1. Introduction

Persistent attention has been devoted to the development of solid state lasers operating close to ~2.0 μm region due to the potential applications in the eye-safe radars, non-invasive medical diagnostics, atmospheric pollution monitoring, remote sensing, military weapons, and effective pump resource for mid-infrared (MIR) optical parametric oscillators (OPO) [1–4]. In this respect, Tm³⁺ has been recognized as suitable laser active ion for its large stimulated emission cross section and long-lived ³F₄ level related to Tm³⁺: ³F₄ → ³H₆ transition. With the advent and rapid development of AlGaAs laser diode (LD), diode-pumped Tm³⁺/Yb³⁺ system is proposed as a promising way to achieve ~2.0 μm laser because of the large absorption of Yb³⁺ at the diode-pumping wavelength 980 nm [5,6]. However, the accompanying intense up-conversion luminescence at 480 and 800 nm originating from the ¹G₄ → ³H₆ and ³H₄ → ³H₆ transitions of Tm³⁺ severely limit the laser output power and slope efficiency [7]. Therefore, there is still a need to explore a new pumping scheme and sensitizing mechanism that would produce an efficient ~2.0 μm emission.

As an alternative, the transition metal Bi ions usually show broad photo-excitation and emission bands, which can offer a variety of selectable pump wavelength and hence be considered as a sensitizer for rare-earth (RE) ions. Recently reported work on the broadband near-infrared (NIR) emission from Bi doped silica glass in the range of 1000–1700 nm has lead to intensive studies on several kinds of Bi doped or Bi-RE co-doped glasses [8]. For example, Sheng and Xu et al. obtained an enhanced broadband ~2.0 μm emission with an energy transfer (ET) efficiency of 47.8% and 72.6% from Bi/Ho³⁺ co-doped borophosphate glass and thin films, respectively [9,10]. Broadband NIR emission in Bi/Tm³⁺ co-doped germanate and chalcohalide glasses were observed and the possible energy transfer processes between Tm³⁺ and Bi were investigated by Ruan et al. [11–13]. Hau et al. achieved a super-broadband emission from 1000 to 2100 nm with a maximum full width at half maximum (FWHM) of 400 nm in the Bi/Tm³⁺ co-doped aluminosilicate glass, which exhibits much broader and longer wavelength than the traditional Tm³⁺ doped fiber amplifiers (TDFA) [14]. Fatehi et al. established an analytical model for Bi/Tm³⁺ co-doped fiber amplifier and confirmed Bi ions can effectively improve the amplification at 1.5–2.0 μm [15]. Besides, Halder et al. fabricated a Bi/Tm³⁺ co-doped lithium-alumino-germanate-silicate glass perform by the MCVD process, they found that the output of ~2.0 μm laser in this co-doped glass fiber has significantly higher efficiency and lower threshold pump power than Tm³⁺ singly doped one. They attributed this that the incorporation of Bi ions in the gain medium which increased the ³F₄ population of Tm³⁺ through ET processes [16]. All of these studies suggest that Bi/Tm³⁺ co-doped glass may be promising for obtaining more high-efficiency, tunable, and broad fiber amplifiers and lasers. Therefore, it is still essential to further study the mechanism between Bi and Tm³⁺ and some interesting results may be obtained.

Herein, we demonstrate an enhanced broadband 1.8 μm emission through Bi sensitization for Tm³⁺ in fluorogermanate glasses. Germanate glass is selected as the host primarily due to its lower phonon energy, broader infrared transparency range, higher solubility for RE ions, better thermal stability, chemical durability and mechanical strength. The underlying Bi → Tm³⁺ ET mechanism is rationally proposed and discussed according to the static-to-dynamic luminescence spectra. The intense sensitization of Bi to Tm³⁺ provides a novel approach for the visible lasers, OPO, commercial 808 and 980 nm LDs, which would greatly boost the potential application of Bi/Tm³⁺ co-doped fluorogermanate glasses for MIR fiber lasers and amplifiers.

2. Experimental

Glasses with a nominal composition (in mol.%) of 70GeO₂–12BaF₂–10Ga₂O₃–3BaO–3La₂O₃–2Y₂O₃ were prepared using conventional melting-quenching technique. The glasses were doped with x wt.% of Tm₂O₃ and y wt.% of Bi₂O₃ (x = 0, 2.0; y = 0, 2.0, 4.0, 8.0). High purity (≥99.99%) GeO₂, BaF₂, Ga₂O₃, BaCO₃, La₂O₃, Y₂O₃, Tm₂O₃, and Bi₂O₃ were used as raw materials. The mixed batches were melted in alumina crucibles at 1350 °C for 1 h and then poured onto a stainless plate followed by annealing at 590 °C for 2 h to remove internal stresses. Subsequently, glass samples were cut and optically polished into a shape of 10 × 20 × 1 mm³ for measurements.

Absorption spectra were measured by Perkin-Elmer Lambda 900 UV/VIS/NIR double beam spectrophotometer (Waltham, MA) with a resolution of 1 nm. The Raman spectra were performed on a Raman spectrometer (Renishaw in Via, Gloucestershire, UK) with a 532 nm laser as the excitation source. Excitation and emission spectra were employed on FLS920 fluorescence spectrometer (Edinburgh Instruments, UK) and computer controlled Triax 320 spectrofluorimeter (Jobin-Yvon Corp.). The 450 W Xe lamp, 808 and 980 nm LDs (Coherent Corp.) were used as the exciting sources. The visible and NIR emission were measured by detectors equipped with R928 photomultiplier tube (Products for research Inc., Danvers, MA) and InGaAs photodetectors, respectively. A PbSe photodetector assembled with lock-in amplifier (Stanford Research Systems, Sunnyvale, CA) and chopper for MIR emission. Additionally, luminescence decay curves were detected with μF900 microsecond lamp. All the measurements were carried out at room temperature.

3. Results and discussions

Figure 1(a) depicts the absorption spectra of Tm³⁺, Bi, and Bi/Tm³⁺ co-doped fluorogermanate glasses in the range of 300–2100 nm. For Tm³⁺ doped sample, six characteristic absorption peaks located at 356, 468, 682, 790, 1210, and 1670 nm can be ascribed to the typical electronic transitions from the ³H₆ ground state to ¹D₂, ¹G₄, ³F₂,₃, ³H₄, ³H₅, and ³F₄ excited states of Tm³⁺, respectively. Energy levels higher than ¹D₂ are not observed due to the intrinsic band gap absorption of the host glass. Four absorption bands at around 500, 730, 800, and 1000 nm are acquired in the Bi doped sample, similar to the characteristic Bi absorption peaks in other glasses [17,18]. In addition, it should be mentioned that glass samples containing Bi show a red-shift relative to Tm³⁺ singly doped glass, which may results from the absorption of Bi centered at around 370 nm. The glass color gradually changes from colorless to reddish-brown with high transparency (above 80%), as shown in the inset of Fig. 1(a). For Bi/Tm³⁺ co-doped sample, all the absorption bands of Tm³⁺ and Bi in singly doped samples can be observed obviously, and the absorptions bands of Tm³⁺ in 300–1000 nm are well overlapped with the broad absorption bands of Bi. It is noted that the intense broadband absorption of Bi nearly covers the near UV-to-NIR wavelength range in both Bi and Bi/Tm³⁺ co-doped samples, which perhaps greatly benefits to the feasible pumping of visible lasers, OPO, commercial 808 and 980 nm LDs if the ET process from Bi to Tm³⁺ occurs efficiently. The structure of the Bi/Tm³⁺ co-doped fluorogermanate glass is analyzed by Raman spectra, as shown in Fig. 1(b). Raman spectroscopy has been considered as a powerful tool in the study of compounds containing subvalent Bi, because many Bi cationic clusters are Raman active, but no detectable single of Raman band from Bi related species in our samples [8]. Only two broad peaks at ~510 and ~850 cm^–1 are detected which attributes to the symmetric bending and stretching mode contributions from Ge–O–Ge or Ga–O–Ga bridging oxygen’s vibrations, as well as the asymmetric stretching vibrations of Ge–O or Ga–O structural units, respectivily [19]. From the Raman spectra, it can be found that the largest phonon energy only extends to 850 cm^–1, which is much lower than that of other germanate glasses (900 cm^–1) and even some tellurite glasses (920 cm^–1) [20,21]. The smaller maximum phonon energy could reduce the probability of non-radiative relaxation and thus is very conducive to Tm³⁺ luminescence.

 

Fig. 1 (a) Absorption spectra of Tm³⁺, Bi, and Bi/Tm³⁺ co-doped fluorogermanate glasses in the range of 300–2100 nm. Inset: the corresponding photographs of the samples; (b) Raman spectra of the Bi/Tm³⁺ co-doped fluorogermanate glass.

Download Full Size | PDF

Figure 2(a) displays the NIR emission spectra from 900 to 1550 nm by exciting the Bi doped sample at 300–800 nm. For the mechanism of the NIR emission from Bi doped glass remains controversial, here we ascribe the origin of the luminescence to “active bismuth,” e.g., Bi⁺ [9]. It is well known that the Bi emission is strongly dependent on excitation. Varying the excitation wavelength within the spectral range of 300–800 nm, overall the NIR emission of Bi exhibits a red-shift from 1175 to 1253 nm. The maximum and minimum wavelengths are 1300 and 1107 nm upon excitation of 800 and 700 nm, respectively, as shown in Fig. 2(b). The emission peak variation with excitation wavelength in germanate glass has been reported by Peng et al., they ascribed this shift to multiple active centers in parallel [18]. The phenomenon investigated herein might be assigned to the same reason. As pointed out by the Raman spectra, multiple structural units coexist in the glass can provide a variety of ligand configurations for the optically active Bi-species. Hence, it is reasonable to suggest that the complex NIR luminescence of Bi may derives from different active centers in the present glasses. In addition, it is noteworthy that the maximum FWHM reaches to 330 nm, which is similar to the Bi-Al co-doped germanate (320 nm) and silica (336 nm) glasses, while much smaller than the Bi-Ta co-doped germanate (400 nm) and Bi doped chalcogenide (600 nm) glasses [22–25]. Considering the asymmetric feature of the emission spectra, they could be deconvoluted into different Gaussian peaks. Here we chose two special peaks to analysis, as shown in Figs. 2(c)-2(d). For the broadest NIR spectra, the fitting results demonstrate that there are three peaks at 1097, 1230, and 1433 nm, with a FWHM of 180, 308, and 457 nm, respectively. For the narrowest NIR spectra, three Gaussian fitting peaks located at 1100, 1300, and 1540 nm are observed, with a FWHM of 85, 232, and 221 nm, respectively. This tunable broadband emission of Bi greatly overlapped with the absorption of Tm³⁺ at 1210 nm, which much benefits the ET process and higher ET efficiency from Bi to Tm³⁺.

 

Fig. 2 (a) NIR emission spectra of Bi for different excitation wavelengths (300–800 nm); (b) Emission wavelength and FWHM as a function of excitation wavelength; Deconvolution of the (c) broadest and (d) narrowest NIR spectra in Bi doped sample.

Download Full Size | PDF

The NIR (a,c) and MIR (b,d) emission spectra of Bi, Tm³⁺ singly doped and Bi/Tm³⁺ co-doped fluorogermanate glasses upon excitation of 808 and 980 nm LDs were measured by the detectors equipped with InGaAs and PbSe, respectively, as shown in Fig. 3. Under 808 nm excitation, intense emission peaks at 1250 (Bi⁺: ³P₁ → ³P₀) and 1450 nm (Tm³⁺: ³H₄ → ³F₄) are observed in the singly doped glasses. When small amounts of Tm³⁺ ions are added into the Bi singly doped sample, a dramatic quenching for Bi luminescence at 1250 nm occurs, while the emission band of Tm³⁺ at 1450 nm shows minimal enhancement. With further increasing Bi₂O₃ concentration, the emission intensity of Tm³⁺ increased monotonously up to the maximum value when y = 8.0 wt.%, as shown in Fig. 3(a). The MIR spectra with different Bi compositions were also measured. For clarity, we present only the representative emission of Bi, Tm³⁺, and Bi/Tm³⁺ co-doped glasses for comparison, as shown in Fig. 3(b). It can be found that the introduction of Bi slightly enhances the 1.8 μm (Tm³⁺: ³F₄ → ³H₆) emission of Tm³⁺, while no apparent emission at 1.8 μm is detected in the Bi-doped glass. This enhancement of Tm³⁺ at 1.8 μm is also found by Li et al [26]. It is interesting that, however, a quenching process was observed by Ruan et al. and Tang et al [11,27]. This difference will be explained in the following discussions. Furthermore, significantly quenching of Bi at 1330 nm and enhanced emission of Tm³⁺ at 1.8 μm are acquired in the Bi/Tm³⁺ co-doped glass upon excitation of 980 nm LD, which confirms the occurrence of ET from Bi to Tm³⁺, as shown in Figs. 3(c)-3(d).

 

Fig. 3 Emission spectra in the region of (a) 1000–1540 nm, (b) 1400–2200 nm upon excitation of 808 nm LD, and (c) 1280–1410 nm, (d) 1400–2200 nm under 980 nm LD excitation.

Download Full Size | PDF

Figure 4(a) illustrates the excitation (λ_em = 1160 nm, dash line) and emission (λ_ex = 550 nm, solid line) spectra of Tm³⁺ and Bi singly doped glasses, respectively. As a result of the limitation of the spectrometer, the excitation spectra at the range of 800–1000 nm cannot be detected. The continuous excitation profile of Bi is composed by three excitation peaks located at 370, 470, and 740 nm in the range of 250–800 nm, which is basically in agreement with the absorption spectrum. From the absorption in Fig. 1(a), we can find that there is no photo-absorption at 550 nm in Tm³⁺ doped glass. Hence, the luminescence decay dynamics of Bi can be studied by upon 550 nm excitation in order to obtain more insight into the existing Bi → Tm³⁺ ET and calculate the related ET efficiency, as shown in Fig. 4(b). It is interesting to find that the lifetime of Bi drops greatly by co-doping with Tm³⁺, which means efficient ET from Bi to Tm³⁺, meanwhile the lifetime becomes slightly longer but still keeps a rapid decay with increasing Bi concentration. Both curves are best fit to a double-exponential decay function [28]:

 

Fig. 4 (a) Excitation (λ_em = 1160 nm, dash line) and emission (λ_ex = 550 nm, solid line) spectra of Tm³⁺ and Bi singly doped glasses; (b) Luminescence decay curves with different Bi concentrations upon excitation by 550 nm of Bi doped and Bi/Tm³⁺ co-doped glasses.

Download Full Size | PDF

The calculated values of $\overline{\tau }$ are 306 and 14 μs for 8Bi doped and 8Bi/2Tm³⁺ co-doped samples, respectively. The energy transfer efficiency $\eta$ can be evaluated by the following equation [9]:

Based on the above experimental results, a schematic representation depicting the possible ET mechanism between Bi and Tm³⁺ ions can be proposed and demonstrated in Fig. 5 [11–16,26,27,30]. Five possible ET processes and energy mismatch in this fluorogermanate glass can be summarized as following:

 

Fig. 5 Schematic energy level diagrams for Tm³⁺ and Bi⁺ ions and the proposed ET mechanism.

Download Full Size | PDF

Upon excitation of 808 nm LD, both Tm³⁺ and Bi ions are excited to the upper levels. The laser excitation populates the ³H₄ level of Tm³⁺ and then relaxes nonradiatively to the ³F₄ level, finally emitting a 1.8 µm photon. Meanwhile, the ¹D₂ level of Bi⁺ is populated and then decays to the ³P₁ level irradiatively. After that, a part of excitation energy from Bi⁺ is transferred to Tm³⁺ via the processes of ①, ② and ③ with emitting a ~1.2 µm photon. The first two ET processes are all in favor of increasing the ³F₄ population and thus improve the emission of Tm³⁺ at 1.45 and 1.8 µm, while process ③ can depopulation of Tm³⁺: ³F₄. It is noted that the energy gap of process ① is ~1704 cm^–1, hence two phonons are required to complete this energy mismatch. Unlike process ①, the energy gaps are only ~572 and ~796 cm^–1 in process ② and ③. Therefore, these two processes can be regarded as resonant ET and occur much easier. Besides, another resonant ET process ④ may appear easily when the Tm³⁺ concentration at a high level. This back ET populates the ¹D₂ level and then quickly decays nonradiatively to builds up the population of the ³P₁ level thus improves all others ET processes. At the same time, the cross relaxation (CR) process ⑤ between Tm³⁺ also takes place and populated the Tm³⁺: ³F₄ level. As can be seen, all ET processes help to increase the population of Tm³⁺: ³F₄ level expects process ③. The competition among them has a dramatic effect on the luminescence prosperities of Bi⁺ and Tm³⁺, which is mainly influenced by the overlap between Bi emission and Tm³⁺ absorption spectra, phonon energy of the glass host, and doping concentration. Aforementioned Ruan et al. and Tang et al observed quenched 1.8 μm emission in germanate and silicate glasses, they attributed it to the depopulation of Tm³⁺: ³F₄, namely the process ③ described above [11, 27]. The maximum doping concentration of Bi in both experiments is only 0.4 and 1 mol.%, and the phonon energy of silicate is much higher than the fluorogermanate glass, these may be the main cause of decreased 1.8 µm emission and low energy transfer efficiency obtained by them. When increase the Bi contents, the ³F₄ and ³H₄ levels of Tm³⁺ are greatly populated through efficient ET processes from Bi⁺ to Tm³⁺, which eventually resulting in the enhancement of 1.45 and 1.8 µm emission. Meanwhile, the Bi luminescent lifetime at 1160 nm sharply reduces and maintains a rapid pace of decay in the presence of Tm³⁺ even at a higher Bi doping concentration, as shown in Fig. 4(b). Upon excitation of 550 or 980 nm, Bi ions are initially excited from ³P₀ to the ¹S₀ or ³P₂ levels, respectively. Subsequently, these ions nonradiatively decay to the ³P₁ and transfer their energy to the adjacent Tm³⁺ ions, promoting the latter to the ³F₄ excited state via a similar ET process and further radiatively decay down to the ³H₆ level with the emission at 1.8 μm.

4. Conclusion

In summary, we demonstrated an enhanced broadband 1.8 μm emission of Tm³⁺ in the fluorogermanate glasses via Bi sensitization pumped by 808 and 980 nm LDs. The ET efficiency from Bi to Tm³⁺ is as high as 95.42%, which may be attributed to the larger overlap between Bi emission and Tm³⁺ absorption spectra, larger doping concentration with optimization, and lower phonon energy in this fluorogermanate glass host. Moreover, it is found that five competitive ET processes involved play an extremely important role in the luminescence properties of Bi and Tm³⁺. The present results manifest that the Bi/Tm³⁺ co-doped fluorogermanate glasses are excellent matrices for 1.8 μm emission and may provide a potential application in high-efficiency MIR fiber lasers and amplifiers.