one of the indispensable elements in advanced DWDM optical networks. Circulators are used as MUX/DEMUX devices, but are also used with the fiber Bragg grating in dispersion compensation, tunable optical Add/Drop, and other applications.
Another application of the circulators is use with a mirror for double passing an optical element to increase efficiency. One example is the reflective erbium-doped fiber amplifier shown in the figure below. In operation, signal light is launched into port 1 of a circulator and passed through port 2 with minimum loss. The signal is combined with the pump light from a pump laser by a WDM coupler, and both lights are launched into an erbium-doped fiber. The amplified signal and residual pump lights are reflected by the mirror and passed through the erbium-doped fiber again so that the signal is amplified twice by the erbium-doped fiber, reducing the required length of the fiber, and the residual pump power is also re-used to increase the pump efficiency.
The idea has been adapted into different devices, such as replacing the coupler and erbium-doped fiber with a dispersion compensation fiber to reduce the required fiber length and adding a Faraday rotator between the mirror and fiber to reduce the polarization-induced effects. Bi-directional fiber amplifiers are also proposed for taking full advantage of the circulator.
With the development of advanced optical networks, applications of optical circulators are expanding rapidly and new functionality and applications are emerging quickly. For example, recently it has been reported that by adding wavelength-selective functions into circulators, a bi-directional wavelength-dependent circulator can be configured, which opens a new dimension of applications in advanced DWDM optical networks.