1. Introduction

As one of the atmospheric transparent windows, single frequency fiber laser operating at 2 μm has attracted many researchers because of its numerous potential applications including eye safe range-finding, LIDAR for heterodyne measurement of wind velocity and as sources for mid-infrared optical parametric oscillators [1–4]. Usually, Tm³⁺ -doped, Ho³⁺ -doped, and Raman fiber lasers are the most widespread way of getting 2-micron laser output. However, Raman fiber lasers requires high power pump sources in the spectral range, which is a shortcoming in the view of practical applications [5]. The Tm³⁺ doped sources are typically limited to efficient operation at <2.05 μm [6], although Tm³⁺ doped fiber lasers with high output power and slope efficiency have been demonstrated in silica [7], silicate [8] and germanate fibers [9]. The transition (⁵I₆→⁵I₇) of Ho³⁺ ions produces radiation in the range of 2.05 μm to 2.2 μm [10], which could totally match the applications which require good atmospheric propagation. Atmospheric transmission spectra provided by ModTran [6] show the advantage of operating at wavelength beyond 2.1μm in comparison to the windows accessible thulium sources.

In order to get a single frequency output, a short laser cavity with enough net gain is required [1]. Therefore, host materials must have high solubility of holmium ions, which are usually exhibited by heavy metal glasses [1] (e.g., germanate, telluride, and fluoride glasses). Another concern for choosing host materials is the phonon energy. High phonon energy results in fast multiphonon relaxation, thereby causing thermal damage and low quantum efficiency [5]. Unlike telluride and fluoride glasses, germanate glasses combine large rare earth solubility, moderate phonon energy (800 cm⁻¹ to 900 cm⁻¹) with strong mechanical strength and high damage threshold. In 2009, A single frequency fiber laser operating at 2.05 μm with over 50 mW output power has been obtained in a 2 cm long Ho³⁺ -doped germanate glass fiber. The maximum doping concentration of Ho₂O₃ in the fiber is 3wt% [9].

In this work, we prepared a series of Ho₂O₃ -doped lead germanate glasses with doping concentration varied from 0.1 wt% to 4 wt%. The optical and spectroscopic properties of the lead germanate glasses with different doping concentrations were presented and studied. The lifetime quenching caused by OH groups was also analyzed, and the energy transfer rates to OH groups under different doping concentrations were obtained. Unclad germanate fibers with 4 wt% Ho₂O₃ were fabricated, and pumped by using an in-house constructed 2 μm fiber laser.

2. Experiment methods

Glass samples of the lead-germanate, 50 mol%GeO₂ + 5 mol%SiO₂-20 mol% PbO-20 mol%CaO + 5 mol% K₂O with x wt% of Ho₂O₃ (x = 0.1, 0.5, 2, 3, 4), denoted as G1 to G5, were prepared by using a conventional quenching method. The commercially available raw materials used were GeO₂ and Ho₂O₃ with 99.999%, and the others were of reagent grade. Well-mixed 20 g batches were melted at 1200 °C for 40 min in a platinum crucible. Dried O₂ was bubbled into the glass melt for 30 minutes to eliminate OH⁻. The melts were then quenched on a stainless steel plate and annealed for 2 h at 500 °C in an muffle furnace. Another glass sample with the same components of G5, named G6, was prepared using the same process described above, except that the batch of G6 was first dried in a vacuum drying oven at 110 °C for 24 h before melting to remove the crystal water content in the raw materials, and the entire preparation process of G6 was conducted in a dried environment (relative humidity ≤5% at 18°C). The annealed samples were cut and polished to dimensions of 10 mm × 10 mm × 0.5 mm for optical and spectroscopic measurements.

A glass batch with components similar to G6, in quantity of 300 g of powder, was melted using the same process applied to G6. Considering the large mass of this glass batch, a longer time of each procedure is required. The well-mixed 300g batch was heated for more than 48 h in the vacuum drying oven before melting. Then, the mixture was melted in platinum crucibles for 3 h in an electrically heated furnace at ~1200 °C with a protective atmosphere with dried oxygen. Three hours later, dried oxygen was bubbled into the melt for 1 hour to further remove the OH- groups in the melt. The stir effect caused by bubbles could homogenize the melt and reduce the cluster of rare earth ions. Finally, the glass batch were quenched on a stainless steel plate and finely annealed at glass transition temperature for 100 h. The entire procedure was also conducted in a drying room. A glass rod was drilled from the bulk glass (BG) and finely polished, then drawn into a 125 μm diameter unclad fiber in a fiber drawing tower.

The refractive index and density of the samples were measured by the prim minimum deviation and the Archimedes method using distilled water as an immersion liquid. The absorption spectra were recorded by using a PerkinElmer Lambda 900 UV/VIS/NIR spectrophotometer. The emission spectra were measured with an Edinburg FLS920 type spectrometer. All measurements were conducted at room temperature.

The concentration of Ho ions was calculated using the measured density and the quoted weight fraction of Ho. The material density was measured as 4.98 g/cm³. For a given weight fraction (W), the Ho density (N_Ho), is calculated as 2Wρ/M, where ρ is the density and M is the mass of Ho₂O₃ (6.2767 × 10⁻²² g). For the 2 wt% sample (G2), the doping concentration of Ho³⁺ ions was calculated to be 3.12 × 10²⁰ cm⁻³.

3.Results and discussion

3.1 Infrared (IR) transmission, absorption spectroscopic properties, and Judd-Ofelt analysis

The infrared transmittance spectrum of G4 is presented in Fig. 1. As is shown, the maximum transmittance is 84.6%, and the IR cutoff wavelength reaches approximately 5.8 μm. The refractive index of G4 at 2.5 μm was fitted by Cauchy dispersion equation and the value is 1.787. This value corresponds to 15.3% Fresnel reflection loss of the two surfaces of the used glass sheet, and such caused the low IR transmittance of the G4 glass. The wide absorption band around 3 μm was caused by the free OH groups [11]. The hydroxyl absorption coefficient was calculated using the equation [12]

 

Fig. 1 IR transmittance spectrum of G4; the inset shows the absorption coefficients of all samples.

Download Full Size | PDF

where l is the sample thickness, T is the transmittance at 3 μm, and T₀ is the transmittance of the host matrix. The calculated results of G1 to G6 are presented in the inset of Fig. 1. The absorption coefficients of G1 to G5 are nearly kept at 2 cm⁻¹, whereas the value of G6 remarkably decreases to 0.45 cm⁻¹. It means that the preparation procedures applied to G6 could effectively eliminate the OH groups. According to reference [11], the content of the free OH groups was estimated from the measured maximum absorption coefficient. The free OH content N_OH (ions/cm⁻³) was acquired as follows [11]

where N_A is Avogadro’s constant, and ε is the molar absorptivity of the free OH⁻ groups in the glass. Here we adopt the molar absorptivity in silicate glasses, 49.1 × 10³ cm²/mol [13], to estimate the OH⁻ content in the germanate glass because of unavailable result for germanate glass by now. As a result, the typical contents of OH groups are about 2.45 × 10¹⁹ cm⁻³ in G1 to G5, and 0.5 × 10¹⁹ cm⁻³ in G6, 4 times smaller than that of G1 to G5. Therefore, it can be concluded that the dehydration method utilized to G6 could efficiently reduce the OH content.

Figure 2 shows the subtracted absorption spectrum of the 3 wt% Ho₂O₃-doped germanate glass from 300 nm to 2200 nm. Eight inhomogeneously broadened absorption peaks were observed from the absorption curve. The bands located at approximately 420 nm, 448 nm, 486 nm, 538 nm, 640 nm, 890 nm, 1148 nm, and 1936 nm were designed to the transition from the ground state ⁵I₈ level to the ⁵G₅, ⁵F₁ + ⁵G₆,⁵F₂ + ³K₈, ⁵F₃,⁵F₄ + ⁵S₂,⁵F₅, ⁵I₅,⁵I₆, and ⁵I₇ levels, respectively. The energy level locations were very similar to those previously reported for telluride [14] and fluoride [15] glass hosts. Owing to the strong intrinsic bandgap absorption located around 370 nm, the absorption from Ho³⁺ cannot be clearly identified below 370 nm. We adopted a 640 nm laser diode to pump the Ho³⁺ -doped glass samples in this work because of the unavailable of 1150 nm pump sources.

 

Fig. 2 Subtracted absorption spectrum of G4.

Download Full Size | PDF

Three Judd-Ofelt intensity parameters Ω_t(t = 2, 4, 6) of Ho³⁺ ions in the germanate glasses, which dominate the radiative transition probabilities, are derived from the least square fit to the measured and experimental oscillator strengths. In the Judd-Ofelt theory, the line strength of the electric-dipole transition between an initial level aJ and a final level bJ'is given by [16, 17]

The squared terms in Eq. (3) are the matrix elements of the doubly reduced unit tensor U^(t) (t = 2, 4, and 6), which are known to be almost independent on host materials. Thus, the U^(t) values calculated for Ho³⁺ in LaF₃ by Weber et al. [18] were used.

The electric-dipole line strength is related to the integrated absorption coefficient over the band and was obtained by a equation described as [17]

Table 1 presents the calculated total radiative transition probability A_r, lifetimes τ_r, and branch ratios β of several energy levels. The results show that the value A_r of the ⁵I₇ level of Ho³⁺ was 94.5s⁻¹, which was larger than that of silicate glasses [21], but smaller than that of tellurite glasses [22]. Given that the radiative transition probability was proportional to n(n² + 2)²/9, and the refractive index in germanate glass was usually intermediate between silicate and tellurite glasses, it was reasonable to obtain the presented result [14].

Table 1. Radiative transition rates (A_r), branching ratios (β), and radiative lifetimes (τ_r) for different energy levels

View Table | View all tables in this article

3.2 Emission properties and lifetimes of the ⁵I₇→⁵I₈ transition with different Ho³⁺ concentrations

The emission spectra of Ho³⁺-doped germanate samples with different doping concentrations were measured at room temperature. A 640 nm laser diode was used to excite Ho³⁺ for fluorescence properties characterization. Considering the large phonon energy of 822 cm⁻¹ of this glass determined from the phonon side spectrum, we supposed that most of the excited Ho³⁺ ions in ⁵F₅ level would finally relax to ⁵I₆ level via multiphonon relaxation after the ions exciting from ground level to ⁵F₅ level. All detailed process are indicated in Fig. 3.

 

Fig. 3 Energy level diagram of Ho³⁺ in the lead-germanate glass; the dash lines represent the non-radiative processes, the solid lines stand for absorption (red line) and emission (A₃₁ and A₂₁) transition of Ho³⁺ ions, and the solid curve means the energy transfer from Ho³⁺ to OH groups.

Download Full Size | PDF

Figure 4(a) shows the normalized fluorescence spectra of G1 to G6. Four major emission peaks were observed: one was at 1890 nm, and three were located at 1942, 2010 and 2050 nm, respectively. The multiple peak profiles of the emission spectrum mainly comes from the rich Stark splits of two transition levels of ⁵I₇ and ⁵I₈. The long wavelength end of the emission spectrum extends well beyond 2100 nm. Besides, no broadening effect is observed from the normalized fluorescence spectra, with, which are usually observed in the glasses with high doping concentrations. In addition, the strongest emission is always located at 2010 nm. The stimulated emission cross-section of Ho³⁺ in this germanate glass was calculated from the fluorescence spectra using Füchtbauer-Ladenburg equation [11]

 

Fig. 4 (a) Normalized emission spectra of G1 to G6; (b) overlaps of absorption and emission cross-section spectra of the 2 μm band; (c) $\left({\tau }_m^{-1}-{\tau }_r^{-1}\right)$(black squares) and integrated 2 μm emission intensity (red squares) as a function of Ho³⁺ ions; ${\tau }_m$and ${\tau }_r$ are the measured and calculated lifetimes of ⁵I₇ level, respectively; (d) fluorescence decay curves of Ho³⁺ doped germanate glasses;.

Download Full Size | PDF

Variations in the integrated emission intensity (red cubic) of the ⁵I₇→⁵I₈ transition with different Ho³⁺ concentrations are presented in Fig. 4(c). The integrated emission intensity increased with increasing Ho³⁺ concentration, although there were some approaches in which could decrease the emission intensity instantaneously such as energy transfer to OH^— or other quenching centers. Especially, the integrated emission intensity of G6 was twice as strong as that of G5, indicating that removal of OH groups could efficiently improve the 2 μm emission intensity.

As Fig. 4(d) shows, the fluorescence lifetime of ⁵I₇→⁵I₈ transition of Ho³⁺ ions near 2-micron decreased from 2.96 ms to 1.85 ms when the doping concentration of Ho₂O₃ increased from 2 wt% to 4 wt%, whereas the lifetime of G6 increased to 4.85 ms, which was higher than that of G5 at the same doping concentration. We did not acquire accurate decay curves for the samples with low doping concentration (≤0.5 wt%) because of the weak fluorescence intensity under short pulse pumping condition. The lifetime is defined as the time for the fluorescence intensity decrease to 1/e of the initial value. The fluorescence decay curves are strait lines respect to a log scale of Y-axis, which indicate there is no other significant nonlinear energy transfer between Ho³⁺ ions involved. Therefore, a possible mechanism for the lifetime decreasing is the energy transfer to the OH groups [23]. Therefore, considering multiphonon decay, the measure lifetime τ_m of Ho³⁺-excited state is finally given by [23]

Taking the values of N_OH, N₀, and ${\tau }_r$ into Eq. (9), and then fitting the Eq. (9) to the data shown in Fig. 4, we obtained the values of W₂₁ and $\alpha$ as 32.9 ± 8.5 s⁻¹, 24.4%, respectively. Thus, the value of W_OH for different doping concentration of Ho₂O₃ also can be calculated. The maximum quantum efficiency (η_m) of the ⁵I₇→⁵I₈ transition of Ho³⁺ ions in this lead-germanate glass expressed as A₂₁/(A₂₁ + W₂₁), was 74.1%. The quantum efficiencies, η, of G3~G6 were defined to be A₂₁/(A₂₁ + W₂₁ + W_OH). Based on the deduced parameters, we also could obtain W_OH, η and ${\tau }_m$ of G1 and G2 samples. All the results are listed in Table 2. The quantum efficiency of G6 is 47.1%, while 17.5% in the silicate glass [21]. A higher η could be reached by further removing the OH groups or other quenching centers.

Table 2. Doping concentrations (N_Ho), energy transfer rate to OH groups (W_OH), quantum efficiency (η), measured lifetimes (τ_m) at ⁵I₇ level, and the product of ${\tau }_m$ × N_Ho of Ho³⁺ in G1 to G6 glasses.

View Table | View all tables in this article

Moreover, using Eq. (7), the W_OH of G6 is calculated as 82.3 s⁻¹, which was close to the value (74.6 s⁻¹) calculated by combining the measured lifetime and W₂₁. Therefore, it can be concluded that the calculations we did match the experimental value well.

3.3 Gain properties and emission spectra in the unclad fibers

The parameter of material gain is an important index for the host glass. From a steady rate equation, the material gain is simply given by [19]

Figure 5 compares the mid-infrared emission spectra of 4 wt% of Ho₂O₃-doped BG glass fibers as a function of fiber length, using the pump excitation at 1.94 μm with a pump power of 500 mW. The spectrum line width in the bulk glass is 154 nm, which is much larger than in the fiber (59 nm). The line width in different fiber length exhibits no obvious change. Thus, the large overlaps between the absorption and emission spectra are proposed to causing the linewidth narrowing. With increasing the fiber length, the center wavelength shifts to the red side, from 2127 nm to 2142 nm, which can be ascribed to the growing radiation trapping along the fiber.

 

Fig. 5 Normalized infrared emission spectra of 4 wt% Ho₂O₃ doped germanate unclad fibers with different fiber length under the excitation of 1940 nm fiber laser.

Download Full Size | PDF

4. Conclusion

In summary, the mid-infrared luminescence properties of Ho-doped germanate glasses and fibers at 2 μm were investigated. Spectroscopic parameters for characterizing Ho³⁺ doped germanate glasses were analyzed and calculated based on Judd-Ofelt theory. The lifetime quenching mechanism of ⁵I₇ level of Ho³⁺ ion was also presented, the quenching rate to OH groups decreased dramatically from 412.2 s⁻¹ to 74.6 s⁻¹ under the same doping concentration (~6 × 10²⁰ ions/cm³) by reducing the OH groups. The index of gain medium, τ_m × N_Ho, was also improved to 30.26 × 10²⁰ ms/cm³ with a quantum efficiency of 47.1%, which predict a high gain and a low pumping threshold.

The emission spectra of 2 μm in bulk glasses and fibers were compared and analyzed. The linewidth was 154 nm in bulk glasses, while 59 nm in the fibers. The large overlap between emission and absorption spectra was proposed for the spectra narrowing. The central wavelength of the emission band shows a red shift with the increment of fiber length and locates around 2120 nm, which can be ascribed to the increasing radiation trapping with the fiber length. The above results indicate that this germanate glasses were attractive and promising laser materials for 2.1 μm single frequency laser output.