1. Introduction

Rare-earth (RE) doped fiber lasers and amplifiers represent one of the major scientific breakthroughs of the 20^th century. Fiber lasers and amplifiers possess various advantages over crystal or gas-based counterparts, such as high-quality beam, high brightness, alignment-free configuration, practicality of operation, and power scalability [1]. Amongst other RE ions, the trivalent erbium, Er³⁺, and thulium, Tm³⁺, hold a special place thanks to their strong emission in the near-infrared (NIR) region, which occurs as a result of transitions between 4f electron energy levels. The erbium-doped fiber amplifiers (EDFA) and lasers (EDFL) provide emission around 1.55 µm, a region also known as 3^rd telecommunication window, and find application especially in telecommunications [2,3]. The thulium-doped fiber amplifiers (TDFA) and lasers (TDFL) exhibit emission further in the NIR region, around 2 µm, also known as “eye-safe” region, thanks to lower risk of damage to retina, and characterized by high absorption of water, which makes them suitable for applications in material processing, sensors, or defence [4].

The fluorescence lifetime of the lowest excited energy level (⁴I_13/2 and ³F₄ in the case of Er³⁺ and Tm³⁺, resp.) is one of the most important parameters when evaluating the suitability of the optical fibers for laser operation. It was consistently shown that the fluorescence lifetime strongly correlates with laser parameters, such as slope efficiency or laser threshold, where fibers with high fluorescence lifetime typically exhibit enhanced laser properties [5–8]. Moreover, the fluorescence lifetime plays an important role in various theoretical tasks, e.g., modelling of laser performance [9], simulations of temperature effects on laser operation [10], or calculation of energy-transfer coefficients [7,11]. The precise knowledge of the fluorescence lifetime is therefore important for the efficient design and evaluation of lasers based on RE-doped silica fibers.

The fluorescence lifetime is highly dependent on the composition, matrix structure, and thermal history of the material. Silica glass remains the most widespread material as a host matrix for RE ions owing to its mechanical, thermal, and chemical stability, which make it a prime choice for the construction of durable, high-power devices. However, pure silica often represents a poor material when it comes to spectroscopic properties; it suffers from several drawbacks such as high phonon energy of the silica network (∼1100 cm^-1) or low solubility of RE ions [12]. These effects contribute to harmful non-radiative decay pathways, such as multiphonon relaxation or concentration quenching. In order to remediate these drawbacks, pure silica needs to be modified by certain additives, which exhibit lower phonon energy and act as network modifiers, breaking the rigid silica network and increasing the solubility of REs, one of the most effective being Al₂O₃.

The most common method of silica fiber fabrication is the Modified Chemical Vapor Deposition (MCVD) combined with solution doping [13,14]. The core is formed by soaking a porous layer inside a pure silica tube with a solution containing Al and RE salts, which is sintered, collapsed and drawn into fiber. The standard solution doping method is limited in the maximum achievable Al₂O₃ concentration, around 5 mol. % Al₂O₃, which makes the preparation of highly RE-doped fibers challenging [15]. In the last decades, a great scientific effort was focused towards the development of methods capable of introducing high dopant concentrations into the fiber, reviews can be found in [12,16]. One of the most promising techniques is the nanoparticle dispersion doping, where the Al or RE salts are replaced in the solution by nanoparticles, either pure Al₂O₃ nanoparticles mixed with RE salts [5,17], or RE:Al₂O₃ nanoparticles directly doped with RE ions [6,8]. Compared to the standard solution doping, the nanoparticle doping method was shown to provide enhanced luminescence and laser performance in some cases, but the compositions of the fibers were often not studied in detail [6,18]. Some authors argued that the Al₂O₃ nanoparticles were preserved in the fibers, providing a beneficial low-phonon environment for RE ions [8], whereas others provided evidence for the complete dissolution of the nanoparticles [5,19]; the issue remains unclear. The principal advantage of the nanoparticle doping method is the higher achievable content of Al₂O₃, up to 10 mol. %, which allows to increase the doping level of RE ions.

The fluorescence lifetime is also strongly dependent on the thermal history and fabrication processing of the material. We recently demonstrated that the various fabrication processes, such as preform sintering, preform elongation, or fiber drawing negatively influence the fluorescence lifetime of various RE ions, such as Tm³⁺, Ho³⁺ and Yb³⁺, but the causes of this effect and links to the matrix structure remain unclear [20]. These effects present a considerable obstacle for the fabrication of specialty optical fibers, which require a complicated fabrication process. For example, the fluorescence lifetime decrease may be one of the causes behind the low efficiency of lasers based on Yb³⁺- or Tm³⁺-doped nanostructured-core fibers [21,22]. Moreover, the fabrication processing effect remains almost completely unexplored in the case of Er³⁺, which is often shown to behave differently than Tm³⁺ or Ho³⁺ [19,23]. Dybdal et al. previously reported that the fluorescence lifetime of Er³⁺ remained nearly unchanged going from preform to optical fiber, but no decay curves were shown in detail, and the study was limited on solution-doped fibers with low content of Al₂O₃ and Er³⁺ ions [24]. For the effective and reliable design and study of laser and amplifier devices based on the highly-doped specialty optical fibers, such as the nanostructured-core fibers, it is necessary to properly investigate the effect of fabrication processing on the matrix structure and the fluorescence lifetime of RE ions.

In this paper, we study the relationship between heat treatment, matrix structure and fluorescence lifetime of RE-doped preforms and optical fibers prepared by the nanoparticle doping. The preforms doped with Er³⁺ and Tm³⁺ ions were prepared and subjected to various types of fabrication processing, such as preform elongation, fiber drawing, or additional heat treatment. The matrix structure was studied by XRD and TEM. The fluorescence lifetime of Er³⁺ and Tm³⁺ was measured in all stages of the fabrication process. The effects of the fabrication processing on the matrix structure and fluorescence lifetime of the Tm³⁺ and Er³⁺ were analyzed and discussed.

2. Experiment

2.1 Sample preparation

Optical fiber preforms were prepared using the MCVD method combined with nanoparticle doping technique. The preforms doped with Tm³⁺ or Er³⁺ ions were fabricated by identical procedure as described in [20]. One preform was prepared for each of the RE ions. The subsequent processing is schematically depicted in Fig. 1 for better orientation. One half of the preform was drawn in a fiber drawing tower, equipped with a graphite resistance furnace (Centorr, USA) at a temperature of 1950 °C into a fiber with a diameter of approx. 125 µm and a core diameter of approx. 17.5 µm. The other half of the preform was elongated in the fiber drawing tower to a diameter of approx. 3 mm. The elongated preform (designated as “cane”) was overcladded with a pure silica tube and drawn in the same manner into a fiber with a diameter of approx. 125 µm and core diameter approx. 5 µm. Furthermore, the samples of preform and cane were subjected to additional heat treatment in the graphite resistance furnace with a slow cooling rate; these samples are designated as HT preform and HT cane.

 

Fig. 1. scheme of the sample preparation.

Download Full Size | PDF

2.2 Sample characterization

Dopant concentrations in the original preform were measured by a JEOL JXA 8230 electron probe microanalyzer (EMPA). The refractive index profiles of the preforms were measured using a Photon Kinetics A2600 refractive index profiler, and the refractive index profiles of the optical fibers were measured using an IFA-100 refractive index profiler. The fibers were characterized regarding their absorption, background losses, and OH^- content. A tungsten halogen lamp was used as a broadband source of radiation. The absorption and background losses were measured using both an ANDO AQ6317 optical spectrum analyzer and a Nicolet 8700 FTIR spectrometer that was adopted in-house for fiber-optic measurements. Background losses were evaluated from absorption spectrum minimum. Er³⁺ and Tm³⁺ contents were estimated from the absorption peaks at 1530 and 1640 nm, as described in [7]. The OH^- content was determined from the OH^- related peak at 1383 nm, as described in [25].

The matrix structure of the preform samples was analyzed by XRD using a Bruker D8 Discover diffractometer in the Bragg-Brentano reflecting geometry. The copper X-ray tube was operated at voltage of 40 kV and current of 40 mA providing Cu-Kα1 radiation (λ = 1.540596 Å). The preforms and the fibers were further analyzed by TEM. The samples for TEM analysis were prepared as follows: in the case of the preform, the cladding was removed and the core was ground into powder. The powder was mixed with isopropanol and dropped on the lacey carbon 300 mesh copper grid (Agar Scientific). In the case of the optical fiber, the cladding was removed by etching in HF and the sample, below 100 nm thick, was cut from the exposed core using GATAN PIPs I. The samples were observed by EFTEM Jeol 2200FS field emission electron microscope in TEM mode. The phase composition was analyzed by electron diffraction and the elemental composition was analyzed by energy-disperse spectroscopy (EDS) in STEM mode with 1 nm spot size.

The samples were characterized regarding fluorescence lifetime. The detailed description of the measurement setup and equipment can be found in [20]. The measured fibers were approximately 1 mm long, and emission was detected from the side in order to suppress the influence of amplified spontaneous emission (ASE) and reabsorption [7]. The decay curves were measured for multiple excitation powers; decay times for each respective excitation power were obtained from the 1/e value of the normalized intensity. The fluorescence lifetime, i.e., the decay time extrapolated to zero pump power, was obtained using Eq. (1):

In addition, selected decay curves were analyzed by fitting a single- or double-exponential function, Eq. (2), in order to retrieve the various components of fluorescence decay:

3. Results and discussion

3.1 General characterization of the preforms and fibers

The general properties of the prepared preforms and fibers are summarized in Table 1. The concentrations of Al₂O₃ and RE ions in the preforms were measured by EMPA, the measured profiles are depicted in Fig. 2. In general, all concentration profiles exhibit typical cylindrical shapes with a flat plateau in the center, typical for doping with Al₂O₃. The preforms possess a high concentration of Al₂O₃ around 8.5 mol. %, which is close to the reliably achievable maximum of the nanoparticle doping. The Tm-doped preform has a doping level of Tm³⁺ ions around 5,000 ppm, the Er-doped preform around 4,000 ppm. The refractive index core-cladding differences were measured in the preforms and as well as both types of optical fiber, standard and OC. The profiles are also depicted in Fig. 2. The results are in a good agreement, the refractive index closely matches the concentration profiles of the dopants.

 

Fig. 2. Concentration and refractive index profiles of the preforms and fibers, a) Tm, b) Er.

Download Full Size | PDF

Table 1. Summary of the general characteristics of the prepared preforms and fibers

View Table | View all tables in this article

The concentrations of RE ions in optical fibers were checked by absorption measurements. In all cases, the measured values are in a reasonable agreement with EMPA results from the preform. It must be noted that the concentration obtained from absorption is only approximate due to the uncertain values of the overlap factor used in the calculation.

The optical fibers exhibit reasonable losses under 0.07 dB/m and OH^- content below 5 ppm, which suggests a good quality of the prepared preforms. The small dimensions of the canes did not allow reliable measurements of EMPA, absorption spectra or refractive index.

3.2 Matrix structure of the preforms and fibers

The XRD method was used for the analysis of matrix structure of the preform samples, i.e., the original and HT preform. The XRD patterns of the Tm-doped preforms are depicted in Fig. 3, the results of the Er-doped preforms are nearly identical (not shown). The original preform exhibits a fully amorphous character with no evidence of crystalline phase, the broad halo band around 20° is characteristic of amorphous silicate matrix. The additional heat treatment followed by slow cooling induced the crystallization of mullite, 3Al₂O₃·2SiO₂.

 

Fig. 3. XRD patterns of the Tm-doped preform and HT preform.

Download Full Size | PDF

The XRD method cannot be used reliably for the analysis of optical fibers due to the small dimensions of the core. In order to get more insight into the changes of matrix structure during fiber drawing, the TEM method was employed for the analysis of the preform and optical fiber; the Tm-doped samples were selected for the analysis. The TEM images of the preform are shown in Fig. 4(a) and (b), the electron diffraction pattern is shown in Fig. 4(c). The TEM images of the fiber are shown in Fig. 4(d) and (e), the electron diffraction pattern is shown in Fig. 4(f).

 

Fig. 4. TEM images of the Tm-doped preform and fiber, a) and b) images of the preform core area, c) ED pattern of the preform core area, d) and e) images of the fiber core area, f) ED pattern of the fiber core area.

Download Full Size | PDF

Nanoparticles of similar size are clearly visible in both the preform and the fiber, the average size of the nanoparticles is around 15 nm. The size of the nanoparticles is significantly lower compared to the original Al₂O₃ nanoparticles used in the doping dispersion (mean size of 50 nm), which confirms at least partial dissolution of the original Al₂O₃ nanoparticles. Moreover, the EDS analysis confirmed that the nanoparticles are not pure Al₂O₃, but alumino-silicate, with the nanoparticles enriched by Al₂O₃, and the surrounding matrix enriched by SiO₂. The concentration of Tm³⁺ ions was below reliable detection limit of EDS analyzer; no reliable information about the Tm³⁺ ions environment could be obtained.

Certain differences are observed between the preform and the fiber. The nanoparticles in the preform have sharp and well-formed boundaries, whereas the nanoparticles in the fiber possess a more diffused character; moreover, the nanoparticle size in the fiber is spread over a wider range. In addition, the nanoparticles in the preform exhibit an electron diffraction pattern with partially crystalline character; several weak spots can be seen in the pattern which were identified as (121) and (210) planes of crystalline mullite phase, 3Al₂O₃·2SiO₂, in trace amounts, which could not be observed by XRD. The nanoparticles in the optical fiber exhibit a fully amorphous character with no trace of crystalline phase. In summary, these effects suggest a diffusion of the constituent oxides and ions during the fiber drawing.

3.3 Luminescence properties of the preforms and fibers

The time-resolved luminescence properties of the Tm-doped samples from emission around 2 µm (³F₄→³H₆ transition) are depicted in Fig. 5. Portion of the results was previously reported in [20]. The representative fluorescence decay curves of the samples are depicted in Fig. 5(a); the values of decay times for all measured powers are shown in Fig. 5(b). The fluorescence decay of all bulk samples is virtually independent on the excitation power; the original preform, and both cane samples exhibit a single exponential character, whereas the HT preform is double exponential in the entire power range. The decay curves of the fibers are single-exponential at low power, but increasing power leads to deviations from single-exponentiality and faster decay. The optical fibers possess a significantly smaller core (approx. 17.5 and 5 µm in diameter) compared to the bulk samples, leading to significantly higher power densities, and therefore increased rate of energy-transfer (ET) processes, such as energy transfer up-conversion (ETU) or cross-relaxation (CR).

 

Fig. 5. fluorescence decay properties of Tm-doped samples, a) representative fluorescence decay curves for low or high excitation power, b) decay time as a function of excitation power, including extrapolation to zero power (grey dashed lines) as average (preform and cane samples) or using Eq. (1) (fiber samples).

Download Full Size | PDF

The fluorescence lifetime decreases significantly as a result of the fabrication processing, evident after extrapolation to zero power, as summarized in Table 2. The original preform exhibits the highest value of fluorescence lifetime, 875 µs, and each processing step causes a decrease of the fluorescence lifetime – the fiber drawing, the preform elongation as well as the additional heat treatment.

The most significant decrease of lifetime can be observed going from the original to the HT preform; moreover the character of the decay curves changes from single- to double-exponential, as seen in Fig. 6. The large core area, and thus low power density, as well as the unpolished and non-transparent nature of the sample should minimize the influence of ASE, ET or reabsorption. The double-exponential character of the HT preform can be therefore ascribed primarily to the incorporation of Tm³⁺ ions in different environments [26,27]. The original preform exhibits a single component of lifetime, around 888 µs value, in a good agreement with the fluorescence lifetime from Table 2. The HT preform contains a slow lifetime component, 661 µs, generally comparable to the original preform, albeit lower, and a fast lifetime component, 298 µs, which is significantly shorter.

 

Fig. 6. fluorescence decay curves of the original and HT preform along with corresponding single- or double-exponential fits.

Download Full Size | PDF

Table 2. Fluorescence lifetimes of Tm-doped samples obtained as fitted parameters of Eq. (1)

View Table | View all tables in this article

The time-resolved luminescence properties of the 1.5 µm emission (⁴I_13/2→⁴I_15/2 transition) of the Er-doped samples are summarized in Fig. 7. The fabrication processing effect on the fluorescence lifetime of Er³⁺ ions is much less pronounced than in the case of Tm³⁺ ions. From Table 3, it is obvious that the values of fluorescence lifetime, τ₀, vary to a much lesser extent, between 9.6 and 10.2 ms, and no specific trends are evident. As shown in Fig. 7(a), the decay curves of all samples recorded under low excitation power are single-exponential and closely overlap. When the power is increased, the decay curves of most bulk samples remain nearly constant and single-exponential, whereas the optical fibers exhibit increasing double-exponential character and decreasing decay time; this behavior is identical as in the case of the Tm-doped samples. The only outlier is the original preform, which also exhibits decreasing lifetime with excitation power, although to a much lesser extent than the optical fibers. Similar behavior in a preform was previously observed, e.g., in a highly-doped preform with Er³⁺ content above 14,000 ppm, and may be ascribed to partial clustering of the Er³⁺ ions, which leads to a higher effect of ETU [28]. The disappearance of this effect after heat treatment and preform elongation suggests modification of the Er³⁺ ion environment. These observations indicate that the environment and luminescence properties of Er³⁺ ions are not completely immune to fabrication processing effects, but the changes are much less significant than in the case of Tm³⁺.

 

Fig. 7. fluorescence decay properties of Er-doped samples, a) representative fluorescence decay curves for low or high excitation power, b) decay time as a function of excitation power, including extrapolation to zero power (grey dashed lines) as average (HT preform and cane samples) or using Eq. (1) (original preform sample and fiber samples).

Download Full Size | PDF

Table 3. Fluorescence lifetimes of Er-doped samples obtained as fitted parameters of Eq. (1)

View Table | View all tables in this article

4. Discussion

In general, the fluorescence lifetime is determined by the environment and bonding of RE ions in the material, which are influenced by the matrix structure. As the results show, the matrix structure undergoes significant evolution during the fabrication processing and heat treatment, which is in a good agreement with the thermodynamic properties of the Al₂O₃-SiO₂ system. The preform sintering and fiber drawing involve temperatures above 2000 °C, well into the liquidus zone of the Al₂O₃-SiO₂ phase diagram [29]. The original Al₂O₃ nanoparticles used for the doping are thus dissolved and react with silica [19]. The Al₂O₃-SiO₂ system, however, exhibits a metastable immiscibility in the solidus region, which leads to the formation of alumino-silicate nanoparticles via the phase separation mechanism [30]. The nanoparticles were previously observed in optical fibers containing at least 20 mol. % of Al₂O₃ prepared by molten-core method [31]. This study confirms the presence of nanoparticles in optical fibers prepared by MCVD method containing below 10 mol. % of Al₂O₃.

The evolution of matrix structure and nanoparticle morphology is in a good agreement with kinetics of the involved fabrication processes. The preform sintering, collapse, or additional heat treatment involve a slow cooling rate due to the large dimensions of the preforms (d ≈ 1 cm) and the presence of crystalline mullite, 3Al₂O₃·2SiO₂, in the preforms is observed by XRD or TEM. On the other hand, the fiber drawing involves a significantly faster cooling rate due to the small dimensions of the fibers (d ≈ 120 µm), which leads to a full amorphization of the matrix and the presence of amorphous, alumino-silicate nanoparticles with less defined boundaries. These observations are also in a good agreement with literature. Various effects, such as dissolution of nanoparticles or changes in their shape and size, were demonstrated during fiber drawing in MgO-SiO₂ [32] or La₂O₃-SiO₂ [33,34] systems. The amorphization of crystalline nanoparticles during fiber drawing was demonstrated in YAG-SiO₂ system [35,36].

Moreover, it was previously shown that the heat treatment and mechanical stresses involved in fiber drawing have a significant impact on the amorphous network itself, e.g., the creation of defects, such as cleavage of bonds or creation of oxygen vacancies [37,38], or changes in bond angles and lengths [39].

The proposed chemical reactions, diffusion effects, or network changes may significantly influence the environment and bonding of RE ions and lead to, e.g., diffusion and placement of RE ions in less favorable phonon environment, clustering of RE ions, or modifications in the coordination sphere of RE ions. However, the XRD and TEM analysis bring no direct evidence about the RE ion environment itself; more detailed studies using advanced methods such as NanoSIMS are necessary.

Nevertheless, more information about the environment of RE ions can be inferred from the fluorescence lifetime. In the preforms and optical fibers presented in this study, the Tm³⁺ ions may be embedded in three distinct environments, i) crystalline mullite, ii) Al₂O₃-rich amorphous nanoparticles, iii) silica-rich matrix. The single-exponential nature of the decay curves suggests that the Tm³⁺ ions are homogenously embedded in only one environment. It was shown previously that Tm³⁺ ions in pure silica or Al₂O₃-deficient matrices, such as solution-doped fibers, possess low values of lifetime due to the high rates of multiphonon relaxation and concentration quenching, generally below 500 µs [7,40]. The measured values of fluorescence lifetime of Tm³⁺ in the 600–900 µs range are thus indicative of RE ion incorporation in the beneficial environment of the amorphous aluminosilicate nanoparticles.

In addition, the presence of crystalline mullite was demonstrated in the preforms, especially the HT preform, as evidenced by XRD. Mullite nanocrystals were shown as a perspective host matrix for Tm³⁺ ions with long fluorescence lifetime above 2 ms, virtually unachievable in silicate matrix [19]. However, no such lifetime was observed in this work, which confirms the low solubility of RE ions in the mullite crystal lattice. A significant incorporation of RE ions into the mullite crystal lattice was previously observed only in a stoichiometric 3Al₂O₃·2SiO₂ glass-ceramic phosphor; any excess of SiO₂ in the nominal composition leads to the RE ions remaining dispersed in the residual amorphous phase [19,41]. The crystallization of mullite in the preforms or optical fibers prepared by MCVD therefore brings no observable benefit. On the contrary, the crystallization of 3Al₂O₃·SiO₂ causes the reduction of volume and a depletion of Al₂O₃ content in the residual amorphous phase. As a result, the fluorescence lifetime of Tm³⁺ ions in the highly crystallized, HT preform is decreased due to multiphonon relaxation, and a second, fast lifetime component below 300 µs appears in the decay curve due to clustering of Tm³⁺ ions and higher rate of ET processes.

The differences in behavior of Tm³⁺ and Er³⁺ can be explained by the influence of phonon energy. The Tm³⁺ ions emit around 2 µm, i.e. approx. 5000 cm^-1. Considering the phonon energy of silica glass, around 1100 cm^-1, less than 5 vibrational phonons are needed to bridge the gap between the ³F₄ and ³H₆ levels and depopulate the excited level non-radiatively – Tm³⁺ ions suffer from high rates of multiphonon relaxation. The fluorescence lifetime of Tm³⁺ ions is therefore highly sensitive to any changes in the material structure and RE ion environment. As previously stated, Tm³⁺ ions in silica-based matrices typically exhibit fluorescence lifetime in the range of 0.3–0.9 ms with quantum efficiencies below 10% [7,42]. Much higher values can be achieved in low phonon energy hosts, e.g., mullite 3Al₂O₃·2SiO₂ [19] or soft glass matrices such as germanate [43], tellurite [44] or ZBLAN [42], routinely above 2 ms.

On the other hand, Er³⁺ ions emit around 1.5 µm, i.e., around 8000 cm^-1, and the multiphonon relaxation rates are much lower. Er³⁺ ions in silicate glass exhibit long lifetimes in the range of 8–15 ms with quantum efficiency close to 100% [6,23]. It was repeatedly shown that the incorporation of Er³⁺ ions in low phonon energy matrices brings no beneficial effect to the fluorescence lifetime, the lifetimes remain similar around 10 ms, e.g., in ZBLAN glass [45,46], or even lower due to a higher influence of other effects, such as concentration quenching or symmetry of the RE ion sites [19,47]. The significantly lower sensitivity of Er³⁺ ions to changes of matrix structure was previously well-demonstrated, e.g., in sodium-silicate glass. Increase of Na₂O concentration had virtually no effect on the fluorescence lifetime of Er³⁺ ions, whereas the lifetime of Ho³⁺ emitting at 2 µm was modified significantly [23].

A comparison should also be made between the Yb³⁺ and Er³⁺ ions. The Yb³⁺ ions emit at 1 µm, i.e., even higher energy than Er³⁺, which makes them virtually unaffected by multiphonon relaxation. However, the fluorescence lifetime decrease during fabrication was observed nonetheless [20]. Yb³⁺ are sensitive to the so-called cooperative up-conversion in closely-coupled ion dimers or trimers, where the excited laser level is depopulated by up-conversion into virtual upper levels, and results in green or UV emission, as described by Qin et al. [48]. This effect was previously described as significantly detrimental to the luminescence and laser properties of Yb³⁺ in silica optical fibers [49,50]. No equivalent effect is however known for Er³⁺ ions.

The fabrication processing and heat treatment effect on the fluorescence lifetime of various RE ions must be considered in the design of specialty optical fibers, e.g., the nanostructured-core fibers, which require multiple steps of heat treatment during fabrication. The Yb³⁺- or Tm³⁺-doped nanostructured-core fibers prepared by the stack-and-draw method were recently demonstrated [21,22], but the laser parameters were below values typically achieved in optical fibers directly fabricated from MCVD preforms [4,51]. The changes in the RE ion environment and the decrease of fluorescence lifetime are one of the possible causes, other possible contributors include higher background loss or unoptimized design of the developed fibers. However, as presented here, the fabrication processing shows no negative effect on the fluorescence lifetime of Er³⁺. The nanostructured-core silica fibers doped with Er³⁺ thus exhibit a great potential for the feasible construction of highly efficient EDFL and EDFA devices, so far demonstrated theoretically [52], or dual-wavelength lasers based on Yb³⁺/Er³⁺ co-doping, where the 1.5 µm emission was observed experimentally in silica fibers [53].

5. Conclusion

We investigated the influence of the optical fiber fabrication process on the matrix structure of the material and the fluorescence decay of the near-infrared emission of Er³⁺ and Tm³⁺ ions. The preforms and optical fibers were fabricated by MCVD method combined with Al₂O₃ nanoparticle doping. It was found that the preform collapse and sintering caused the dissolution of the doped Al₂O₃ nanoparticles, and the presence of alumino-silicate nanoparticles was observed in both the preform and the fiber. The preform elongation and fiber drawing caused a significant decrease of fluorescence lifetime in the case of Tm³⁺ ions, from 875 µs in the preform down to 610 µs in the overcladded optical fiber, whereas no significant effect was observed in the case of Er³⁺ ions, all samples exhibited fluorescence lifetime in the 9.6–10.2 µs range with no specific trends. The low sensitivity of fluorescence lifetime of Er³⁺ to heat treatment is potentially beneficial for the fabrication of Er³⁺-doped nanostructured-core fibers and other specialty optical fibers.