1. Introduction

Photoluminescence (PL) from Eu³⁺-activated glasses and glass ceramics is characterized by various sharp PL bands which result from the intra-configurational 4f ↔ 4f parity-forbidden electronic transitions from ⁵D₀ to ⁷F_J (J = 0, 1, 2, 3, 4) [1–6]. The local environment around Eu³⁺ does not affect ⁵D₀ →⁷F₁ PL, but the probability of ⁵D₀ →⁷F₂ relaxation is strongly affected by the ligand symmetry. As a consequence, Eu³⁺ dopants can also be used as sensitive probes of the structural symmetry in the vicinity of Eu³⁺ [1,2]. Technical applications rely on the broad absorption band of Eu³⁺ in the spectral region of 350–420 nm, associated with a relatively high conversion efficiency to the red spectral range, which have made Eu³⁺ (and its divalent counterpart) one of the most prominent optical activators in phosphors, e.g., for light conversion [7–14]. In this context, glasses as host materials offer the advantages of very high compositional flexibility, simple and low-cost manufacture and versatile forming technologies, superior thermal stability and an epoxy-free assembly process [3,4, 15–20]. This, however, usually goes at the expense of absorption cross section and quantum efficiency of luminescence, what results in potentially low quantum yield. Controlled crystallization of the supercooled liquid to generate a glass ceramic material where one or more crystalline phases are deposited in a matrix of residual glass may be employed to overcome this problem: the presence of crystallite species may, on the one side, provide interfaces for multiple scattering of the incoming light which improves luminescence yield. On the other side, the crystallization procedure and the precipitated crystalline phase can be chosen in such a way that during crystallization, the active ion at least partially partitions into the crystalline matrix to attain the spectral properties of a polycrystalline phosphor. The type of the precipitated crystallite species is typically adjusted by choosing the appropriate chemical composition of the precursor glass. The partitioning of Eu³⁺ ions is determined primarily through the relation of ionic radii, valence states and coordination numbers of available lattice sites. An increase in internal quantum efficiency (η_QE) can be obtained when the lifetime of the excited state is increased and the probability of non-radiative energy transfer is decreased [11,12]. However, such glass ceramics which are suitable as a host material for Eu³⁺ can be obtained only from a rather limited number of chemical systems.

Here, we report on Eu³⁺-activated Na₃YSi₃O₉. Polycrystalline materials of this type have previously been demonstrated as efficient red-emitting phosphors by various authors [21–23]. We now discuss an Eu³⁺-activated Na,Y - silicate glass and the corresponding glass ceramic, focusing on the changes in spectroscopic properties which are associated with the transition from glass to glass ceramic. We then consider this material as a promising alternative for photoluminescent light conversion to the deep red spectral range.

2. Experimental

Precursor glass samples with nominal compositions (in mol%) 30Na₂O-8Y₂O₃-53SiO₂-5Al₂O₃-2TiO₂-2P₂O₅-0.5Eu₂O₃ (NYSA) were prepared by conventional melting of 50 g batches of analytical grade reagents Na₂CO₃, Y₂O₃, SiO₂, TiO₂, NH₄H₂PO₄ and Eu₂O₃, where NH₄H₂PO₄ acts as an oxidizing agent to prevent the formation of divalent europium. The thoroughly mixed raw material mixtures were melted at 1600 °C in ambient air for 2 h, using alumina crucibles. Glass slabs were obtained after pouring the melt into a graphite mould which was preheated to 500 °C. The molded glass was subsequently annealed at 600 °C for 2 h to relax residual thermal stress. From the slabs, disks of 15 × 15 × 1.5 mm³ were cut and polished on both sides for further investigation. Crystallization experiments were conducted on individual disks which were placed on alumina substrates and isothermally annealed at 800 °C for time spans between 4 and 16 h (in ambient air). The selection of this crystallization temperature was done in accordance with the results of non-isothermal differential scanning calorimetry (DSC, Netzsch, Ar atmosphere) which was conducted at a heating rate of 20 K/min. To identify and further describe the crystalline phases which occurred during crystallization, X-ray diffractometry (XRD, Siemens Kristalloflex D500, Bragg-Brentano, 30 kV/30 mA, Cu Kα) was employed (using a step width of 0.02°/s over 2θ range of 10-60°). Static and dynamic PL were studied with a high-resolution spectrofluorometer and through time correlated single photon counting (TCSPC, Horiba Jobin Yvon Fluorolog FL3-22) using a static Xe lamp (450 W) and a Xe flashlamp (75 W) as excitation sources. Photoluminescence excitation (PLE) spectra were corrected over the lamp intensity with a silicon photodiode and PL spectra were corrected by the spectral response of the employed PMT. CIE color coordinates were subsequently determined from the integrated PL spectra. The external quantum efficiency η_eQE, i.e., the number of PL photons divided by the number of PLE photons [24] was obtained from three individual measurements using a BaSO₄-coated integration sphere,

3. Results and discussion

Figure 1(a) shows a DSC curve of the Eu³⁺-activated NYSA glass after baseline correction. The onset temperature of glass transition T_g and two non-isothermal crystallization peak temperatures T_c1 and T_c2 were identified at 635.5 ± 0.5, 818.6 ± 0.5 and 905.7 ± 0.5 °C, respectively. The appearance of two crystallization peaks indicates that at least two crystalline phases precipitate at elevated temperature. Based on this temperature, we have chosen the annealing temperature of 800 °C which lies close to T_c1 and only in the onset of the second crystallization step. We may readily assume that for different time spans of treatment (4-16h), the crystallization step which corresponds to T_c1 is always completed, and the step which corresponds to T_c2 is partially completed for the lower annealing times, and fully completed for the 16 h treatment. The difference in the glass ceramics which is obtained through the different annealing times therefore arises from different phase assemblage and different stages of crystal growth (i.e., different crystallite sizes).

 

Fig. 1 (a) DSC curve of the NYSA glass (obtained at 20 K/min). (b) Ex situ XRD patterns of the as-melted NYSA glass and glass ceramic, exemplarily shown for 16 h annealing at 800 °C. The crystal structures of Na₃YSi₃O₉, NaYSiO₄ and NaAlSiO₄ are schematically shown in (c-e).

Download Full Size | PDF

Figure 1(b) represents XRD patterns of the NYSA precursor glass and the corresponding glass ceramic after 16 h of annealing at 800 °C. The purely glassy state of the precursor material to within the limitations of laboratory XRD is confirmed by the absence of any sharp diffraction peaks. Thermal annealing leads to the appearance of multiple diffraction peaks. Due to the structural similarity within the system of Na_xYSi_yO_{1/2x + 3/2 + 2y}, accurate phase assignment is rather complicated. The following indexing was therefore done to obtain a simplified experimental overview. That is, we assigned the observed diffraction peaks to two primary Na_xYSi_yO_{1/2x + 3/2 + 2y}-type-crystalline species, (x = 1;y = 1) and (x = 3; y = 3). The corresponding structures are described on JCPDS card no. 00-022-1049 (NaYSiO₄) and JCPDS card no. 00-036-0127 (Na₃YSi₃O₉). We further observe another minor crystalline phase, NaAlSiO₄ (JCPDS card no. 00-011-0220). The crystal structures of all three phases are depicted in Fig. 1(c)-(e). Orthorhombic NaYSiO₄ belongs to the space group Pbn2₁ (33) and is composed of [YO₆] and [NaO₆] octahedra and [SiO₄] tetrahedra [25]. Orthorhombic Na₃YSi₃O₉ belongs to space group P2₁2₁2₁ (19), where Na⁺ and Y³⁺ again occupy octahedral lattice sites, while Si⁴⁺ is fourfold coordinated [26]. Hexagonal NaAlSiO₄ has a nepheline-type structure which belongs to space group P6₃ (173) and comprises [NaO₆] octahedra and [AlO₄] and [SiO₄] tetrahedra [27].

Duffy’s optical basicity Λ of the glass provides a simplistic means to evaluate the oxygen field strength and to estimate the prevailing redox equilibrium between Eu²⁺ and Eu³⁺ [28–30]. It is obtained through linear mixing of the partial molar basicity Λ_i of each component i according to its molar fraction X_i,

Figure 2(a) shows PLE spectra (monitoring the electric-dipole transition ⁵D₀ →⁷F₂ of the Eu³⁺ at 612 nm in the range of 300–500 nm) of the as-melted Eu³⁺-activated NYSA glass and corresponding glass ceramics. The inset of Fig. 2(a) is the energy diagram of Eu³⁺ ions, shown to illustrate the origin of the PLE and PL lines. The obtained PLE spectra cover a broad spectral region, i.e., 360–470 nm and consist of five sharp peaks centered at 361, 382, 394, 415 and 464 nm which are readily attributed to the intrinsic intra-configurational 4f ↔ 4f transitions from the ground state (⁷F₀) to the labeled excited states of Eu³ [32]. The position of the excitation peaks remains unaffected by crystallization. However, with thermal annealing of the sample, the bands sharpen notably and their intensity generally increases. Furthermore, the transition of ⁷F₀ →⁵L₆ (with a full width at half maximum FWHM ~3 nm) exhibits the highest excitation intensity of all bands and for all samples, regardless of annealing time. Consequently, the peak of this band was selected to record the Eu³⁺-related PL spectra.

 

Fig. 2 Room-temperature photoluminescence excitation (a, PLE, monitoring the PL line of Eu³⁺ at 612 nm) and emission (b, PL, excited at 394 nm) spectra of the as-melted NYSA glass and corresponding glass ceramics (labels: annealing time at 800 °C). (c) Integrated PL intensity over the full spectrum and peak PL intensity of 612 nm as a function of annealing time. Lines in (b) are guides for the eye. Inset of (a): Energy level diagram of the Eu³⁺ ion. Inset of (b): Zoom at the spectral region of 500-570 nm. Lines in (c) are guides for the eye.

Download Full Size | PDF

In Fig. 2(b), the corresponding PL spectra are shown. They range from orange to the far red spectral region (i.e., from 575 to 720 nm) and consist of the five lines which are typical for Eu³⁺-related photoluminescence, i.e., the intra-configurational 4f ↔ 4f transitions from the first excited level ⁵D₀ to the indicated crystal field split levels ⁷F_{J (J = 0, 1, 2, 3 and 4)} (also shown in the inset of Fig. 2(a). The three major emission bands are located at 591 nm (magnetic-dipole transition ⁵D₀ → ⁷F₁), 612 nm (electric-dipole transition ⁵D₀ →⁷F₂) and 702 nm (⁵D₀ →⁷F₄). The broad PL band which would be typical for Eu²⁺ is not detected in any of the samples, confirming the absence of Eu²⁺ within detection limits [33].

As expected, thermal annealing of NYSA glass samples results in significant changes of PLE as well as PL spectra. Firstly, for short annealing time, i.e., ≤ 4 h, the red PL line at 612 nm (⁵D₀ →⁷F₂) is dominating the emission spectrum. For longer annealing (≥ 8 h), the bands at 591 and 702 nm strongly increase in intensity relative to the band at 612 nm (Fig. 2(b) and 3(a)). A second observation is that with increasing annealing time, also the absolute intensity of all PL bands increases notably, which is consistent with the PLE observation. That is, the peak PL intensity at 612 nm and the integrated PL intensity over the whole spectrum increase ~11-fold and ~20-fold, respectively, after crystallization (Fig. 2(c)). As mentioned before, this enhancement of PL as well as PLE intensity may be caused by either multiple scattering, increased absorption cross-section or enhanced lifetime of the excited state (increased internal quantum efficiency and reduced probability of non-radiative energy transfer as a result of the incorporation of Eu³⁺ into the crystalline environment). Thirdly, as a result of crystallization, all PL lines become sharper and components are better resolved. For example, the FWHM of PL line at 712 nm is narrowed from 10 to 4 nm after crystallization. Likewise, the other PL lines show a similar change as a function of annealing time.

In the glass ceramic samples, all PL lines exhibit strong Stark-splitting (702 nm into 702 and 688 nm contributions, 652 nm into 652 and 657 nm, 612 nm into 614 and 624 nm and 591 nm into 589 and 593 nm, see Fig. 2(b) and 3(a)). This observation of general sharpening and well-resolved splitting is a strong indication for the incorporation of Eu³⁺ species into a crystalline environment.

Besides PL from the first excited Eu³⁺:⁵D₀ level, only very weak PL lines from the higher levels of ⁵D_{J (J = 1, 2, 3)} are observed in the glass sample. This is related to strong non-radiative relaxation of the ⁵D_{J (J = 1, 2, 3)} levels through a fast phonon-assisted process (inset of Fig. 2(b)) [34]. However, after crystallization, the PL lines which are attributed to the higher lying excited levels (with peaks at 511, 537 and 553 nm) are strongly enhanced (inset of Fig. 2(b)).

Chromaticity coordinates of all samples were calculated from the PL spectra. For the glass sample, the emission color is (0.655, 0.345), i.e., pure red. These coordinates do not change notably upon crystallization, what is surprising on the first view. That is, the photoemission occurs in pure red color for the glass as well as for the glass ceramics, and can readily be observed with the naked eye.

As mentioned in the introduction section, the electronic-dipole-allowed transition of ⁵D₀ →⁷F₂ depends strongly on the structural symmetry of the Eu³⁺ ligands. At the same time, the magnetic-dipole allowed transition of ⁵D₀ →⁷F₁ is independent on local environment. Thus, the PL intensity ratio of both transitions can be used as a measure to evaluate the ligand symmetry of the Eu³⁺ sites. In general, a low value of R represents a high ligand symmetry and a low bond covalency of Eu³⁺ sites. As for the present case, a value of R = 4.8 is obtained for the NYSA glass sample (Fig. 3(b)). This indicates that in the glass, the Eu³⁺ ions occupy highly non-centrosymmetric lattice sites. With increasing degree of crystallization, the value of R decreases down to about 1.4 (Fig. 3(a)-(b)), what indicates that the Eu³⁺ ions locate in a more and more symmetric and less covalent environment.

 

Fig. 3 PL spectra (a, normalized to the PL peak at 612 nm) and annealing-time-dependent asymmetry ratio R of the PL intensities of ⁵D₀ →⁷F₂ and ⁵D₀ →⁷F₁ for the NYSA glass and corresponding glass ceramics (b).

Download Full Size | PDF

Figure 4 shows the normalized PL decay curves of the NYSA glass and glass ceramics, monitoring the PL lines at 591 nm (⁵D₀ →⁷F₁), 612 nm (⁵D₀ →⁷F₂) and 702 nm (⁵D₀ →⁷F₄). All decay curves can be well-fit to a single exponential decay function, with the lifetime τ after which the emission intensity reached 1/e of its initial value. The lifetime of the first excited state ⁵D₀ of Eu³⁺ increases with increasing annealing time, i.e., the lifetimes of Eu³⁺ PL lines at 591, 612 and 702 nm increase from 2.04 to 3.22 ms, from 1.93 to 2.87 ms and from 2.12 to 3.22 ms, respectively (insets of Fig. 4(a)-(c)). This enhancement further reflects the lower probability of non-radiative energy transfer and, thus, higher internal quantum efficiency. Consistent with this observation, also the directly measured value of η_eQE increases, i.e., from 19.5 to 36.2% after crystallization.

 

Fig. 4 Normalized PL decay curves of the Eu³⁺-activated NYSA glass and glass ceramics annealed at 800 °C for different times under excitation at 394 nm and monitoring PL at (a) 591 nm (⁵D₀ →⁷F₁), (b) 612 nm (⁵D₀ →⁷F₂) and (c) 702 nm (⁵D₀ →⁷F₄) (labels: annealing time at 800 °C). Inset of (a), (b) and (c): PL lifetimes of Eu³⁺ dependent on annealing times at 800 °C. Lines in the inset of (a), (b) and (c) are guides for the eye.

Download Full Size | PDF

The PL intensity of Eu³⁺, the observation of Stark splitting, the asymmetry ratio R, PL lifetimes and the directly observed external quantum efficiency η_eQE are strongly dependent on partitioning of the Eu³⁺ species. All present results support the conclusion that during crystallization, Eu³⁺ ions are incorporated into a crystalline lattice. Although four different crystalline lattice sites (^VIY³⁺, ^VINa³⁺, ^IVSi⁴⁺ and ^IVAl³⁺) are available, ^VIEu³⁺ ions are most probably incorporated on ^VIY³⁺ sites in NaYSiO₄ and Na₃YSi₃O₉ due to the similarity of ionic radii (the ionic radius mismatch between ^VIY³⁺ and ^VIEu³⁺ is about 5%) and equivalent valence state, as compared to ^VINa⁺, ^IVSi⁴⁺ and ^IVAl³⁺ sites (Table 1 [35],). In sodium disilicate glass, Eu³⁺ is reported to be (6 ± 0.5)-fold coordinated in a highly distorted oxygen environment [36]. While corresponding data are not available for exactly the present glass composition, we assume that we face a very similar situation also in the present case with 6-fold coordinated Eu³⁺ in a highly-distorted coordination shell. This explains the high asymmetry ratio of 4.8 which we observed here and leads to spectral dominance of electric-dipole allowed red emission in the NYTS glass host. In NaYSiO₄ and Na₃YSi₃O₉ crystalline phases, octahedral [YO₆] groups are located in a distorted octahedral environment without centrosymmetry [25,26]. The standard deviation of Y³⁺-O^2˗ distance in [YO₆] groups is used to evaluate the degree of distortion for octahedral [YO₆] groups, yielding values of 0.07 and 0.09 Å in NaYSiO₄ and Na₃YSi₃O₉, respectively, which indicates that the degree of local distortion of the [YO₆] groups is similar in NaYSiO₄ and Na₃YSi₃O₉ crystals. After crystallization, Eu³⁺ ions are embedded in the NaYSiO₄ and Na₃YSi₃O₉ crystalline phases, resulting in a less asymmetric environment. The degree of crystallization can be controlled by annealing time at 800 °C, i.e., the crystalline volume fraction increases with annealing time resulting in more Eu³⁺ ions incorporating into the NaYSiO₄ and Na₃YSi₃O₉ crystalline phases with increasing annealing time. Consequently, the asymmetry ratio decreases with increasing annealing time from 4.8 to 1.4, emission lines of Eu³⁺ are more and more Stark split, and the emission intensity, lifetime and η_eQE are enhanced. Further consideration of the evolution of the asymmetry ratio with increasing annealing time reveals that the major part of gain in symmetry occurs already after relatively short annealing, i.e., after 4 h at 800 °C. Considering the previous discussion of the presence of two crystallization steps, we may assume that in this regime, the first crystallization step is well completed while the second has occurred, if at all, to an unknown extend. Considering further the composition of the precursor glass, we have a ratio of Y³⁺:Si⁴⁺ of about 1:3, a ratio of Na⁺:Si⁴⁺ of about 1:1 and a ratio of Na⁺:Y³⁺ of > 3:1. Crystallization occurs towards the general join of Na_xYSi_yO_{1/2x + 3/2 + 2y} where the coordination numbers in the glass phase are already close to the coordination numbers of the individual species in the crystalline phase. The most mobile species is the monovalent Na⁺ ion which is, in addition, present in relative abundance with respect to the general stoichiometry of Na_xYSi_yO_{1/2x + 3/2 + 2y} when x = y. The formation of a phase with specific stoichiometry is therefore determined by the abundance and mobility of Y³⁺ and Si⁴⁺, where the stoichiometry of the precursor glass dictates the formation of Na₃YSi₃O₉. Any phase with higher or lower Y³⁺:Si⁴⁺ ratio would require a certain degree of Y³⁺ diffusion which becomes an increasing barrier with increasing degree of crystallization (through increasing consumption of available Y³⁺). We therefore conclude that crystallization occurs first into Na₃YSi₃O₉, characterized by T_c1. During this process, especially Al₂O₃ and Na₂O enrich in the residual liquid phase, what results in further precipitation of NaAlSiO₄, characterized by T_c2. However, as NaAlSiO₄ does not provide suitable lattice sites for the incorporation of Eu³⁺, its contribution to the emission characteristics is only through affecting the scattering behavior of incoming light. Controlled precipitation of this phase through prolonged thermal treatment or a second annealing step around T_c2 can therefore be employed to tune light scattering without affecting Eu³⁺ partitioning.

Table 1. Relative difference in ionic radii (D_r (%) = 100 × [R_m (CN) –R_d (CN)]/R_m (CN)) between matrix cations and dopant Eu³⁺ [2]

View Table

4. Conclusions

In conclusion, we reported on photoluminescence from Eu³⁺-activated Na,Y-silicate glasses and glass ceramics of the Na₃YSi₃O₉ type. Glass ceramics were obtained through controlled crystallization of the super-cooled precursor liquid. We demonstrated that during crystallization, a substantial amount of Eu³⁺ partitions on octahedral Y³⁺ lattice sites. An evaluation of the asymmetry ratio of the PL band intensities of ⁵D₀ →⁷F₂ and ⁵D₀ →⁷F₁ indicates that this incorporation occurs primarily in Na₃YSi₃O₉, which is also the first phase to precipitate. Controlled precipitation of NaAlSiO₄ through prolonged thermal treatment or a second annealing step can be employed to tune light scattering and, hence, absorption cross section and light extraction without affecting Eu³⁺ partitioning. As a result of crystallization, we obtain a 20-fold gain in absolute emission intensity, depending on crystallization conditions, and a 50%-gain in external quantum efficiency. Due to a strong shift of the spectral symmetry to the ⁵D₀ →⁷F₄ emission band at 702 nm with increasing thermal annealing time, the pure red emission color is kept unaffected by glass crystallization. The material may therefore be considered as an alternative red-emitting phosphor for light conversion.