Cart 0

Materials and Fabrication Technologies in Optical Fiber Manufacturing

This is a continuation from the previous tutorial - fiber nonlinearities.


Double Crucible Technique

The first attempt at producing high-purity glass, the so-called ‘‘double-crucible technique,’’ proceeded along the lines of conventional glass melting but used specially prepared constituents.

Soda-lime-silicate and sodium-borosilicate glasses were made from materials purified to parts-per-billion (ppb) levels of transition metal impurities by ion exchange, electrolysis, recrystallization, or solvent extraction.

These starting glasses were melted, fined, drawn to cane, and fed into an ingenious continuous casting system composed of concentric platinum crucibles, shown in Fig. 3.1.

A thin stream of core glass flowed from the upper crucible, passed through the reservoir of cladding glass, and was concentrically surrounded by the cladding as it flowed through the orifice of the lower thimble.

The time and temperature of core-cladding contact in the cladding reservoir were controlled, enabling diffusion to produce the index gradient needed to minimize intermodal dispersion.


Figure 3.1  Double-crucible method for fiber fabrication.


Despite its elegance, overwhelming problems beset this method from the start.

First, contamination during processing raised the impurity level from the ppb level in the constituents to parts-per-million (ppm) levels in the fiber. Many attempts were made to eliminate this contamination and a partial solution was achieved by using the oxygen partial pressure of the atmosphere during processing to control the redox conditions within the molten glass.

Absorption by iron and copper, the two principal contaminants, could thus be minimized by altering their valence state. Iron could be oxidized primarily to the Fe3+ state and copper retained the monovalent state by processing in a controlled oxygen atmosphere.

Thus, strong absorptions by Fe2+ and Cu2+ at near-infrared wavelengths were diminished. Fibers adequate for commercial systems of the time were made this way. Losses as low as 5 dB/km were achieved at 0.9 μm, but the lower losses offered by the 1.3- to 1.5-μm spectral window were unattainable using this technique.

Fundamental electronic vibrations and severe OH- contamination were intrinsic to the starting materials and could not be appreciably lowered by improved processing. The method was stillborn as it was introduced to the market due to the advent of a superior technology.


Vapor-Deposition Techniques

The double-crucible technique was short-lived because vapor-deposition techniques soon appeared, which were capable of lower losses from the visible into the infrared. These techniques appeared in the early 1970s and may be categorized as either inside or outside processes.

Both use oxidation of silicon tetrachloride vapor to produce submicron amorphous silica particles. Other chloride vapors such as germanium tetrachloride and phosphorus oxychloride are used as sources of dopants in the silica.

Outside deposition uses flame hydrolysis whereby chloride vapors pass through a propane-oxygen or hydrogen-oxygen flame to produce a ‘‘soot’’ of SiO2 particles. The particles partially sinter as they collect on a mandrel.

The inside process uses these same reactants together with oxygen, but the reaction occurs inside a silica tube in the absence of hydrogen. The high temperatures needed for reacting halide vapors with oxygen are provided by an oxygen-hydrogen burner, which traverses along the tube as it rotates on a glass working lathe.

The reactions produce particles by oxidation rather than hydrolysis. These particles are deposited on the inside wall of the tube downstream of the torch and are sintered to form a vitreous layer as the torch moves past the deposit.


Outside Vapor Deposition

Two versions of outside processes have been developed. These are the ‘‘outside vapor deposition’’ (OVD) process developed by Coming Glass Works and the ‘‘vertical axial deposition’’ (VAD) version developed by a consortium of Japanese cable makers and Nippon Telephone and Telegraph Corporation.

The OVD process is one of the most common techniques used for optical fiber fabrication. A schematic of the steps involved in using the OVD technique is shown in Fig. 3.2.

During the soot-deposition step of the OVD process, silica and doped silica particles are generated in a methane/oxygen flame via hydrolysis reaction. Silica soot preforms are formed by multilayer deposition of vapors and particles on a rotating cylindrical target rod by traversing the soot-containing flame along the axis of the cylindrical target.


Figure 3.2  Schematic of the outside vapor deposition (OVD) process.


During the soot-deposition process, dopants are added to certain segments of the silica-based preform to modify the refractive index of these soot layers. Typical dopants used in the OVD process include germania, boron, fluorine, phosphorus, and titania.

The most widely used dopant is germania (GeO2), which is often used in the central core region of the fiber. Germania doping of the silica preform is achieved by the introduction of a dopant precursor (e.g., GeCl4) to the deposition burner.

The precursor undergoes oxidation in the flame, resulting in the formation of GeO2 particles and GeO vapors. This is in contrast to the case of SiO2 in the flame, which is predominantly in particulate form.

The relative rates of the germania particle and vapor deposition, as well as the morphology of the resulting deposit (i.e., amorphous or crystalline germania), are controlled by the chosen process conditions.

Following the soot-deposition step, the porous soot preform is treated with a drying agent (e.g., chlorine) for removal of water and metal impurities.

Preform drying isotopically performed at temperatures between 950 and 1250°C, where the diffusion rates and reaction kinetics of the drying reaction with impurities (e.g., Fe reaction with chlorine) are fast and the preform is not significantly densified.

The reaction of the drying agent with the hydroxyl species in the soot may be written as


While the hydroxyl content of the soot preform may be reduced to some degree by merely heating the preform to a high temperature, the remaining OH content is still not sufficient to enable acceptable absorption contribution to attenuation for telecom-based applications.

Following the drying process, the porous soot preforms are sintered into glass blanks at temperatures ranging from 1200 to 1600°C.For silica-based soot preforms, surface energy–driven viscous flow is the dominant mechanism of sintering.

A ‘‘unit cell’’ model was provided by Mackenzie and Shuttleworth to predict the rate of increase of density during surface energy–driven viscous flow sintering of granular solids.

The final step of the OVD process involves drawing the sintered glass blanks into 125-μm diameter optical fibers for use in telecommunication systems. During the fiber draw process, the glass preforms are heated (typically in an inert atmosphere) to temperatures above the softening point of the glass (2000–2200°C for silica-based fibers), followed by drawing into fibers by applying axial tension to the samples.

Precision control of the fiber draw conditions (i.e., draw speed, tension, furnace temperature, fiber diameter, etc.) is of great importance, as this plays a large role in both the physical and chemical characteristics of the glass fiber and the overall optical system performance of the waveguides. 


Vertical Axial Deposition

The VAD process is a variant on the OVD method, where the core and clad glasses may be deposited either simultaneously or separately. The VAD process also forms a cylindrical body using soot, but deposition occurs end-on, as shown in Fig. 3.3.

Here, a porous soot cylinder is formed without a hole by depositing the core and cladding simultaneously using two torches. When complete, the body is sintered under conditions similar to those used for OVD.

A fundamental difference between the two processes is that while the composition profile of the OVD preform is determined by changing the composition of each layer, the VAD profile depends on subtle control of the gaseous constituents in the flame and the shape and temperature distribution across the face of the growing soot boule.


Figure 3.3   Vertical axial deposition (VAD) fiber fabrication process. (a) End-on growth of boule; (b) profile of soot preform after removal of mandrel; (c) preform sintering; (d) fiber drawing.


Critical to the development of VAD was the design of a torch composed of up to 10 concentric silica tubes. Typically, reactant vapors pass through one or more of the central passages where they are protected from premature reaction by a ring of inert shield gas.

The outer series of tubes alternate between hydrogen and oxygen to compose the flame. By manipulation of gas flows, the temperature and particle distribution in the flame can be controlled to determine the surface temperature distribution and the shape of the boule.

In spite of this rather fragile control of composition, VAD had one significant advantage over first-generation OVD. Recall that at this time, transmission systems were using graded index multimode fiber. The high refractive index differences—between core and cladding required by such fiber—were obtained with heavy core doping.

This produced a large mismatch in thermal expansion between core and cladding and caused cracking of consolidated OVD preforms at the inner surface as the preform cooled below the glass transition temperature. Because VAD preforms do not have a central hole, they can better withstand thermal stress.

The major challenge to VAD was how to create an optimized index profile to minimize mode dispersion. Initially it was thought that control of the GeO2 distribution across the boule required several GeCI4 sources, each of different composition.

However, it was found that such grading could be accomplished by control of the boule surface temperature distribution. Eventually, process development focused critically on the shape of the growth face and the temperature profile across it.

Figure 3.4 shows GeO2 incorporation into silica as a function of the temperature of the boule end-face. Below 400°C, GeO2 is lost by vaporization of discrete crystalline particles when the boule is sintered at high temperature.


Figure 3.4  Relation between substrate temperature and GeO2 concentration in the vertical axial deposition (VAD) process.


Direct Nanoparticle Deposition

The Direct Nanoparticle Deposition (DND) technology, developed and commercialized by Liekki Corporation (Finland), is a new fiber manufacturing process, ideally suited for the demanding needs of advanced high-power fiber
laser applications.

DND provides the flexibility to engineer the glass matrix into which the rare-earth ions are dissolved. This capability makes it possible to increase the rare-earth (RE) concentration without sacrificing the fiber performance with effects resulting from too high local RE ion concentration.

Furthermore, DND technology inherently provides radial control of dopants, resulting in excellent flatness of the refractive index profile, a key attribute in large, low numerical aperture cores. This ability also facilitates the fabrication of advanced waveguiding structures in which the gain provided by rare-earth ion is independently controlled. 

The DND process is based on the combustion of gaseous and atomized liquid raw materials in an atmospheric oxy-hydrogen flame. The flexibility in how raw materials are fed to the process gives the freedom of incorporating materials with very different vapor pressures.

The glass is doped in-flame where the glass particles are formed, thus the clustering tendency is low. The DND process makes it possible to mix the refractive index effecting materials (e.g., alumina, germanium, phosphorous, etc.) with other doping materials (e.g., Yb, Er, Nd, etc.) already during the deposition of the glass particles.

This improves the homogeneity of the glass composition prior to the sintering phase. Various parameters affect the formation process, for example, the vapor pressure of the metals, temperature of the flame, gas velocities, droplet route through the flame and the Gibbs free energy of the raw materials.

The qualitative results obtained from the particle size distribution measurement shows a single-peaked particle-size distribution, which indicates that the particles are formed through evaporation-condensation process. Rapid quenching and a short residence time produce small particles with narrow size distribution.

Fiber manufacturing using the DND process can be described as a special form of OVD where nano-size particles of the typical glass former materials and the gain dopants are deposited simultaneously onto a target alumina rod mounted on a rotating glass lathe to form the fiber’s core and cladding regions.

The glass modifiers and RE dopant elements are fed into the process—in liquid or vapor phase—directly into the reaction zone through independent and controlled channels of a specially designed flame burner (Fig. 3.5), while a SiCl4 gas bubbler is used as a source for the silica glass former.

The atomized particles leaving the burner are uniform in composition and can be controlled in size from 10 to 100 nm. The rapid particle formation and deposition process allows extremely high doping levels with reduced clustering and photo darkening, as well as a higher fiber damage threshold.

After the glass formation and doping stages, the alumina mandrel is gently removed from the deposited outside preform and handling tubes are attached to it. The preform is then inserted into a furnace where the first step is drying and cleaning. Finally, the porous glass is sintered into a solid clear core fiber preform.


Figure 3.5  Direct nanoparticle deposition process (DND).


In contrast to the multistep modified chemical vapor deposition (MCVD) solution doping process (soaking and diffusion)—which produces core absorption values up to 1200 dB/m at 976 nm—the DND process can reach 2000 dB/m and beyond.

Furthermore, as a result of this new preform fabrication process, core-to-clad ratios of up to 0.5 are attainable with DND, whereas MCVD-produced fibers are limited to values of 0.16.

In addition, the DND process is fast, efficient, and particularly well suited for producing large-mode-area double-clad (DC) fibers with a large core-to-clad ratio (e.g., a highly ytterbium-doped DC fiber with 20-μm core and 125-μm cladding diameters).

The DND process is applicable to single-mode, DC, and DC-PM fiber, as well as to more complex fiber designs, such as fibers with rectangular (or other noncircular) cores and claddings, multicore fibers, or coupled multiple waveguiding-element fibers.

With the control and flexibility DND provides, it is possible to integrate new functionality into the active fiber. The DND is directly applicable to complicated fiber designs such as polarization maintaining (PM) PANDA-fibers, all-glass coated fibers, multilayered core and cladding designs for spectral filtering and reduction non-linear effects as well as other advanced fiber designs.

The combination of these features will ultimately result in low-cost, fully functionally integrated, all-glass active fibers for high-power fiber lasers and amplifiers. Such truly monolithic (un-spliced) structures provide a common platform for a range of fiber applications and the necessary means to reduce the cost of fiber lasers dramatically.


Modified Chemical Vapor Deposition

Inside processes, such as MCVD, had a different origin. Following the tradition of the electronics industry, chemical vapor deposition (CVD) techniques were used to produce doped silica layers inside silica substrate tubes.

As in CVD, the concentration of reactants was very low to inhibit gas phase reaction in favor of a heterogeneous wall reaction that produced a vitreous particle-free deposit on the tube wall.

The tube was collapsed to a rod and relatively low loss fiber obtained. However, deposition rates were impractically low and attempts to increase them always produced silica particles that deposited on the tube wall and resulted in excess loss.

The solution was to exactly reverse the CVD practice: intentionally produce a gas phase reaction by increasing the reactant flows by more than 10 times. Submicron particles were, thus, produced that deposited on the tube wall and were fused into clear pore-free glass as the torch traversed along the tube.

MCVD was, thus, developed as the process shown in Fig. 3.6.


Figure 3.6  The modified chemical vapor deposition (MCVD) process consists of deposition of glass layers inside a silica tube, collapse of tube to a solid rod, and drawing of preform into fiber.


High-purity gas mixtures are injected into a rotating tube, which is mounted in a glass working lathe and heated by a traversing oxy-hydrogen torch. Homogeneous gas phase reaction occurs in the hot zone created by the torch to produce amorphous particles, which deposit downstream of the hot zone.

The heat from the moving torch sinters this deposit to form a pure glass layer. Typical torch temperatures are sufficiently high to sinter the deposited material, but not so high as to deform the substrate tube. The torch is traversed repeatedly to build up, layer by layer, the core or cladding.

Composition of the individual layers is varied between traversals to build the desired fiber index structure. Typically, 30 to 100 layers are deposited to make either single-mode or graded index multimode fiber.


Chemical Equilibria: Dopant Incorporation

After the initial demonstration of feasibility, fundamental investigations established the knowledge required to create a commercial process. For instance, it was necessary to better understand the chemistry of the MCVD process in order to control the incorporation of GeO2 and limit hydroxyl impurities.

In addition, to increase fabrication efficiency, it was necessary to understand the mechanism by which particles deposit on the substrate tube, as well as the manner in which the silica particles are sintered into pore-free glass. Although process development preceded quantitative understanding, optimization of the commercial process required this knowledge.

The chemistry of SiCl4 and GeCl4 oxidation was investigated by infrared spectroscopy. Samples of effluent gases from typical MCVD reactions demonstrated that as the maximum hot-zone temperature reached 1300°K, SiCl4 began to oxidize to Si2OCl6 (Fig. 3.7).


Figure 3.7  Modified chemical vapor deposition (MCVD) effluent composition as a function of hot-zone temperature. Starting reactants: 0.5 g/min SiCl4, 0.05 g/min GeCl4, 0.016 g/min POCl3, 1540 cm3/min O2.


Up to 1450°K, the amount of oxychloride increases to a maximum, whereas at higher temperatures the SiCl4Si2OCl6, and POCl3 contents decrease until their concentration in the effluent is insignificant above about 1750°K. Above this temperature, all reactants are converted to oxides.

The behavior of GeCl4 is different. Its concentration in the effluent gas stream decreases between 1500 and 1700°K, but above 1700°K remains approximately 50% of its original value. It is clear that the majority of the initial germanium is unreacted and escapes in the effluent.

These results indicate that at low temperatures (T < 1600°K), the extent of the reaction for SiCl4, GeCl4, and POCl3 is controlled by reaction kinetics, while at higher temperatures thermodynamic equilibria become dominant. It is clear from rate studies that the residence times in the hot zone are sufficient to produce equilibrium above 1700°K. The SiCl4and GeClconcentrations at high temperatures are strongly influenced by the equilibria:




Equilibrium constants for these reactions may be written



where \(P_i\) are the partial pressures of gaseous species and \(a_i\) represents the chemical activities of the solid species.

The activities can be approximated by \(\gamma_ix_i\), where \(x_i\) is the mole fraction of the particular species in the solid and \(\gamma_i\) is the activity coefficient.

An activity coefficient of unity implies an ideal solution obeying Raoult’s law. The equilibrium constants for these reactions have been determined as a function of temperature and indicate that Eq. (3.2) strongly favors the formation of SiO2 at high temperature, as verified by the experiments described earlier.

Oxidation of GeCl4 by Eq. (3.3), on the other hand, is incomplete because the equilibrium constant, \(K_{\text{GeO}_2}\), is less than unity at temperatures higher than 1400°K.

This means that only a fraction of the germanium starting composition will be present as GeO2. The presence of significant Cl2 concentration resulting from the complete oxidation of SiCl4 shifts the equilibrium further toward GeCl4 by the law of mass action. Low oxygen partial pressure has the same effect.


Purification from Hydroxyl Contamination

A second important aspect of MCVD chemistry is the incorporation of the impurity \(\text{OH}^-\) because reduction of \(\text{OH}^-\) in optical fibers to ppb levels is essential for realization of low attenuation in the 1.3- to 1.55 μm region.

Hydrogen species originate from three sources: diffusion of \(\text{OH}^-\) from the substrate tube during processing, impurities in the starting reagents and carrier oxygen gas, and contamination from leaks in the chemical delivery system.

The \(\text{OH}^-\) level in the fiber is controlled by the reaction


with equilibrium constant


The concentration of \(\text{OH}^-\) incorporated into the glass, \(C_{\text{SiOH}}\), is described by


During deposition in MCVD, \(\text{Cl}_2\) is typically present in the 3–10% range because of oxidation of the chloride reactants. This is sufficient to reduce \(\text{OH}^-\) by a factor of about 4000.

However, chlorine is typically not present during collapse and significant amounts of \(\text{OH}^-\) can be incorporated by diffusion of torch byproducts through the silica tube.

Figure 3.8 shows the dependence of the \(\text{SiOH}\) concentration in the resultant glass as a function of typical \(P_{\text{O}_2}\) and \(P_{\text{Cl}_2}\) concentrations used during MCVD deposition and collapse with 10 ppm \(\text{H}_2\text{O}\) in the starting gas. Figure 3.8 also shows typical consolidation of the VAD and OVD soot processes.


Figure 3.8  Typical incorporation of \(\text{OH}^-\) during processing stages of modified chemical vapor deposition (MCVD), for 10 ppm \(\text{H}_2\text{O}\) in chemical precursors.



Turning now from the reaction equilibria, we consider the mechanism of deposition of particles on the tube walls. The SiO2 particles produced by vapor phase reaction have diameters in the range 0.02–0.1 μm and are, thus, entrained in the gas flow.

Without the imposition of a temperature gradient, they would remain in the gas stream and exit from the tube end. However, temperature gradients in the gas stream produced by the traveling torch give rise to the phenomenon of thermophoresis.

Here, particles residing in a thermal gradient are bombarded by energetic gas molecules from the hot region and less energetic molecules from the cool region. A net momentum transfer forces the particle toward the cooler region.

Within an MCVD substrate tube, because the wall is cooler than the center of the gas downstream of the torch, particles are driven toward the wall where they deposit.

The MCVD process is shown schematically in Fig. 3.9 in terms of (1) heat transfer in the hot zone, (2) reaction, (3) particle formation, (4) particle deposition beyond the hot zone where the tube wall becomes cool relative to the gas stream, and (5) consolidation of previously deposited particles in the hot zones as the torch traverses to the right.


Figure 3.9  Particle formation and thermophoretic deposition in modified chemical vapor deposition (MCVD).


A mathematical model for thermophoretic deposition, experimentally verified, concluded that deposition efficiency (ratio of SiO2 equivalent entering tube to that contained in exhaust) may be expressed as \(e=0.8(1-T_\text{e}/T_\text{rxn})\), where \(T_\text{rxn}\) is the gas reaction temperature and \(T_\text{e}\) is the temperature downstream of the torch at which the gas and the tube wall equilibrate. Typically, \(T_\text{e}\) is about 400°C and \(T_\text{rxn}\) about 2000°C, giving an efficiency value on the order of 60%. Note that the efficiency is not a function of the maximum tube temperature.

Examination of the process of consolidation of the soot layer on the inner surface of the silica tube revealed the mechanism to be viscous sintering. By this mechanism, the rate of consolidation is proportional to the sintering time and surface tension and inversely proportional to the void size, initial soot density, and glass viscosity.


Plasma Chemical Vapor Deposition

A second inside process, plasma chemical vapor deposition (PCVD), is similar to MCVD in that it uses the same reactants inside a silica substrate tube that is collapsed after deposition and drawn into a fiber.

The primary difference between the two methods is that the oxidation of reactants in the tube is initiated by a non-isothermal microwave plasma inside the tube rather than by heating the exterior of the tube, as shown in Fig. 3.10. In the PCVD process, most of the deposition is in vapor form, instead of soot form, as is seen via the MCVD technique.


Figure 3.10  Schematic representation of plasma chemical vapor deposition (PCVD) process.


The generation of the plasma requires a reactant vapor pressure of only a few torr. A microwave cavity—operating at 2.45 GHz, which produces the plasma heat source—traverses along the substrate tube and promotes chemical reaction.

During the PCVD step, the vapors diffuse to the wall of the tube and undergo heterogeneous reaction to form deposit. Soot formation, while possible, is not desired during the PCVD deposition process, as this can result in glass defects (bubbles, etc.).

Furthermore, the reaction and deposition of both GeO2 and SiOis much more efficient than in MCVD, approaching 100%. However, this leads to a high level of water retention in the glass, which results in higher optical losses. Typical methods to address the hydroxyl issues include the use of fluorine-based dopants (C2F6, SF6, etc.), along with the reactants during the PCVD step.

Another advantage—especially for multimode preforms—is that because the plasma involves no latent heat, it can be traversed very rapidly to produce hundreds of layers. The resulting deposit, thus, has a very smooth and precise index profile essential for minimizing intermodal dispersion.


Sol-Gel Processes 

Optical fiber drawing technology and each of the processes associated with preform fabrication—OVD, VAD, MCVD, and PCVD—have been developed to the point at which they can easily yield both multimode and single-mode fiber in long lengths, with losses limited only by the intrinsic properties of fused silica, their principal constituent.

In their initial form, each produced preform typically yields only about 10 km of fiber. The quantity of silica glass necessary to make up the core and primary cladding of any conventional single-mode or multimode optical fibers is small, totaling less than 20% of the fiber’s eventual mass. The remainder is composed of the substrate tube and overclad cylinder.

To reduce the cost of the latter (conventionally made by flame deposition and machined to desired dimensions), and to increase the yield of drawn fibers from each preform, an ‘‘overcladding’’ technique based on sol-gel technology was developed and introduced commercially into fiber production.

With overcladding, after a preform core rod with an oversized core region is fabricated by conventional means, the outer diameter is built up to attain the proper proportions and increase the effective glass mass content to facilitate multikilometer drawing of fiber.

The outer diameter is increased either by jacketing the preform with a second silica tube or by using it as a bait rod for subsequent OVD or VAD soot deposition. This procedure increases the length of fiber produced from a vapor-deposited preform, yielding up to a hundred kilometers of fiber. 


Alkoxide Sol-Gel Processing

Suitable overcladding material may be prepared by sol-gel and powder forming techniques. In one instance, a chemical precursor, typically a silicon alkoxide such as Si(OC2H5)4, is reacted with water in the presence of ethanol and an acid catalyst.

The ‘‘sol’’ is cast into cylindrical molds and poly-condensation of the resulting silanol groups produces a filamentary siloxane gel network. The gel body is dried and consolidated to form silica glass as films or bulk bodies.

Alternatively, commercial colloidal powders obtained from flame hydrolysis, commonly known as ‘‘fumed silica,’’ are formed into bodies by mechanical compaction, centrifugation, or casting/gelation.

In this last approach, the silica particles (generally 0.05–0.5 μm) are dispersed in water to form a sol. Control of pH or addition of surface active agents is used to promote electrostatic and steric stabilization to inhibit interparticle attractive forces, which cause agglomeration.

The dispersed colloid containing up to 60 weight percent silica is cast after the stabilizing forces are dissipated. Gelation by van der Waals forces soon follows to produce a semirigid body. After drying, the porous silica body can be sintered to glass much like the soot boules formed in the OVD and VAD processes.

Drying of the gel body is accompanied by large stresses due to shrinkage and capillary forces, which generally cause the body to fracture. To date, bodies as large as overcladding tubes have not been successfully fabricated using this approach.

However, there has been much effort to fabricate all-gel preforms because the chemical precursors are available in high purity and the refractive index of the glass may be altered by adding dopant alkoxides.

Fibers with a raised index core, doped with alkoxides such as Ge(OC2H5)4, have not yielded fiber with losses comparable to that produced by the vapor technique because the germanium dioxide either dissolves in the liquor or precipitates in some crystallized form. The usual product of consolidation is bubbled silica, whose index is raised only marginally.

The alkoxide route has achieved its best success in fabricating fibers with a silica core and lowered index cladding. Hydrolysis and polycondensation of Si(OC2H5)3F lowers the index by incorporating fluorine.

The sol is cast, gelled, and dried to yield a porous silica body with a surface area of 200–650 m2/g. Such high surface area allows consolidation at low temperatures in a fluorine containing atmosphere.

The result is a down-doped tube (\(\Delta=-0.62\%\)), which is collapsed with a stream of oxygen flowing down the center. This removes the fluorine from the inner tube wall and produces a core region with a higher refractive index than the fluorine-doped cladding. Losses as low as 0.4 dB/km have been reported for such fiber. 


Colloidal Sol-Gel Processing

The colloidal approach has achieved more success in producing large bodies for overcladding. The colloidal process implemented for this purpose differed from that of the more publicized tetraethyl orthosilicate (TEOS) process.

This process produces low-density gels (volume fraction of silica) and limits the size of monolithic bodies produced. For commercial fiber production, bodies weighing more than 10 kg are needed and are produced by higher silica loading of the gel.

This was achieved by using nano-dimensioned silica particles produced commercially by flame hydrolysis of silica tetrachloride. These are marketed by various names; one commonly used is OX-50.

It is dispersed in water with loading of 50wt% or more and can be gelled to yield large cylinders strong enough to survive processing to net-shape and large glass bodies of waveguide-quality silica.

Briefly, the colloidal silica is dispersed in water containing quaternary ammonium hydrochloride. This produces a high pH (>12) and a negative and repulsive surface charge on the particles. The resulting fluid sol is centrifuged.

Gelling results from addition of an ester—typically methylformate—which lowers pH and dissipates the surface charge. After a few hours in the mold, the cylinder is removed and carefully dried to prevent warping or stress-cracking.

Finally, the cylinder is suspended in a sintering/purification furnace, where it is ‘‘burned out’’ to remove organics and purified in a chlorine containing atmosphere, and then consolidated into glass at about 1500°C in a helium gas atmosphere.

For fiber production using this method, extraordinary care must be exercised in the equipment design; high precision molds and driers are required to achieve uniform dried bodies without warping and cracking.

Likewise, purification requires similar care: Water and transition metal impurities are removed in inert, oxidizing, and chlorine containing atmospheres. Purification from transition metals may be enhanced by use of a low oxygen partial pressure atmosphere, as indicated by the reaction:


By firing in an atmosphere protected from air intrusion, oxygen partial pressures can be in the range of 10-6 atm. Thus, at temperatures between 600 and 1000°C, iron and other impurities are effectively removed. This was demonstrated by intentionally contaminating a gel body with 1wt% hematite. After a two-step dehydration and consolidation treatment, the residual iron content was only 40 ppb.

Finally, an additional equilibration in thioryl chloride is used to remove minute quantities of zirconia present in less than the parts/trillion (ppt). These, if present, cause low strength breaks in the hundreds of kilometers drawn from the gel-silica preforms.

The process of overcladding with gel-derived material may be accomplished using two strategies, as shown in Fig. 3.11.


Figure 3.11   A hybrid sol-gel strategy in which (right side) gel is cast into tubes and used to overclad a core rod. And, alternatively (left side), gel is granulated and then fusion-sprayed onto a preform to accomplish overcladding.


On the left side, in the ‘‘rod-in-tube’’ process, an overcladding tube is formed from gel and then consolidated directly onto a core rod. Tubes for the rod-in-tube process are formed by dispersion, milling, casting, and gelation of colloidal silica.

After removal from the mold and air-drying, they are placed over a core rod and the assembly is dehydrated, consolidated, and drawn into fiber. A satisfactory interface, free of bubbles and other defects, must be obtained between the core rod and the gel-derived overcladding tube.

By proper cleaning of the core rod and appropriate consolidation conditions, the loss of the eventual fiber can be as low as that of the original core rod.

On the right side of the graph, instead of casting a tube, the wet gel is granulated into particles, which are fed through an oxygen plasma torch to deposit glass droplets onto the core rod.

Because these particles are 100 μm in diameter, they deposit on the rod by impaction, rather than by weak thermophoretic forces. Deposition efficiency is, thus, quite high.



Vapor phase synthetic silica processes have led the way toward extremely low-loss index-guided transmission optical fiber. Although these methods have demonstrated of high purity and precise control over the index structure, these methods are generally constrained to cylindrically symmetric structures and modest levels of index contrast (\(\Delta{n}\le0.03\)).

Microstructured optical fibers consist of an array of holes, which extended longitudinally along the z-axis of the fiber. The large index (\(\Delta{n}=0.45\)), combined with the ability to pattern structures with dimensions similar to the wavelength of light, yields novel waveguide properties such as photonic band-gap guidance, such as endlessly single-mode behavior, low bend loss, high birefringence, high nonlinearities, and dispersion control.

Increased sophistication in fiber fabrication has yielded losses as low as 0.28 dB/km for index-guided structures and 1.7 dB/km for hollow core structures, making these fibers suitable for both transmission and device applications.

Several methods have been adopted for fabrication of microstructured fibers such as the stacking and drawing of glass capillaries, extrusion of soft glasses, preform drilling, and sol-gel casting.

Here, we describe the sol-gel casting method for the fabrication of microstructured fiber, developed originally at Bell Laboratories and continued at OFS Laboratories.  

The sol-gel casting process, originally developed for the fabrication of large precision overcladding tubes for optical fiber preforms, has been adapted for the fabrication of microstructured fiber designs.

The sol-gel process offers advantages in the fabrication of microstructured fiber including low-cost starting materials, high purity, design flexibility, reusable mold and mandrel elements, and the ability to scale up to large bodies for the generation of low-cost long lengths of microstructured optical fiber.

Figure 3.12 outlines the steps involved in the fabrication of an exemplary sol-gel microstructured preform. A mold containing an array of mandrel elements is assembled. The mandrels are individually tensioned to ensure uniformity along the length of the mold. The mold is subsequently filled with colloidal silica dispersed at high pH with an average particle size of 40 nm.


Figure 3.12  Fabrication processing of microstructured preforms using sol-gel casting. (1) Casting and gelation, (2) mandrel removal, and (3) drying, purification, and sintering of gel body.


The pH is lowered by the addition of an organic ester, causing the sol to gel. At the wet gel stage, the mandrel elements are removed, leaving air columns within the gel body. The gel body is then dried, purified thermochemically to remove organic and transition metal contaminants, and sintered into vitreous silica.

Because contaminants are removed in the dried gel body, the sol-gel process is reasonably insensitive to contaminants introduced during mold assembly, unlike stack-and-draw, or drilling methods, which may potentially introduce contamination at the glass stage.

The sintered preform is then available for draw. If required, additional glass processing steps such as etching, overcladding, or stretching may be performed on the microstructured preform.

The air holes are pressurized during draw to obtain the desired size and air-fill fraction in the fiber, and the hole size is monitored during draw using online measurements and offline measurements with an optical microscope.

Images of several microstructured fibers fabricated using the sol-gel process are displayed in Fig. 3.13, demonstrating the wide range of fiber designs afforded by the casting process.


Figure 3.13   Examples of microstructured optical fibers fabricated using the sol-gel casting method. Examples include (a) small-core high delta fiber, (b) circular single-mode, (c) birefringent fiber, and (d) air-core fiber.


Unlike the stack-and-draw technique, the sol-gel process allows the hole size, position, and shape to be adjusted independently in non–closest-packed structures such as circular arrays.

Furthermore, the sol-gel technique can generate structures consisting of several hundreds of holes required for low confinement losses in both index-guided and photonic band-gap fibers, which are expensive and challenging to fabricate by methods such as preform drilling.

To date, continuous lengths of more than 10 km of sol-gel microstructured fibers have been drawn with variations in hole size of approximately less than 2% over kilometer length with optical losses of roughly 1 dB/km at 1550 nm and OH peak absorption loss at 1385 nm of 1.5 dB/km, values that are competitive with microstructured fibers produced from high-quality VAD capillaries.

Continued refinements in powder source, processing, and glass surface finish will continue to lower the optical loss. Preform dimensions can be readily scaled to preform sizes of more than 200 km with reusable molds and mold assemblies, providing a route toward large-scale production of low-cost microstructured fiber. 


Fiber Drawing

Typical preforms produced by any of the previously described methods are about a meter in length and between 2.0 and 7.5 cm in diameter. These preforms are drawn into 125-μm diameter fiber by holding the preform vertically and heating the end of the preform above the glass softening temperature until a glob of glass falls from the end.

This forms a neck-down region, which provides transition to a small-diameter filament. Uniform traction on this filament results in a continuous length of fiber. Before this fiber contacts a solid surface, a polymer coating is applied to protect the fiber from abrasion and preserve the intrinsic strength of the pristine silica. The fiber is then wound on a drum.

Although the basic principles of fiber drawing were established before the advent of optical fiber technology, stringent fiber requirements necessitated improvements in process control and understanding of the effects of draw conditions on optical performance. Fiber is now drawn without inducing excess loss while maintaining high strength and dimensional precision and uniformity.

The essential components of a draw tower, shown schematically in Fig. 3.14, are a preform feed mechanism, a furnace capable of 1950–2200°C, a diameter monitor, a polymer coating applicator, a coating curing unit, a traction capstan, and a take-up unit.


Figure 3.14  Schematic of a fiber draw tower.


The furnace is typically either a graphite resistance type or an inductively coupled radiofrequency zirconia furnace. The former requires an inert atmosphere to prevent oxidation of the graphite element.

The zirconia furnace may be operated in air but must be held above 1600°C, even when not in use, because the volume change associated with the crystallographic transition of zirconia at this temperature can cause stress-induced fracture. The advantage of this furnace is that it generally has less contamination by particles emitted from the heater element.

The uniformity of the fiber diameter depends on control of the preform feed rate, the preform temperature, and the pulling tension. Over long lengths of fiber (>100 cm), diameter variations can result from changes in preform diameter and drifts in furnace temperature and the speeds of the feed and capstan motors.

Diameter variations with shorter length period arise from perturbations in the temperature of the neck-down region caused by thermal fluctuations. These may be minimized by control of convective currents and nonuniform gas flows inside the furnace, as well as acoustical and mechanical vibrations.

The diameter monitor positioned below the furnace typically provides feedback to the capstan, adjusting the draw tension to maintain constant fiber diameter.


The next tutorial discusses about laser output-beam properties.


Share this post



Sold Out