Multi-core Fibers

Author: the photonics expert Dr. Rüdiger Paschotta

Most optical fibers have a single fiber core, which is usually located on the fiber axis. However, there are also specialty fibers containing multiple cores, which may e.g. be arranged on a ring around the fiber axis or on some 2D grid. (For example, a seven-core fiber may have six cores on the edges of the hexagon and a central core in addition – see Figure 1.) Such fibers are called multi-core fibers (or multicore fibers). In the case with two cores only, one may also use the term dual-core fiber.

Fabrication

Multi-core fibers can be realized with all-glass fiber technology or alternatively as photonic crystal fibers containing air holes. In the first case, one may fabricate an all-glass preform which contains multiple cores, or combine multiple single-corporate forms to a branch from which the fiber is drawn (bunch fiber). For photonic crystal fibers, no special extension of the fabrication technology is required; one can simply assemble a more complex bundle of rods and/or tubes to obtain a preform with multiple cores.

Guiding in Multiple Cores

In principle, each of the fiber cores in such a fiber can act as a separate waveguide, so that light can independently propagate through those cores. However, there can be some mode coupling between the cores (see Figure 2) if the distance between two cores is so small that the corresponding mode fields have a significant spatial overlap. This means that light which is initially coupled into one core can eventually couple over to other cores; that effect is similar as in fused fiber couplers.

For such a situation, one can compute so-called supermodes, i.e., field configurations which are stationary despite the coupling. Note, however, that supermodes calculated in mode coupling theory for an idealized situation may not be the true modes of a fiber subject to random fluctuations in fabrication and/or due to operation conditions. Light propagation in such fibers can also be investigated with methods of numerical beam propagation, where one may also take into account random fluctuations.

Although in some cases the mentioned kind of coupling is desired, in many others it is avoided or minimized by using large enough spacings between the cores. For example, one often wants to avoid substantial crosstalk in optical fiber communication systems. Often, the fiber cores have rather small diameters of a few micrometers, so that single-mode propagation is obtained in some wavelength range. A distance of the order of 30 to 40 μm between each pair of cores may then be sufficient to avoid significant coupling even within kilometers of fiber. The usual fiber diameter of 125 μm then allows only for a quiet limited number of cores in such weakly-coupled fibers. There is obviously a trade-off between a high core density and low crosstalk e.g. in telecom systems. That trade-off can be mitigated in various ways, e.g. by reducing the coupling by stronger mode confinement (e.g. with refractive index trenches or air holes around the cores), or with heterogeneous designs where the different core modes have different effective refractive indices. In principle, one can also increase the fiber diameter, but that is often not practical because the sensitivity of the fiber to bending is then increased.

Note that the coupling between fiber cores will generally critically depend on the exact distance. This means that it can be affected even by tiny fabrication tolerances.

Twisted Multi-core Fibers

Some multi-core fibers obtain a twist. This is achieved by rotating the fiber around its axis during the pulling from the fiber preform. The twisting can provide benefits in different application areas:

• One can make fiber-optic sensors which can measure twisting (torsion). The built-in twist (from fiber fabrication) causes mechanical strain, and that will be increased or decreased by additional strain applied to the sensor, depending on the direction of the applied torsion. The strain effects can be measured interferometrically.
• In telecom applications with space division multiplexing (see below), twisting can enhance mode coupling, which can be beneficial for signal transmission.

Tapered Multi-core Fibers

In some cases, multi-core fibers are tapered. Each fiber core will then be subject to the same taper ratio. However, as at most one core can be on the fiber axis, the others will also experience a lateral shift, which implies bending in the taper region. That may lead to additional bend losses. Another problem may be unwanted coupling of light between the cores in the region with small fiber diameter. One may need to optimize the taper design in order to minimize such effects – for example, based on numerical simulations of flight propagation in such devices.

Telecom Applications

In optical fiber communications, there is a long-term trend towards more and more expanding the transmission capacities, as data traffic keeps growing at a large rate. Obviously, there is thus an interest in maximizing the transmission capacity per fiber, and one of the technological options is using multiple cores in one fiber, so that multiple signals can be simultaneously transmitted with spatial separation (space division multiplexing, SDM). That principle can also be realized with few-mode fibers or multimode fibers, but multi-core fibers allow for realization with much weaker coupling between the channels. The two approaches may even be combined when using multiple multimode cores, which can result in a larger number of transmission channels.

When using fibers with negligibly weak coupling between the cores over the full transmission distance, the system can be conceptually simpler. However, fibers with relatively strong mode coupling can also be employed, using techniques of multiple-input multiple-output (MIMO) digital signal processing. In the latter case, the spacing between the cores can be much smaller, so that more fiber cores can be placed in a single fiber, and the overall transmission capacity can be higher. The same kind of techniques can be used in conjunction with few-mode fibers and multimode fibers. Compared with those, single-mode multi-core fibers have the advantage that the spread of group velocities is much smaller, which allows the use of MIMO receiver with smaller complexity.

A substantial technical challenge for the industrial use of multi-core fibers is the need to couple light for multiple signal channels into the different cores of the fiber, and to handle outputs from multiple cores. Suitable coupler devices also have to satisfy a number of practical requirements. One of the proposed solutions is to use laser-inscribed 3D waveguides in a small glass block, which connect different cores of a MCF with the cores of a set of output or input fibers which are arranged in a linear sequence [3].

Another challenge is that splicing of multi-core fibers is obviously more difficult than for ordinary single-core fibers: the cores need to be carefully aligned.

For long transmission distances, fiber amplifiers are often employed. Special erbium-doped fiber amplifiers for multi-core fibers have been developed, where simultaneous amplification for all the cores is achieved, in some cases even using only a single pump source. Further research and development of such multi-core EDFAs [6] is required, paying attention to technical issues like core-dependent gain. There are realizations based on cladding-pumped active fibers where only one or two pumped lasers are required and the differential model gain can still be quite limited [17].

The practical difficulties for introducing multi-core fiber telecom systems are substantial, also concerning cost; it is therefore not clear how far this development will go. It is also considered just to use multiple fibers, or a larger number of such fibers within one fiber cable. As a simpler alternative solution, it is relatively easy to further increase the number of fibers, for example by using thinner glass fibers with thinner polymer coatings.

Other Applications

Multi-core fibers can also be utilized for other (non-telecom) applications. An example is the area of fiber-optic sensors. Possible operation principles can exploit the sensitivity of coupling between multiple cores to external influences such as strain or temperature changes. Such sensors may be realized in the form of interferometers involving light passes through different cores.

Another example is the application of multi-core fibers in high-power fiber lasers and amplifiers. In order to mitigate nonlinear effects, one distributes the optical power over the different fiber cores and may at the end recombine the light (→ coherent beam combining). The relative optical phases of the light which is propagated through the different fiber cores must be carefully controlled. This, however, is less challenging than when using separate fibers because externally induced phase changes in the different cores are normally quite similar, and only their differences count for beam combining.

Simulation of Light Propagation in Multi-core Fibers

It is possible to simulate the propagation of light in multi-core fibers. With such simulations, one can address many questions, for example of the following types:

• How strong is the optical coupling (crosstalk) between different cores? How does that coupling depend on design parameters such as the distance between the cores and their refractive index profiles?
• How sensitive is the design to unavoidable imperfections, for example fluctuations of the core positions and diameters, or to bend-induced effects?
• Could nonlinear effects substantially impact the light propagation?
• How would the observed effects influence the transmission of data signals, or the operation of a fiber-optic sensor?
• How to optimize the fiber design accordingly, taking into account both coupling phenomena and other aspects such as favorable mode areas and chromatic dispersion? What are the required fabrication tolerances?
• How well does the coupling of light between different multi-core fibers work, for example in terms of insertion loss?

Different numerical methods can be used, for example the following:

• One can use numerical beam propagation, starting with a certain input amplitude profile (for example, with input beams focused to one or several of the fiber cores). One can then investigate, for example, how light is coupled into the other cores, or potentially stays in the cladding. The optical powers in the guided modes of all cores can be calculated at any longitudinal position by evaluating all their intercourse between the amount functions and the calculated amplitude profile.
• With such methods, it is also easily possible to introduce additional influences such as imperfections of the refractive index profile (e.g., random changes of the distance between different cores) and arbitrary bending of the fiber.
• A disadvantage is that the computation times can be considerable if the fields need to be propagated over substantial distances.
• Note that beam propagation is often done for monochromatic light. Simulations for multiple optical frequencies may be combined for a more comprehensive analysis, e.g. concerning the often pronounced wavelength dependence of coupling effects.
• A numerically far more efficient approach is to simulate mode coupling with a set of coupled differential equations for the mode amplitudes.
• However, such methods are more difficult to implement and validate, particularly if things like random changes of the refractive index profile must be introduced.

The proper choice of method will depend on the circumstances, in particular on the exact questions to be answered by the simulations.

More to Learn

Encyclopedia articles:

• fibers
• fiber core
• few-mode fibers
• specialty fibers
• fiber amplifiers