1. Introduction

Tunable broadband luminescent materials have been of great interest owing to their promising applications in fiber-optic communication, imaging, and remote sensing [1–3]. With the ever-growing demand for high-capacity telecommunication, it is necessary to develop superior luminescent materials with broadband, efficient, and tunable NIR luminescence, for the next generation of communication systems [3,4]. Extensive studies of the luminescent materials have been concentrating on the glass systems doped with rare-earth (e.g., Pr³⁺, Nd³⁺, Tm³⁺, and Er³⁺) or transition-metal ions (e.g., Cr³⁺, Ni²⁺, Bi°). However, the emission bandwidth in rare-earth (RE) doped glasses is still not satisfying because the f-f transitions of RE ions lead to narrow absorption and emission bands [5]. Furthermore, the f-f transitions spectral bandwidth is relatively independent of glass composition and crystal field strength. Compared to the RE-doped glasses, transition-metal doped glasses exhibit broader NIR emission [6–8]. Despite great progress in developing a new class of gain media for optical amplifiers, there are still some critical issues that remain unsolved. For example, in Bi-doped glasses, the mechanism of the luminescence of active emission centers has not been really revealed, and this hinders tailoring light absorption and luminescence [9]. For Ni-doped glasses, the polycrystals, in which Ni²⁺ ions are octahedrally coordinated to their ligands, are required to realize NIR emission [10]. The scattering loss from the glass-crystal interface results in the degradation of spectral performance of fiber amplifiers [11].

As an alternative, Te-doped glasses with broadband NIR luminescence have been developed [12,13]. Unlike the above-mentioned single active ion doped glasses, the broadband luminescence of Te-doped glasses arises from Te clusters (e.g., Te₂^-, Te₄, and Te₄²⁺) [14–16]. The formation process of clusters involves both the reduction of cations to the atoms and the incorporation of multi-components into a composite entity in an ordered manner, and this process strongly depends on the surrounding ligand field [17]. Moreover, Te can exist as various types of species (from single ions to ions in cluster) and in different valence states (from positive to negative). The conversions can easily occur among different species or among different valence states [18–21]. Thus, Te NIR emission (originating from Te clusters) is rather sensitive to the local chemistry (e.g., redox states, defects and stoichiometric deviation) around Te. Both the configuration and valence states of Te clusters are controlled by the local chemistry, and the former factors affect the formation of Te clusters and the optical performances in glasses.

Several approaches have been proposed to adjust local glass chemistry, e.g., by introducing oxygen or hydrogen into glass melts, or by high-energy radiation) [22,23]. High-energy radiation could induce the formation of defects in glasses, leading to an increase in optical loss [24]. It is difficult to control the degree of reduction or oxidation of glass melts just by applying gases since the redox reaction is a spontaneous and rapid process.

In addition to the aforementioned reduction methods, oxygen-deficient structural defects (e.g., E’ centers, i.e., unpaired electrons at network forming cations bonded to three oxygens [25,26]) can affect the local glass chemistry, and thereby modify the physical properties of glasses [27]. This is a powerful tool for adjusting the local glass chemistry, and hence for tuning the optical performances of Te-doped glasses. Here the three key questions arise: how can the oxygen-deficient structural defects be created in Te-doped glasses? What determines the number of structural defects? Is structural defect engineering an effective approach to realize the tunable broadband luminescence in Te-doped glasses? In this study, we answered these questions by taking the potassium-alumino-germanate glass as an example. The introduction of Al₂O₃ into the [GeO₄] tetrahedral network leads to an unstable structural state with higher potential energy in the glass host, and thereby generating structural defects [28,29]. Also, the present study gives insight into the relationship between the structural defects and the optical performances of Te-doped germanate glass, as well as into the mechanism of Te NIR luminescence.

2. Experimental section

2.1 Sample preparation

Glass samples with molar composition of (91.9-x) GeO₂-8Al₂O₃-xK₂O-0.1TeO₂ (with x = 5, 10, 15, 20, 25, 30, 35, 40) were fabricated by melt-quenching method in air without any annealing. High pure GeO₂ and TeO₂ (99.99 mol%) and analytical pure reagents Al(OH)₃ and K₂CO₃ were used as raw materials to avoid glass-melt contamination. A batch of ∼30 g for each glass composition was weighed and uniformly mixed in an agate mortar and then melted in high purity corundum crucibles at the temperature range from 1500 to 1580 °C for 30 min. The amount of volatiles is considered to be fixed. The melt was subsequently cast onto a cold brass plate and immediately pressed with another plate to increase the cooling rate and promote glass formation. The obtained bulk glass was then cut, ground, and finely polished for subsequent optical measurement. The thickness of samples is almost the same.

2.2 Material characterization

Structural analyses of all samples were conducted by using a Renishaw InVia Raman spectrometer with a 532 nm laser source at an output power of ∼5 mW for excitation. Structural defect in germanate glasses was detected by electron paramagnetic resonance (EPR) spectra using an X-band spectrometer (ELEXSYS, Bruker A300) with a microwave power of 0.5 mW, and modulation amplitude of 3 G. The transmittance spectra of samples were measured using a Perkin Elmer Lambda-900 Ultraviolet-Visible-Near-infrared (UV-Vis-NIR) spectrophotometer in a spectral range from 200 to 3200 nm. Static excitation spectra and emission spectra were taken using an Edinburgh FLS920 spectrofluorometer equipped with a liquid nitrogen-cooled photomultiplier (Hamamatsu R5509-72). A 450 W xenon lamp was used as excitation sources. NIR luminescence spectra under 808 and 980 nm laser diode (LD) excitation were recorded through a Zolix Omni λ3007 spectrometer equipped with an InGaAs photodetector and an SR830 Stanford Research lock-in amplifier.

3. Results and discussion

Figure 1(a) shows the Raman spectra of the studied glasses, reflecting the evolution of microscopic structure with varying K₂O content. As K₂O content increases, the strongest band at 420 cm^-1 is gradually shadowed by the band at 515 cm^-1. The former band originates from the symmetric stretching vibration of Ge-O-Ge bonds in six-GeO₄-membered rings (rings compose of six GeO₄ tetrahedra), while the latter one arises from that in superstructural units such as four- or three-GeO₄-membered rings, respectively [30]. In parallel, the bands around 973 and 860 cm^-1, which are induced by the asymmetric stretching of the Ge-O-Ge bonds, shift toward lower frequency [30,31]. It is also seen that a band around 830 cm^-1 and a shoulder band around 770 cm^-1 develop with increasing K₂O content, which are attributed to the asymmetric stretching vibrations of Ge-O^- bonds in Q³ and Q² tetrahedra (where Qⁿ is a tetrahedron with n bridging oxygen), respectively [31]. This trend indicates that the glass network is depolymerized, and a decrease in the size of topological cage (i.e., dissociation of six-GeO₄-membered rings into smaller rings). Above 30 mol% K₂O, Ge-O^- bonds in Q² tetrahedra lowers network connectivity. In addition, the decrease of network connectivity with K₂O content is unfavorable for the formation of structural defects, i.e., positively charged oxygen vacancies, since Ge-O^- bonds are charge-balanced and stabilized by K⁺ cations. In contrast, if there is no K₂O present in the structural network, the Al₂O₃ will act as modifiers and thereby non-bridging Ge-O^- bonds will form, which are charge-balanced by Al³⁺. Yet, such network is energetically unstable, and this is confirmed by the fact that the glass consisting of network-forming oxide and alumina has extremely low glass-forming ability [29]. Thus, oxygen vacancies may easily occur in the unstable network. The presence of these structural defects (i.e., singly ionized oxygen vacancies) is confirmed by an intense EPR peak at g = 1.91 in the K₂O free sample (Fig. 1(b)) [32]. It is also seen that the EPR peak gradually diminishes with increasing K₂O content to 40 mol%, indicating a decrease and even disappearance of oxygen vacancies, and therefore a less concentration of singly ionized oxygen vacancies (..). Owing to the fact that NIR emission of Te clusters is very sensitive to variations in local chemistry around Te, the topology and structural defect engineering is a powerful approach to control Te cluster configurations and chemical states, and thereby to tune their luminescence behavior in the glass system.

 

Fig. 1. Structural analyses of the germanate glasses with various K₂O contents. (a) Raman spectra; (b) EPR spectra.

Download Full Size | PDF

The inset of Fig. 2(a) shows photographs of the samples with various K₂O contents. As K₂O content increases from 5 to 20 mol%, glass color changes from reddish-brown to dark-green. When further K₂O is added, the dark-green becomes light-green, finally colorless (see the 35 mol% K₂O containing sample). Such a trend is shown by the transmission spectra in Fig. 2(a) for the samples with K₂O = 5, 10, 15, 20, 25, 30, 35, 40 mol%. As the K₂O content increases from 0 to 20 mol%, the strongest absorption peak near 960 nm, originating from nonluminous Te₄ (NL-Te₄) clusters [33], gets weaker, while the absorption band near 640 nm appears, which is attributed D_2h-symmetric Te₄ clusters (Fig. 2(b)) [34,35]. This indicates that the NL-Te₄ clusters transform into the D_2h-symmetric Te₄ clusters with increasing K₂O content. As K₂O content increases to 30 mol%, a new absorption peak near 500 nm emerges, which is accompanied by a decrease of the absorption band near 640 nm, indicating the presence of a new luminescence center. With increasing K₂O content from 0 to 30 mol%, the absorption bands at 500, 640, and 960 nm slightly shift towards longer wavelengths (Fig. 2(c)). When K₂O increases to 35 mol%, no peak can be observed.

 

Fig. 2. Control of the configurations and chemical state of Te clusters by varying K₂O content. (a) Transmission spectra, Inset: Photographs of glass samples; b) Varying absorption intensity; (c) Varying position of the absorption peak; (d) Excitation spectra monitored at 980 nm of Te-doped glasses with various K₂O content compared with 20 mol% K₂O-containing glass without Te doping.

Download Full Size | PDF

As shown in Fig. 2(d), no excitation peak can be observed in the 20 mol% K₂O-containing glass without Te doping, suggesting that these absorption bands are attributed to Te centers. With an increase of K₂O content from 0 to 20 mol%, the excitation band at 330 nm, which has been attributed to C₁-symmetric Te₅, becomes evidently weaker, whereas the excitation bands at 435 and 645 nm, which are assigned to D_2h-symmetric Te₄ [34], are enhanced. This suggests that C₁-symmetric Te₅ is transformed into D_2h-symmetric Te₄. Further increasing K₂O content to 30 mol% leads to weakening of excitation bands at 435 and 645 nm, and the appearance of a new excitation band at 515 nm, indicating that D_2h-symmetric Te₄ is converted into a new luminous center. When K₂O content rises to 35 mol%, all excitation bands weaken and even disappear at 40 mol%.

The strong dependence of the emission intensity on K₂O content is consistent with that of excitation band intensity (Fig. 3). The absorption edge of glass matrix without TeO₂ is far below that of the Te-doped glass [34]. The absorption edge is beyond 330 and 435 nm excitation band because short wavelength light (below the absorption edge) is absorbed by the Te clusters. Thus, these excitation lights can penetrate the glass matrix and excite Te cluster for emission. Upon 330 nm excitation, the emission peak appears at 855 nm, with an FWHM of 410 nm, in free K₂O sample (Fig. 3(a)). With the increase of K₂O content, the emission intensity decreases and its FWHM becomes narrower, down to around 384 nm. Upon 435 nm excitation, the emission at 925 nm increases rapidly with K₂O content from 15 to 20 mol%, and then decreases with further increasing K₂O (i.e., reaching the maximum in sample with 20 mol% K₂O) (Figs. 3(b) and (e)). As the K₂O content increases from 15 to 35 mol%, the emission peak shifts from 925 to 968 nm and its FWHM broadens from 288 to 322 nm. Upon 645 nm excitation, the variation of luminescence behavior with K₂O content is similar to those of the luminescence excited by 435 nm. Specifically, the strongest emission appears in the 20 mol% K₂O containing sample, and the peak position shifts from 920 to 968 nm and its FWHM broadens from 288 to 340 nm with increasing K₂O content from 15 to 35 mol% (Figs. 3(c) and (e)). Upon 515 nm excitation, as the K₂O content increases from 15 to 35 mol%, the emission peak shows a large redshift from 920 to 1026 nm and its FWHM significantly broadens from 288 to 396 nm, while the intensity of luminescence reaches the maximum when the K₂O content is increased to 30 mol% (Figs. 3(d) and (e)). Both the redshift and the increased FWHM might result from the emergence of new active Te centers. It is because the emission excited by 515 nm with large FWHM is located longer wavelength. When the excitation center appears, the emission, which is excited by the near wavelength that overlaps with this new excitation band, will shift to longer wavelength and become wider.

 

Fig. 3. Tuning of the luminescent behavior of germanate glasses by varying K₂O content. (a) Emission spectra excited by the 330 nm light; (b) 435 nm; (c) 515 nm; (d) 645 nm; (e) K₂O content dependence of the emission peak position, intensity, and full width at half maximum (FWHM); (f) Emission spectra obtained by the 808 nm LD excitation; (g) K₂O content dependences of both the peak position and the FWHM for the emission excited by 808 nm LD; (h) Emission spectra obtained by the 980 nm LD excitation.

Download Full Size | PDF

For practical application in fiber laser, Fig. 3(f) illustrates the NIR luminescence spectra of glass samples with different K₂O contents upon 808 nm LD excitation. A super-broadband luminescence at 1160 nm with an FWHM of ∼310 nm can be observed in samples with K₂O content of 0∼10 mol%. As K₂O content increases from 0 to 10 mol%, the C₁-symmetric Te₅ fraction decrease, and therefore the emission intensity drops (Figs. 3(f) and (g)). This indicates that the luminescence located at 1160 nm is related to active C₁-symmetric Te₅. With increasing K₂O content from 10 to 20 mol%, the emission intensity is greatly enhanced, because of the conversion of C₁-symmetric Te₅ into D_2h-symmetric Te₄. In parallel, the emission peak shifts from 1160 to 985 nm. Such a blue-shift indicates an increase in the contribution of the D_2h-symmetric Te₄ clusters to the emission at 980 nm. Accompanied by the blue-shift of the emission, the FWHM becomes narrowed from 310 to 258 nm, since the contribution of C₁-symmetric Te₅ becomes less. From 20 to 30 mol% K₂O, the emission intensity increases, whereas from 30 to 35 mol% the luminescence decreases and finally vanishes at 40 mol%. When increasing K₂O from 20 to 35 mol%, the emission peak shifts from 985 to 1073 nm and its FWHM broadens from 258 to 363 nm. These phenomena are attributed to the appearance of a new cluster center (discussed below).

Upon 980 nm LD excitation, the ultra-broadband emission at 1215 nm with an FWHM of ∼250 nm can be observed in glass samples with K₂O=15 to 35 mol% (Fig. 3(h)). From 15 to 20 mol% K₂O, the emission intensity is enhanced by four times. A further increase of K₂O content, however, leads to a decrease in the emission intensity until NIR luminescence disappears for 40 mol% K₂O.

Based on the above results, we propose a scheme to describe the possible mechanism for the stabilization and conversion of active Te cluster centers by engineering structural defects (i.e., singly ionized oxygen vacancies $\textrm{V}_\textrm{O}^ \bullet $) and network topology (Fig. 4). TeO₂ as the raw material offers the source to form Te clusters (Fig. 4(a)). In the aluminogermanate glass, 4 oxygen vacancies ($\textrm{V}_\textrm{O}^ \bullet $) give electrons to Te⁴⁺ to yield one elemental tellurium (i.e., Te⁴⁺ + 4$\textrm{V}_\textrm{O}^ \bullet $= Te°). The six-GeO₄-member rings acting as topological cages can accommodate Te atoms to form Te₅ emission centers and NL-Te₄ (Fig. 4(b)). When K₂O content is increased to 20 mol%, the decrease in $\textrm{V}_\textrm{O}^ \bullet $ content weakens the reduction effect. On the other hand, when K₂O content is increased, the absorption edge shifts to the lower wavelength signify a stabilization of oxidation states (Fig. 2(a)). The stabilization of oxidation states is favored by increasing optical basicity leading to higher valence states of multivalent metal ions [36,37], and thus less multivalent Te⁴⁺ towards Te atoms and aggregate into clusters. Moreover, the six-GeO₄-membered rings transform into smaller rings, leading to smaller topological cages, and hence promoting the conversion of larger tellurium clusters (C₁-symmetric Te₅) into smaller D_2h-symmetric Te₄ clusters (Fig. 4(c)). This conversion enhances Te NIR emission not only because of the formation of the D_2h-symmetric Te₄ clusters, but also due to the isolation of the Te clusters from each other. The latter effect weakens the interaction between those clusters, and hence lowers the non-radition transitions (Figs. 3(b), (c), and f). With further increasing K₂O content, $\textrm{V}_\textrm{O}^ \bullet $ rapidly decreases and even disappears, thereby suppressing the reduction effect, and hence, leading to the conversion of D_2h-symmetric Te₄ into higher valence state (Fig. 4(d)). This may result in the enhancement of the NIR emission excited by 515 nm (Fig. 3(d)). Therefore, the emission may originate from high valence D_2h-symmetric Te₄⁺ or Te₄³⁺, which correspond to the excitation bands at 508 and 518 nm (according to TDDFT calculation), respectively [15]. This agrees with the excitation at 515 nm in our experiment. Above 35 mol% K₂O, the network connectivity is drastically lowered, and hence, the luminous Te clusters have a low probability to be confined in the topological cages, leading to the disappearance of Te NIR emission (Figs. 2(d), 3(f) and 3 h). The presence of the new active emission center induced by $\textrm{V}_\textrm{O}^ \bullet $ allows both for extending the tunable emission range and for improving luminescence efficiency.

 

Fig. 4. Structural defect engineering of the configurations and chemical states of Te clusters. Schematic for the stabilization and conversion of active Te cluster centers with the evolution of structural defect and topology network.

Download Full Size | PDF

4. Conclusions

This work demonstrated the unique correlation between the configuration and chemical state of dopant Te and oxygen-deficient structural defects (i.e., singly ionized oxygen vacancies) in germanate glass. The results made it possible to manipulate local chemistry around Te through engineering structural defects of glass by varying network modifier K₂O content. By controlling the local glass chemistry of the glass, the configuration and chemical state of Te centers are finely tuned over a wide range of valence states, from single ions and aggregated clusters to positively charged clusters. The defect engineering could not only improve the optical performances of luminescent materials with efficient emission, but also give new functionalities (e.g., extension in optical amplification bandwidth and improvement of fiber laser wavelength-tunability). Furthermore, the new approach could be applicable to other multivalent dopants in disordered systems in order to tailor a variety of material properties such as optical, magnetic and electrical characteristics.