Ce³⁺/Yb³⁺/Er³⁺ triply doped bismuth borosilicate glass: a potential fiber material for broadband near-infrared fiber amplifiers

Abstract

Erbium doped bismuth borosilicate (BBS) glasses, possessing the broadest 1.55 μm near infrared (NIR) emission band among oxide glasses, stand out as excellent fiber material for optical fiber amplifiers. In this work, we demonstrate that both broadened and enhanced NIR emission of Er³⁺ can be obtained by sensibly combining the effects such as mixed glass former effect, phonon-assisted energy transfer (PAET) and de-excitation effect induced by codopant. Specially, by codoping CeO₂ in a controlled manner, it leads to not only much improved optical quality of the glasses, enhanced NIR emission, but also significantly suppressed energy transfer up-conversion (ETU) luminescence which is detrimental to the NIR emission. Cerium incorporated in the glasses exists overwhelmingly as the trivalent oxidation state Ce³⁺ and its effects on the luminescence properties of Er³⁺ are discussed. Judd-Ofelt analysis is used to evaluate gain amplification of the glasses. The result indicates that Ce³⁺/Yb³⁺/Er³⁺ triply doped BBS glasses are promising candidate for erbium doped fiber amplifiers. The strategy described here can be readily extended to other rare-earth ions (REs) to improve the performance of REs doped fiber lasers and amplifiers.

Introduction

Since Payne et al. first demonstrated the feasibility of erbium doped fiber amplifiers (EDFAs) working at 1.55 μm near-infrared (NIR) wavelength¹, a wide range of applications in telecommunications, medical surgery, eye-safe measurement and spectroscopy have been realized^2,3. Nowadays, research is mainly oriented in two directions: either broaden the gain spectrum or enhance the output power of Er³⁺ doped fibers, to meet the insatiable demand for optical data transmission capacity and high power lasers. However, the gain width of conventional SiO₂-based EDFAs is restricted to 40 nm⁴, limiting the wavelengths that can be simultaneously amplified⁵. In contrast, Er³⁺ doped fluoride, tellurite and especially bismuth glasses exhibit gain spectra wider than 50, 70 and 80 nm, respectively^6,7,8, thus are more promising candidates for broadband EDFAs.

When it comes to lasing efficiency, although codoping with Yb³⁺ has been routinely applied⁹, the quasi-three-level scheme of Er³⁺ still suffers from the problems limiting the NIR lasing efficiency, such as the population inversion bottle neck between the 1.55 μm lasing level ⁴I_13/2 and the ground level ⁴I_15/2, the undesirable energy transfer up-conversion (ETU) emission and excited state absorption (ESA). Recent studies have focused on methods speeding up the non-radiative transition (multi-phonon relaxation, MPL) between the 980 nm pumping level ⁴I_11/2 and the lasing level ⁴I_13/2 to increase the population of the latter. Since the multi-phonon relaxation rate is exponentially proportional to the maximum phonon energy of host materials¹⁰, utilizing comparatively high phonon energy (HPE) materials is a sensible choice, for example, conventional SiO₂-based EDFAs have representative vibrational phonon energy 1100 cm⁻¹, higher than that of tellurite (700 cm⁻¹) and fluoride (500 cm⁻¹) fibers, it can provide gain amplification up to more than 1000 times easily. Borate crystals of even higher phonon energy (~1350 cm⁻¹), such as Na₂B₄O₇¹¹, Ca₄YO(BO₃)₃ (YCOB) and YAl₃(BO₃)₄ (YAB), have been demonstrated as excellent laser materials too^12,13. The HPE also brings in an added merit that it effectively restrains the deleterious ETU emission and/or ESA.

Another way to increase the population of the lasing level ⁴I_13/2 is by addition of a suitable codopant, such as Ce³⁺, Eu³⁺ or Tb³⁺, to accelerate the non-radiative decay ⁴I_11/2 → ⁴I_13/2 via phonon-assisted energy transfer (PAET). It has been found that all of the three codopants can decrease the lifetime of ⁴I_11/2 level, however, only does Ce³⁺ hardly influence the lifetime of the lasing level ⁴I_13/2, thus standing out as the optimal choice¹⁴. Codoping of Ce³⁺ has been widely applied to improve the NIR lasing efficiencies of optical fibers¹⁵, glasses^14,16, glass-ceramics¹⁷ and crystals¹⁸. It also helps to suppress the ETU emission and ESA, for example, the very strong ETU emission from Er³⁺ doped NaGd(WO₄)₂ crystal almost disappeared completely, and the lasing properties were strongly improved upon the codoping of Ce^3+ 19.

Having briefly reviewed the recent studies on attaining the broadened and enhanced NIR emission of Er³⁺ in various hosts, we chose bismuth borosilicate (Bi₂O₃-B₂O₃-SiO₂, BBS) glass system as potential fiber material for EDFAs. This system has some attractive properties, such as raw materials being relatively cheap, large refractive index (>2.0), excellent fiber drawing ability, being fusion splicable to silica-based fibers and intrinsically photosensitive that entails direct writing of fiber Bragg gratings in the same gain media. Besides, since the system contains HPE compounds B₂O₃ and SiO₂, it is expected that the average phonon energy of the glasses will be large.

Cerium oxide (CeO₂) was added into the system, serving as an oxidant to against darkening/crystallization of BBS glasses which limits to a great extent the optical applications of these glasses²⁰, but also as the ion source for Ce³⁺. The latter point deserves some comments, as is known that cerium incorporated into the glass network is possibly present in two oxidation states, Ce³⁺ and Ce⁴⁺, and their ionic equilibrium depends on the condition of glass formation and the type of glass system involved²¹. However, we will show in this work that cerium presented overwhelmingly in trivalent oxidation state Ce³⁺ in the BBS glasses. Our results also demonstrate that by resorting to the mixed glass former effect (inhomogeneous broadening) a very broad NIR emission can be obtained in the designed glasses, and when CeO₂ was doped in a controlled manner both enhanced NIR and dramatically weakened ETU emissions can be observed. Based on Judd-Ofelt (JO) analysis, it is concluded that the lasing properties of the glasses can be improved upon the codoping of CeO₂.

Experiments

Glass samples of (40-x)Bi₂O₃-50B₂O₃-10SiO₂-xCeO₂-1Yb₂O₃-0.5Er₂O₃ (x = 0, 0.1, 1, 3, 5, 10, 15, 20 in mol.%) were prepared by conventional melt-quenching method. The raw materials were weighed 20 g, mixed in an agate mortar for at least 15 min., stored in an alumina crucible and then put in an electric furnace. To prevent evaporation of Bi₂O₃, the materials were pre-sintered at 450 °C for 30 min., samples with varying content of CeO₂ were melted at different temperatures in a range from 950 to 1550 °C. The reason to increase the melting temperature is to dissolve more concentration of CeO₂, otherwise the samples suffer from serious crystallization (see Results part below). All the obtained glasses were annealed around T_g for 180 min. Finally, the glasses were cut and polished carefully to meet the requirements for optical measurements.

Absorption spectra were measured by a Perkin-Elmer Lambda 950 UV-VIS spectrophotometer over the spectral range of 300–2000 nm. Emission spectra were measured by an Edinburgh FS920 fluorescence spectrometer equipped with a liquid-nitrogen cooled steady state InSb detector and luminescence decay was recorded by a TDS3000C digital phosphor oscilloscope. Sample density was obtained by Archimedes method with alcohol as the immersion liquid. Er³⁺ concentration was calculated from the measured density and initial composition. The X-ray photoelectron spectra spectroscopic (XPS) was measured by ESCALAB250Xi (Thermo Fisher U.S.), the glass samples were broken in the XPS chamber and only those newly created cross-sections were measured. Raman spectra were measured by RenishawInvia Raman microscope (Renishaw, Gloucestershire, UK) with the excitation wavelength of 515 nm. All the measurements were carried out at the room temperature.

Results

Figure 1 shows coloration of the glasses with varying content of CeO₂ and melted at different temperatures. Depending on the content of CeO₂ and melting temperature (MT), the figure can be divided into four parts. At high MT (>1200 °C) and low CeO₂ content (<5 mol.%) region (b), the glasses suffer from serious darkening/devitrification and cannot be obtained due to the evaporation of Bi₂O₃; at high MT and CeO₂ content region (c), the addition of CeO₂ markedly improves the transparency and avoids the devitrification of the glasses. Obviously, the solubility of CeO₂ increases with MT such that up to 20 mol.% CeO₂ can be incorporated into the glasses. At low MT and CeO₂ content region (a), on the other hand, the addition of CeO₂ increases the absorption of the glasses in the visible region. Finally, glasses cannot be prepared at low MT and with high CeO₂ content (d) due to severe crystallization resulted from the undissolved CeO₂.

Figure 1: Digital photo of the obtained glasses doped with different concentration of CeO₂ (increasing from the left to the right) and melted at different temperatures (increasing from the bottom to the top).

The empty space implies that it is not possible to obtain the glasses with the indicated compositions and MT due to severe devitrification (crystallization).

The XRD patterns of the samples sitting on the border separating those that can form glass and those cannot were measured (cf. Fig. S1). The amorphous nature of the glasses was confirmed. DSC curves of some representative samples (showing the best NIR emission properties) were also measured (cf. Fig. S2 and Table S1). The ΔT = T_x − T_g (T_x, onset temperature of crystallization and T_g, glass transition temperature) values were considerably larger than 100 °C, indicating excellent thermal stability and fiber drawing potential. On the other hand, as is known, traditional bismuth glasses have relatively low melting (<1000 °C) and glass transition temperatures (<450 °C), the glasses volatilize and the optical properties worsen terribly when melted or handled at high temperatures, all of which pose a serious barrier to fiber drawing and fusion with conventional silicate optical fibers⁸. As shown in Fig. 1, addition of a moderate amount of CeO₂ (e.g., 3 mol.%) makes excellent glasses even when melted at temperature as high as 1350 °C. And the glass transition temperature can be significantly increased up to 638 °C for the glass containing CeO₂ up to 20 mol.% (Table S1). Therefore, the advantage to increase the melting and glass transition temperatures is to mitigate the difficult in the fiber drawing and fusion with conventional silicate fibers.

Absorption spectra of all the studied glasses were measured (cf. Fig. S3). Here, we show only spectra of the samples enclosed by two rectangles in Fig. 1. At low MT (e.g., 1050 °C), the addition of CeO₂ increases the absorption in the visible region such that the cut-off edge of the glasses undergoes a red-shift to longer wavelength (Fig. 2(a)). At high MT (e.g., 1250 °C), the transparency of the glasses improves markedly upon the addition of CeO₂, while the cut-off edge again shifts to longer wavelength with CeO₂ (Fig. 2(b)).

Figure 2

Absorption spectra of the samples marked by the rectangles in Fig. 1. (a) at low MT 1050 °C; and (b) at high MT 1250 °C.

The NIR emission spectra of all the studied glasses were measured (cf. Fig. S4), in all cases a very broad emission band with the full width at half maximum (FWHM) about 80 nm (Table 1), covering S (1460~1530 nm), C (1530~1565 nm) and L (1565~1625 nm) silica fiber communication bands, can be clearly observed. Codoping of CeO₂ resulted in the desired variations of the emission spectra. On one hand, both the intensity and bandwidth of the emission increase upon the finite codoping of CeO₂, for example, no more than 1 mol.% for the glasses melted at MT lower than 1050 °C. Further increase in CeO₂, however, exerts an adverse effect on the NIR emission intensity (Fig. 3(a)). We would like to stress that even though the enhancement of the NIR emission shown in Fig. 3(a) is not that breathtaking, the result is quite reliable and reproducible, because we have tested different batches of the glasses with the same designed composition and melted under approximately the same condition, the similar enhancement was observed in every run. The CeO₂ induced enhancement is more evident at high MT, for example, a two-fold increase in the NIR emission intensity was observed for the glasses melted at 1150 °C (cf. Fig. S4(c)). A thorough study also show that the optimal doping concentration of CeO₂ leading to the largest enhancement is MT dependent, and increases with MT (Fig. 3(d)). The variation of the NIR emission lifetime with the content of CeO₂ is shown in Fig. 3(c), it decreases with CeO₂ very mildly.

Table 1 JO parameters Ω_t (t = 2, 4, 6, ×10⁻² cm⁻²)*, FWHM (nm), σ_e (×10⁻² cm²), τ (ms), σ_e × FWHM (×10⁻ nm³) and σ_e × τ (×10⁻² cm²·s) of the samples.

Figure 3

NIR (a), ETU (b) emission spectra and measured lifetime (c) of samples melted at the same MT (1050 °C) but doped with different concentration of CeO₂. (d) Optimal doping concentration of CeO₂ leading to the largest enhancement of the NIR emission versus MT.

On the other hand, the ETU emission almost disappears completely upon even slight co-doping of CeO₂, and decreases gradually with increasing content of CeO₂ (insert in Fig. 3(b)). It is noted that the red 660 nm ETU emission is unusually stronger than that at the green 532 nm wavelength which will be discussed later on.

We also studied the MT dependence of the NIR and ETU emission spectra of the glasses. Fixing the doping concentration of CeO₂ at 1 mol.%, both the NIR and ETU emissions decrease with increasing of MT (Fig. 4(a,b)), similar to our previous work²². However, the full width at half maximum (FWHM) of the NIR emission band remains to be 80 nm (Table 1). On the other hand, the lifetime of the NIR emission varies arbitrarily with MT (Fig. 4(c)).

Figure 4

NIR (a), ETU (b) emission spectra and measured lifetime (c) of samples doped with the same concentration CeO₂ but melted at different MT.

Since the NIR emission is subject to phonon energy of glasses too, the distribution of phonon energies (lattice vibrations) of the studied glasses was examined by Raman scattering experiment, and the results are shown in Fig. 5. There are five major peaks located at ~140, 367, 721, 914, 1324 cm⁻¹, respectively. The strongest peak (130 cm⁻¹) attributes to the vibrational mode of Bi³⁺ in [BiO₆] octahedron. The 367 cm⁻¹ peak is caused by the distorted vibration of Bi-O-Bi and Si-O-Si in [BiO₆] and [SiO₄]²³. The 721 cm⁻¹ band is due to the distorted vibration of B-O-B in [BO₃] and [BO₄]²⁴. The vibration of Bi-O-Si accounts for the peak located at 914 cm⁻¹. The 1324 cm⁻¹ peak is induced by the asymmetric telescopic vibration of B-O⁻ in [BO₃] triangle and [BO₄] tetrahedron²⁴. Apparently, the network of BBS glasses is mixed with different glass forming units. With the substitution of CeO₂ for Bi₂O₃, the low phonon energy modes related to [BiO₆] decreases, while the maximum phonon energy of the glasses is invariable.

Figure 5

Raman spectra of samples doped with different concentration of CeO₂.

Discussion

In the present work, the effects of CeO₂ are two-fold. Firstly, it works as an effective oxidant to against the darkening/crystallization of the BBS glasses due to the precipitation of Bi metal or clusters during glass melting²². The above processes can be expressed by the following equations:

It is the release of oxygen gas (Eq. 2) at high MT that oxidizes the Bi metal or clusters (Eq. 3) and improves the transparency of the BBS glasses (Fig. 1)²⁵. According to Eq. 2, CeO₂ itself goes into Ce₂O₃ during the melting, the consumption of oxygen gas by the Bi metal or clusters favors the formation of Ce³⁺ ions, especially at high MT. The XPS measurement clearly backs up this assumption (Fig. 6). Although both Ce³⁺ and Ce⁴⁺ can co-exist in glasses, the XPS spectra show that trivalent Ce³⁺ dominates in the studied BBS glasses, as confirmed by the resemblance of the XPS spectra to that of the standard reference CeF₃ crystal and the absence of the characteristic doublet related to Ce⁴⁺ at ~916 eV^26,27. However, either the presence of Ce³⁺ or Ce⁴⁺ can increase the absorption in the visible region up to 500 nm (Fig. 2), which is induced by either the 4f-5d transitions of Ce³⁺ or the charge transfer between O²⁻ and Ce⁴⁺.

Figure 6: XPS of Ce 3d spin orbit doublet of samples doped with 5 mol.% of CeO₂ but melted at different MT.

Also included are the standard references CeO₂ and CeF₃ crystals. The XPS peaks were fitted to Gaussian function provided by Origin software.

Secondly, the role of codoping CeO₂ is to improve the NIR luminescent properties of Er³⁺. The variation of the NIR emission with the content of Ce³⁺ is complex; however it can be analyzed based on the phonon assisted energy transfer (PAET) between the relevant electronic energy levels of Er³⁺ and Ce³⁺ ions. The PAET may involve one, two or more than two phonons depending on the phonon energy of the host, concentration of and critical radii between rare earth ions (REs) in solids²⁸. Generally, one phonon assisted process accounts for 20~30%, while two and more than two phonons assisted processes only 10⁻⁶~10⁻³ of the total transition probability of REs. Referring to the energy level diagrams of Er³⁺ and Ce³⁺ (Fig. 7), it is seen that the following two PAET processes which may influence the NIR emission: (a) Er³⁺: ⁴I_11/2 + Ce³⁺ : ²F_5/2 → Er³⁺: ⁴I_13/2 + Ce³⁺: ²F_7/2, and (b) Er³⁺: ⁴I_13/2 + Ce³⁺: ²F_5/2 → Er³⁺: ⁴I_15/2 + Ce³⁺ :²F_7/2 have energy mismatches approximately 1400 cm⁻¹ and 4300 cm⁻¹, respectively. The maximum phonon energy of the glasses is 1350 cm⁻¹ (Fig. 5), which means that the former process (a) leading to the increased population of the NIR lasing level ⁴I_13/2 of Er³⁺ and enhanced NIR emission requires one phonon, while the latter (b) more than two phonons. Although the latter may quench the NIR emission, it hardly plays a role for low doping concentration, because the PAET involving more than two phonons generally has extremely low efficiency. In fact, the luminescence decay measurement (cf. Table S2) shows that the lifetime of the NIR luminescence varies by only 5% when the content of Ce³⁺ is low (<1 mol.%).

Figure 7: Energy diagram of Yb³⁺, Er³⁺ and Ce³⁺, and possible ET processes between these ions.

The solid arrows represent radiative transitions and the curly ones non-radiative transitions (multi-phonon relaxations).

However, at too high concentration of Ce³⁺ (e.g., >5 mol.%), one phonon assisted process may also occurs between one Er³⁺ ion and a pair of Ce³⁺ ions (via Er³⁺: ⁴I_13/2 + Ce³⁺: ²F_5/2 + Ce³⁺: ²F_5/2 → Er³⁺: ⁴I_15/2 + Ce³⁺: ²F_7/2 + Ce³⁺: ²F_7/2). The energy transfer efficiency (η) between Er³⁺ and Ce³⁺ can be calculated according to the equation: η = 1 − τ/τ₀, where τ and τ₀ are lifetimes of the donor (Er³⁺) with and without acceptor (Ce³⁺). A rough estimate of η shows that it equals to 30~60% for high doping concentration of Ce³⁺, suggesting that one phonon assisted process indeed occurs. A more rigorous calculation of the PAET efficiency requires knowledge of the absorption cross-section of the Ce^{3+ 2}F_7/2 → ²F_7/2 transition (~2200 cm⁻¹), which is impossible to obtain because the intrinsic phonon cut-off edge of the samples lies at shorter wavelength (~3000 cm⁻¹). Incidentally, the PAET has also been found a very successful means to improve the lasing efficiency of Yb³⁺/Ho³⁺ and Yb³⁺/Tm³⁺ pairs in solids²⁹. We have failed to measure the lifetimes of ⁴I_11/2 level of Er³⁺ before and after codoping of CeO₂ because the 1000 nm and 2.7 μm emissions from ⁴I_11/2 level were too weak to be recorded, thus we are not able to deduce the PAET efficiency between Er³⁺ and Ce³⁺.

One may have noticed that the lifetime of the 1.55 μm emission of Er³⁺ was rather short (~1 ms) in the studied glasses (Fig. 3(c)). This is because the glasses contain a large amount of B₂O₃ as high as 50 mol.%, giving rise to much faster multi-phonon relaxation between the lasing level ⁴I_13/2 and the ground level ⁴I_15/2 as compared to glasses containing a less amount of B₂O₃ (~3 ms)³⁰. In this respect, it seems preferable to use glasses containing a limited amount of B₂O₃, however, when referring to our previous work²² and other published papers³⁰, it appears that the broadband NIR emission with a FWHM over 80 nm can only be obtained in glasses containing relatively high content of B₂O₃⁸. Since the emission bandwidth of rare-earth (RE) ions is largely determined by the variations of ligand fields around RE ions from site to site (inhomogeneous broadening), and the ligand fields around RE ions are mainly influenced by glass formers, consequently, the broadband NIR emission observed in the BBS glasses can be partially accounted for by the so-called “mixed glass former effect” due to the presence of a multiplicity of glass former motifs (Fig. 5)³¹. However, a compromise indeed exists between the emission lifetime and the bandwidth regarding the optimized concentration of B₂O₃. Fortunately, because Bi₂O₃-B₂O₃-SiO₂ ternary system has a very broad glass forming region, providing a rich tuning range of glass compositions, the lasing and other relevant physical/ optical properties are amenable to being tailored for specific requirements. Work is now under way to further refine the lasing properties of the glasses and to understand the line broadening mechanism induced by B₂O₃.

Previous report showed that when the content of B₂O₃ increased to 8 mol.% in Er³⁺/Yb³⁺ codoped tungsten-tellurite glasses, the red ETU emission was already getting stronger than the green one³². This phenomenon is also owing to the presence of HPE compound B₂O₃, which tends to decrease the lifetime of ⁴I_11/2 level and as such increase the population of ⁴I_13/2 level, consequently, the ETU emission involving the intermediate ⁴I_11/2 level, viz., the green one, was reduced, whereas the red one involving the intermediate ⁴I_13/2 level enhanced. As mentioned above, the studied glasses contain a high concentration of B₂O₃, it is no wonder that the red ETU is stronger than that of the green one (Fig. 3(b)).

Judd Ofelt (JO) analysis is the most useful tool in estimating the forced electric dipole transitions of RE ions. JO parameters can be derived from the absorption spectra and the results are shown in Table 1. According to the theory, three JO intensity parameters (Ω_t, t = 2, 4, 6), which depend on the local chemical environments, determine the radiative transition rate of RE ions. Of the three parameters, Ω₂ is the most sensitive to local structure, and previous studies have suggested that it increases with the degree of asymmetry around RE ions^33,34. The other two parameters, Ω₄ and Ω₆ are not so much susceptible to the environmental changes as Ω₂. However, Ω₆ has been found to increase with the extent of the overlap between the 4f and 5d orbitals of Er³⁺ ions, or instead, it decreases as the 5d electron density diminishes. The substitution of CeO₂ for Bi₂O₃ reduces the number of highly polarizable Bi-O bonds (Fig. 5), and according to our previous study²², this may in turn decrease to some extent the microsymmetries around Er³⁺ ions, which partly accounts for the increased Ω₂.

Gain coefficient of the NIR emission, G(λ), was calculated according to , where P is the population inversion given by the ratio between the population of Er³⁺ ions at the lasing level (⁴I_13/2) and total concentration () of Er³⁺ ions, σ_a(λ) is the absorption cross-section, σ_e(λ) is the emission cross-section which can be calculated according to Fuchtbauer-Ladenburg equation:, where c is the speed of light, n is the refractive index and A_rad is the radiative transition rate obtained from JO analysis, Δλ_eff was calculated by . Positive gain already appears when only twenty percent of Er³⁺ ions are populated to the lasing level (insert in Fig. 8(a)). The maximum gain increases with the concentration of CeO₂. It should be mentioned that the gain may also depend on the ESA of Er³⁺ as in the case of Er³⁺-doped silica fibers^35,36, so in reality, the calculated G(λ)-values can suffer from overestimating. On the other hand, with the fixed doping concentration of CeO₂, the gain coefficient barely changes with MT (Fig. 8(b)).The sudden drop in the gain value is due to the crystallization of the samples melted at high MT.

Figure 8

Variation of gain coefficient of samples as a function of CeO₂ (a) and MT (b). Insets: typical gain spectra for demonstration.

Conclusion

Depending on the melting temperature, a large amount of CeO₂ (up to 20 mol.%) can be doped into the bismuth borosilicate (BBS) glasses, and cerium ions present as trivalent oxidation state Ce³⁺ overwhelmingly. The benefits of the doping of CeO₂ include, on one hand, improving significantly the transparency of the glasses especially for those melted at high temperatures; on the other hand, enhancing the 1.55 μm near-infrared (NIR) emission without sacrificing the lifetime due to the phonon-assisted energy transfer (PAET) between Er³⁺ and Ce³⁺. However, the doping concentration must be limited because otherwise both the intensity and lifetime of the NIR emission degrade. In addition, the undesirable energy transfer up-conversion (ETU) emission and excited state absorption of Er³⁺ are severely quenched by doping of CeO₂ stemmed from the accelerated non-radiative transition ⁴I_11/2 → ⁴I_13/2 of Er³⁺. The unique combination of enhanced NIR emission intensity, broadened NIR emission bandwidth and suppressed ETU emission makes Ce³⁺/Yb³⁺/Er³⁺ triply doped BBS glass a potential fiber material for broadband EDFAs. And because Bi₂O₃-B₂O₃-SiO₂ system has a very broad glass forming region, providing a rich compositional tuning range, the lasing and other related physical/optical properties can be tailored for specific requirements. The strategy described here, namely, de-excitation by PAET between rare-earth (RE) ions, increasing non-radiative transition by using high phonon energy compounds and line broadening through mixed glass former effect, can be readily extended to other RE pairs such as Yb³⁺-Tm³⁺ and Yb³⁺-Ho³⁺ to enhance the 2.0 μm emissions of Tm³⁺ and Ho³⁺ doped solids. More work has to be done on optimizing the concentration of B₂O₃ and exploiting the physical basis for the line broadening induced by B₂O₃.