Mid-IR and Infrared Fibers

This is a continuation from the previous tutorial - optical fibers and fiber optic communications.

1. Introduction

Infrared (IR) optical fibers are fibers that transmit radiation from 2 to approximately 20 μm. The first IR fibers were fabricated in the mid-1960s from a rather special class of IR transparent glasses called chalcogenide glasses. It was well known that mixing chalcogen elements, for example, arsenic and sulfur, can form a dark red glass that is transparent well beyond 2 μm.

In 1965, this arsenic trisulfide (As₂S₃) glass was first drawn into crude optical fiber by Kapany et al., but the losses were more than 10 dB/m from 2 to 8 μm. A loss of 10 dB/m means that a 1-m length of As₂S₃ fiber would transmit only 10% of the incident light. Energy would also be lost because of reflection from the fiber end-faces, which for this high refractive index glass (\(n=2.3\)) amounts to an additional 31%. Furthermore, their As₂S₃ fiber was quite brittle.

During the mid-1970s, there was interest in developing an efficient and reliable IR fiber to link broadband long wavelength radiation to remote photodetectors in military sensor applications. For example, researchers at Hughes Research Laboratories in Malibu, California, studied various IR fibers that could be used in a surveillance satellite application to transmit IR radiation in the 3- to 5-μm, 8- to 12-μm, and longer wavelength bands to an interior IR detector array.

In addition, there was an ever-increasing need for a flexible fiber delivery system for transmitting CO₂ laser radiation in surgical applications. Based on these needs, various IR materials and fibers were developed. These fibers include the heavy metal fluoride glass (HMFG) and polycrystalline fibers, as well as hollow rectangular waveguides. Although none of these fibers had physical properties even approaching those of conventional silica fibers, they were, nevertheless, useful in lengths less than 2–3 m for a variety of IR sensor and power delivery applications.

IR fiber optics may logically be divided into three broad categories: glass, crystalline, and hollow waveguides. These categories may be further subdivided based on the fiber material, structure, or both, as shown in Fig. 13.1.

Over the past 30 years, many novel IR fibers have been made in an effort to fabricate a fiber optic with properties as close to silica as possible, but only relatively few have survived. An excellent review of the types of IR fibers may be found in a book by Harrington. Other sources of general information on IR fiber types may be found in the literature.

In this tutorial, only the best, most viable, and in most cases, commercially available IR fibers are discussed. In general, both the optical and the mechanical properties of IR fibers remain inferior to silica fibers, so the use of IR fibers is still limited primarily to non-telecommunication short-haul applications requiring only tens of meters of fiber rather than kilometer lengths common to telecommunication applications.

The short-haul nature of IR fibers results from the fact that most IR fibers have losses in the few dB/m range. An exception is the fluoride glass fibers, which can have losses as low as a few dB/km. In addition, IR fibers are much weaker than silica fiber and, therefore, are more fragile. These deleterious features have slowed the acceptance of IR fibers and restricted their current use to applications in chemical sensing, thermometry, and laser power delivery.

The obvious key property of IR fibers is their ability to transmit wavelengths longer than most oxide glass fibers. While some IR fibers can transmit well beyond 20 μm, most applications do not require the transmission of IR wavelengths longer than about 12 μm. A summary of the spectral loss for five of the six subcategories of fibers listed in Fig. 13.1 is shown in the composite data in Fig. 13.2.

Some of the important optical and mechanical properties of IR fibers are listed in Table 13.1. For comparison, the properties of silica fibers are also listed. The data in Table 13.1 and in Fig. 13.2 reveal that, compared to silica, IR fibers usually have higher loss; larger refractive indices and dn/dT; lower melting or softening points; and greater thermal expansion.

From these data, it may also be seen that there is a wide variation in the range of transmission for the different IR fibers and that the loss of most of the IR fibers is quite high compared to silica fibers. Remembering that 1 dB/m is a bulk loss of about 20% per meter, it is again evident that this high loss will restrict applications to meter-long lengths.

All of the solid-core fibers shown in Fig. 13.2 have a much lower theoretical or intrinsic loss. The reason that the losses are so high is that the fibers contain impurities and imperfections that give rise to a large extrinsic absorption and scattering.

Some of these extrinsic absorption bands are evident in Fig. 13.2. The absorptions shown in Fig. 13.2 for the hollow waveguide are not due to impurities, rather they are due to interference effects resulting from the thin-film coatings used to make the guides.

Finally, many IR fibers do not have a proper cladding analogous to conventionally clad oxide glass fibers. Nevertheless, core-only IR fibers such as sapphire and chalcogenide fibers can still be useful. This is because their refractive indices are sufficiently high that there is less evanescent wave energy outside the core. As long as the unclad fiber does not come in contact with an absorbing medium, the fiber can operate reasonably well, as there will be very little leakage of light from the core to the surrounding air.

The motivation to develop a viable IR fiber stems from many proposed applications. A summary of the most important current and future applications and the associated candidate IR fiber that will best meet the need is given in Table 13.2.

There are several noteworthy trends seen in this table. The first is that hollow waveguides are an ideal candidate for laser power delivery at all IR laser wavelengths.

The air core of these waveguides gives an inherent advantage over solid-core fibers because IR materials used in solid-core fibers have laser damage thresholds that are frequently very low. The air-core waveguides are capable of delivering close to 3000 W of cw CO₂ laser power, far in excess of any IR solid-core fiber.

However, solid-core IR fibers are ideal evanescent-wave sensors for monitoring chemical processes in the sensitive fingerprint region of the infrared spectrum. In these applications, the chemical or biological agent surrounds the fiber core and some portion of the light is coupled out of the core into the surrounding medium.

This type of chemical sensor is potentially very sensitive and selective. Chalcogenide and silver halide fibers are particularly good for this application, as they are quite inert and their high refractive index means that only a small portion of the light is out-coupled from the core into the absorbing medium.

A good fiber for gas sensing is the hollow waveguide, as the core of this fiber can be filled with gas so that light propagating through the waveguide is partially absorbed by the gas. Temperature measurements using long wavelength transmissive fibers like the silver halides or hollow waveguides are possible over a large temperature range.

Normally blackbody radiation from a source is transmitted through the fiber and the temperature determined by calibration to a blackbody of known temperature. Because blackbody radiation from room-temperature objects is peaked near 10 μm, IR fibers are excellent candidates for use in measuring temperatures less than 50\(^\circ\)C.

2. Halide and Heavy Metal Oxide Glass Fiber Optics

There are two IR transmitting glass fiber systems that resemble conventional silica-glass fibers in that they are drawn from glass preforms and they have a core region surrounded by a cladding layer.

One is the heavy metal fluoride glasses (HMFGs), or fluoride glass for short, and the other is heavy metal germanate glass fibers based on GeO₂. The germanate glass fibers generally do not contain fluoride compounds. They also do not contain silica (SiO₂), rather they contain heavy metal oxides to shift the IR absorption edge to longer wavelengths.

The advantage of germanate fibers over HMFG fibers is that germanate glass has a higher glass transition temperature and, therefore, a higher laser-damage threshold. But the loss for the HMFG fibers is lower.

2.1. Fluoride Glass Fibers

Poulain et al. discovered the HMFGs based on zirconium fluoride, also called fluorozirconate glasses, accidentally in 1975 at the University of Rennes. In general, the typical fluoride glass has a glass-transition temperature, \(T_g\), four times less than silica, is considerably less stable, and has practical failure strains of only a few percent compared to silica’s greater than 5%.

While an enormous number of multicomponent fluoride glass compositions have been fabricated, comparably few have been drawn into fiber. This is because the temperature range for fiber drawing is normally too small in most HMFGs to permit fiberization of the glass.

The most popular HMFGs for fabrication into fibers are the fluorozirconate and fluoroaluminate glasses, of which the most common are ZBLAN (ZrF\(_4\)-BaF\(_2\)-LaF\(_3\)-AlF\(_3\)-NaF) and AlF\(_3\)-ZrF\(_4\)-BaF\(_2\)-CaF\(_2\)-YF\(_3\), respectively. The key physical properties that contrast these two glasses are summarized in Table 13.3. An important feature of the fluoroaluminate glass is its higher \(T_g\), which largely accounts for the higher laser damage threshold for the fluoroaluminate glasses compared to ZBLAN at the Er/YAG laser wavelength of 2.94 μm.

The fabrication of HMFG fiber is similar to any glass fiber–drawing technology except that the preforms are made using some type of melt-forming method rather than by a vapor-deposition process common with silica fibers.

Specifically, a casting method based on first forming a clad glass tube and then adding the molten core glass is used to form either multimode or single-mode fluorozirconate-fiber preforms.

The cladding tube is made either by a rotational casting technique in which the clad tube is spun in a metal mold or by merely inverting and pouring out most of the molten clad glass contained in a metal mold to form a tube.

The clad tubing is then filled with a higher index core glass. Other preform fabrication techniques include rod-in-tube and crucible techniques. The fluoroaluminate fiber preforms have been made using an unusual extrusion technique in which core and clad glass plates are extruded into a core/clad preform.

All methods, however, involve fabrication from the melted glass rather than from the more pristine technique of vapor deposition used to form SiO₂-based fibers. This process creates inherent problems such as the formation of bubbles, core–clad interface irregularities, and small preform sizes.

Most HMFG fiber drawing is done using preforms rather than the crucible method. A ZBLAN preform is drawn at about 310\(^\circ\)C in a controlled atmosphere (to minimize contamination by moisture or oxygen impurities which significantly weaken the fiber) using a narrow heat zone compared to silica.

Either UV acrylate or Teflon coatings are applied to the fiber. In the case of Teflon, heat shrink FEP fluoride is generally applied to the glass preform before the draw.

The theoretical or intrinsic attenuation in HMFG fibers is predicted to be about 10 times less than that for silica fibers. Based on extrapolations of the intrinsic losses resulting from Rayleigh scattering and IR multiphonon absorption, the minimum loss is projected to be about 0.01 dB/km at 2.55 μm. Refinements of the scattering loss have modified this value slightly to be 0.024 dB/km or about eight times less than that for silica fiber.

In practice, however, extrinsic loss mechanisms still dominate fiber loss. The lowest measured loss for a 60-m long ZBLAN fiber is 0.45 dB/km at 2.3 μm. This loss is dominated by extrinsic loss mechanisms due to scattering (crystallites, oxides, and bubbles) and impurities such as Ho\(^{3+}\), Nd\(^{3+}\), Cu\(^{2+}\), and OH\(^-\).

In Fig. 13.3, losses for two ZBLAN fibers are shown. The data from British Telecom (BTRL) represents state-of-the-art fiber 110 m in length. The other curve is more typical of commercially available (Infrared Fiber Systems, Silver Spring, MD) ZBLAN fiber. The lowest measured loss for a BTRL, 60-m long fiber is 0.45 dB/km at 2.3 μm.

Some of the extrinsic absorption bands that contribute to the total loss shown in Fig. 13.3 for the BTRL fiber are Ho\(^{3+}\) (0.64 and 1.95 μm), Nd\(^{3+}\) (0.74 and 0.81 μm), Cu\(^{2+}\) (0.97 μm), and OH\(^-\)(2.87 μm).

Scattering centers such as crystals, oxides, and bubbles have also been found in the HMFG fibers. In their analysis of the data in Fig. 13.3, the BTRL group separated the total minimum attenuation coefficient (0.65 dB/km at 2.59 μm) into an absorptive loss component equal to 0.3 dB/km and a scattering loss component equal to 0.35 dB/km.

The losses for the fluoroaluminate glass fibers are also shown for comparison in Fig. 13.3. Clearly, the losses are not as low as those for the BTRL-ZBLAN fiber, but the AlF\(_3\)-based fluoride fibers do have the advantage of higher glass-transition temperatures and, therefore, are better candidates for laser power delivery.

The reliability of HMFG fibers depends on protecting the fiber from attack by moisture and on pretreatment of the preform to reduce surface crystallization. In general, the HMFGs are much less durable than oxide glasses. The leach rates for ZBLAN glass ranges between 10\(^{-3}\) and 10\(^{-2}\) g/cm\(^2\)/day. This is about five orders of magnitude higher than the leach rate for Pyrex glass.

The fluoroaluminate glasses are more durable, with leach rates that are more than three times lower than those for the fluorozirconate glasses. The strength of HMFG fibers is less than that for silica fibers. From Table 13.1, we see that Young’s modulus E for fluoride glass is 51 GPa compared to 73 GPa for silica glass.

Taking the theoretical strength to be about one-fifth that of Young’s modulus gives a theoretical value of strength of 11 GPa (\(R=1.198r\frac{E}{\sigma_\text{max}}\)) for fluoride glass. The largest bending strength measured has been about 1.4 GPa, well below the theoretical value. To estimate the bending radius \(R\), we may use the approximate expression, where \(\sigma_\text{max}\) is the maximum fracture stress and \(r\) is the fiber radius.

2.2. Germanate Glass Fibers

Heavy metal oxide glass fibers based on GeO₂ have shown great promise as an alternative to HMFG fibers for 3 μm laser power delivery. Today, GeO₂-based glass fibers are composed of GeO₂ (30–76%)–RO (15–43%)–XO (3–20%), where \(R\) represents an alkaline-earth metal and \(X\) represents an element of Group IIIA. In addition, small amounts of heavy metal fluorides may be added to the oxide mixture.

The oxide-only germanate glasses have glass-transition temperatures as high as 680\(^\circ\)C, excellent durability, and a relatively high refractive index of 1.84. In Fig. 13.4, loss data are given for a typical germanate glass fiber. Although the losses are not as low as they are for the fluoride glasses shown in Fig. 13.3, these fibers have an exceptionally high damage threshold at 3 μm. Specifically, more than 20 W (2 J at 10 Hz) of Er/YAG laser power has been launched into these fibers.

2.3. Chalcogenide Glass Fibers

Chalcogenide glasses are composed of two or more chalcogen elements normally selected from the small group including As, Ge, Sb, P, Te, Se, and S. When these elemental materials are heated and mixed in an oxygen-free environment, some very stable and simple glasses can result.

One of the oldest chalcogenide glasses studied is the binary glass arsenic trisulfide, As₂S₃. This glass is deep red in color, and it is very stable. In the mid 1960s, this glass was drawn into the first IR fiber by Kapany. It was not until some 10–15 years later, however, that these materials were studied seriously as viable IR fiber candidates.

The reticence to pursue these materials in the early days came in part from the toxic nature of some of the elements used in the glasses. Today they are a popular IR fiber material, as they are readily drawn into fiber with a broadband IR transmission but are much more delicate in nature than the oxide glass fibers. They are finding many applications in chemical and temperature sensor systems and as IR image bundles.

Chalcogenide fibers fall into three categories: sulfide, selenide, and telluride. Within these categories, one usually finds that the binary and ternary glasses are excellent choices for fiberization. That is, unlike the fluoride glasses where it is commonplace to have five or more components, most chalcogenide glasses have only two or three elemental components.

In general, these glasses have softening temperatures comparable to fluoride glass. They are very stable, durable, and largely insensitive to moisture. A distinctive difference between these glasses and the other IR fiber glasses is that they do not transmit well in the visible region and their refractive indices are quite high. Additionally, most of the chalcogenide glasses, except for As₂S₃, have a rather large value of dn/dT. This fact limits the laser power handling capability of the fibers.

Chalcogenide glass is made by combining highly purified (>6 nines purity) raw elements in a sealed ampoule that is heated and mixed in a rocking furnace for about 10 hours. After melting and mixing, the glass is quenched and a glass preform fabricated using rod-in-tube or rotational casting methods.

Fiber can be drawn using a preform or from the melt using the double-crucible method. As in the fluoride glass fibers, a buffer polymer coating is applied over the cladding using a UV acrylate or by first applying a Teflon heat-shrink tube over the preform and then drawing into fiber.

The transmission range for chalcogenide fibers depends heavily on the mass of the constituent elements. The lighter element glasses such as arsenic trisulfide have a transmission range from 0.7 to about 6 μm. This glass and some phosphorous-containing and Ge-S–based glasses are the only ones transmitting visible radiation.

Longer wavelength transmission is possible through the addition of heavier elements like Te and Se. When these elements are present, the glasses take on a silvery metallic appearance, and they become essentially opaque in the visible region. This trend is evident from the loss spectra shown for the most important chalcogenide fibers in Fig. 13.5.

A key feature of essentially all chalcogenide glasses is the strong extrinsic absorption resulting from contaminants such as hydrogen, H\(_2\)O, and OH . For example, there are invariably strong absorption peaks at 4.0 and 4.6 mm due to S-H or Se-H bonds, respectively, and at 2.78 \(\mu\)m and 6.3 \(\mu\)m due to OH\(^-\) (2.78 \(\mu\)m) and/or molecular water. As a result, typical chalcogenide loss spectra are normally replete with extrinsic absorption bands, as is clearly seen from the data in Fig. 13.5.

This would seem at first glance to be sufficiently deleterious that the applications for these fibers would be limited. However, many applications for these fibers are possible simply by working outside these extrinsic bands. Another important feature of most of the chalcogenide fibers is that their losses are usually much higher than the fluoride glasses. In fact at the important CO\(_2\) laser wavelength of 10.6 \(\mu\)m, the lowest loss is still slightly above 1 dB/m for the Se-based fibers.

A more recent chalcogen-type glass is based on a combination of chalcogen elements mixed with halides such as iodine. These so-called chalcohalide glasses afford the advantage of longer wavelength transmission than pure chalcogenide glass fibers.

The most popular compositions studied today are the quaternary systems based on tellurium. These are the TeX glass systems and one of the most popular is Te-Se-As-I. For this TeX glass, the halide \(X=I^-\). In thin window-type samples, TeX glasses transmit from 1 to 20 \(\mu\)m. Thus, their transparency range extends further than the standard telluride glasses fabricated without any halogen. Most of these glasses have rather low values of \(T_g\) usually about 150\(^\circ\)C or less.

The preparation of the TeX glasses is very similar to that for the chalcogenides. The starting materials are first purified and then a rod-in-tube preform is made by a rotational casting method.

Blanchetiere et al. at the University of Rennes has done much of the work on these glasses and on fiber drawing. For the core glass composition, they have chosen Te\(_2\)Se\(_{3.9}\)As\(_{3.1}\)I, and for the clad glass, Te\(_2\)Se\(_4\)As\(_3\)I. The refractive indices for the two glasses are 2.8271 and 2.8205 at 10.6 \(\mu\)m for the core and clad, respectively, giving a fiber numerical aperture (NA) of about 0.2.

Fiber drawing from the preform was done at about 200\(^\circ\)C, with a drawing speed of 0.5–3.5 m/min. An online, UV acrylate coating was applied to the fiber for protection against moisture and, of course, to improve the strength of the fiber.

The losses for a core-only and a core/clad TeX fiber, made by Lucas’ group at the University of Rennes, are presented in Fig. 13.6. From the data in Fig. 13.6, it can be seen that the minimum attenuation occurs between 7 and 9 \(\mu\)m.

The minimum loss for the core/clad fiber is about 1 dB/m at 9 \(\mu\)m. A core-only fiber with an acrylate coating has a slightly lower loss of 0.5 dB/m at 9 \(\mu\)m, but the acrylate coating on this fiber exhibits absorption especially in the 8- to 10-\(\mu\)m region. The core/clad TeX fiber shows some strong extrinsic absorption bands.

From Fig. 13.6, there is strong absorption due to OH\(^-\) (3 \(\mu\)m), Se-H (4.6 \(\mu\)m), and H\(_2\)O (6.3 \(\mu\)m). The data also show that the loss at 10.6 \(\mu\)m is rather large, but it is significantly better at 9.3 \(\mu\)m where a CO\(_2\) laser can operate. Using a 9.3 \(\mu\)m CO\(_2\) laser, they were able to transmit 2.6 W of laser power through a 600-\(\mu\)m diameter, 1-m long fiber.

3. Crystalline Fibers

Crystalline IR fibers are an attractive alternative to glass IR fibers because most nonoxide crystalline materials can transmit longer wavelength radiation than IR glasses and, in the case of sapphire, exhibit some superior physical properties.

One disadvantage over glass fibers is that crystalline fibers are somewhat difficult to fabricate because crystalline materials do not have a glassy region, so they cannot be drawn into fiber as is done with glasses.

Crystalline fibers must be fabricated either using modified crystal-growth techniques in which a fiber is pulled from the melt or by heating the crystal to temperatures below the melting point and then applying significant pressure to extrude the material through a die.

There are two types of crystalline fiber: polycrystalline (PC) and single-crystal (SC) fiber. Historically, the first crystalline fiber made was the PC fiber, KRS-5 (TlBrI). This fiber was fabricated by a hot-extrusion technique at Hughes Research Labs in 1976.

KRS-5 was chosen because it is very ductile and because it can transmit beyond the 20 \(\mu\)m range required for the intended military surveillance satellite application. Today the best PC fibers, made from silver halide crystals, have losses in the 0.3 dB/m range at 10.6 \(\mu\)m. Nevertheless, the Ag-halide PC fibers continue to be popular today for short-length applications in sensor systems and for limited use in low-power laser delivery.

There has been comparatively less work on SC fiber optics. One reason for this is that they are much harder to fabricate than PC fibers. Only a few crystalline host materials have been studied, with the most important being the refractory oxides. Of these, sapphire fiber (Al\(_2\)O\(_3\)) is the most studied SC fiber, and it has the lowest loss.

4. Polycrystalline (PC) Fibers

There are many halide crystals that have excellent IR transmission, but only a few have been fabricated into fiber optics. The technique used to make PC fibers is hot extrusion. As a result, only the silver and thallium halides have the requisite physical properties such as ductility, low melting point, and independent slip systems to be successfully extruded into fiber.

In the hot extrusion process, a single-crystal billet or preform is placed in a heated chamber and the fiber extruded to net shape through a diamond or tungsten carbide die at a temperature about one-half the melting point. The final PC fibers are usually from 500 to 900 \(\mu\)m in diameter with no buffer jacket. The polycrystalline structure of the fiber consists of grains on the order of 10 microns or larger in size.

The billet may be clad using the rod-in-tube method. In this method, a mixed silver halide such as AgBrCl is used as the core and then a lower index tube is formed using a Cl\(^-\) rich AgBrCl crystal. The extrusion of a high-quality core–clad fiber is not easy because the extrusion process distorts the core–clad interface, often leaving a highly irregular core region.

Artjushenko et al. at the General Physics Institute in Moscow have achieved excellent clad Ag-halide fibers with losses nearly as low as the core-only Ag-halide fiber. At Tel Aviv University, high-quality multimode and single-mode Ag-halide fibers have been produced.

Today, the PC Ag-halide fibers represent the best PC fibers. KRS-5 is no longer a viable candidate largely because of the toxicity of Tl and the greater flexibility of the Ag-halide fibers.

The optical losses in PC fibers are well above the intrinsic loss of the bulk material. In general, the best PC fibers made have losses between 0.3 and 0.5 dB/m around 10 \(\mu\)m. This is typically the lowest loss region, which is fortuitous because many applications for these fibers involve the transmission of CO\(_2\) laser radiation. The losses for both core-only and core–clad silver halide fibers are shown in Fig. 13.7.

The core-only fiber (A) is the AgBrCl fiber extruded by Moser et al. at Tel Aviv University, while fiber (B) is a core–clad AgBrCl fiber fabricated by Artjushenko in Moscow and ART Photonics in Berlin. Both loss curves represent the current technology of silver halide fibers and, in the case of fiber (B), the quality of commercially available fiber. The core–clad fiber is available with an NA of 0.15 or 0.3. The core diameters of the fibers range from 500 to 900 \(\mu\)m. The lengths of the fibers typically do not exceed 3 or 4 m but may extrude fibers as long as 20 m.

There are interesting features of the loss data in Fig. 13.7. First, there are several impurity absorption bands due to water at 3 and 6.3 \(\mu\)m and sometimes a SO\(_4^{--}\) absorption near 9.6 \(\mu\)m. These bands are seen in the core-only fiber but are less evident in the core–clad fiber. Presumably this is due in part to the presence of the clad layer, which protects the core from contamination by water and other ions during extrusion.

Furthermore, we note the increasing attenuation as the wavelength decreases. This is a result of \(\lambda^{-2}\) scattering from strain-induced defects in the extruded fiber. An important feature of the data is that the minimum loss is near 10.6 \(\mu\)m. These fibers have been used to transmit about 100 W of CO\(_2\) laser power, but the safe limit seems to be 20 to 25 W. The higher powers can more easily damage the fiber as a result of the low melting point of the fibers.

There are several difficulties in handling and working with PC fibers. One is an unfortunate aging effect in which the fiber transmission is observed to decrease in time. Normally the aging loss, which increases uniformly over the entire IR region, is a result of strain relaxation and possible grain growth as the fiber is stored.

Another problem is that Ag halides are photosensitive and exposure to visible or UV radiation creates colloidal Ag, which in turn leads to increased losses in the IR. Finally, the AgBrCl is corrosive to many metals. Therefore, the fibers should be packaged in dark jackets and connectorized with materials such as Ti, Au, or ceramic materials.

The mechanical properties of these ductile fibers are quite different from those of glass fibers. The fibers are weak, with ultimate tensile strengths of about 80 MPa for a 50/50 mixture of AgBrCl. The main difference, however, between the PC and glass fibers is that the PC fibers plastically deform well before fracture.

This plastic deformation leads to increased loss as a result of increased scattering from separated grain boundaries. Therefore, in use, the fibers should not be bent beyond their yield point; too much bending can lead to permanent damage and a high loss region in the fiber.

5. Single-Crystal (SC) Fibers

The most common and viable SC fiber developed is sapphire. Sapphire is an insoluble uniaxial crystal with a melting point of 2053\(^\circ\)C. It is an extremely hard and chemically inert material that may be conveniently melted and grown in air.

The usable fiber transmission is from about 0.5 to 3.2 \(\mu\)m. Other important properties include a refractive index of 1.75 at 3 \(\mu\)m, a thermal expansion about 10 times higher than silica, and a Young’s modulus approximately 6 times greater than silica.

These properties make sapphire an almost ideal IR fiber candidate for applications less than about 3.2 \(\mu\)m. In particular, this fiber has been used to deliver more than 10 W of average power from an Er/YAG laser operating at 2.94 \(\mu\)m. Oxide materials like Al\(_2\)O\(_3\) have the advantage of high melting points and chemical inertness, and they may be conveniently melted and grown in air.

Sapphire fibers are fabricated using either the edge-defined, film-fed growth (EFG) [38] or the laser-heated pedestal growth (LHPG) techniques. In either method, some or all of the starting sapphire material is melted and an SC fiber is pulled from the melt.

In the EFG method, a capillary tube is used to conduct the molten sapphire to a seed fiber, which is drawn slowly into a long fiber. Multiple capillary tubes, which also serve to define the shape and diameter of the fiber, may be placed in one crucible of molten sapphire so that many fibers can be drawn at one time.

The LHPG process is a crucibleless technique in which a small molten zone at the tip of a SC sapphire source rod (<2 mm diameter) is created using a CO\(_2\) laser. A seed fiber slowly pulls the SC fiber as the source rod continuously moves into the molten zone to replenish the molten material.

Both SC fiber growth methods are very slow (several mm/min) compared to glass fiber drawing. The EFG method, however, has an advantage over LHPG methods because more than one fiber can be continuously pulled at a time.

LHPG methods, however, have produced the cleanest and lowest loss fibers because no crucible is used, which can contaminate the fiber. The sapphire fibers grown by these techniques are unclad, pure Al\(_2\)O\(_3\) with the C axis usually aligned along the fiber axis.

Fiber diameters range from 100 to 300 \(\mu\)m and lengths are generally less than 2 m. In general, the fibers are all unclad, but it is possible to add a polymer coating such as Teflon using heat-shrink tubing.

The optical properties of the as-grown sapphire fibers are normally inferior to those of the bulk starting material. This is particularly evident in the visible region and is a result of color-center type defect formation during the fiber draw. These defects and the resulting absorption can be greatly reduced if the fibers are postannealed in air or oxygen at about 1000\(^\circ\)C.

In Fig. 13.8, the losses for LHPG fiber grown at Rutgers University and EFG fiber grown by Saphikon, Inc. (Milford, NH), now St. Gobain Crystals, are shown. Both fibers have been annealed at 1000\(^\circ\)C to reduce short-wavelength losses.

We see that the LHPG fiber has the lowest overall loss. In particular, LHPG fiber loss at the important Er/YAG laser wavelength of 2.94 \(\mu\)m is less than 0.3 dB/m compared to the intrinsic value of 0.15 dB/m. There are also several impurity absorptions beyond 3 \(\mu\)m that are believed to be due to transition metals like Ti or Fe. Sapphire fibers have also been used at temperatures up to 1400\(^\circ\)C without any change in their transmission.

Young’s modulus for sapphire is very high. In fact, the modulus for sapphire is about six times greater than that for silica. In practice, this means that SC sapphire fibers are rather stiff, a feature readily observed when one bends equal-diameter sapphire and silica fibers.

There has been only limited strength data taken on the optical SC sapphire fibers. Wu et al. have measured 110-\(\mu\)m diameter LHPG fiber under tension. Their measurements yielded failure strains between 1.20 and 1.85%. These are measurements on only two samples, so their statistics are quite limited. Photran, Inc. (formerly Saphikon) claims that its 325 \(\mu\)m diameter fiber can be bent into a 60-mm loop. Jundt et al. indicate that their 150-\(\mu\)m diameter LHPG fiber can be bent to a 4-mm radius.

6. Hollow-Core Waveguides

The first optical frequency hollow waveguides were similar in design to microwave guides. Garmire et al. made a simple rectangular waveguide using aluminum strips spaced 0.5 mm apart by bronze shim stock. Even when the aluminum was not well polished, these guides worked surprisingly well. Losses at 10.6 \(\mu\)m were well below 1 dB/m and Garmire et al. demonstrated early on the high-power handling capability of an air-core guide by delivering more than 1 kW of CO\(_2\) laser power through this simple structure.

These rectangular waveguides, however, never gained much popularity primarily because their overall dimensions (~ 0.5 x 10mm) were quite large in comparison to circular cross-section guides and because the rectangular guides cannot be bent uniformly in any direction. As a result, hollow circular waveguides with diameters of 1 mm or less fabricated using metal, glass, or plastic tubing are the most common guide today.

In general, hollow waveguides are an attractive alternative to conventional solid-core IR fibers for laser power delivery because of the inherent advantage of their air core. Hollow waveguides not only enjoy the advantage of high laser power thresholds but also low insertion loss, no end reflection, ruggedness, and small beam divergence.

A disadvantage, however, is a loss on bending that varies as \(1/R\), where \(R\) is the bending radius. In addition, the losses for these guides vary as \(1/a^3\), where \(a\) is the radius of the bore. Unfortunately, this means that the flexibility of the very small bore (approximately <250 \(\mu\)m) guides is somewhat negated by their higher loss. However, the loss can be made arbitrarily small for a sufficiently large core.

The bore size and bending radius dependence of all hollow waveguides is a characteristic of these guides not shared by solid-core fibers. Initially these waveguides were developed for medical and industrial applications involving the delivery of CO\(_2\) laser radiation, but they have been used to transmit incoherent light for broadband spectroscopic and radiometric applications. They are today one of the best alternatives for power delivery in IR laser surgery and industrial laser delivery systems with losses as low as 0.1 dB/m and transmitted cw laser powers as high as 2.7 kW.

Hollow-core waveguides may be grouped into two categories: (1) those whose inner core materials have refractive indices greater than one (leaky guides) and (2) those whose inner wall material has a refractive index less than one (attenuated total reflectance, i.e. ATR, guides).

Leaky or \(n\gt1\) guides have metallic and dielectric films deposited on the inside of metallic, plastic, or glass tubing. ATR guides are made from dielectric materials with refractive indices less than one in the wavelength region of interest. Therefore, \(n\lt1\) guides are fiber-like in that the core index (\(n\approx1\)) is greater than the clad index. Hollow sapphire fibers operating at 10.6 \(\mu\)m (\(n=0.67\)) are an example of this class of hollow guide.

6.1. Hollow Metal and Plastic Waveguides

The earliest circular cross-section hollow guides were formed using metallic and plastic tubing as the structural members. Tubing made from stainless steel or nickel was used in many of the early guides.

One of the most successful approaches was used by Miyagi and his group in Japan. In their method they first sputter deposited Ge, ZnSe, and ZnS coatings on aluminum mandrels. Then a layer of Ni was electroplated over these coatings before the aluminum mandrel was removed by chemical leaching. The final structure was then a flexible Ni tube with optically thick dielectric layers on the inner wall to enhance the reflectivity in the infrared.

These guides had losses near 0.2 dB/m at 10.6 \(\mu\)m, but the hardness and springy character of the nickel tubing can be a disadvantage because it is less flexible than glass or plastic tubing.

Plastic tubing seems almost ideal in that it is very flexible and inexpensive. At Tel Aviv University, Dahan et al. applied Ag followed by AgI coatings on the inside of polyethylene and Teflon tubing to make a very flexible waveguide. The problem with these plastic materials is that they tend to be too soft and the inside surfaces are somewhat rough.

The softness leads to deformation of the circular cross-section on bending and the roughness increases the scattering losses. Better results are obtained when harder polymers like polycarbonate tubing is used. George and Harrington made very low loss (0.05 dB/m at 10.6 \(\mu\)m) waveguides using polycarbonate tubing.

6.2. Hollow Glass Waveguides

One of the most popular hollow waveguides today is the hollow glass waveguide (HGW) developed by Harrington at Rutgers University. The advantage of the hollow glass structure over other hollow structures is that it is simple in design, flexible, and most important, it has a very smooth inner surface. HGWs have a metallic layer of Ag on the inside of silica glass tubing and then a dielectric layer of either AgI or a polymer-like cyclic olefin polymer (COP) is deposited over the metal film.

Figure 13.9 shows a cross-section of an HGW with Ag/AgI coatings. The fabrication of the Ag/AgIHGWbegins with silica tubing, which has a polymer (UV acrylate or polyimide) coating on the outside surface. A liquid-phase chemistry technique is used to deposit the Ag and AgI films inside the glass tubing. This technique is similar to that used by Croitoru et al. to deposit metal and dielectric layers on the inside of plastic tubing.

The first step involves depositing a silver film using standard Ag plating technology. Generally, the Ag film is between 0.2 and 1 \(\mu\)m thick, and it is deposited slowly over about 1 hour. Immediately after the silver is deposited, an iodine solution is pumped through the tubing, and through a subtraction process, a layer of AgI is formed.

By controlling the concentration of the iodine solution and the reaction time, an AgI film of the correct optical thickness can be deposited. Using these methods, HGWs with bore sizes ranging from 100 to 1200 \(\mu\)m and lengths as long as 13 m have been made.

The measured spectral response for two different 700-\(\mu\)m bore, 1-m long HGWs is given in Fig. 13.10. One guide has a 0.3-\(\mu\)m thick AgI film and the other a 0.8-\(\mu\)m film. The spectral response of the HGW with the thinner AgI layer is appropriate for the shorter IR wavelengths as, for example, for transmission of Er/YAG laser energy.

The thicker AgI film gives the lowest loss at 10.6 \(\mu\)m. Clearly, for broadband IR applications, it is desirable to have a thin AgI layer. It would seem, therefore, that merely making even a thinner dielectric layer would yield a waveguide that would transmit well into the visible region of the spectrum.

It is possible in principle to tailor the optical response to achieve short wavelength transmission, but this is very difficult using AgI films. The reason is that AgI (purple color) does not transmit well in the visible region. A better approach for visible wavelengths is to use a transparent polymer dielectric film over the Ag layer.

Miyagi and his group have successfully used COP for visible and IR transmission. An advantage of this approach is that the hollow waveguide will transmit visible radiation from, for example, a visible laser and this will provide an aiming beam along with the IR energy.

The straight-guide loss data for several bore sizes using CO\(_2\) and Er/YAG lasers are shown in Fig. 13.11. The solid curves are theoretical calculations of the losses for the lowest order HE\(_{11}\) mode showing the \(1/a^3\) dependence predicted by Marcatili and Schmeltzer.

At the CO\(_2\) laser wavelengths, the calculated losses agree quite well with those measured. However, at 3 \(\mu\)m the calculated losses are much lower than the measured values. This is a result of increased scattering losses at the shorter wavelengths and the multimode character of the Er/YAG laser used in the measurements.

The increased loss on bending HGWs is shown for two 530-\(\mu\)m bore guides in Fig. 13.12. These data show that the loss increases as the curvature or \(1/R\) increases. All data were taken with a constant length of waveguide under bending. A curvature of 20 m\(^{-1}\) represents a bend diameter of only 10 cm! This is sufficiently small for most applications.

It is important to note that while there is an additional loss on bending for these hollow guides, it does not necessarily mean that this restricts their use in power delivery or sensor applications.

Normally most fiber delivery systems have rather large bend radii and, therefore, a minimal amount of the guide is under tight bending conditions and the bending loss is low.

From the data in Fig. 13.12, one can calculate the bending loss contribution for an HGW link by assuming some modest bends over a small section of guide length. An additional important feature of hollow waveguides is that they are nearly single mode.

This is a result of the strong dependence of loss on the fiber mode parameter. That is, the loss of high order modes increases as the square of the mode parameter so even though the guides are very multimode, in practice only the lowest order modes propagate.

For example, a less than 300-\(\mu\)m bore guide will operate virtually single mode at 10 \(\mu\)m. HGWs have been used quite successfully in IR laser power delivery and, more recently, in some sensor applications. Modest CO\(_2\) and Er/YAG laser powers below about 80 W can be delivered without difficulty.

At higher powers, water-cooling jackets have been placed around the guides to prevent laser damage. The highest CO\(_2\) laser power delivered through a water-cooled, hollow metallic waveguide with a bore of 1800 \(\mu\)m was 2700 W and the highest power through a water-cooled 700-\(\mu\)m bore HGW was 1040 W.

Sensor applications include gas and temperature measurements. A coiled HGW filled with gas can be used in place of a more complex and costly White cell to provide an effective means for gas analysis.

Unlike evanescent wave spectroscopy in which light is coupled out of a solid-core– only fiber into media in contact with the core, all of the light is passing through the gas in the hollow guide cell, making this a sensitive, quick response fiber sensor. Temperature measurements may be aided by using an HGW to transmit blackbody radiation from a remote site to an IR detector. Such an arrangement has been used to measure jet engine temperatures.

7. Summary

During the past 30 years of the development of IR fibers, there has been a great deal of fundamental research designed to produce a fiber with optical and mechanical properties close to that of silica.

IR fibers are still far from that Holy Grail, but some viable IR fibers have emerged that can be used to address some of the needs for a fiber that can transmit greater than 2 \(\mu\)m. Yet the current IR fiber technology is still limited by high loss and low strength.

Nevertheless, more applications are being found for IR fibers as users become aware of their limitations and, more importantly, how to design around their properties.

There are two near-term or short-length applications of IR fibers: laser power delivery and sensors. An important future application for these fibers, however, is in active fiber systems like the Er- and Pr-doped fluoride fibers and doped chalcogenide fibers.

As power-delivery fibers, the best choice seems to be hollow waveguides for CO\(_2\) lasers and either SC sapphire, germanate glass, or HGWs for Er/YAG laser delivery.

Chemical, temperature, and imaging bundles make use of mostly solid-core fibers. Evanescent wave spectroscopy (EWS) using chalcogenide and fluoride fibers is quite successful.

A distinct advantage of an IR fiber EWS sensor is that the signature of the analyte is often very strong in the infrared or fingerprint region of the spectrum. Temperature sensing generally involves the transmission of blackbody radiation.

IR fibers can be very advantageous at low temperatures, especially near room temperature where the peak in the blackbody radiation is near 10 \(\mu\)m. Finally, there is an emerging interest in IR imaging using coherent bundles of IR fibers. Several thousand chalcogenide fibers have been bundled by Amorphous Materials (Garland, TX) to make an image bundle for the 3- to 10-\(\mu\)m region.

The next tutorial discusses about fundamentals of laser oscillation.