Highly Er/Yb-Co-Doped Photosensitive Core Fiber for the Development of Single-Frequency Telecom Lasers

Abstract

A highly erbium- and ytterbium-co-doped photosensitive fiber with a germanophosphosilicate glass core was fabricated by the MCVD method, utilizing an all-gas-phase deposition technique developed “in-house”. Due to doping with germanium oxide (GeO₂), this fiber revealed high-grade photosensitivity (without hydrogen loading) to UV laser radiation at a 193 nm wavelength. The short (28 mm) Fabry–Perot laser cavity was designed by inscribing two fiber Bragg gratings (highly and partially reflective FBGs) directly in the core of the fabricated fiber sample. The stable single-frequency operation regime of the designed laser was observed. The laser emission peak was centered at 1540 nm, with a linewidth of 50 kHz. The slope efficiency of the laser was 10%, and the maximal output power reached a level of 35 mW.

1. Introduction

The initial studies devoted to the development of single-frequency fiber lasers were carried out using optical fibers made of quartz glass activated by rare-earth ions [1,2,3]. Although silica glass has a number of advantages that make it the most widely used optical fiber material, the maximum doping concentration of rare-earth ions in silica glass is limited, due to the clustering effect leading to undesirable cooperative interactions among active ions. Therefore, after the demonstration of a number of single-frequency fiber lasers based on silica optical fibers, the interest of researchers has shifted to the area of creating single-frequency fiber lasers utilizing optical fibers made of multicomponent glass, which provides more efficient laser generation due to the relatively high concentration of active rare-earth ions in multicomponent glass [4,5]. However, despite the fact that optical fibers made of silica glass are inferior in lasing efficiency compared to multicomponent fibers due to the lower concentration of rare-earth ions, nevertheless, a combination of other positive qualities makes silica fibers a promising active medium for single-frequency lasers when high powers are not required. The advantages of silica optical fibers include, for example, the fact that the technology for manufacturing active optical fibers by chemical vapor deposition (CVD) is quite common, affordable, and well developed. The CVD technique provides a high level of purity and homogeneity of the composition of the synthesized glasses and, in turn, this provides high reproducibility of the parameters of optical fibers. The technology of creating birefringent fibers made of quartz glass is also well developed. In addition, when splicing silica fibers with one another, there is no problem of reducing the mechanical reliability of the splicing points, which occurs when joining multicomponent (for example, phosphate) and silica-based fibers due to the difference in the thermal expansion coefficients of these glasses [6].

For the fabrication of laser resonators designed for operating in the telecommunications wavelength range of 1500–1600 nm, silica fibers with a core co-doped with erbium and ytterbium ions have proven themselves fairly well [7]. In addition, the core glass of these fibers should contain phosphorus oxide, since the transfer of excitation energy from ytterbium ions to erbium ions is most efficient in the phosphosilicate matrix [8]. It should be noted that phosphosilicate glass (including those activated by erbium and ytterbium ions) has a weak photosensitivity to UV irradiation; therefore, it is very difficult to inscribe Bragg gratings with a high reflection coefficient in active phosphosilicate glass-core fibers using standard techniques [9,10]. The low initial photosensitivity limits the use of phosphosilicate fibers for the fabrication of high-Q resonators of single-frequency lasers, in which Bragg gratings occupy a significant part of the fiber core material.

An urgent challenge of achieving high efficiency in laser resonators is focusing on minimizing intra-cavity losses, taking into account that the resonator (including Bragg gratings) is preferably made entirely on a segment of the active fiber in order to avoid additional losses associated with the joining points of the active optical fiber and fiber Bragg gratings. Possible ways to solve this problem have been proposed previously. The method of increasing (by more than an order of magnitude) the photosensitivity of phosphosilicate fibers to UV irradiation by pre-saturating them with molecular hydrogen (“hydrogen-loading”) has become the most widespread [9,10]. The negative side of this approach is the growth of photoinduced optical losses in the fiber, associated with the guidance properties of the core doped with hydroxyl centers in the glass matrix [10]. In addition, the increased probability of non-radiative relaxation of the excited state of erbium and ytterbium ions as a result of the dipole–dipole interaction with hydroxyl centers also negatively affects the laser emission characteristics of the active fiber [11].

It is possible to increase the initial photosensitivity of Er/Yb phosphosilicate fibers in another way, at the stage of a fiber preform fabrication. In such a fiber, the core has a two-layer structure (a confined, doped core combined with a photosensitive cladding design): the central active part is surrounded by a photosensitive layer doped with germanium and boron oxides [12,13]. The disadvantage of this approach is the small overlap between the Bragg grating inscription area and the guided fundamental mode of the optical fiber core. The manufacture of an optimal two-layer structure of the fiber core, in which the refractive index of the outer photosensitive layer should be close to the refractive index of undoped quartz glass, is also of great technological complexity.

Within the framework of this research, a fiber with a germanophosphosilicate core doped with a high concentration of Er₂O₃ and Yb₂O₃ oxides was manufactured. In our previous work [14], we found a close-to-optimal ratio of the concentrations of Er₂O₃ and Yb₂O₃ dopants in the phosphosilicate matrix (0.1 and 0.8 mol%, respectively) in order to develop active Er/Yb fiber with a competitive gain at ~1550 nm (0.2 dB/cm). The disadvantage of the phosphosilicate fiber presented in [14] was the extremely low initial photosensitivity of the fiber’s core glass host, which was compensated by hydrogen loading. In the developed germanophosphosilicate fiber, the addition of only 1.5 mol% GeO₂ to the phosphosilicate glass ensured higher photosensitivity of the fiber core to UV irradiation from an excimer laser (193 nm). A fiber laser Fabry–Perot resonator was fabricated on a segment of this fiber by inscribing a pair of matched Bragg gratings directly in the fiber core without its preliminary treatment with molecular hydrogen. The laser demonstrated stable single-frequency generation at a wavelength of 1540 nm. The following main characteristics of the radiation emerging from the laser were studied: optical power dependence (slope efficiency), emission linewidth, and relative intensity noise.

2. Materials and Methods

2.1. Optical Fiber Preform Preparation Details

The fiber preform was fabricated by the MCVD method, with deposition of the Er₂O₃-Yb₂O₃-GeO₂-P₂O₅-SiO₂ glass core components using an all-gas-phase deposition technique. High-purity volatile chlorides POCl₃, GeCl₄, SiCl₄, and CCl₄, as well as the low-volatility chelates Er(thd)₃ and Yb(thd)₃, were used as precursors (thd = 2,2,6,6-tetramethyl-3,5-heptanedicionate) containing less than 10⁻⁵ wt% transition metal impurities. The light-guiding structure was deposited using a Heraeus Suprasil F300 reference quartz tube with an outer diameter of 15 mm and a wall thickness of 1.3 mm. Before the core glass deposition, the silica tube was preliminarily pre-collapsed to an inner diameter of 5–6 mm. The use of the thin-walled tube and its preliminary collapse made it possible to reduce the temperature for the sealing of a tube preform into a rod from 2200 °C to 1950–2000 °C and reduce the time of high-temperature treatment to only 1 burner pass. Through these approaches, we were able to minimize the loss of P₂O₅ and GeO₂ from the center of the core and reduce the region of the central dip in the refractive index profile.

A serious problem in the manufacture of active fibers is the low volatility of REE precursors. In our case, the chelate containers were heated to a maximum temperature of 160 °C, limited by the sintering of the powder. However, even at such a high thermostatting temperature, the vapor pressure of chelates (~1 mm Hg) is 2 orders of magnitude lower relative to other precursors. For this reason, it is impossible to fabricate fibers with high contents of Er₂O₃ and Yb₂O₃ using the standard MCVD technique, with simultaneous dosing of all necessary precursors into the deposition tube. A high concentration of erbium and ytterbium oxides in the core was secured by an original technique for the separate deposition of glass components [15]. This deposition technique is as follows: First, at a temperature of 1300–1400 °C, a layer of GeO₂/P₂O₅/SiO₂ glass is deposited in the form of soot. Then, a layer of Er₂O₃/Yb₂O₃ oxides is deposited on its surface, maintaining 5–6 burner passes along the preform’s length. Finally, the layer impregnated with active dopants at a temperature of 1850–1900 °C is sintered into a transparent Er₂O₃-Yb₂O₃-GeO₂-P₂O₅-SiO₂ glass layer in the presence of CCl₄ flow to withdraw the water impurities formed during the impregnation stage. The manufacture of the entire glass core with a diameter of ~ 1.5 mm by depositing 6 thin porous layers, with long-term impregnation of each of them, provides the necessary concentration of the active dopants and homogeneity of the glass over the cross-section of the core.

Figure 1 shows the parameters of an ASL273 fiber sample synthesized by the developed method. The refractive index profile in the preform was measured using a PK2600 preform analyzer (Photon Kinetics, Beaverton, WA, USA). The core glass composition was determined by X-ray microanalysis (JEOL 5910LV electron microscope, Tokyo, Japan). A single-mode fiber (cutoff wavelength: 0.92 µm) with an outer diameter of 125 µm was drawn from the preform. The measurement of the fiber’s optical losses was carried out with the help of a Yokogawa AQ6370D optical spectrum analyzer in the spectral range of 900–1650 nm, with a spectral resolution of 1 nm, by means of a cutback technique. The maximum concentration of P₂O₅ in the fiber core was 12 mol%, that of GeO₂ was ~1.5 mol%, and those of Yb₂O₃ and Er₂O₃ were 1.33 and 0.25 mol%, respectively. The difference between the refractive indices of the core and cladding (Δn) measured in the preform was 0.021, which, within the measurement error, is consistent with the composition of Er₂O₃-Yb₂O₃-GeO₂-P₂O₅-SiO₂ glass. The absorption value of Er³⁺ ions at a wavelength of 1535 nm was 87 dB/m, and that of Yb³⁺ ions at a wavelength of 976 nm was ~1400 dB/m. The level of minimal optical loss in the fiber at a wavelength of 1150 nm (far from the absorption peaks of erbium and ytterbium) did not exceed 70 dB/km, indicating high purity and homogeneity of the obtained glass, in which there were no scattering centers (i.e., second-phase inclusions).

2.2. Fiber Laser Ultrashort-Cavity Fabrication Technique

In the manufactured optical fiber, the induced refractive index was measured depending on the dose of UV radiation, with a wavelength of 193 nm. An Optosystems CLS-5000 ArF excimer laser (pulse duration of ~10 ns and repetition rate of 10 Hz) was used as a source of this radiation. To determine the amplitude of the modulation of the induced refractive index, the method from [16] was used, which is based on the mathematical processing of the experimental spectrum of a fiber Bragg grating. The Bragg grating was inscribed using a standard phase mask with a period of 1064 nm. The measurement results are shown in Figure 2. The maximum recorded value of the induced refractive index (refractive index modulation) was 3 × 10⁻⁴ at a total exposure dose of 1200 J/cm². In [17], in a fiber co-doped with P₂O₅ and GeO₂, the same value of modulated refractive index was achieved with doses of UV irradiation at a wavelength of 193 nm, with a corresponding exposure dose above 3000 J/cm². It is important to note that the Bragg grating formation dynamics presented in this work had a rather complex form due to the competition between two different mechanisms of photoinduced refractive index inscription, namely, type I and type IIa. The principal difference between the refractive index induction dynamics obtained in our experiment and that observed in [17] was our relatively monotonic growth, corresponding predominantly to the type I mechanism. The weak contribution of the type IIa mechanism, which manifests itself when gratings are inscribed in fibers with a GeO₂ concentration of about 20 mol% or higher [18], can be explained by the low concentration of GeO₂ in our fiber sample (only 1.5 mol%).

The Fabry–Perot fiber laser (also called a “Distributed Bragg Reflector” or “DBR” laser) cavity studied in this work was formed entirely on a short segment (28 mm long) of the active fiber. The FBG inscription was carried out using a standard technique of UV laser irradiation through a phase mask. The active fiber segment was irradiated in a “pristine” state (i.e., without hydrogen loading or any other treatments). Using a phase mask with a period of 1064 nm and 193 nm pulsed radiation from the Optosystems CLS-5000 ArF excimer laser, two wavelength-matched uniform FBGs (highly reflecting, or “HR”, and partially reflecting, or “PR”) of the same length (9 mm) were induced in the core of the active fiber at a distance of 10 mm from one another. First, a PR grating with a length of 9 mm was recorded, with a full width at half-maximum (FWHM) of 0.2 nm and a reflection coefficient of 13 dB. The PR FBG was inscribed with a total UV irradiation fluence of 0.5 kJ/cm².

The transmission spectrum of this grating is shown in Figure 3 (curve 1). Then, an HR FBG grating, also 9 mm long, was recorded. The HR FBG was inscribed using a 5 times greater dose of exposure: ~2.5 kJ/cm². The dose of UV radiation required to inscribe a Bragg grating with a reflection coefficient of more than 99.9% was calculated based on the measured dependence, as presented in Figure 2. The length of the segment of the active fiber between the edges of the gratings was 10 mm.

Thus, the total length of the fabricated laser cavity, including the HR and PR FBGs, was 28 mm. The transmission spectra of the total laser cavity, measured immediately after its inscription using the Yokogawa AQ6370C optical spectrum analyzer, are depicted in Figure 3. As can be seen from Figure 3, the maximum intensity of the Bragg peaks of the FBGs is located near the wavelength of 1540 nm. The interference peaks on the mutual HR and PR FBGs transmission spectrum are related to the Fabry–Perot cavity resonances. The experimentally measured distance between the neighboring peaks is 0.06 nm, and this is reliably consistent with the numerically simulated mode-spacing value (7.9 GHz) of this Fabry–Perot cavity using the formula from [19].

3. Results

The laser cavity was fusion-spliced to the pieces of passive single-mode germanosilicate optical fibers (core diameter of 4 μm and a cutoff wavelength of 920 nm). The cavity was located on a massive metal substrate, without additional cooling. Thermally conductive paste was used for better heat dissipation from the laser to the substrate. The experimental setup for testing the characteristics of the laser is shown in Figure 4. The laser was backward-pumped with a single-mode laser diode operating at a wavelength of 976 nm through the 980/1550 nm wavelength division multiplexer (WDM). The output port of the WDM was fusion-spliced to the PR FBG. To prevent undesired back-reflection, two optical isolators were used on both sides of the laser cavity. To measure the laser characteristics, the laser’s output signal after the isolator was alternately fed to a scanning Fabry–Perot interferometer (FPI), an optical spectrum analyzer (OSA), a photodiode power sensor, and a photodetector (PD) connected to a radio frequency spectrum analyzer (ESA) or an oscilloscope (Osc.), as well as into a Mach–Zehnder interferometer to measure the linewidth using a self-heterodyne technique (Figure 4).

In our setup, we used a scanning Fabry–Perot interferometer with a free spectral range of 750 MHz and a resolution of 11 MHz. Despite the fact that several maxima were observed in the transmission spectrum of the fiber laser cavity (Figure 3), the laser operated in a single longitudinal mode (Figure 5). Single-frequency operation was stable without mode-hopping. An examination of the laser signal using a photodetector connected to an oscilloscope showed that the lasing was continuous-wave without self-pulsation.

The lasing wavelength measured with the optical spectrum analyzer was 1540 nm (Figure 6). The Thorlabs Photodiode Power Sensors S122C was used as a power meter to measure the output laser power. The slope efficiency with respect to the launched pump power was 10%, and the laser threshold was 65 mW (Figure 7).

The laser linewidth was measured using a self-heterodyne technique based on the Mach–Zehnder interferometer (Figure 4). The output of the laser was divided between two arms of the interferometer, one of them contained an acousto-optic modulator (AOM) with a carrier frequency of 110 MHz, while the other contained a 25 km delay line. The radio frequency (RF) signal was analyzed using a Rohde & Schwarz FPC1000 Spectrum analyzer. The measured beat-signal spectral width was about 1 MHz at the level of −20 dB (Figure 8); this corresponds to a laser linewidth of 50 kHz.

The relative intensity noise (RIN) was measured using a fast photodetector with a bandwidth of 200 MHz. The measurements were carried out at a laser output power of 25 mW. The resolution of the RF spectrum analyzer’s RBW was 10 kHz. The measured RIN spectrum in Figure 9 shows a maximum associated with relaxation oscillations at a frequency in the vicinity of 2.3 MHz. Towards the higher frequencies, the noise decreases monotonically. The measured maximum value of RIN (at the frequency of relaxation oscillations) was −103 dB/Hz.

The laser output power stability was investigated in long (~30 min) and short (80 microseconds) time intervals. The time dependencies of power are shown in Figure 10a,b for the long and short time intervals, respectively. Figure 10a reveals two time dependencies, which corresponds to the two different values of laser power: 2 and 9.5 mW. The calculated standard deviation values are 0.04 and 0.07 mW, respectively. In Figure 10b, there are two oscillograms plotted, associated with a “before lasing threshold” state and a “normal lasing” mode at the power of 2 mW.

4. Discussion

The differential lasing efficiency obtained in this work was 10%. It exceeded the efficiency of the resonator (0.3%) by more than an order of magnitude, which was studied in our earlier work [20] and was also fabricated on a 25 mm long Er/Yb phosphosilicate fiber using a similar grating inscription technique. In our opinion, such a significant difference in the slope efficiency of two resonators with approximately the same lengths (28 and 25 mm) can be mainly attributed to the hydrogen-loading treatment of the Er/Yb phosphosilicate fiber sample used in [20]. According to [11,21], hydrogenation of the core glass together with UV irradiation of Er-doped fibers induces additional losses and increases the probability of non-radiative relaxation of the excited state of erbium ions, which leads to degradation of the fiber gain characteristics and, potentially, of the laser slope efficiency. A detailed study of the effects of hydrogen loading on the amplifying and laser properties of Er/Yb phosphosilicate fibers will be the subject of further research.

The detailed analysis of the laser signal oscillograms (Figure 10b) shows clearly that the laser operates strictly in continuous-wave (CW) mode, i.e., without any periodic oscillations. It should be taken into account that single-frequency fiber lasers based on the “single” Er-doped (i.e., without Yb co-doping) fibers usually have a problem with their CW operation mode stability. The nature of this issue is a break in the CW mode’s operation by the spontaneous excitation of the self-pulsing (or self-Q-switching) mode in the laser cavity [22,23]. Thus, additional Yb co-doping of Er-doped fibers plays a crucial role in the CW mode stabilization of the laser’s short cavities based on these fibers. However, single-frequency lasers with cavities based on “single” Er-doped fibers have a much narrower linewidth in comparison with the investigated laser (50 kHz)—i.e., several kilohertz [24], or even sub-kilohertz [25] values. Taking this fact into account, we assume that the broadening of the laser’s linewidth in highly Er/Yb-doped fibers is caused by thermal effects due to the non-uniform pump absorption along the length of the laser cavity.

5. Conclusions

A highly doped Er/Yb fiber with a high initial photosensitivity to UV irradiation was developed, which is intended for use as the basis for resonators of single-frequency fiber lasers generating in the telecommunications wavelength range of 1500–1600 nm. The concentrations of Er₂O₃ and Yb₂O₃ in the fiber core region were 0.25 and 1.33 mol%, respectively. The main dopant of the core was P₂O₅, with a concentration of 12 mol%, which ensured efficient transfer of excitation energy from Yb³⁺ ions to Er³⁺ ions. Additionally, the fiber core was co-doped with ~1.5 mol% GeO₂. This concentration of germanium oxide was sufficient to ensure the high initial photosensitivity of the core, necessary for the fabrication of highly reflective Bragg gratings in the fiber using a standard phase mask and laser UV radiation at a wavelength of 193 nm. Thus, a Fabry–Perot fiber laser resonator was fabricated on a segment of the active fiber with a total length of 28 mm, without a hydrogen-loading stage. The developed laser, which had an effective cavity length of ~13 mm, demonstrated stable single-frequency lasing at a wavelength of 1540 nm. The differential efficiency (slope efficiency) of laser generation was 10%, which is several times higher than that of resonators fabricated using a similar technique for inscribing Bragg gratings in hydrogen-loaded fibers.

The main advantage of the developed Er/Yb optical fiber with a germanophosphosilicate glass core is its high initial photosensitivity to UV irradiation. Due to this, it is not necessary to pre-saturate the fiber with molecular hydrogen before inscribing the Bragg gratings, which greatly simplifies the fabrication of resonators for single-frequency lasers with good slope efficiency.