1. Introduction

Broadband infrared emission of rare-earth dopants is of primary importance in several high tech areas such as optical amplifiers and laser sources operating in the wavelength range of the extended telecommunication window of 1.2 to 1.7 µm [1–10]. The most successful dopant has been the Er³⁺ due to its efficient radiative transition ⁴I_13/2→⁴I_15/2, Fig. 1(a), with long-living lasing level ⁴I_13/2, which occurs at about 1.53 to 1.56 µm and coincides with the wavelength minimum of optical loss in currently used telecommunication optical fibres (C-band) [1–4 and refs therein]. The Tm³⁺ dopant has also been considered for optical amplification in telecommunication fibres at wavelengths from about 1.45 to 1.48 µm (S-band) corresponding to the ³H₄→³F₄ transition of the Tm³⁺, Fig. 1(a), however the terminating level ³F₄ has very long life-time preventing efficient population inversion for the ³H₄→³F₄ transition [3,4 and refs therein].

Since the actually available telecommunication window expands now from 1.2 to 1.7 µm [1], a search is carried out for broadening the amplification band of Er³⁺, [4–8 and refs therein], increasing amplification gain of Tm³⁺ [3,4], or using infrared emission bands of other rare-earth dopants, e.g. of Ho³⁺ [9] or Dy³⁺ [10 and refs therein]. An 1.55 µm emission of Er³⁺ broadens when doped in low phonon energy hosts, such as fluoride [5–8] and tellurite [6] glasses and oxyfluoride nano-glass-ceramics [5,7,8] but may suffer from a prolonged lifetime of the feeding level ⁴I_11/2 of the Er³⁺ (Fig. 1(a)). Perhaps the most suitable option to decrease the lifetime of the feeding level ⁴I_11/2 of the Er³⁺ is the co-doping with another rare-earth dopant, i.e. with Ce³⁺, which accepts the energy from the ⁴I_11/2 of the Er³⁺ and thereby quickly depopulates the ⁴I_11/2 [5].

Yb³⁺ is known as useful co-dopant both for the Er³⁺ and Tm³⁺-dopants due to the unique high absorption cross-section transition ²F_5/2←²F_7/2 of the Yb³⁺ (Fig. 1) coincident with a traditional pump wavelength of telecommunication systems at 980 nm [2–4]. The excited Yb³⁺ ion efficiently transfers the energy both to the Er³⁺ and Tm³⁺ co-dopants [11,12 and refs therein], which consequently emit.

In this work we show that Yb³⁺-Er³⁺-Tm³⁺ co-doped oxyfluoride glass-ceramics 32(SiO₂)9(AlO_1.5)31.5(CdF₂)18.5(PbF₂)5.5(ZnF₂): 1.5(YbF₃)1.0(ErF₃)1.0(TmF₃), mol%, provides overlapping emission bands of the Er³⁺ (⁴I_13/2→⁴I_15/2 transition) and the Tm³⁺ (³F₄→³H₆ transition) resulting in an entire infrared emission band covering range between about 1.5 to 1.7 µm with a common lifetime of about 9 ms. Contrary and fortunately, the lifetimes of all higher levels of the Er³⁺ and Tm³⁺ have been found to be shorter than 70 µs due to efficient energy transfer to the ⁴I_13/2 of the Er³⁺ and ³F₄ of the Tm³⁺, which may result in high population inversion of these known lasing levels ⁴I_13/2 of the Er³⁺ and ³F₄ of the Tm³⁺. This has a potential for optical amplification/lasing of this nano-glass-ceramics at the wavelengths range from about 1.5 to 1.7 µm, which covers a substantial part of the available extended telecommunication window of 1.2 to 1.7 µm [1]. The mechanism for the efficient energy transfer between excited levels of the Yb³⁺, Er³⁺, Tm³⁺ co-dopants has been discussed by comparing the room and low temperature emission spectra and lifetimes of the dopants in the nano-glass-ceramics and its precursor glass.

 

Fig. 1. a) Energy level diagram of the Tm³⁺, Yb³⁺ and Er³⁺. The excitation and emission transitions of interest are indicated by arrows, respectively. b) Absorbance spectra of the GC (thick line) and precursor glass (thin line) at room temperature; some absorption transitions are postsigned.

Download Full Size | PDF

2. Experimental

The precursor oxyfluoride glass 32(SiO₂)9(AlO_1.5)31.5(CdF₂)18.5(PbF₂)5.5(ZnF₂): 1.5(YbF₃)1.0(ErF₃)1.0(TmF₃) mol% has been prepared by melting the batch consisting of the respective oxides and fluorides at about 1000°C for 1 hour [7]. The melt has been quenched in air atmosphere into the brass mould. The corresponding nano-glass-ceramics (GC) has been obtained on heat-treatment of the precursor glass at about 440°C for 30 minutes. The alternative procedure for preparation of GC consists in irradiation of the precursor glass by tightly focused laser beam (direct laser writing) resulting in local heating of the glass and precipitation of the GC in the irradiated areas [13].

On the heat-treatment of this precursor glass, the rare-earth dopants nucleate the size-restricted growth of nano-crystals of nearly cubic β-PbF₂ [5,7,9,10] to the fixed diameter of about 8 nm, as measured by means of transmission electron microscopy (TEM) [7], x-ray diffraction (XRD) [7,9,10]. Since the dopants nucleate the growth of the nano-crystalline phase [7], these nano-crystals are very heavily doped with rare-earth atoms and may thereby form stoichiometric compound of tveitite type, e.g. Pb_1-xLn_xF_2+x [5,10]; where Ln stands for a rare-earth ion, e.g. Yb³⁺, Er³⁺, Tm³⁺, and x≈0.3 [10], which has a structure similar to the face centered cubic β-PbF₂ [14,15]. Thus, we assume that the actual chemical composition of the nano-crystals may approach the Pb₃(Yb-Er-Tm)₁F₉, where the ratio of Yb:Er:Tm probably nearly follows the ratio 1.5:1:1 corresponding to their ratio in the batch. The super-ionic phase transition observed in the β-PbF₂ and tveitite-type crystals near to 440°C [14] does obviously help in diffusion of F- ions into the Pb_1-xLn_xF_2+x nano-crystals for required charge compensation since the valence of Ln³⁺ substitute is higher than of Pb²⁺. It suggests that the heat-treatment of the precursor glass should be carried out at about 440°C, which actually corresponds to the nano-crystallization peak in the differential thermal analysis (DTA) curve of the precursor glass [7,10].

3. Results and discussion

Figure 1(b) shows room temperature absorbance spectra of the GC and its precursor glass; some absorption bands corresponding to the Yb³⁺, Er³⁺and Tm³⁺ dopants have been post-signed. It is seen that the broad absorption band, ²F_5/2←²F_7/2, of a characteristic shape of the Yb³⁺ at about 0.98 µm dominates the absorption spectrum pointing out its use for optical pump. This band masks a substantially weaker and narrower absorption band ⁴I_11/2←⁴I_15/2 of the Er³⁺, which is also present at about 0.98 µm [2–8]. Other characteristic bands of the Er³⁺ and Tm³⁺, post-signed in Fig. 1(b), are also present in the absorption spectrum indicative that all dopants have been quite well dissolved both in the precursor glass and corresponding GC. Also, a characteristic change in the shape of the absorption bands with nano-ceramming of the precursor glass can be seen in Fig. 1(b), e.g. for the band ⁴I_13/2←⁴I_15/2 of the Er³⁺ [7] and for the ³F₄←³H₆ band of the Tm³⁺ [16] indicative that in the GC the dopants partition in the nano-crystalline phase of β-PbF₂ type [5,7–10,16] making the yttrofluorite type nano-crystals [10,14,15]. The percentage of rare-earth dopant in the nano-crystalline phase in this GC has been estimated to be as high as about 90% [8,9].

 

Fig. 2. a) room temperature emission spectra of the Yb³⁺-Er³⁺-Tm³⁺ co-doped GC (thick solid line) and precursor glass (thin solid line) excited at 900 nm. b) emission spectra of the Yb³⁺-Er³⁺-Tm³⁺ co-doped GC at 77 K excited at 900 nm (into Yb³⁺ absorption band, dot line), 522 nm (into Er³⁺ absorption band, thick line) and 463 nm (into Tm³⁺ absorption band, thin solid line).

Download Full Size | PDF

Figure 2(a) shows steady state room temperature emission spectra of GC and precursor glass when excited at 900 nm, i.e. at the high energy onset of the absorption band of the Yb³⁺ (neither Er³⁺ nor Tm³⁺ absorb at this wavelength, e.g. [7,16]). We see in Fig. 2(a) that in addition to the emission band of the Yb³⁺ (1.0 µm) and weak emission band of the Tm³⁺ (observed only in GC at 1.2 µm), the spectrum is dominated by the broad emission band covering in particular the range between 1.5 to 1.7 µm and consisting of the overlapping emission bands of the Er³⁺ (1.55 µm transition ⁴I_13/2→⁴I_15/2), and Tm³⁺ (the 1.64 µm high-energy shoulder of the transition ³F₄→³H₆ [16]). The shape of the bands of the Er³⁺ and Tm³⁺ co-dopants is similar to the shape of the respectively single Er³⁺ [7,8] and Tm³⁺ [16] doped glasses and GC’s of the same host chemical composition. Of special importance/interest here is a high energy shoulder of the Tm³⁺ band at 1.64 µm, which increases drastically in GC compared to the precursor glass, as it has been noted also for the single Tm³⁺-doped GC [16]. The weak shoulder at 1.60 µm belongs to the emission band of the Er³⁺ [8]. The response function of the spectrometer has been taken into account in this work, while we used for detection of emission a liquid nitrogen cooled photomultiplier and InSb photodetector. A stronger emission of the Tm³⁺ at 1.64 µm than of the Er³⁺ at 1.55 µm may be practically useful due to higher optical loss of the telecommunication fibres at 1.64 µm compared to 1.55 µm [1]. We found the ratio of these bands of the Tm³⁺ and Er³⁺ can be adjusted according to the purpose by changing the ratio of Tm³⁺ to Er³⁺ in the batch. Actually, current attempts to broaden the amplification band of existing telecommunications focuses on use of the L-band of telecommunication fibres, especially at about 1.58 to 1.65 µm [2–4], this range is covered by emissions of the Er³⁺ and Tm³⁺ in GC shown Fig. 2 (a).

The excitation spectra for the ⁴I_13/2→⁴I_15/2 (measured at 1.505 µm) and the ³F₄→³H₆ (measured at 1.640 µm) emission bands of the Er³⁺ and Tm³⁺, respectively, are shown in Fig. 3 for the GC (a) and for its precursor glass (b). A very important result is seen in Fig. 3(a) that the excitation spectra are the same for these emission bands of the Er³⁺ and Tm³⁺ in GC. Moreover, the lifetimes of these two bands of the Er³⁺ and Tm³⁺, 8.8 and 9.6 ms, respectively, are also similar in GC (see Table). Therefore the entire band in GC between 1.5 to 1.7 µm behaves as a band of single dopant pointing out that it could be used as a single emission band for optical amplification/lasing in the extended telecommunication window.

 

Fig. 3. Excitation spectra for the emission bands of the Er³⁺, ⁴I_13/2→ ⁴I_15/2 transition at 1505 nm (thick line) and of the Tm³⁺, ³F₄→ ³H₆ transition at 1640 nm (thin line) in GC (a) and precursor glass (b).

Download Full Size | PDF

Remarkably, such a good emission performance of the Er³⁺-Tm³⁺ couple, Fig. 2(a) can be achieved only in the case of GC, while in the precursor glass the lifetimes (1.1 ms for the Er³⁺ and 6.1 ms for the Tm³⁺, see Table) and excitation spectra, Fig. 3(b), of the Er³⁺ and Tm³⁺ differ substantially. The difference in excitation spectra, Fig. 3(b), is seen in particular at about 463 and 680 nm, where the absorption spectrum of the precursor glass is dominated by absorption bands of the Tm³⁺, the ¹G₄←³H₆ and ³F₃←³H₆ transitions, respectively, Fig. 1. The difference in excitation spectra and substantially longer lifetime of the Tm³⁺ in the glass point out that energy transfer between co-dopants is suppressed in the precursor glass, while it is reinforced in the GC, as seen from Fig. 3(a), where the excitation spectra for Er³⁺ and Tm³⁺ are identical. Therefore, we conclude that a substantial shortening of distance between co-dopants takes place in the GC compared to the precursor glass, in agreement with [5,7,17]. While in GC, the most of rare-earth dopants are trapped in nano-crystals, in the precursor glass, the dopants distribute homogeneously across the entire glass network.

An interesting implication can be drawn from the low temperature (77 K) emission spectra of GC presented in Fig. 2(b). Here we see that the energy transfer between Yb³⁺, Er³⁺ and Tm³⁺ co-dopants has been vanished even in case of GC, because individual pump of each of the dopants (in 980 nm absorption band for the Yb³⁺; 522 nm band for the Er³⁺; and 463 nm band for the Tm³⁺) results in emission mostly from the same co-dopant, which was pumped. The lifetimes for the Er³⁺ (6.8 ms) and Tm³⁺ (12.0 ms) also differ substantially at 77 K in case of GC (see Table). Therefore, we conclude that the energy transfer between Yb³⁺, Er³⁺ and Tm³⁺ co-dopants is quite suppressed at 77 K even in GC, and this indicates the importance of vibronic contribution to the energy transfer at the room temperature. We suggest that the phonon, which mediates the energy transfer between the co-dopants is a 140 cm^-1 “ungerade” (asymmetric) vibration a_u [18] typical of PbF₂ crystal ([19] and refs therein). While this phonon vibration is well populated at room temperature (300 K=210 cm^-1), it is almost completely frozen at 77 K (77 K=54 cm^-1). A vibronic contribution of this “ungerade” phonon mode to the intensity of optical transition of rare-earth dopants in this GC host has been noted earlier, e.g. in [9] for the case of the Ho³⁺. Other typical vibrations of PbF₂ are higher than 210 cm^-1 (i.e. a dominating symmetric vibration t_2g at 240 cm^-1) and could not substantially contribute to the energy transfer at room temperature. Also, the “ungerade” asymmetric vibration induces more efficiently a dipole-dipole interaction of Van der Waals type between the co-dopants, which allows an energy transfer between co-dopants, e.g. [12] and refs therein.

Table. 1. Lifetime of the energy levels of the Er³⁺ and Tm³⁺ at room temperature and at 77 K, where indicated, in 3.5 mol% (Yb³⁺-Er³⁺-Tm³⁺) co-doped GC and precursor glass, and in single, 3.5 mol% Er³⁺ and Tm³⁺-doped GC and precursor glass.

View Table

The lifetimes of the ⁴I_13/2→⁴I_15/2 (1.55 µm) transition of the Er³⁺ and of the ³F₄→³H₆ (1.64 µm) transition of the Tm³⁺ in Yb³⁺-Er³⁺-Tm³⁺ co-doped and in Er³⁺ and Tm³⁺ single doped samples with similar doping level are compared in the table. A simple correlation between lifetimes of the Er³⁺ and Tm³⁺ in co-doped and single doped samples is absent indicating that it is the energy transfer between Yb³⁺, Er³⁺ and Tm³⁺ co-dopants, which defines lifetimes of the co-dopants in co-doped samples.

The Er³⁺ and Tm³⁺ co-dopants seem to play an important role in mutual depopulation of their higher energy levels. This has been confirmed by measurement of lifetimes of majority of the upper levels of the Er³⁺ and Tm³⁺ using time-resolved up-conversion luminescence spectroscopy. The lifetime of all levels was shorter that 70 µs. Apparently, the lifetimes of the feeding levels of the Er³⁺ (⁴I_11/2→⁴I_15/2, 0.98 µm transition) and Tm³⁺ (³H₅→³H₆, 1.2 µm transition) are also shorter than 70 µs, as seen from the Table. This has a positive impact on the population inversion of the lasing levels of the Er³⁺ (⁴I_13/2) and Tm³⁺ (³F₄).

4. Conclusion

Yb-Er-Tm co-doped GC has a potential for emission/light amplification devices in the tunable wavelength range from 1.5 to 1.7 µm. 9 ms lifetime of the lasing transitions of the Er³⁺ (⁴I_13/2→⁴I_15/2 transition) and the Tm³⁺ (³F₄→³H₆ transition), which overlap and cover the 1.5 to 1.7 µm range, and very short lifetime of their upper feeding levels, indicate that population inversion for those lasing levels may be achieved in GC. Short-length chip amplifier/lasers due to high doping levels of pump co-dopant Yb³⁺ and emitting co-dopants Er³⁺ and Tm³⁺, and high pump cross-section for the Yb³⁺, may be possible.