Introduction to Photosensitive Fibers

This is a continuation from the previous tutorial - introduction to integrated optics.

 

1. Introduction

Photosensitivity is another amazing feature of most specialty optical fibers (SOFs). The photosensitivity phenomenon is different from photo-darkening and radiation-darkening, which induces excess losses. There is then no added background losses due to the fiber Bragg gratings (FBGs) inscribed in the fiber’s core besides filtered wavelengths.

Photosensitivity of a medium is defined as its capacity to have its refractive index permanently changed by a modification of its physical or chemical properties through light exposition.

Photosensitivity is a complex phenomenon because of the diversity of both parameters and effects that are observed. Fiber composition, fabrication process, operation wavelength, and even light source are all different parameters that can have a significant influence on the photosensitive properties of a fiber. Effect of interest in this tutorial is the photosensitivity of Ge-doped silica fiber for which the core refractive index can be permanently modified by an ultraviolet (UV) irradiation.

In 1978, photosensitivity was first observed by Hill et al. at the Communication Research Centre (CRC) in Canada. To improve the performance of his tunable fiber Raman laser, Ken Hill ordered two spools of high numerical aperture (HNA) single-mode SOF from Bell Northern Research (BNR). These fibers turned out to be inadequate for use in the fiber Raman laser but accidentally led to the discovery of the photosensitivity.

The experiment consisted of injecting light from a single-frequency Argon laser (514 nm) into the core of a doped silica fiber. Hill observed that a fraction of the input power was being reflected by the fiber itself and this phenomenon was attributed to the formation of a permanent index grating permanently inscribed in the fiber core by the light propagating in the fiber.

Progress in optical fiber photosensitivity research was initially slow, probably because the scientific community perceived the phenomenon as a peculiarity of the BNR fibers.

An important development was realized 11 years later by Meltz et al., who reported the formation of FBGs by a transverse holographic method. It consisted of illuminating the core from the fiber’s side with the interference pattern of two beams of coherent UV radiation in the 244-nm germania oxygen-vacancy defect band.

By using this holographic method, the periodic gratings could then be photo-imprinted (to resonate at wavelengths in the infrared [IR] region of the spectrum) to operate as Bragg reflectors at any wavelength longer than the writing wavelength, rather than be resonant near the writing wavelength.

The work of Meltz et al. motivated the optical fiber communication community because of the increased possibility that grating-based devices could be fabricated in the IR region, which is their main spectral region of interest. Moreover, side-written gratings can be produced in a few minutes (in contrast with relief gratings), with low loss, and without any other modification of the fiber.

With this new technology and its potential applications, Meltz et al. generated a huge amount of interest, resulting in worldwide research into the fabrication and applications of side-photo-imprinted fiber gratings.

In less than 5 years, progress has been extremely rapid on many fronts. One key point of the technology development is the use of phase masks for writing FBGs instead of using the transverse holographic method, with first a static writing through the mask and next with the translation of the UV beam through the mask into the fiber core. The phase mask method simplified the fabrication of FBGs and permitted them to get longer gratings. Using a phase mask, Hill wrote the first fiber gratings in the spring of 1992. Currently, the phase mask method is still the principal method for manufacturing fiber gratings.

With the emergence of multiple applications based on FBGs, it appeared necessary to enhance the photosensitivity of optical fibers. A lot of work has been performed in order to increase the intrinsic photosensitivity of Ge-doped fibers by modifying its chemical composition with the addition of constituents like boron, phosphorous, aluminum, cerium, and others.

New methods relying on postfabrication techniques, such as flame-brushing and hydrogen-loading, have also been developed to enhance the Ge-doped fibers’ photosensitivity. At this time, the molecular hydrogen diffusion method, which has been developed at AT&T Bell Labs in 1993, is one of the most widely used methods to enhance the two-photon absorption mechanism involved in the 244- to 248-nm writing.

Photosensitivity in Ge-doped silica fibers is now a well-known and mature subject. Several mechanisms contribute to and explain this phenomenon: changes of color centers, volume, or stress.

The discovery of fiber photosensitivity led to the realization of optical components based on FBGs or long-period gratings. Reflectors for fiber lasers, wavelength division filters, gain-flattening filters for fiber amplifiers, chirped FBGs for pulse compression and dispersion compensation, short- or long-period gratings for temperature, strain, and pressure sensors are all examples showing that photosensitivity revolutionized both the optical fiber communication and the optical fiber sensor field. Before the advent of photosensitive fibers and FBGs, these components were made out of bulk optics.

 

2. Design and Fabrication

Since the discovery of photosensitivity in optical fibers, many designs driven by the desire to extend the photosensitive fiber’s limited performances have been proposed.

First, the enhancement of the photosensitivity of standard single-mode fibers (SSMFs) was demonstrated to be efficiently realized by hydrogen (H₂) loading. The HNA fiber design, which requires a high Ge concentration in the core, turned out to be intrinsically photosensitive; moreover, it offsets the FBGs cladding modes in transmission. This type of fiber led Hill’s discovery of the photosensitivity phenomenon. The cladding mode suppression (CMS) design further enhanced the FBG bandwidth in transmission.

The rare earth–doped (RED) photosensitive fiber design facilitated the building of short-cavity fiber lasers. The polarization maintaining (PM) photosensitive fiber was designed for applications requiring a photosensitive fiber that maintains the polarization plane of the transmitted light in the fiber.

Furthermore, the photosensitivity of soft glass fluoride fibers also facilitated the building of short-cavity fiber lasers. Polymeric fiber, for which FBGs can be tuned over a wider range than glass FBGs, was also investigated for its sensor applications.

In each case, the fiber design consisted of simulating the optimal cross-section, in terms of refractive index profile (RIP), geometry, and chemical composition, to achieve the requested fiber performances.

The photosensitive chemical elements are nonmetal in groups IIIA, IVA, and VA of the periodic table, combined with oxygen to form oxide glasses (Table 9.1). The different photosensitive fiber designs are treated separately in subsequent sections in this tutorial.

 

 

Photosensitive fibers like all other SOFs are fabricated by drawing optical preforms. The fibers are then coated on-line by standard UV-cured acrylate. The know-how rests in preform design and fabrication.

The silica fiber preforms are fabricated by different chemical vapor-deposition (CVD) processes including modified CVD (MCVD), plasma CVD (PCVD), outside vapor deposition (OVD), and vertical axial deposition (VAD). Soft-glass fluoride fibers are fabricated by glass melting and preform assembly called the ‘‘rod-in-tube’’ method.

The photosensitivity level of silica fibers is critically dependent on several MCVD preform fabrication parameters, in particular the collapsing procedure, the atmospheric conditions during collapsing, and the concentration of the codopants in the fiber core.

In the industry, preform collapsing is commonly performed under reduced atmosphere to favor the germanium-related oxygendeficient centers (GODCs). Another MCVD approach developed by Dianov et al. is the deposition of white soot layers and their sintering under a reduced atmosphere.

 

3. Standard Numerical Aperture Fibers

Over the years, a standard numerical aperture (NA) of about 0.12 was established for SSMFs used by the optical communication industry. The germanium dioxide (GeO₂) concentration necessary to obtain such an NA is about 4 mol% and the fiber RIP is of the step-index type with a \(\Delta{n}\) (defined by \(n_\text{core}-n_\text{clad}\)) of about 0.005.

The core dopant choice, RIP, and NA of the SSMF are the result of minimization of the fiber losses at 1310 and 1550 nm, but the enhancement of the photosensitivity was not considered when optimizing these fiber parameters.

The development of a highly photosensitive fiber having the same optical characteristics as the SSMF was realized by boron doping of germano-silicate fibers, as well as antimony doping and tin doping. The design, fabrication particularities, characteristics, and applications of these types are described in the following subsections.

 

3.1. Standard Single-Mode Fibers

Since the introduction of FBG as a commercial product in 1995, the use of FBG has increased in the telecommunications and the sensors fields, with the development of new components built with SSMFs.

Indeed, the filtering capabilities (small insertion loss, low transmission loss, high wavelength selectivity) of FBGs make them an ideal candidate as well for add–drop filters and dispersion compensators than for temperature or strain sensors.

Most of the FBG applications require gratings with high reflectivity, so fiber photosensitivity is an important factor in the development of specific FBGs. For example dense wavelength division multiplexers (DWDMs) require highly photosensitive fibers for efficient inscription of several gratings.

Moreover, the performances of a system depend on some FBG characteristics like bandwidth. FBGs for WDM filters have very narrow spectral widths, whereas FBGs for dispersion compensators have wider bandwidths.

Intrinsically, the SSMF photosensitivity is too small for most of these applications. Typically, without any sensitizations, UV-induced index changes are limited to about \(3\times10^{-5}\). However, it is possible to overcome this problem by enhancing the photosensitivity of SSMF through hydrogen (H₂) loading and then increase refractive index modulation to the order of \(10^{-3}\).

Similar results have been obtained using other MCVD and VAD SSMFs. The mechanism is, therefore, not dependent on fiber or preform processing, but on the interactions between GeO₂ and H₂ molecules, coupled with the UV exposure conditions.

Various sensitization methods have been developed including UV hypersensitization, \(\text{OH}^-\) flooding, and under strain. From a practical point of view, the H₂ loading method remains the best technological solution for writing low-loss, strong short-period gratings with a superior Bragg wavelength stability.

 

3.2. Boron-Doped Germano-Silicate Fibers

Fiber photosensitivity can be enhanced by increasing the germanium concentration. Boron co-doping is then used to lower the RIP of the Ge-enriched fibers, so a larger amount of GeO₂ can be incorporated without increasing the NA.

Boron co-doped fiber has an excellent photosensitive response. Konstantaki et al. studied the effects of Ge concentration, boron co-doping, and hydrogenation on FBG characteristics.

They showed that an unloaded boron co-doped fiber increased the photosensitivity only slightly. In contrast, hydrogenation with the presence of boron in the fiber core significantly enhances the photo-induced increase rate of the refractive index and results in FBGs with wider bandwidths.

The modulation depth of the induced refractive index change is stronger and results in FBGs with high reflectivity. Different photosensitivity-enhancement mechanisms observed individually due to boron co-doping and hydrogenation contribute additively to the final photosensitivity in hydrogenated boron codoped fibers.

Typical B₂O₃ and GeO₂ concentrations are 8 and 12 mol%. INO’s fiber model PS-SMF-30 corresponds to this type of fiber (Table 9.2). The photo-induced refractive index change is about \(3\times10^{-3}\).

 

 

This type of fiber is mainly fabricated using the MCVD technique. First, a standard P₂O₅ and F co-doped silica cladding is deposited. Afterwards, SiCl₄, BCl₃, and GeCl₄ vapors carried by O₂ gas are used as core precursors for the MCVD process. The collapse is under nitrogen (N₂) to favor the GODCs.

A typical preform chemical composition profile obtained by electron probe microanalysis is shown in Fig. 9.1.

 

 

For SSMF, the NA or \(\Delta{n}\) is the same for the preform and fiber (i.e., there is no change in the \(\Delta{n}\) during the drawing process). This is not the case for boron-doped germano-silicate fibers.

The \(\Delta{n}\) is lowered during the drawing process because of some stress effect, which means that the \(\Delta{n}\) of the preform must then be higher to compensate for the \(\Delta{n}\) reduction during the drawing.

To illustrate this fact, a typical preform RIP having a maximum \(\Delta{n}\) of 0.012 is shown in Fig. 9.2 and the corresponding fiber RIP with a maximum \(\Delta{n}\) of 0.009 is shown in Fig. 9.3. Hence, the \(\Delta{n}\) was lowered by 0.003 during the drawing.

Another difference that can be observed from Fig. 9.3 is that the index dip observed in the PS-SMF-30 is wider and deeper than the one typically observed for SSMF.

 

 

 

As can be seen on Fig. 9.4, the background loss is also higher for this fiber compared to the SSMF, especially in the telecommunication window of 1.55 μm due to the higher phonon energy of the boron dopant. However, it is worth noting that the higher background loss does not limit the applications because of the short length required per component. All other characteristics are similar to SMF.

 

 

3.3. Antimony-Doped Fibers

A standard NA antimony (Sb) doped silica optical fiber was developed by Oh et al..The RIP and the index difference (\(\Delta{n}\)) between the core and cladding were similar to the SSMF.

In D₂-loaded samples, they observed UV photosensitivity with an initial refractive index growth rate six times higher than in SSMF. The fiber core diameter was 8.5 μm and the Sb₂O₃ concentration was 3.5 mol%.

The attenuation at 1550 nm was 700 dB/km, which is very high, and it was attributed mainly to the OH absorption induced in the wet sol-gel process. Antimony is a promising element, but the background losses problem would need to be solved.

 

3.4. Tin-Doped Fibers

A standard NA tin (Sn)-doped silica optical fiber was developed by Brambilla et al.. The fiber preform was fabricated by the MCVD process by depositing a soot rich in SnO₂ at 1300\(^\circ\)C and then sintered at 1800\(^\circ\)C.

During the collapse process, the temperature was kept under 2200\(^\circ\)C to avoid massive SnO₂ volatilization. The core NA was about 0.1 with a large index dip and the cutoff wavelength was about 1300 nm. The fiber exhibited a moderate photosensitivity with an index modulation of up to \(2.5\times10^{-4}\).

 

4. High Numerical Aperture Fibers

The HNA refers to silica optical fibers with an NA from about 0.2 to 0.4 maximum. This design, requiring a high Ge concentration in the core, turned out to be intrinsically photosensitive, which historically played an important role in Hill’s discovery of the photosensitivity phenomenon.

Because of the HNA, the fiber needs a smaller core diameter than with the SSMF to be single mode in the operating wavelength between 1310 and 1610 nm. The number of modes supported by an optical fiber is described by the fiber parameter \(V\), given by

\[\tag{9.1}V=\frac{2\pi{a}}{\lambda}\text{NA}\]

where \(a\) is the core radius and \(\lambda\) is the wavelength.

For a given design, the fiber is single mode when the \(V\) parameter is smaller than 2.405 (i.e., for wavelengths longer than the LP\(_{11}\) cutoff wavelength [\(\lambda_c\)] of the fiber).

For optimal performance, \(\lambda_c\) should be lower than the operating wavelength by about 50 nm. Depending on the NA, the core diameter will range from 2 to 7 μm. The higher the NA, the smaller the core diameter has to be for the fiber to remain single mode.

The mode-field diameter (MFD) will also be smaller for the HNA fiber than for the SSMF, leading to a higher splice loss to SSMF due to mode mismatch. Typically, the splice losses between an HNA fiber and an SSMF are < 0.2 dB, whereas they are less than 0.05 dB between two SSMFs.

Another interesting feature of HNA fibers is that they push the cladding mode coupling loss on the short wavelength’s side from the Bragg wavelength in transmission. When the grating first starts to grow, wave vector matching requirements lead to a gap between the Bragg wavelength and the onset of the cladding mode coupling. The width of this initial offset is given by

\[\tag{9.2}\Delta\lambda_\text{off}=\lambda_\text{B}-\lambda_\text{L}=\Lambda(n_\text{eff}-n_\text{clad})\]

where \(\lambda_\text{B}\) is the Bragg wavelength grating defined by \(\lambda_\text{B}=2n_\text{eff}\Lambda\), \(\lambda_\text{L}\) is the longest wavelength of the cladding mode, \(n_\text{clad}\) is the cladding index, \(\Lambda\) is the grating pitch, and \(n_\text{eff}\) is the effective index of the propagating mode.

\(n_\text{eff}\) can be approximated by

\[\tag{9.3}n_\text{eff}\approx{n_\text{core}}+\Delta{n}_\text{UV}/2\]

where \(n_\text{core}\) is the refractive index of the core and \(\Delta{n}_\text{UV}\) is the change in the core index caused by the UV radiation, and, such that considering the \(\Delta{n}\), \(\Delta{\lambda}_\text{off}\) is now expressed as

\[\tag{9.4}\Delta\lambda_\text{off}\approx\Lambda(\Delta{n}+\Delta{n}_\text{UV}/2)\]

Therefore, \(\Delta\lambda_\text{off}\) becomes larger as \(\Delta{n}\) is increased. A larger \(\Delta\lambda_\text{off}\) causes the excess loss region, due to the cladding modes, to be shifted far from the Bragg reflection wavelength. For a moderate NA of 0.25, the \(\Delta\lambda_\text{off}\) is 4.5 nm and for an ultrahigh NA of 0.40, the \(\Delta\lambda_\text{off}\) is 12 nm. This feature is very important in developing DWDM components.

Furthermore, other elements were found to enhance the photosensitivity of HNA fibers including tin (Sn) and indium (In). The design, fabrication particularities, characteristics, and applications of these three types are described in the following subsections.

 

4.1. Heavily Ge-Doped Silica Optical Fibers

Intrinsically, the photosensitivity of heavily Ge-doped silica fibers is sufficient for most FBG applications. INO’s fiber model PS-HNA-40 is a photosensitive step-index HNA fiber, whose physical characteristics are listed in Table 9.3. The photo-induced refractive index change is \(1\times10^{-3}\).

HNA fibers are fabricated by drawing silica optical preforms obtained by the CVD processes, including MCVD, PCVD, OVD, or VAD. The preform fabrication by MCVD requires special care during the core deposition, precollapse, and collapse. Only SiCl₄ and GeCl₄ with an O₂ carrier gas are used as core precursors.

Compared to SSMF fabrication, the GeCl₄ flow must be increased and deposition temperature lowered to favor the Ge deposition. The precollapse and collapse steps are performed under nitrogen (N₂) atmosphere to favor the GODCs.

On the preform, the maximum \(\Delta{n}\) measured was 0.022. The RIP of the corresponding fiber is shown in Fig. 9.5. Note that this fiber has a maximum \(\Delta{n}\) of 0.021, which indicates that contrary to the standard NA case, the \(\Delta{n}\) difference between the preform and the fiber in the HNA scenario is negligible. The background loss is higher than an SSMF but lower than the boron co-doped germano-silicate fiber (Fig. 9.6). All other characteristics are similar to SSMF.

 

HNA fiber is intrinsically photosensitive and its primary purpose is to offset radiation and cladding modes in the transmission spectra of FBGs. Its low birefringence is also an attractive specification in DWDM applications.

 

4.2. Tin-Doped Germano-Silicate Fibers

An HNA tin (Sn)-doped germano-silicate optical fiber was developed by Dong et al.. The preform was fabricated by introducing SnCl₄ vapor in the MCVD process. The required extra SnCl₄ bubbler needs to be heated at 39\(^\circ\)C because of the low vapor pressure of SnCl₄.

Nitrogen instead of oxygen was used as the carrier gas for SnCl₄. Two SnO₂-GeO₂-SiO₂ porous soot layers were deposited at about 1250\(^\circ\)C and they were sintered into a transparent glass at about 1600\(^\circ\)C. The preform was then collapsed into a solid rod in the conventional manner.

The core diameter of the resulting fiber was 4.8 μm, the NA was in the 0.20 range, and the cutoff wavelength was about 1250 nm. Background losses at 1300 and 1550 nm were 3 and 2 dB/km, respectively, which are much lower than those with boron-doped germano-silicate fibers (25 dB/km and 115 dB/km at the same wavelengths).

The photo-induced refractive index change was \(3\times10^{-3}\), which is three times larger than that for heavily Ge-doped silica fibers but about the same as boron-doped germano-silicate fibers. Sn is an interesting photosensitive element but requires an extra SnCl₄ bubbler for the preform fabrication, which is not common among fiber manufacturers.

 

4.3. Indium-Doped Germano-Silicate Fibers

An HNA indium (In)-doped germano-silicate optical fiber was developed by Shen et al.. The In was chosen for its large cation size (80 pm) compared to that of Sn (71 pm). Higher temperature sustainability is, thus, expected for the FBG written in these fibers.

The preform was fabricated by the MCVD process followed by the solution doping technique. The solution was prepared from In₂O₃ powder instead of commonly used salts. The resulting preform core was brown instead of transparent as usual. The preform was drawn into a 125-μm fiber. The core diameter was 5 μm, the core NA was about 0.21, and the calculated cutoff wavelength was 1375 nm.

The In₂O₃ concentration was about 0.05 mol%, and the GeO₂ was about 9 mol%. The fiber exhibited a refractive index modulation of \(3.2\times10^{-4}\). Annealed FBGs in this fiber type survived a high temperature of 900\(^\circ\)C for 24 hours and even 1000\(^\circ\)C for more than 2 hours.

Unfortunately, the background loss was not reported by the authors, but the brown color of the core suggests it might be quite high. This specification needs improvement to classify indium as a good photosensitive element.

There are numerous applications for HNA fibers in areas such as filters, narrow-band reflectors for fiber lasers, optical strain/temperature sensors, and modal couplers. HNA fibers are also perfectly suitable to adapt the cladding mode offset in order to optimize the channel spacing in telecommunications applications.

 

5. Cladding Mode Suppression

The CMS design enhances the FBGs bandwidth in transmission. Add–drop filters or multiplexer–demultiplexer components can be fabricated at potentially low cost using this technology.

However, an FBG written in an SSMF induces coupling to the cladding modes in addition to coupling to the fundamental counter propagating mode LP\(_{01}\). This coupling to higher order cladding modes induces several resonance dips in the transmission spectrum on the short-wavelength side of the Bragg reflection.

The performance of such devices can be improved by using appropriately designed fibers. Several methods have been proposed to reduce this effect, but they do not give optimal performances.

The first design was an HNA fiber that allows shifting of the cladding modes to a shorter wavelength. A second design is the depressed cladding fiber, where the cladding mode coupling loss is reduced, but improvement is modest and the field distribution is different from an SSMF. The third design, proposed by Delevaque et al., consists of a photosensitive cladding to obtain a uniform photosensitivity over the spatial extent of the guided orthogonal mode that ensures a negligible coupling. This last design is still considered the best technical solution. The core of the CMS design can have a standard NA or an HNA.

The standard NA PS-RMS-28 fiber was fabricated by the MCVD process and the design consists of a matched photosensitive cladding made of F and Ge co-doped silica and a Ge-doped silica core. The core refractive index had a profile similar to that of SSMF and F has been incorporated into the cladding in sufficient concentration to lower the index of refraction to the silica level.

The PS-RMS-50 design consists of an HNA photosensitive fiber with the cladding mode suppression feature. The matched photosensitive cladding is made of B and Ge co-doped silica and the silica core is doped with Ge. The advantages of the PS-RMS-28 over PS-RMS-50 include an NA, MFD, and attenuation within the tolerances of the SSMF. The average splice loss of a PS-RMS-28 to an SSMF is about 0.03 dB at 1550 nm. Typical specifications for SSMF, PS-RMS-28, and PS-RMS-50 fibers are compared in Table 9.4.

To compare the cladding mode losses between different fibers, the FBG photo-inscription method is used to study the photosensitivity of the fibers. All of the gratings presented here have been imprinted using a CW frequency-doubled argon laser at 244 nm with a standard holographic phase mask.

The PS-RMS-50 fiber is intrinsically photosensitive and does not require H₂ loading. A 30-dB grating has been photo induced in this fiber and in an H₂-loaded SSMF. The exposure time was 2 minutes for the H₂-loaded SSMF and 1 minute for the PS-RMS-50, meaning that this fiber is twice as photosensitive, even without H₂ loading.

The cladding mode coupling losses in the PS-RMS-50 are below 0.1 dB, compared to almost 1.5 dB for the SSMF fiber. This substantial improvement is obtained by making the fiber cladding photosensitive.

Limitations of the PS-RMS-50 are the unwanted propagation losses, which are as high as 50 dB/km at 1550 nm. These high losses are due to the high concentration of B in the cladding, which is necessary to counteract the increase of the refractive index attributable to the high Ge content.

To overcome these limitations, a new type of fiber (PS-RMS-28) has been designed. The goal was to reduce the transmission losses while preserving the same low cladding mode losses. The low Ge concentration of the new fiber required hydrogen loading to make it more photosensitive.

The cladding modes coupling losses measured on an H₂-loaded PS-RMS-28 fiber were lower than 0.1 dB. For comparison, the transmission spectrum of the gratings imprinted in the SSMF and PS-RMS-28 is shown in Fig. 9.7. 

The first strong cladding mode coupling loss observed on the short-wavelength side of the Bragg peak in Fig. 9.7 often referred as the ‘‘ghost-dip’’ is likely due to coupling to the LP\(_{11}\) mode. Because of the photosensitive cladding around the core and lateral exposure, the UV light is preferentially absorbed on one side, inducing an asymmetry in the RIP, as illustrated in Fig. 9.8.

It has been shown by Poulsen et al. that in fibers having a highly UV-sensitive core, the transverse UV absorption allows coupling to asymmetric cladding modes, even in the absence of a blaze angle during grating writing. Their model emphasizes that the index asymmetry, due to the UV light absorption, can explain the strength of the first cladding mode dip that cannot be explained by an unintentionally induced blaze angle.

 

6. Rare Earth-Doped Photosensitive Fibers

The RED photosensitive fiber is an important component for the development and fabrication of fiber laser cavities.

Unfortunately, most RED fibers are not photosensitive because of the replacement of the Ge by Al and/or P necessary to reduce the effect of quenching and lifetime shortening. It is well known, however, that the presence of P bleaches the absorption band centered on 240 nm and, thus, reduces the photo-induced index change.

The first reported photosensitive RED fiber was by Bilodeau et al. on an Er/Ge–doped fiber. The Er concentration of that fiber was too low for fiber laser and amplifier applications but showed some weak photosensitivity with a \(\Delta{n}_\text{uv}\) of about \(3\times10^{-5}\).

A higher Er concentration without quenching requires an aluminosilicate glass host, which reduces the photosensitivity. Since then, many photosensitive RED fibers have been reported (Table 9.5). Different rare earth elements, fiber designs, fabrication particularities, characteristics, and applications are described in the following subsections. 

 

6.1. Germano-Alumino-Silicate Glass Host Core

The RED Ge/Al/Si glass host core fiber is intrinsically photosensitive as long as the GeO₂ concentration is 20 mol% or more. Below that and depending on the application, the fiber might need to be H₂ loaded before the FBG fabrication.

The Ge/Al/Si core design was applied to the Er-doped silica fiber. The GeO₂ concentration was 20 mol%, which makes this fiber intrinsically photosensitive. This single clad design was developed for fiber laser operating at 1550 nm. This fiber corresponds to fiber model Er 304 (Table 9.6).

 

The Ge/Al/P/Si core design was also applied to the Nd-doped silica fiber. The P₂O₅ concentration being lower than 1 mol% did not reduce the fiber photosensitivity. The GeO₂ concentration was only 8 mol%, which is not sufficient for the fiber to be intrinsically photosensitive.

This fiber then needed to be H₂ loaded to become photosensitive. The single clad fiber was designed for the development of fiber lasers operating at 1064 nm. This fiber corresponds to fiber model Nd 100 (Table 9.7).

The RED photosensitive fibers are fabricated by drawing silica optical preforms obtained by a combination of MCVD process and solution doping technique. 

 

6.2. Confined Core

The confined core design was developed to achieve efficient Er/Yb–co-doped photosensitive fibers. Actually, the Er and Yb ions are confined in the guiding core. The photosensitivity is obtained from the Ge-doped ring surrounding the Er/Yb–co-doped phospho-silicate glass host core necessary to make an efficient energy transfer between the Yb and Er ions and to avoid ion quenching.

The confined core ring design was applied to the all-silica double-clad hexagonal Er/Yb co-doped single-mode photosensitive fiber. The all-silica composition and geometry allow this fiber to be fusion-spliced with fiber pigtailed laser diodes. Moreover, it ensures long-term reliability.

The double-clad (DC) design allows the coupling of high pump power lasers to the active fiber core where the hexagonal shape enables efficient mode mixing. This all-silica guiding structure has the advantage of producing a more robust fiber, with better resistance to heating at the pump launching surface than the double-clad fibers, which use a low-index polymer coating to form the outer cladding. This fiber corresponds to fiber model EY 701 (Table 9.8). The confined core design could be applied to all RED fibers.

 

 

The mother preform of this fiber was fabricated by a combination of MCVD process and solution-doping technique. The Ge-doped ring core was deposited by MCVD and the Er/Yb–doped phospho-silicate core was fabricated by the solution-doping technique.

The mother preform was then sleeved to enlarge its diameter and meet the targeted LP\(_{11}\) cutoff wavelengths. For the double-clad design, the preform was then shaped into a hexagon.

Meanwhile, the silica core of a Fluosil preform was drilled. The NA of the Fluosil preform was 0.264. The final double-clad preform was obtained by fusing the drilled Fluosil preform onto the hexagonal mother preform. After drawing, the resulting all-silica fiber was round with a hexagonal first clad and a round core (Fig. 9.9).

This double-clad fiber was characterized in terms of refractive index profile, core diameter, LP\(_{11}\) cutoff wavelength, effective NA, spectral attenuation and absorption, photosensitivity, and splicing losses.The RIP of the fiber was measured by the refracted near-field technique (Fig. 9.10).

 

 

From this fiber index profile, the measured core diameter at 50% of maximum \(\Delta{n}\) is 5.6 μm, the calculated effective core NA is 0.17, and the calculated LP\(_{11}\) cutoff wavelength is 1250 nm. The estimated background losses at 1550 nm is 210 dB/km and the Er/Yb absorption of the multimode pump guide at 976 nm is 1.2 dB/m.

To measure the photosensitivity of the fiber, it was hydrogen loaded for 10 days at 1500 psi. A 10-mm-long FBG was then written into that fiber and into hydrogen-loaded SSMF for comparison. The photosensitivity results presented in Fig. 9.11 clearly show that this all-silica Er/Yb fiber was indeed photosensitive, which indicates that it would be suitable for FBG writing to realize a cladding pumped fiber laser cavity. 

The single-mode splice loss has been measured between this all-silica Er/Yb fiber and an SSMF. The multimode splice loss was measured between this all-silica Er/Yb fiber and a 125 μm all-silica fiber pigtail having a 113 μm core diameter and a 0.242 NA.

The results of the splice losses measurements at a 1310-nm wavelength are given in Table 9.9. The splicing losses to an SSMF were 0.09 dB, and they were 0.07 dB to a multimode fiber pigtailed laser diode pump. Gratings of more than 20 dB were obtained on a hydrogen-loaded Er/Yb fiber compared to 16 dB for a hydrogen-loaded SSMF. Both gratings were written under the same conditions. 

 

6.3. Photosensitive-Clad

The photosensitive clad design was developed by Dong et al. to achieve efficient Er/Yb–co-doped silica fibers. The fiber is single clad with a highly photosensitive B/Ge–doped silica clad surrounding the single-mode Er/Yb– doped phospho-silicate core.

B and Ge were chosen because both elements increase the photosensitivity. Furthermore, B lowers the refractive index of silica, whereas Ge increases it. By using the right concentrations of both elements, it is possible to obtain a B/Ge clad having the same refractive index as the pure silica clad.

The Er/Yb–co-doped core is not affected by the B/Ge– co-doped silica cladding. The highly photosensitive cladding allows strong (>99%) 10-mm-long gratings to be easily achieved in these fibers, despite the reduced overlap between the grating and the guided optical field. The high absorption at 980 nm permits efficient pump absorption over a short fiber length.

The photosensitive clad design was also applied to the all-silica double-clad, single-mode, Yb-doped photosensitive optical fiber. This fiber corresponds to fiber model Yb 708, whose physical characteristics are presented in Table 9.10. The photosensitive clad design could be applied to all RED fibers.

 

6.4. Confined Core and Photosensitive Clad

The confined core and photosensitive clad design is a combination of the two aforementioned designs. It was developed to achieve efficient Er/Yb–co-doped photosensitive fibers.

The central core is single mode over a 1500-nm operating wavelength. The ring around the Er/Yb confined core is co-doped with Ge, making the core photosensitive. This feature allows fabrication of FBGs in the ring, which will make fabrication of a fiber laser cavity possible. This fiber corresponds to fiber model EY 305 (Table 9.11). This design could be applied to all RED fibers. 

 

6.5. Antimony-Doped Alumino-Silicate

A germanium-free antimony (Sb) and thulium co-doped alumino-silicate optical fiber was designed and developed by Sahu et al..

The fiber was fabricated by the MCVD process, along with the solution-doping technique. The fiber, being not intrinsically photosensitive, was H₂ loaded for 2 weeks before the FBG fabrication.

Afterwards, the fiber was annealed at 100\(^\circ\)C for 24 hours to outgas any residual hydrogen in the loaded sample and to stabilize the index modulation at room temperature.

The FBGs were fabricated using an illuminating beam at a 244-nm wavelength, having an intensity of 300 W/cm\(^2\). The resulting index modulation achieved in these conditions was \(1\times10^{-3}\).

The fiber had a 120-μm diameter, an NA of 0.16, a cutoff wavelength of 1500 nm, and a Tm concentration of about 1000 ppm. These preliminary results suggest that Sb might become a promising photosensitive element for the development of RED photosensitive fibers, which are much needed for the fiber laser industry.

 

7.  Polarization Maintaining

The polarization maintaining (PM) photosensitive fiber design was developed for applications requiring a photosensitive fiber that maintains the polarization plane of the transmitted light in the fiber.

PM fibers are used for various applications because they can be designed as well to optimize the performances of optical fiber sensors than to get optimal specifications for telecommunication components.

The requirements for PM photosensitive fibers include good photosensitivity, low crosstalk, low transmission loss, and good mechanical bending characteristics. The PM design selected for the photosensitive fiber is a PANDA type. Many designs of PM fibers have been developed. INO’s fiber model PS-PM-60 corresponds to an HNA photosensitive fiber with a PANDA design (Table 9.12). 

 

The photosensitive core is fabricated by the MCVD process. The PANDA feature is obtained by drilling two holes on both sides of the core and inserting stress-applying part (SAP) rods in the holes.

The SAPs are also manufactured by the MCVD process. The SAPs have a larger thermal expansion coefficient and a lower vitrification temperature than the other parts of the preform.

Hence, a residual stress remains in the fiber after the drawing procedure and the presence of this stress across the fiber core region induces a large birefringence in the fiber.

The RIP for the X-scan perpendicular to the SAPs and Y-scan along the SAPs is given in Fig. 9.12. The background loss is affected by the insertion of the SAPs, as shown in Fig. 9.13. For this fiber, the crosstalk is \(-44\) dB on a 4-m long fiber, which is sufficient for most applications. The beat length is 4.9 mm at 1550 nm, corresponding to a birefringence of \(3.2\times10^{-4}\).

 

 

8. Other Photosensitive Fiber Types

This section covers other photosensitive fiber types including polymer optical fibers (POFs), cerium-doped fluoride glass fibers, and heavily P-doped silica fibers. The fabrication processes and applications are varied and are presented in the following subsection.

8.1. Polymer Optical Fibers 

Silica fiber photosensitivity finds applications in many fields such as optical signal processing, communications, and sensors. Many types of silica FBGs like chirped gratings, apodized gratings, or long-period gratings have been extensively studied over the past 2 decades. The technology is now completely mature and relatively inexpensive.

The POF was developed at about the same time as the silica fiber by the U.S. company Dupont in 1968. During the following years, POF was neglected in front of silica fiber. POF technology was mostly developed in Japan with a new fabrication process to get a reduction of the attenuation and increase the bandwidth for telecommunications applications. Zubia et al. present in their review the historical evolution of the most important landmarks related to POF from its first fabrication to the year 2000.

Photosensitivity of polymethylmethacrylate (PMMA) was discovered 35 years ago with the work of Tomlinson et al. who showed that the refractive index of PMMA could increase after a 325- or 365-nm UV irradiation. After subsequent work by Kaminov et al., Welker et al., and Peng et al., it was demonstrated that a significant photosensitivity could be induced in POFs not only with doping materials but also with non-doped basic material that could give significant photosensitive effects at UV wavelengths, so that POF grating may be possible to fabricate. The first POF grating was written with a few-moded POF and a writing wavelength of 325 nm. After the development of single-mode POF by Koike, 1-cm-long gratings were written with a reflection bandwidth reduced to approximately 1 nm for a maximum reflectivity of about 80%.

The photorefractive effect involved in polymer fiber gratings is the photopolymerization of the fiber. This phenomenon is different from the photosensitive phenomenon observed in silica fibers. Indeed, the incident UV light launched in the fiber causes polymerization of the unreacted monomer, which increases the polymer density and hence the refractive index of the fiber. Photopolymerization in PMMA POF is induced at a wavelength of 325 nm. The method to write FBG can be the same as for silica fiber with the use of a phase mask. An example of the growth dynamics of a polymer-based FBG is presented in Fig. 9.14.

 

 

Both types I and II can be observed for polymer FBGs. Indeed, there is a distinctive threshold in UV exposure during the grating growing. This threshold separates the two grating formation stages. An example of their specificities is summarized in Table 9.13. Note that researchers have found that because of the thermal stress induced by UV irradiation, polymer FBG growth is a writing-power–dependent process.

 

 

The photosensitivity in CYTOP POF is smaller than in PMMA POF. The CYTOP fiber, composed of perfluorinated polymer, has been developed by Keio University and Asashi Glass Company in Japan. Its photosensitive properties have been investigated by Tanio et al.. It is possible to write FBGs in CYTOP fiber. Although the index modulation is smaller (0.6 times the PMMA), gratings are much more thermally stable than with PMMA fiber. The interest in using perfluorinated polymer is its low attenuation of 0.3 dB/km at 1550 nm, which is close to the silica fiber attenuation, and is of interest for short-distance communications.

Strain and temperature have been extensively studied for silica-based FBGs because of the number of applications based on these properties. Despite their high polymer fiber attenuation, PMMA-based FBGs have interesting potential applications because their grating lengths are short (~ 1 cm long). Indeed, it has been demonstrated that the Young modulus of polymer fibers is about 30 times less than that of the silica, so that mechanical tunability of polymer-based FBGs can be 30 times larger than for a silica-based FBG. Moreover, because the thermal strain coefficient in polymer fibers is higher compared with the silica fiber, the thermal tuning range of polymer Bragg gratings can be significantly wider. Table 9.14 summarizes mechanical and thermal properties of silica and PMMA materials. A great difference between silica-based and PMMA-based FBGs has been pointed out by Liu et al. when they demonstrated that both strain and thermal sensitivities of types I and II PMMA-based FBGs are similar.

 

 

Applications of PMMA-based FBGs are various. Because of their wide tuning range, polymer FBGs have interesting applications in the sensors field, with simultaneous strain and temperature sensors. Figure 9.15 shows a simple sensor constructed by combining two FBGs, one made with a silica fiber, and the other with a polymer fiber. The setup uses a broadband source. To sense the temperature change and a strain change independently but simultaneously, the Bragg wavelength shift due to each grating is recorded. These quantities can be expressed in terms of the temperature change and the strain change. With one grating made of silica and the other made of polymer, it is possible to separate the contribution of temperature and strain and sense both changes more accurately.

 

 

Optical add–drop multiplexer is an important device in WDM communication and sensing. Its basic configuration consists of an FBG inserted between two broadband optical circulators. With a silica-based FBG, the device is not easily tunable, but by replacing the silica-based FBG with a polymer-based FBG, the wavelength is then tunable (mechanically or thermally) over 20 nm. Another configuration is the Mach-Zehnder interferometric configuration with two identical FBGs. Tunability of the device can be obtained by replacing the silica FBGs with polymer FBGs.

Dispersion compensator is also a very important device in communication systems. An interesting means to get a higher tunability would be with the use of a polymer-based chirped Bragg grating.

Bragg gratings in POF are more tunable than in silica fibers. Indeed, a temperature step of about 50 degrees will give a 10-nm variation of the POF FBG wavelength compared with 1 nm for silica FBG. Variations of the same order of magnitude can be obtained by stretching the POF FBG.

 

8.2. Fluoride Glass

Fluoride (or nonoxide) glasses were discovered in 1975 by Marcel and Michel Poulain. At that time, one of the most important companies making fluoride fibers was ‘‘Le Verre Fluore´.’’

The principal interest of these glasses is their transmission spectra because they are transparent to light from 300 to 6500 nm. ZBLAN is the most common composition (ZrFM4-BaF2-LaF3-AlF3-NaF), but other compositions permit to reach different mechanical or optical properties.

Fluoride fibers can be used in a variety of scientific and industrial applications such as infrared and near-infrared fiber spectroscopy, ultrafast fiber pyrometry, astronomical interferometry, or telecommunications with fluoride glass optical amplifier modules.

Lasers can be developed with fluoride fibers for the visible spectrum and for infrared wavelengths. Fluoride fibers can also be interesting for medicine and surgery, more particularly for Er/YAG lasers. IRphotonic has developed a special fluoride fiber with low moisture suitable for Er/YAG transmission.

Until now, research on the photosensitivity of fluoride glasses was not very extended, with only a few studies on this topic and no available commercial product. The first studies on the photosensitivity of fluoride glasses or fibers were done with trivalent rare earth ion cerium (Ce\(^{3+}\))-doped glasses. They present the influence of the cerium concentration on the photosensitivity of the fluoride fibers with 1550 nm Bragg grating inscription with a 246-nm UV light exposure and prove that enhancement of the Ce\(^{3+}\) concentration contributes to increase the photo-induced refractive index change up to \(4\times10^{-4}\) in a single-mode Ce\(^{3+}\)-doped fluoride fiber.

Other studies have been performed on RED fluorozirconate (FZ) glasses with permanent holographic gratings written at 248 nm in bulk samples of fluoride glasses doped with Tb\(^{3+}\), Pr\(^{3+}\), Tm\(^{3+}\), or Ce\(^{3+}\), with refractive index changes at less than \(1\times10^{-5}\) for cerium and lower for the others.

Other work has been performed by Zeller et al., who report on undoped fluoride glass slides exposed to pulsed 193 nm UV irradiation. Photosensitivity of fluoroaluminate (FA), FZ, and fluorozircoaluminate (FZA) glasses is compared. FA and FZ glasses provide small index changes (\(2.0\times10^{-6}\) for FA and \(2.6\times10^{-6}\) for FZ), but the index change up to \(1.75\times10^{-4}\) has been evaluated in FZA glass at 1550 nm.

 

8.3. Heavily P-Doped Silica Fibers

The low-loss heavily P₂O₅-doped silica fiber is a good candidate for the development of an efficient high-power Raman fiber laser (RFL) used as a pump source for Raman fiber amplifiers (RFAs) or erbium-doped fiber amplifiers (EDFAs).

The RFL gain block consists of a P-doped fiber as the gain medium; the laser cavity is obtained by FBG inscribed in Ge-doped silica fibers. The two FBGs are then spliced to the gain medium fiber. The splice losses could be reduced or even avoided by inscribing the FBGs directly into the heavily P-doped silica fibers.

The number of cavities depends on the Stokes shift. For a Ge-doped silica fiber, the Stokes shift is limited to 13.2 THz. For example, five laser cavities are needed to shift the Yb-doped fiber laser pump emitting at 1060 nm to the pump wavelength of 1480 nm for EDFAs.

Using a heavily P-doped silica fiber, the number of cavities is reduced to only two, thanks to the Stokes shift of 39.9 THz. This shift is three times as large as the Ge-doped silica fiber. In only one step or one pair of FBGs, a pumped P-doped silica fiber at 1060 nm will convert the pump light to 1240 nm. This wavelength is suitable to pump RFAs for the 1310-nm window.

Although the P-doped silica fiber is very promising, it is not intrinsically photosensitive and needs to be treated to become so. Once applied to the P-doped silica fiber, the standard hydrogen-loading process has many disadvantages such as significant OH formation, hydrogen out diffusion issues, and a contribution to component instability from an undesired index change.

The OH formation is due to the hygroscopic nature of P, and consequently, it increases the background losses, reducing the efficiency of Raman converters. A proposed solution is the photolytic hypersensitization or two-stage exposure process.

The first stage consists of irradiating the hydrogen-loaded P-doped fiber, which is the photolytic hypersensitization, with a low-exposure dose homogeneous beam. Then, the remaining free hydrogen is removed from the fiber by leaving it at room temperature for many days.

The second stage consists of writing the FBGs as usual. The FBGs were written directly through a phase mask at 193 nm from an ArF source. The 193-nm source wavelength was found to be the most efficient writing wavelength with the lowest induced OH. The fiber was fabricated by the MCVD process. The P₂O₅ concentration was 15 mol%. Other specifications are given in Table 9.15.

 

 

The next tutorial explains laser amplification in detail.