New designs increase functionality of specialty fiber


Abdel Soufiane, Keith Dowling, and Carter Houghton

Specialty optical fibers have been key elements in WDM components and modules. Advanced fiber designs will further improve component performance, enhance reliability, simplify manufacturing processes, and reduce material costs.

While the use of specialty optical fiber is widespread throughout the communications industry, component developers often overlook it as an enabling technology. Unlike standard, single-mode transmission fibers, specialty optical fiber is custom-designed for specific applications and has been integral to the development of devices such as couplers and optical amplifiers.

In the mid-1990s, the commercial introduction of the erbium-doped fiber amplifier (EDFA) made dense wavelength-division multiplexing (DWDM) possible. At the heart of the EDFA is a specialty optical fiber—erbium-doped fiber. Next-generation DWDM systems will demand more from EDFAs in terms of performance and cost. Continued innovation of erbium-doped fiber is essential to meeting these demands.

More consistent doping. Most erbium-doped fibers do not have uniform doping profiles because of the doping techniques currently used in the industry. Doping nonuniformity presents a challenge to EDFA manufacturers because identical lengths of fiber do not necessarily provide the same performance, resulting in rework and high scrap costs.

In the modified chemical-vapor deposition (MCVD) process, which is widely used in the fiberoptics industry to make the glass preforms from which optical fiber is drawn, the most prevalent method for incorporating erbium ions into the glass is solution-doping (see Fig. 1). In solution-doping, the unsintered oxide soot is immersed in a solution that contains erbium ions. The soot is then dried and sintered, incorporating the erbium into the glass matrix. Controlling the soot density is challenging and leads to a variation in the erbium solubility, resulting in inconsistent concentration of erbium in the fiber.

An alternative technique to solution-doping is to sublimate an erbium precursor by heating it in a chamber separate from the deposition substrate. This generates an erbium-containing vapor that is oxidized simultaneously with the host-glass precursors. The resulting soot is an erbium-doped silica glass. However, the low vapor pressure of the erbium precursor causes the concentration of erbium to decrease along the length of the deposition tube. This also results in inconsistent concentration of erbium in the fiber.

While the use of erbium-doped fiber is widespread, doping nonuniformity contributes significantly to the overall cost of an EDFA. Specialty fiber manufacturers are exploring several approaches to overcome this fundamental flaw, including novel precursors and fabrication processes.

Higher doping concentrations. Erbium-doped fiber is one of the major material cost drivers in an EDFA, which may contain several hundred to several thousand dollars worth of erbium-doped fiber. One way to lower costs is to dope the fiber with higher concentrations of erbium. Although this technique requires significant glass-composition development, it can result in amplifiers that achieve specified performance levels with shorter fiber lengths, thus lowering the material cost of the EDFA by as much as 15% (see Fig. 2). This is especially critical for emerging applications such as L-band EDFAs, which require significantly more erbium-doped fiber than their C-band counterparts. Some L-band EDFA designs require as much as 100 m of erbium-doped fiber.

The incorporation of arbitrary amounts of erbium into a silica-based glass, however, is not practical because of the effects of clustering. The matrix of the silicon oxide (SiO2) host is rigid, which prevents erbium ions from readily bonding with the nonbridging oxygen ions, thus causing the erbium ions to cluster together. When clustering occurs, the close proximity of erbium ions enables "upconversion." In upconversion, an erbium ion that is in an excited state (having absorbed a pump photon) can transfer its energy to a neighboring erbium ion that is also in an excited state. This upconversion provides a mechanism for one of the ions to give up its energy without providing amplification, thus reducing the efficiency of the amplifier. In addition, the upconverted erbium ion has alternative routes back to the ground state, further reducing the efficiency of the amplifier.

One way to prevent erbium clustering is to include additional dopants in the host glass. The incorporation of aluminum, for example, modifies the glass structure by increasing the distance between the erbium ions, thus minimizing the erbium-erbium energy transfer. While aluminum enhances the dissolution of erbium in silica glass, alternative dopants could further increase the amount of erbium in silica.

While advances in erbium-doped fiber are constantly improving the performance and cost of EDFAs, future optical networks will require amplifiers for new wavelength bands such as the S-band. Specialty optical-fiber developments are under way to address this trend. New doping candidates such as thulium appear promising for S-band amplifiers. Other amplification approaches include fiber optimized for discrete Raman amplification, a technology that allows amplification at any wavelength.

A common theme in optical-component design and manufacturing is to minimize component count and human intervention. One way of achieving this is to bring functionality that was previously performed by discrete optics inside the fiber itself.

Photosensitive fiber is a specialty optical fiber optimized so that its refractive index changes when exposed to ultraviolet (UV) light. The magnitude of the refractive-index variation depends on many factors such as UV wavelength and intensity, fiber pre- and post-processing, and core glass composition. The main application of photosensitive fibers is in fiber Bragg gratings (FBGs). In FBGs, a periodic perturbation of refractive index along an optical fiber acts as an in-fiber optical filter.

Examples of devices incorporating FBGs include pump lasers, optical add/drop modules, gain-flattening filters, and dispersion-compensation modules. These components are facing the same cost and performance challenges as EDFAs, and several innovations in photosensitive fiber design exist to treat these problems.

Cladding mode suppression. Cladding mode (or short-wavelength) loss is an effect that occurs because of coupling between the fundamental core mode of the fiber and backward propagating modes supported by the cladding. The result of this coupling is a loss of transmitted light at wavelengths shorter than the center wavelength of the FBG (see Fig. 3). This loss limits the performance of all FBG-based devices, especially those designed for broadband operation.

Modification of the waveguide structure or to the composition of the fiber can reduce or eliminate cladding mode loss. One approach is to design a fiber with a high numerical aperture, which increases the distance between the FBG center wavelength and the cladding-mode-loss region. However, this approach does not eliminate cladding mode loss altogether and the spacing may not be sufficient for broadband devices.

An alternative approach is to uniformly extend the photosensitive region of the fiber beyond the core and into the inner cladding. This design corrects the nonuniform refractive-index modulation of the fiber cross section, preventing coupling to the cladding modes, thus eliminating cladding mode loss. Despite the potential of this design, maintaining comparable photosensitivity in both the core and cladding while achieving the required refractive-index difference presents a significant challenge.

Athermal fiber. Variations in the temperature of the fiber can lead to changes in the refractive index and to the thermal expansion of the fiber, causing a slight shift in the FBG center wavelength. For silica-based glass, the FBG center-wavelength shift is due to the variation of the refractive index with temperature through the thermo-optic coefficient, which is typically 7 x 10-6/°C, and the coefficient of thermal expansion, which is typically 5 x 10-7/°C (note that the thermo-optic coefficient is the predominant contributor).

Athermal fibers can preserve FBG performance over a wide range of temperatures, using novel glass compositions and/or waveguide structures. One approach to mitigating the shift of the FBG center wavelength with temperature is to incorporate materials with opposite thermo-optic coefficients in the core of the fiber. For example, combining the appropriate amounts of germanium oxide (GeO2) with boron oxide (B2O3) in silica fiber can create a core composition with a negligible thermo-optic coefficient.

The focus today is on discrete fiber types. Many optical components use different types of fiber spliced together to achieve a desired functionality. For example, a pump-laser pigtail can incorporate both specialty pigtail and photosensitive fiber.

Tomorrow's fibers will combine multiple functions into one fiber, reducing final product cost and preserving or even improving performance. A photosensitive, polarization-maintaining coupler fiber with a tailored cutoff wavelength is one such fiber. This three-in-one fiber could be used in Raman applications such as grating-stabilized pigtails and polarization-beam combiners. Furthermore, having this specialty fiber as a common pigtail for these components would reduce the loss when they are spliced together.

Abdel Soufiane is founder, chairman, and CTO; Keith Dowling is product manager; and Carter Houghton is director of business development at IntelCore Technologies, 8 Saint Mary's Street, Boston, MA 02215. Carter Houghton can be reached at

More in Transport