1. Introduction

With the rapid growth of information demand, which will reach 100-1000 Pbit/s per fiber in 20 years, it’s a big challenge for the present optical fiber communication system to realize a super-big-capacity optical communication. One of the approaches that have been proposed is widening the spectral region for the information transmission [1], for example, effort has been done to broaden the presently widely used rare-earth ions doped fiber amplifiers’ amplification bandwidth. Although some valuable results have been obtained [2], the potential of this route is still limited due to the native luminescence nature of 4f-4f electronic transition of rare-earth ions, therefore leaving most of the optical fiber’s low optical loss spectra region of 1300-1700 nm being not used [1, 3].

In recent years, Bi-doped glasses have gained increasing attention due to its broadband and tunable near-infrared (NIR) emission in the range of 1000-1700 nm with the full width at half maximum (FWHM) even over 300 nm in some cases [1, 4], which make it a promising material for the application in broadband optical amplification. Besides, when co-doped with proper active ions, its luminescence could even cover the entire low optical loss spectra region [5, 6]. Although NIR luminescence with high efficiency have been reported successively in various types of bulk glass systems such as silicate [7–10], germanate [11–13], borate [14], phosphate [15], chalcogenide [16], and germanium silicate [17, 18], the amplification gain and laser performance in Bi-doped glass fiber leaves much to be desired [19–21]. One possible reason is that Bi related centers present rich and strong absorption bands in the whole visible and NIR wavebands, which leads to the serious self-absorption during resonance oscillation.

In stark contrast to the rare-earth and transition-metal ions with relatively simple ionic state in glass matrix, Bi has been identified to be present in various forms, including Bi³⁺, Bi²⁺, Bi⁺, Bi atom, Bi cluster and Bi nanocrystal. We recognized that the activation energy of the chemical transformations of these species is relatively low. So the thermal activation during fiber drawing process is anticipated to be sufficient enough to trigger the transformation of Bi species. From this point of view, the previous simple optimization of the optical performance of Bi-doped bulk glass is far from content and the study of thermal annealing dependence of optical properties is an inseparable part of the research. However, there is seldom investigation about the effect of heat-treatment on the NIR luminescence properties of Bi-doped glasses. In the present contribution, we carefully studied the thermal stability of NIR luminescence of Bi-doped borate glasses. The NIR luminescence intensity decreased sharply and almost disappeared when heat-treated at 690 °C for 2 h. We proposed a way of improving the thermal stability of borate glasses by regulating the structure rigidity of glass. The NIR luminescence degradation can be firmly suppressed with the addition of La³⁺ for enhancement of the structure rigidity. The results provide valuable guidance for suppressing performance degradation of Bi-doped glass during fiber drawing process.

2. Experimental

The glass samples with the compositions of 75B₂O₃-20BaO-5Al₂O₃-1Bi₂O₃ (0La) and 75B₂O₃-15BaO-5La₂O₃-5Al₂O₃-1Bi₂O₃ (5La) (in mol%) were prepared by the conventional melting-quenching technique. Analytical grade reagents H₃BO₃, BaCO₃, Al₂O₃, La₂O₃ and Bi₂O₃ were used as raw materials. 30 g batches were mixed homogeneously and then melted in a corundum crucible at 1550 °C for 20 min in air and then cast onto a stainless steel plate. The obtained glasses were cut into pieces. Then the glass samples were heat-treated at different temperatures between 400 and 700 °C for 2 hours and polished into slices of an appropriate size with a thickness of 1.5 mm for optical measurements.

The photoluminescence (PL) spectra were recorded using a ZOLIX SBP300 spectrophotometer with an InGaAs detector excited by 800 nm LD. The differential thermal analysis (DTA) was carried out by a CRY-Z Differential Thermal Analyzer at a heating rate of 10 °C/min. The photoluminescence excitation (PLE) spectra and luminescence decay curves were recorded by a FLS920 fluorescence spectrophotometer (Edinburgh Instrument Ltd, UK). The structure of the glasses was analyzed with a Raman spectrometer (Jobin Yvon Corp., France) using 514.5 nm radiation from an argon ion laser for excitation. Since the Bi centers have a strong absorption around 514 nm and small amounts of Bi exert little effect on the glass structure, the Raman spectra were performed on corresponding Bi free glasses. X-ray diffraction (XRD) analysis was carried out on a D/MAX-2550pc diffractometer with Cu Kα as the incident radiation source. Transmission electron microscopy (TEM) was performed on a FEG-TEM (Tecnai G2 F30 S-Twin, Philips-FEI, Netherlands). All the measurements were performed at ambient atmosphere.

3. Results and discussion

Thermal properties of the glass were measured by DTA, and the glass transition temperature (T_g) of samples 0La and 5La were estimated to be around 588 °C and 593 °C, respectively. A slight increment of T_g with the addition of La₂O₃ indicates the enhancement of the rigidity of glass structure. This results from the relative strong cationic field strength of La³⁺ ions (CFS_La3+ = 2.67 A⁻²) which is beneficial for stabilization of glass structure.

PL spectra were measured to study the effect of heat-treatment conditions on the luminescence properties of the two glass samples under excitation with an 800 nm LD, as is depicted in Figs. 1(a) and 1(b). It can be observed that both glass samples show broad emission with the central wavelength at around 1250 nm. Upon heat-treatment at 690 °C for 2 h, NIR luminescence of sample without La₂O₃ (0La) almost disappears, while the glass sample with La₂O₃ (5La) shows a different scenario which only decreases about half.

 

Fig. 1 Photoluminescence spectra of the glass samples 0La (a) and 5La (b) before and after heat-treatment when excited by an 800 nm LD, (c) dependence of the NIR luminescence intensity on the heat-treatment temperatures. The insets show the corresponding photographs of the glass samples heat-treated at various temperatures.

Download Full Size | PDF

We further systematically investigate the effect of heat-treatment conditions on the NIR emission of the glasses, both samples were heat-treated at different temperatures and their optical properties were investigated. From the insets of Fig. 1(c), it can be observed that the color of these glasses gradually become dark with the increasing heat-treatment temperature and the color change of sample 0La is more obvious. The NIR luminescence intensity as a function of heat-treatment temperature is shown in Fig. 1(c). Both samples experience emission loss at around T_g. The NIR emission intensity of glass 0La decreases nearly 88% when heat-treated at 690 °C for 2 h. In contrast, the luminescence degradation of glass with La₂O₃ is suppressed and the luminescence decrement of sample 5La is estimated to be around 46%. Thus, the thermal stability of luminescence can be improved by nearly 4.5 times with addition of 5 mol% La₂O₃. In a further study, we found that with La₂O₃ up to 10 mol%, the emission intensity after heat-treatment is comparable to that of the as-made sample.

In order to investigate the origin of NIR luminescence degradation during heat-treatment, PLE and PL spectra and the corresponding decay curves of the as-made and heat-treated samples were measured. As can be clearly seen from Fig. 2(a), except the intensity decrement, the excitation and emission spectra remain unchanged after heat-treatment. Besides, the corresponding decay curve of heat-treated sample follows the same decay mode with the as-made sample, indicating a similar behavior of Bi-related active center in the samples before and after heat-treatment, as is shown in Fig. 2(b). The same phenomena have also been observed in glass 5La, see Figs. 2(c) and 2(d). This firmly suggest that the NIR luminescence of the as-made and heat-treated glasses are all derived from the same Bi-related active center, which is most probably ascribed to low valent Bi ions [12, 4, 22–25]. These active Bi centers are supposed to easily capture the electrons released from the wrong bond during thermal activation, and then gradually aggregate into nonluminous Bi metallic colloids, resulting in the decrement of luminescence intensity and darking phenomenon of the glass samples [4, 26, 27].

 

Fig. 2 PLE (λem = 1150 nm) and PL (λex = 468 nm) spectra of as-made and heat-treated glass samples 0La (a) and 5La (c). Corresponding decay curves of glass samples 0La (b) and 5La (d) when excited at 468 nm and monitored at 1150 nm.

Download Full Size | PDF

To expound the possible transformation mechanism of Bi active center in the glass matrix and reveal the differences in thermal stability of NIR luminescence between glass 0La and 5La, Raman spectra were measured to analyze the structure of these two glasses. As shown in Fig. 3(a), the intense peak at 798 cm⁻¹ is assigned to the symmetric ring-breathing vibration of the boroxol ring [28]. Peak at ~663 cm⁻¹ is attributed to stretching of metaborate [BO₂]^- [29], and the band near 723 cm⁻¹ belongs to the bending vibrations of three coordinated boron units [28]. The origin of Raman band at ~1215 cm⁻¹ is ascribed to the stretching modes of [BO₃] triangular units [30], while the vibrational modes of B-O- terminal bonds of [BO₃] metaborate triangles contribute to the high-frequency Raman envelop (1350-1600 cm⁻¹) [31]. Another peak at ~900 cm⁻¹ can be assigned to the symmetric stretching vibration of the planar orthoborate units [32]. According to literature [29], peaks at around 477 cm⁻¹ and 767 cm⁻¹ can be ascribed to the symmetric breathing vibrations of six-membered rings with one BO₄ tetrahedron and ring type metaborate groups. From the inset of Fig. 3(a), peak at around 767 cm⁻¹ is more prominent in glass 5La than that in glass 0La, demonstrating a relative higher content of BO₄ tetrahedron groups in glass 5La. The successful introduction of rich BO₄ tetrahedron groups with the addition of La₂O₃ may lead to more rigid glass structure. This is highly beneficial for stabilization of Bi active centers and suppression of NIR emission degradation.

 

Fig. 3 (a) Raman spectra of as-made glass 0La and 5La. Inset is the amplified spectrum region from 755 to 790 cm⁻¹, (b) schematic illustration of transformation mechanism of Bi active centers in glass 0La and 5La, and (c) XRD pattern and TEM photograph of glass 0La.

Download Full Size | PDF

The transformation mechanism of Bi active centers and the improvement of thermal stability can also be understood based on the topological constraint theory [33]. As shown in the schematic diagram of Fig. 3(b), for glass without La₂O₃, when heat-treated at temperature above T_g, there is ample thermal energy to overcome the bond constraints, and the network becomes floppy. As a result, the Bi ions distributed inside the glass network can easily migrate to form Bi metallic colloid through the ‘microchannel’ formed during the crack of the floppy network at high temperature. And this formation process of Bi-colloid can be directly evidenced by XRD and TEM analysis of glass 0La before and after heat-treatment [34], see Fig. 3(c). In contrast, the glass with La₂O₃ has a relative more rigid network and becomes more difficult to get floppy. More specifically, due to the large cationic field strength and high coordination number of La³⁺ ions (CFS_La3+ = 2.67 A⁻²) compared with Ba²⁺ ions (CFS_Ba2+ = 1.11 A⁻²), the mobility of the floppy network at high temperature is restricted to a certain degree, and the ‘microchannel’ big enough to transit Bi ions is harder to form, which, in turn, imposes restrictions on the migration of Bi ions, including the Bi NIR luminescence active centers, which is shown in the lower part of Fig. 3(b). Consequently, the glass with La₂O₃ presents higher thermal stability of NIR luminescence than that of the glass without La₂O₃.

4. Conclusion

In summary, we systematically studied the heat-treatment temperature dependent NIR luminescence in Bi-doped borate glass. It was found that the NIR luminescence intensity generally decreases with the increase of heat-treatment temperature and experiences a dramatic decrement at around glass transition temperature. The possible mechanism is discussed, and can be attributed to the agglomeration of Bi active centers. We proposed a way of improving the thermal stability of NIR luminescence of borate glasses by regulating the structure rigidity of glass. We demonstrated a great improvement of the thermal stability of NIR luminescence by addition of La₂O₃. This is benefited from the large cationic field strength and high coordination number of La³⁺ ion which lead to the enhancement of the rigidity of glass structure. The results not only provide valuable guidance for suppressing performance degradation of Bi-doped glass during fiber drawing process, but also present an effective way to control the luminescence properties of main group elements in glasses from the perspective of glass structure.