Specialty Single-Mode Fibers

This is a continuation from the previous tutorial - polarizing beam-splitter prisms.

 

1. Introduction

The last two decades of the twentieth century saw an immense increase in the number of applications in which optical fibers of different kinds were used. The explosive development of fiber optics for communication was a major driving factor behind this progress, providing the necessary tools and enabling technologies.

Contrary to popular belief, however, the development of the optical fiber was not initially driven by the needs of the communications sector but has a much longer history.

For instance, the ‘‘controlled’’ guiding of light in a transparent water jet was first described in 1841. The technology subsequently found use for the spectacular illumination of water fountains at the great exhibitions of the late nineteenth century.

A short notice in The Lancet issue of 1889 described the use of a glass light guide for purposes of medical examination, and the 1920s and 1930s saw the development of the concept of fiber bundles for image transfer in medical and other applications.

These were the driving forces behind such important inventions as, for example, the glass clad fiber, which gained commercial success in medical endoscopes as well as faceplates for image intensifier devices in the early 1960s.

Up to this time, optical fibers had been of the multimode type. The single-mode optical fiber was discovered and described about the same time.

With the invention of the laser and the techniques for making low-loss fibers in the early 1970s, made available the basic prerequisites for efficient fiber optic communication.

The enormous inherent technical and economical advantages of fiber optic communication spurred a huge global research and development (R&D) effort into commercializing such systems and large-scale deployment started in the late 1970s—at first with multimode fibers, but from the beginning of the 1980s onwards with increasing deployment of single-mode fibers to meet the requirements of high-speed long-haul communications.

With an in-depth understanding of the special properties of single-mode fibers and the tools available for manufacturing them, the field was open for extending their application beyond the one of pure light transport.

Specialty single-mode fibers have, therefore, been developed to incorporate several functionalities (i.e., multifunctional fiber), and today, they find important uses in a diverse range of applications such as optical amplifiers and lasers, sensors, signal restoration, and optical filtering.

To illustrate this diversity, we devote this tutorial to components based on some unusual multifunctional, specialty single-mode fibers, namely macrohole fibers, multicore fibers, fibers with internal electrodes, and fibers for high-temperature–resistant fiber Bragg gratings.

This set of examples is by no means exhaustive, but it is a broad set of the most common specialty SM fibers used in the past.

 

2. Macrohole Fiber

Macrohole fibers belong to the group of microstructured fibers, which encompass a wide variety of fibers with air holes or other structures extending in the axial direction.

There are several classes of microstructured fibers, and the terminology is not standardized. Here, we choose to categorize the fibers by their physical appearance and use the terms macrohole and microhole fibers, where the macroholes typically are several times larger than the wavelength of light.

Microhole fibers, commonly called ‘‘holey’’ or ‘‘photonic crystal fibers’’ (PCFs), are treated in detail in a later tutorial.

While silica PCFs are typically fabricated by the stack-and-draw method, which is a quite labor-intensive and complex procedure, the manufacturing of a macrohole fiber can be significantly less demanding.

In a simple case, it involves three steps: manufacturing of the starting preform, preparation of the required holes and structures in the preform, and drawing of the fiber. The preform manufacturing does not differ from other fiber types.

The holes are machined into the preform using high-precision diamond drills, laser ablation, or ultrasonic tools. Hole sizes and positions depend on the desired structure of the final fiber, a typical hole is 3 mm in diameter and machined into a 25-mm diameter preform. Features such as grooves or flats are machined onto the preform.

The processed preform is carefully cleaned from any residual debris and contamination and can be further stretched and sleeved if necessary to meet the geometrical requirements.

Finally, the preform is mounted in the fiber drawing tower, and drawn under accurate control of the drawing conditions. The holes are carefully pressurized using dry, inert gas during drawing, to maintain the desired geometry of the fiber.

The manufacturing procedures allow for flexibility in the fiber design, in terms of shape, size and position of holes, material composition, and size of the fiber core. The use of two or more cores may also be advantageous. A choice of protective fiber coatings can be applied in the drawing, typically acrylates or polyimides.

Figure 6.1 shows two examples of macrohole fibers manufactured using the described technique.

 

 

The fiber to the left was designed for use in all-fiber electrooptic devices. The core has a high numerical aperture, to allow for tight confinement of the mode, and the positions of the holes are chosen so that one hole is further away from the core than the other. To prepare such components, the holes are metal-filled in a post-processing procedure described later in this tutorial.

The second image depicts a one-hole fiber, with an additional groove machined on the outside of the fiber. The groove allows for simple alignment of the fiber in post-processing procedures.

There are a large number of proposed and published component applications for macrohole fibers. Although these fibers in some cases compete with PCFs, the macrohole fibers have two unique properties:

• The relatively large size of the holes enables efficient introduction of various materials into the holes. These materials can be used for interaction with the cladding modes or evanescent field of the guided mode of the light or to perform active functions.
• The larger structures make the fibers easier to manufacture, compared to PCFs, and they are, therefore, attractive to cost-sensitive applications. However, unique guiding properties, such as those enabled by photonic band-gap structures, are not possible to implement in these fibers.

Functions performed using macrostructured fibers include supercontinuum generation in tapered hole fibers, dispersion management, fibers with decreased bend loss for compact optical fiber wiring, and fibers for polarimetric sensing and lasers; this latter one is shown in Fig. 6.2.

 

 

The introduction of materials in the holes adds the attractive possibility to manipulate the guiding properties of the fiber. This can be achieved by interaction of the guided mode with actively controllable materials in the holes or by using the inserted materials for other active functions, such as a metal electrode to implement electro-optic control of the fiber.

The holes in macrostructured fibers can be filled with different materials with relative ease. The materials are introduced into the holes by pressure. Short sections can be filled using capillary forces only.

In many cases, the force achievable by mere vacuum pumping (<1 bar) is sufficient to fill a long enough section of the fiber. If a larger force is needed, materials are dispensed from a pressurized vessel.

In this manner, fibers can be filled with gases and liquids. Solids, such as metals or polymers, are introduced to the holes in a molten or a precursor state and post-solidified. The filling procedure is described in some detail later in this tutorial.

 

2.1 Microfluidic Devices

Eggleton et al. introduced a class of hybrid devices using macrostructured fibers with movable plugs of fluids. The fiber, commonly called the ‘‘grapefruit’’ fiber, has six large air holes forming a circular inner cladding with a diameter of approximately 34 μm, around a single-mode germanium-doped core.

The holes are placed far from the single-mode core so there is no significant interaction of the fundamental mode with the material in the holes in the unaltered state.

The interaction is achieved either by tapering the entire structure so that the guided mode expands into the holes or by using long period gratings (LPGs) to couple light between core and cladding.

A liquid plug is inserted into the holes, and the holes are sealed by splicing to a standard fiber on both ends. The structure is hence an air-hole fiber with a short section of the holes filled with a liquid plug.

On-fiber heaters are used to thermally control the air-filled sections of the hole fiber. When heated, the air expands and displaces the fluid plug. In this manner, the position of the fluid plug along the hole-fiber device can be controlled.

The tapered devices are formed by heating and stretching a section of the fiber to an outer diameter of approximately 50 μm, taking care that the holes do not collapse, and that the transition is adiabatic. In the tapered section, the mode is guided by total internal refraction on the holes.

When the liquid plug, in this case methylene iodide, which has an index much higher than silica, is positioned in the tapered section, the mode leaks into the liquid and is attenuated. A thermally controlled variable optical attenuator (VOA) with an extinction of 45 dB is demonstrated.

By selectively filling only one of the holes with a movable liquid plug, an adjustable polarizer is described. A more simple VOA using a solid UV-cured acrylate in the tapered section was demonstrated, by thermally controlling the refractive index of the polymer.

A series of low-index fluid plugs are inserted into the holes, acting as a LPG in the tapered section. By thermally expanding air in the holes, the fluid LPG is compressed, changing the period of the LPG, hence, changing the resonance peak of the LPG filter.

In an untapered device, the interaction between core and cladding is performed by inscribed LPGs in the core of the fiber. The loss spectrum of the LPG is determined by the propagation properties of the core and cladding modes. By displacing the high- or low-index liquid plug over the LPG, the filter properties of the LPG are altered. A similar approach, using a solid UV-curable polymer.

 

3. Fibers with Internal Electrodes

The explosive growth of the field of microstructured fibers has been accompanied by the development of devices based on the insertion of various materials in the holes running parallel to the fiber core as described earlier.

Besides applications with liquids and gases, new applications of fibers with internal electrodes are emerging. With a long electrode running parallel to the fiber length, one can subject the core of the fiber to a very strong electric field, because the isolation capability of silica is excellent (typical practical breakdown field \(\gt3\times10^8\text{ V/m}\)). Continuous electrodes of tens up to hundreds of meters in length have been reported.

Applications of fibers with electrodes include active control of the refractive index through the electro-optical effect, control of the fiber birefringence through the passage of current in the electrode, and ‘‘poling,’’ a process after which the fiber gains an effective second order nonlinearity.

Lithography of electrodes in D-shaped and twin-hole fibers has also been reported to produce a periodic structure for quasi-phase matching. The sections to follow review these devices.

 

3.1 Electrode Incorporation

In the early 1980s the PANDA fiber was invented, where a pair of stress elements were inserted in the cladding on opposite sides of the core introducing birefringence, which results in polarization maintenance.

Soon after, a similar fiber was designed where the space occupied by the stress elements was instead left open, resulting in a ‘‘side-hole’’ or ‘‘hollow-section’’ fiber, which found applications in pressure sensing.

Based on such geometry, in 1986 Luksun Li et al. described a technique to fabricate long ‘‘internal electrodes.’’ In that pioneering work, they used a fiber with two holes.

Aided by a few atmospheres of overpressure, they pumped an indium/gallium alloy—a liquid metal—into the holes. Metal-filled pieces as long as 30 m were reported, and the authors also mentioned preliminary attempts to directly draw a preform incorporating a metal. Such fibers were then used for Kerr modulation, although many practical aspects of that experiment are unavailable.

The same group also showed that a metal-filled fiber (BiSn alloy, in this case) could be used to polarize light, because the optical loss for the TE and TM polarizations can differ by more than 40 dB for a few centimeters-long device. This early work was, however, discontinued and the techniques presented were left unexploited for many years.

For almost a decade, the use of fibers with electrodes laid dormant, until the discovery in 1991 of the possibility of inducing second-order nonlinearity in silica glasses by thermal poling.

By subjecting a silica disk—and as later shown— an optical fiber to a high-voltage bias (~ 4 kV) at a temperature in the neighborhood of 280°C, the displacement of cations led to the creation of a permanent strong electric field distribution in the sample. This paved the way to making fiber components such as Pockel’s cells and frequency doublers.

The second-order nonlinearity induced by poling is in general confined to a thin layer of approximately 10-μm thickness adjacent to the anode electrode, and therefore, the core of the fiber has ideally to be inside that region (i.e., very close to the electrode).

Furthermore, the need for a high-voltage bias during poling and the relatively weak nonlinearity induced favored poling of long pieces of fiber with the electrodes inside the glass to prevent electrical breakdown of the material.

The internal electrode configuration was the natural choice. An important improvement to the arrangement was demonstrated by researchers at Sydney University, who by side-polishing the fiber could gain access to the metal and still be able to splice the active fiber from both ends.

A simple but time-consuming technique of manually inserting a thin metal wire into the holes was used by the various groups working in the field. Wire insertion requires skill, is time consuming (i.e., costly), and is not appropriate for fabrication of long devices (tens of centimeters or more).

Furthermore, the position of the wire in the holes varies along the fiber and from device to device, leading to a non-uniform field distribution, uncertainty in the performance, reproducibility problems, and impedance variation along the device.

Nevertheless, long pieces of fiber have been manufactured with an electrode inserted during drawing. A 200-m long piece of fiber with one internal electrode became available.

For poling, the need for an outer electrode behaving as ground led to the interesting development of electrically conductive coatings. Two types of coatings were reported, one involving polyimide and the other one carbon-loaded acrylate, with better performance in terms of conductivity and uniformity. Metal-filled capillaries for applications in medical microprobing have also been developed, and silica-isolated micrometer-size electrodes are available.

The technique of pumping liquid metal into a ready drawn fiber with holes was redeveloped in 2002, and a schematic of the arrangement used is shown in Fig. 6.3.

 

 

One end of the fiber with holes is inserted into the molten metal and the other end is kept free in the atmosphere. The liquid metal is contained in a small crucible in a sealed cell that is pressurized, forcing the melt into the fiber and filling the holes up to where the fiber is cold (and the metal solidifies).

The entire cross-section of the hole is filled with metal, improving reproducibility. Up to 14 pieces of fiber are filled at a time, so the fabrication becomes quicker and cheaper.

Various types of alloys have been used for fiber components. A euthetic alloy of Bi (43%) and Sn (57%) melts at 137°C, which allows the standard acrylate coating to be intact while filling the holes.

The euthetic Au (80%) and Sn (20%) melts at 282°C and can be inserted in liquid form at 300°C and used as a solid electrode for poling at 260°C.

The time to fill the fiber depends on hole size, type of alloy, temperature, and pressure. Typically, a 1-m long fiber device is filled in less than 1 minute. The technique is compatible with 125-μm fibers (as shown in the insert of Fig. 6.3) and the hole size ranges from about 20 to 40μm.

The internal electrodes are accessed by side-polishing in approximately 1 minute, and the fibers are polished with the primary coating still on. Electrical contact is made with a thin wire or with conductive epoxy.

Splicing the fibers with holes to standard telecom fibers can give losses of approximately 0.1 dB, but the end to be spliced needs to be free of metal. This is achieved by initially leaving approximately a 20-cm long piece of the fiber outside the oven, as shown in Fig. 6.3.

By inserting half this length into the oven and removing the supply of metal, the molten metal column can be displaced about 10 cm further along the fiber, freeing also the other end from metal.

In some cases, it is advantageous to use thin conductive films on the surface of the holes, rather than a solid alloy electrode, because the stress introduced by a thin film is small, and periodic electrodes for quasi-phase matching can be easily fabricated.

Silver film electrodes were deposited on the inside of a two-hole fiber by mixing a silver nitrate solution with a reducer flowing in the holes. To limit the size of the particles, and to prevent clogging of the approximately 30-μm holes, the two solutions were mixed at low temperature (~8°C) to reduce the reaction rate and inserted by high pressure.

The films were usually conducting for fiber lengths more than 2 m, and devices with electrodes longer than 3 m were produced. Such hole-coated fibers could be spliced to standard fibers without further processing.

The thickness of the Ag layer was measured to be in the range approximately 0.1-1.0 μm. Periodic internal electrodes could be fabricated by point-by-point side-exposure to 0.53-μm radiation through the acrylate coating, causing laser ablation.

The fiber was translated and the top electrode ablated, after which the beam and lens were translated by 40 μm along the fiber and a new period recorded. Only the surface of the hole closest to the core of the fiber was exposed to the ablating beam, so that the continuity of the electrode was not jeopardized, the film remaining continuous on the back part of the hole.

A typical pattern produced is shown in Fig. 6.4, with openings approximately 15 μm in diameter and center-to-center separation of approximately 40 μm, close to the beat length of 1064 and 532 nm in standard telecom fibers.

In applications demanding periods of a few microns only, side exposure of a photoresist with lower power light through the acrylate coating and conventional lithography inside the fiber should be possible.

 

 

3.2. Applications

A component for tunable polarization control based on a metal-filled fiber has been developed, exploiting the tight physical contact between the metal alloy and the surface of the fiber hole.

Current is run through the alloy causing heating through Ohmic dissipation. Because of thermal expansion, the heated alloy exerts pressure on the glass, which strains the core asymmetrically, leading to birefringence and a change in the polarization state of the signal in the fiber.

Complete coverage of the Pointcare Sphere can be accomplished with two metal filled fiber components spliced at 45°C. Most applications of internal electrode fibers, however, are related to their use as electro-optical modulators.

In this case, a voltage is applied between the electrodes, causing an electric field to be established across the core and a change in the refractive index and, thus, optical path.

A Mach-Zehnder interferometer comprising an active fiber in one arm and a passive fiber in the other is sufficient means to provide for switching, transforming phase into amplitude modulation.

When the phase changes by \(\pi\) radians, the interference of the signals in the two arms changes from constructive to destructive and light is switched from one output fiber to the other (2 x 2 switching).

Interferometers were built exploiting the Kerr effect in 1-m long pieces of twin-hole fiber prepared with BiSn electrodes. Figure 6.5 illustrates the quadratic dependence of the Kerr switch when driven by voltages as high as 4 kV.

 

 

The switching voltage was a few hundred volts and the phase excursion becomes a more rapidly varying function of the applied voltage as the voltage is increased. For example, with a 3.8-kV DC bias, the required voltage for a \(\pi\) phase shift is approximately 100 V.

If the fiber is first poled at approximately 280°C for a few minutes with high voltage applied between the two electrodes, a large permanent electric field can be recorded across the core and the refractive index gains a linear dependence on the applied voltage.

After poling, tens of volts are sufficient for full switching in the interferometer. Although this is still an order of magnitude larger than for LiNbO₃ modulators, all-fiber devices exhibit potentially lower loss and higher optical power handling capability.

One such switch has been used for video transmission (i.e., as a modulator) and as a 2 x 2 switch for protection of a fiber network operating at 10 Gbps without degradation of the signal quality.

Fiber interferometers are long devices (typically 1-m long arm length) whose transmission function exhibits a sinusoidal wavelength dependence if the optical paths are unequal.

The application of a control voltage signal to the active fiber results in electro-optical tuning of the sinusoidal spectral response. Therefore, electro-optical filtering can be accomplished in an unbalanced Mach-Zehnder interferometer incorporating a poled fiber in one of the arms.

This has been exploited in the construction of a stepwise tunable ring fiber laser. The ring incorporated an Er-doped fiber, a circulator with a sampled fiber Bragg grating and an unbalanced Mach-Zehnder interferometer.

The laser could be made to operate at any of the 16 wavelengths of the sampled grating. The interferometer is tuned to transmit one of the sampled grating’s wavelengths by the control voltage and the laser will oscillate at this frequency since it exhibits the lowest loss.

Both mode-locking and Q-switching operation have also been accomplished with poled fibers with internal electrodes.

Internal electrode fibers have also been exploited for wavelength conversion and in particular frequency doubling. D-fibers were poled with a film electrode defined by lithography on the flat surface of the fiber and an internal wire electrode.

Up to 20% conversion efficiency was accomplished with high-power femtosecond optical pulses. Furthermore, quasi–phase-matched fibers have been constructed by periodically illuminating a continuously poled fiber with UV light selectively erasing the permanent field.

The period was carefully adjusted for phase matching. The highest conversion efficiency demonstrated was 2.5% achieved in an 11.5-cm long device with pump peak power of only 108 W.

Higher conversion efficiency is, however, expected with an improved laser source. By bending the periodically poled fiber, it was possible to achieve 27-nm tunability.

The electrode deposition techniques described here also opens a number of opportunities for steering active media inside the fiber, such as liquid crystals and magnetic powders.

Components based on fibers with internal electrodes are versatile and can be used to perform a number of functions, such as optical switching, wavelength conversion, and active polarization control.

The most attractive feature of fiber components is that they to a large extent inherit the characteristics of standard telecom fibers in terms of low loss, ease of splicing, and competitive price. It is, therefore, likely that we will see a growth in the number of applications of fibers with internal electrodes.

 

4. Multicore Fibers and Components

An optical fiber is generally conceived as consisting of a light-guiding core concentrically positioned in a surrounding cladding structure. The concept of embedding two or more cores in a common cladding structure was, however, launched quite early in the history of single-mode fibers.

Although multicore fibers have since attracted considerable attention and a large variety of applications and designs have been proposed, the commercial success of such fibers has been quite limited.

The proposed applications for multi-core fibers span over lasers and amplifiers, transport fibers for broadband communications, passive and active fiber optic components such as filters, multiplexers, and so on, and various kinds of sensors.

The ultimate multicore fiber is, of course, the image fiber used in endoscopes and other such devices. It may contain tens of thousands of individual cores in a fiber of a diameter of less than 2 mm. These cores are, however, generally not single mode but support a limited number of propagating modes of visible wavelengths.

Multi-core fibers can be divided into two categories. In the first category, the fibers are designed for a controlled coupling of the guided modes between the cores, whereas for the second category, the design is such that the modes in the different cores are practically decoupled.

 

4.1. Coupled Cores

The coupling of energy between the modes guided in the various cores in multi-waveguide structures, including fibers, has been extensively analyzed theoretically.

The most common fiber is the twin-core fiber where two identical single-mode cores, waveguide 1 (WG1) and waveguide 2 (WG2) in Fig. 6.6, are symmetrically located in the fiber cladding.

Illuminating one of the cores will equally excite two transversal modes in the fiber, one symmetrical and one asymmetrical, as indicated in Fig. 6.6, with slightly different propagation constants.

The intensity distribution in the fiber structure is the result of the summation of the two modes: Energy is periodically transferred back and forth between WG1 and WG2 as a result of the beating between the two modes.

The first core will be completely depleted of its energy after a certain length. This energy is now guided in the second core and will start to transfer back to the original core. The process continues ad infinitum as light travels down the fiber.

The fiber length necessary for completion of one cycle is called the beat length and depends on core separation, core geometry, refractive index, and wavelength of the light.

 

 

Figure 6.7 shows a BeamProp simulation of the coupling of light, with a beat length of 32 mm, between two cores 16 μm apart in a fiber.

The coupling of energy between cores is, of course, extensively used in various fiber coupler schemes. The use of coupled multicore fibers has been proposed for many other applications and some typical examples are presented in the following subsection.

 

 

4.1.1 Optical Amplifiers

The possibility to ‘‘split’’ light from one core and propagate it for a certain distance in a parallel core has been exploited in active fibers for use in amplifiers and lasers.

The use of a twin-core erbium-doped fiber for channel gain equalization was proposed early by Zervas et al.. The pump and WDM signals are launched into one of the cores. Since the beat length is wavelength dependent by approximately \(\lambda^3\), the various WDM channels will travel partly different roads down the fiber and thus interact and saturate different subsets of erbium ions.

The gains of the different channels are, thus, spatially decoupled, resulting in inhomogeneous broadening. If the power in one channel increases with respect to the others, the corresponding gain will ultimately decrease, leading towards spectral gain equalization.

A slightly different approach was proposed by Lu and Chu, in which case only one of the cores was doped with erbium ions while the other remains passive. Since the cores are different, with one core being amplifying, there is a nonreciprocity in the coupling between them.

For the C-band amplifier, the pump and signal are launched into the erbium-doped core. The fiber is designed so that at the peak of the erbium gain (1533 nm) the coupling from the active to the passive core is at maximum while the coupling from the passive core is at minimum and subsequently excess energy at 1533 nm is wasted. The net result is an overall flattening of the gain in the C-band.

In a similar manner, a flattened L-band amplifier can be constructed. In this case, the pump and signal are coupled into the passive core. The coupling coefficient from the passive to the active core increases with wavelength, thus compensating for the decrease in the erbium gain.

For Raman amplifiers, Kakkar et al. suggested in a theoretical paper an asymmetrical twin-core fiber with a high NA central core close to a low NA core with a larger radius.

Both cores are single mode and have the same propagation constant at a selected phase-matching wavelength. Signal and pump are launched into the central core. By appropriate fiber design, the pump and signals will be confined to the central core at wavelengths much shorter than the phase-matching wavelength, resulting in a high pump and signal overlap.

As the signal wavelength approaches the phase-matching wavelength, a larger part of the signal’s power will be confined in the second core, decreasing the pump and signal overlap and increasing in the effective mode area (\(A_\text{eff}\)).

As the fiber’s effective Raman gain coefficient (\(\gamma_\text{eff}\)) is related to the material’s Raman gain coefficient (\(g_\text{eff}\)) and the effective area as \(\gamma_\text{eff}=g_\text{eff}/A_\text{eff}\), an increase in \(g_\text{eff}\) can be compensated for by an increase in \(A_\text{eff}\).

 

4.1.2 Fiber Lasers

Winful and Walton proposed the use a of a twin-core fiber for passive mode locking of a fiber laser. The fiber is similar to the ones proposed by Lu and Chu but, in this case, with the active core enclosed in a cavity consisting of a high reflector and an output coupler.

The length of the cavity is half a beat length in the absence of amplification. A low-intensity pulse will completely couple into the passive core where it is lost.

However, the central portion of the high-intensity pulse will induce changes in the active cores index, de-tuning the coupler, while the wings of the pulse will couple o the passive core. The system was expected to operate in the pulsed mode because high-intensity pulses minimize losses.

A similar approach for a ring laser was proposed by Oh et al. and the concept was further elaborated by Martı´-Panamen˜o et al.. Graydon et al. practically demonstrated a triple-frequency erbium ring laser where the individual gains of the different lasing wavelengths are partially decoupled from the others due to the inhomogeneous broadening introduced by the twin-core design.

Other practical implementations of lasers were demonstrated by Kanˇka et al. and Peterka et al. who used erbium-doped dual-core fibers for line narrowing and wavelength stabilization as well as high-speed pulse generation.

Wrage et al. presented a multicore fiber laser array for high power. Eighteen single-mode Nd-doped cores were equally distributed close to the annulus of a multimode fiber carrying the pump modes.

The bad beam quality, usually associated with laser arrays, caused by the different beams interacting incoherently, was overcome by phase locking the 18 resonators in a Talbot cavity.

Yanming et al. presented an alternative solution based on an isometric concept with up to 19 equidistant Yb-doped cores. These cores couple to each other via the cladding through the evanescent fields facilitating in-phase oscillations for all cores. These modes combine to supermodes with 80% of the energy in the lowest order (in-phase) mode with a Gaussian-like appearance.

 

4.1.3 Miscellaneous Applications

Ortega and Dong demonstrated a tuning procedure for adjusting the coupling wavelength of a twin-core fiber. By tapering the fiber, the propagation constants for the two cores can be tuned to coincide at a predetermined wavelength.

At wavelengths on both sides of the coupling wavelength, the mismatch in propagation constants is sufficiently large to suppress coupling and the fiber constitutes an optical filter with a high temperature and mechanical stability.

Another way to tune the coupling was demonstrated by Atkins et al.. Ge-doped cores are inherently photosensitive, that is, the refractive index changes in response to exposure to UV light. By illuminating one of the cores while monitoring the coupling, the index is trimmed until a maximum response is obtained.

An et al. have reported the use of a LPG–assisted twin-core fiber for an add–drop filter. An LPG is written into one of the cores and the photosensitive cladding surrounding the core. The mismatch of the cores prevents coupling while the LPG at the coupling wavelength ensures that the overlap of the two core modes for this wavelength is large enough for efficient coupling to occur.

Finally, Chu and Wu used the large nonlinear effect in a symmetrical erbium-doped twin-core fiber to demonstrate optical switching.

Coupled twin-core fibers have also been proposed for sensing applications. An early, but perhaps not so practical, attempt was published in 1983 by Meltz et al., who demonstrated the temperature-dependent coupling between the cores in a twin-core fiber.

 

4.2 Uncoupled Cores

If the difference in propagation constants of a multicore fiber or the distances between the cores are sufficiently large, the coupling of energy between the waveguides will be so weak that effectively no coupling occurs over the length of the fiber.

The use of such fiber for transmission purposes for cost-effective access networks was proposed early and was later continued by, for example, Le Noane et al. at CNET in the mid-1990s and Rosinski. The technique, however, does not seem to have been commercially implemented on a larger scale.

The two uncoupled cores of a dual-core fiber can constitute the two arms of a Mach-Zehnder interferometer. Compared to an ordinary Mach-Zehnder interferometer, in which the two light paths are separated in two physically different fibers, this approach facilitates a much higher stability because common mode disturbances, such as temperature drift and external vibrations, will have a similar effect on the two cores and, thus, be effectively canceled.

The coupling of light into the two arms and the subsequent recombination can quite easily be achieved by tapering a small part of the fiber, thus creating, for example, an in-fiber 3-dB coupler. By manipulating with different means the refractive index difference between the cores, the transmission through the interferometer can be controlled for various purposes.

 

4.2.1 Switching and Multiplexing

Nayar et al. demonstrated all-optical switching in a 200-m nonlinear Mach-Zehnder interferometer in 1991. The signal is added to the input port (Fig. 6.8) and split 30:70 by the coupler into each of the two interferometer arms.

 

 

The two propagating beams are recombined by the output coupler and the energy is partitioned between output 1 and output 2 depending on the phase difference. By suitable phase-shifting means (not shown), the energy can be directed to only one of the ports.

Increasing the intensity will induce an unequal index change in the two cores through the optical Kerr effect. When the corresponding phase shift between the propagating beams is \(\pi\), the energy will have been switched to the other output.

In an implementation of an optical add–drop multiplexer, Yvernault et al. inscribed a Bragg grating in each of the two arms of a twin-core fiber Mach-Zehnder interferometer (Fig. 6.9).

 

 

The gratings were phase matched through UV exposure. The unit constitutes a four-port device: ‘‘In’’ and ‘‘Drop’’ ports on one side of the interferometer and ‘‘Out’’ and ‘‘Add’’ ports on the other.

Wavelengths present at the In port are equally split by the integral 3-dB coupler to the two arms. Signals at the Bragg wavelength are reflected in both arms and interfere constructively at the Drop port, while other signals are transmitted through and interfere constructively at the Out port where they appear unattenuated.

Similarly, a signal at the Bragg wavelength present at the Add port will appear at the Out port. The unit can be wavelength tuned through Ohmic heating caused by passing an electrical current through a thin metal layer deposited on the fiber surface. The same group also proposed a variable optical attenuator based on a thermally tuned twin-core Mach-Zehnder interferometer.

 

4.2.2 Fiber Sensors

Uncoupled multicore fibers have been proposed for several sensing applications in a large variety of configurations and some examples are given here.

Noda et al. already demonstrated how the twisting angle of a fiber could be measured in a Mach-Zehnder–like twin-core interferometer by monitoring the far-field interference fringes created by the two emerging beams.

Tanak et al. constructed a system for quench detection of superconducting magnets using a similar setup. The small temperature increase of a superconducting magnet mockup was detected by observing the far-field interferogram from a double-core fiber monitoring the cooled magnet before the temperature rise could cause loss of superconductivity.

Khotiaintsev et al. demonstrated the use of a twin-core fiber with an integrated coupler as a small-sized probe for invasive laser Doppler anemometry. Parasitic phase modulation due to external disturbances was reported to be greatly attenuated because of the integrated design.

Wosinski et al. in 1994 proposed the use of a dual-core fiber with different strain and temperature dependencies for the two cores for monitoring strain in, for example, composites.

Changes in temperature and/or strain introduce refractive index changes that can be detected with, for example, fiber Bragg gratings inscribed in the cores. Strain and temperature are then easily obtained from two independent equations. The sensor examples so far cited employ dual-core fibers.

Gander et al. have demonstrated how the far-field interference pattern from a four-core fiber can be used for the simultaneous measurement of bending about two orthogonal axes.

The strain caused by bending the fiber will introduce different amounts of phase shifts for the propagating light in the four cores depending of the orientation of the fiber relative to the bend direction. By inscribing gratings in the cores, similar information can be obtained from the changes in reflection spectra upon bending the fiber.

Bulut and Inci used a four-core fiber for creating a stable light pattern illuminating an object for three-dimensional Fourier transform profilometry. The interference pattern, created by the interference of the wave fronts emitted from the four-fiber cores, is projected onto an object. The resulting deformed pattern contains information about the surface’s topography, which can be retrieved through Fourier analysis.

An unusual application of multicore fiber was presented by Watson et al.. Through laser ablation, the material around each core of a four-core fiber was removed to create pits about 15 μm deep. The pits were covered by thin reflective films so each pit finally constituted a pressure-sensitive Fabry-Perot interferometer and the whole assembly an ultra-miniature four-channel pressure sensor. 

 

4.3. Manufacturing Multicore Fibers

There are basically two approaches to manufacture an optical fiber with several cores.

Dorosz and Romaniuk described a multicrucible technique. This method is schematically illustrated in Fig. 6.10.

 

 

An outer crucible contains the molten glass that will eventually constitute the cladding of the fiber to be (A), while an inner crucible contains a glass melt that will make up the cores (B). The core glass flows through the nozzles at the bottom of the inner crucible into the cladding glass and the combined glass streams jointly exit the outer crucible through the bottom nozzle without intermixing of the streams. The fiber is then drawn and coated in the normal fashion.

The multicrucible technique is best suited for soft glasses, silica, or nonsilica based, with low process temperatures. Very complex fiber structures can be manufactured with a large range of different glass compositions. The attenuation of these fibers is, however, quite high, 0.1–1 dB/m. This is mostly because of impurities in the raw materials used for making the bulk glass.

For glasses with higher processing temperatures (\(\ge\)2000°C) such as the silica glasses today used in most optical fibers, the multicrucible technique is less suitable. Instead such fibers, or more correctly preforms, are manufactured by assembling the core rods in a cladding structure by various methods and then fusing the assembly into a homogenous preform from which the fiber is drawn in the normal fashion.

The core rods can be manufactured with, say, standard modified chemical vapor deposition (MCVD) or outside vapor deposition (OVD) techniques. In this case, the core, doped with, say, Ge, Er, Al, and so on, must be separated from the surrounding cladding structure, which can be done with, for example, etching with hydrofluoric acid. Cores can also be manufactured from bulk glass if the higher attenuation usually associated with such glass can be tolerated.

The cladding can be in the form of a solid rod into which are drilled holes at the appropriate places to house the core rods. After thorough cleaning and drying, the rods are inserted into the holes and the assembly is fused at high temperature into a solid preform rod. Stretching and overcladding techniques, as mentioned earlier, can also be used where appropriate. Figure 6.11 depicts a cladding rod with holes drilled into which core rods (front) will be inserted.

 

 

An alternative technique for making multicore preforms is to machine from ordinary MCVD or OVD preforms the core and some of the surrounding cladding, which subsequently is inserted into properly shaped grooves in the multicore preform, as illustrated in Fig. 6.12.

From a core preform (Fig. 6.12a), the core and part of the cladding is removed by sawing, yielding a core piece (Fig. 6.12b). A matching groove is machined out of a preform, leaving the central core intact (Fig. 6.12c) and the core piece is inserted into the groove.

The assembly is inserted into a silica tube to keep it together (Fig. 6.12d). Upon heating, the tube collapses on the rod and the whole assemble fuses into one solid dual-core preform (Fig. 6.12e) subsequently drawn into a fiber with standard methods.

 

 

5. Fibers for High-Temperature-Resistant Gratings

Fibers with a high photosensitivity, for an efficient fabrication of fiber Bragg gratings, are presented in a later tutorial. Here, specialty fibers developed for obtaining high-temperature–resistant fiber gratings are reviewed. The chosen grating fabrication techniques have a determinant role and will, therefore, also be reviewed.

High-temperature–resistant fiber gratings find an obvious application for fiber-sensing measurements at high temperatures and in particular the measurement of the high temperatures themselves.

Optical fiber thermometry is especially advantageous in hostile environments, such as in the presence of corrosive chemicals, high electric, magnetic, or strong radiofrequency fields.

Examples are the measurement of temperature and pressure in oil wells where the temperature can exceed several hundreds degrees centigrade, measurement of the temperatures reached in engines or electrical transformers, and the measurement of temperatures in industrial processes involving highly exothermic chemical reactions.

Another situation where high-temperature–resistant gratings are required is when the fiber containing the grating needs to be metallized for subsequent welding or embedded in a metallic structure.

Fiber gratings known to be high-temperature–resistant have been found successfully stable in the presence of ionizing radiations. One hypothesis, which has not yet been tested or confirmed as far as we know, is that high-temperature– resistant fiber gratings also exhibit a higher stability when subject to very large optical fields, as is the case in high-power fiber lasers.

Although the photosensitivity of optical fibers is usually defined as a permanent refractive index change following an exposure to light, fiber gratings are not stable at elevated temperatures and can always be totally erased when a sufficiently high temperature is applied.

Even for room temperature use, a grating needs to be annealed at an elevated temperature to stabilize its refractive index modulation. Afterwards, at temperatures lower than the annealing temperature, the grating generally does not experience substantial thermal decay.

The thermal annealing operates an accelerated aging process, which removes the lesser stable contributions to the index change. The thermal stability depends not only on the type of fiber used and on the fabrication conditions (irradiation wavelength, intensity and dose applied), but also on presensitization techniques.

The latter techniques help in getting quicker index changes but, not so surprisingly, give rise to significantly lower stability. The strength of gratings in hydrogen-loaded fibers, for example, is already strongly reduced at temperatures of only about 100°C.

The first fiber gratings, which appeared to have good stability at elevated temperatures, were the so-called ‘‘type II gratings.’’ The formation of these gratings, first obtained in 1992, is caused by a highly increased nonlinear absorption when applying a large UV intensity single pulse.

They usually appear at the first core–cladding interface as a corrugation of the glass due to remelting. These gratings have shown to be stable at temperatures as high as 800°C for a period of 24 hours. However, type II gratings are difficult to manufacture in a controlled manner and exhibit huge loss due to coupling to cladding modes.

Another type of grating with increased temperature resistance are the so-called ‘‘type IIa gratings.’’ These gratings are obtained by high-dose irradiation at 193 nm in high germanium concentration fibers (never obtained in hydrogenloaded fibers). They are stable up to about 300°C.

Let us now review the effect of the fiber composition on the thermal stability of fiber gratings. Most fiber gratings are obtained in fibers containing a substantial concentration of germanium. However, strong gratings could be obtained in 1996 in a germanium-free fiber based on a nitrogen-doped silica core.

These gratings when written with 193-nm light and of type IIa, exhibited a good thermal stability up to temperatures of about 600°C. Later and still in the same type of fibers, gratings with a post-exposure increasing strength at elevated temperatures were reported.

This new phenomenon was associated to the thermodiffusion of nitrogen in the UV-irradiated regions of the fiber. These gratings after proper annealing were shown to be stable up to 900°C.

Gratings written with 248-nm light in tin-doped silica fibers also exhibit very high thermal stability. No major degradation can be observed at 850°C over a few hours and extrapolations from experimental data indicate that gratings operating at 500 K for 10 years will retain more than 99% of their initial strength.

Tin-doped silica fibers are low loss at 1.55 μm and exhibit large photosensitivity, but they usually have a large numerical aperture.

Finally, an antimony-germanium (Sb-Ge) co-doped fiber with high-temperature sustainability has been developed. Gratings written in this fiber with 248-nm light have been shown to be quite stable at temperatures up to 500°C and the decrease as a function of temperature above that temperature is not very steep (still a significant reflectivity at 950°C).

The underlying mechanisms of the photosensitivity are the creation or modification of defects and structural modifications of the glass matrix itself either by relaxation of internal stresses or by compaction, which is a natural tendency of amorphous solids.

The highest temperatures for thermal stability of induced defects are between 100 and 600°C. Structural changes are more stable and the highest temperatures in these cases are between 400 and 1000°C.

The next mechanism of even higher thermal stability is that which is limited by viscous flow or diffusion of dopants. In 1996, a method was proposed to create high-temperature stable fiber gratings by periodically modifying the concentration of fluorine in the core of the exposed fiber.

This type of grating is usually called chemical composition gratings (CCGs). They are manufactured by writing a type I grating in a hydrogen-loaded fiber containing a fluorine-doped core.

Exposure to UV light induces a periodic variation of OH bondings in the core due to a photo-induced reaction between molecular hydrogen and the glass matrix. When the fiber is heated at a temperature about 900 to 1100°C, the original type I grating totally disappears, as expected, and a new one appears after several tens of minutes and remains stable after development.

It is believed that a chemical reaction takes place at these elevated temperatures between the OH bondings and the fluorine atoms, leading to the formation of HF molecules. The latter have a high mobility and rapidly diffuse out of the core region, leading to a periodic extraction of fluorine.

With simple diffusion models, it has been shown that the stability of these gratings at an elevated temperature only depends on the diffusion process between the dark regions (low exposure to UV light) with larger concentration of fluorine and the bright regions having experienced a depletion of fluorine.

These CCGs have an excellent stability at temperatures below approximately 800°C, as illustrated by Fig. 6.13. A comparison between different fiber grating types, in particular types I and IIa and CCGs, has been realized in 2002 and clearly shows the significantly higher thermal stability of the CCGs.

A similar type of CCG, with similar stability behavior, has been obtained in erbium-doped fibers, which did not contain any fluorine. A new type of process that produces gratings with even higher temperature stability has been discovered.

The gratings are obtained with large exposure doses and, therefore, large concentration of OH bondings. It is believed that the CCG development leads to the formation of molecular water and, hence, periodic extraction of oxygen in the irradiated cores in that case.

The strength of these gratings was basically constant at more than 1100°C for 60 minutes. A totally new fabrication technique, with a huge potential, was demonstrated in 2003. It apparently works in any type of fiber and without any pre-photosensitization.

It consists of irradiating the fibers with ultrashort pulses from near-infrared (IR) lasers. The refractive index changes occur through a multiphoton absorption (most probably four or five photons) and the spatial locations where they appear can, hence, be chosen by properly focusing the writing laser beam.

Gratings fabricated in these conditions are highly thermally stable. Here, also type I and type II gratings can be distinguished depending on the intensity of the IR laser beam, and the latter ones, obtained above a certain ionization threshold, are shown to be stable up to the glass transition temperatures.

In 2004, a grating fabricated with this technique in a multimode crystalline sapphire fiber was reported and it was shown that no reduction in the grating reflectivity or any hysteresis in the Bragg resonance was detectable up to 1500°C.

 

 

 

Summary

The single-mode optical fiber has been shown to have great potential beyond the one of mere light transport for use in a diverse range of applications. The combination of several technologies in one physical fiber allows the construction of multifunctional entities with the inherent advantages of small size, light weight, and high stability.

This tutorial has demonstrated a large number of components and solutions based on multifunctional single-mode fibers. Given the rapid development in material sciences and of manufacturing technologies, the future is likely to continue to bring forward new exciting fiber-based solutions to new and old problems.

 

The next tutorial discusses in detail about atomic rate equations.