1. Introduction

Broadband conversion of ultraviolet (UV) radiation to visible (VIS) light is highly desirable in a variety of applications, ranging from solar energy harvesting (UV-A and UV-B conversion), luminescent lighting, UV-detection and imaging to, e.g., agricultural lighting. Thereby, the converter may either be used to directly emit VIS radiation or as a sensitizer for a secondary conversion or energy transfer process. In principle, a large variety of optically active materials can be employed for this purpose. However, several limiting factors have to be considered, such as low excitation bandwidth, high stokes shift, photostability or, for example, undesired excitation in the blue spectral region. In this respect, knowledge on the optoelectronic properties of vanadium ions is still relatively limited. In inorganic oxide matrices, vanadium is usually present in three different oxidation states, V³⁺ ([Ar] 3d²), V⁴⁺ ([Ar] 3d¹) and V⁵⁺ ([Ar]), whereby the trivalent ion is readily oxidized to higher valence. Photoluminescence is attributed to the presence of V⁵⁺ ions and covers a broad band ranging from 400 to 700 nm [1–7]. Excitation (~200-400 nm) may be understood as an energy transfer process from O^2- to V⁵⁺ which occurs intrinsically in the vanadyl group. Emission then results from relaxation of ≡V-O^- to ≡V = O.

Generally, synthesis of vanadium-bearing materials on the melting route offers unique advantages with respect to dopant concentration, chemical homogeneity and compositional versatility. It provides several tools to adjust the equilibrium between V⁵⁺ and V⁴⁺. Moreover, it enables facile fabrication of specific geometries such as fiber and/or fleeces, microspheres, sheet, tape etc. In this respect, vanadium has a long tradition in melt-processed glasses, where it may be present in quantities of several tens of ppm (e.g. as coloring agent) to up to tens of mol% (e.g. in electron-conducting vanadate glasses). By choosing a suitable matrix composition and, eventually, employing agents for heterogeneous nucleation, glasses can subsequently be converted into glass ceramics with controlled crystallite size and volume fraction [8]. In this way, a crystalline environment can be provided that may at least partially host the dopant species [9–12]. At the same time, all advantages of glass processing can be made use of. However, comparably little is known on the luminescence of V-doped glasses and, in particular, V⁵⁺-doped glass ceramics. Here, we report on luminescence and vanadium partitioning in a V-doped zinc silicate model glass and the corresponding nanocrystalline Li₂ZnSiO₄ glass ceramic.

2. Experimental

In accordance with previous work, lithium zinc silicate glasses were chosen as a model system due to their facile convertibility into Li₂ZnSiO₄ glass ceramics [10,11,13,14]. The availability of relatively large tetrahedral Zn²⁺ lattice sites (ionic radius r ~0.60 Å) and the proximity of mobile Li⁺ for eventual charge compensation appear favorable for the incorporation of various dopant species [11]. Samples with nominal composition (mol%) 48SiO₂-24Li₂O-16ZnO-8Al₂O₃-3K₂O-1P₂O₅-xV₂O₅ (SLZAKP, x = 0, 0.1, 0.2, 0.5, 1, 2 and 4) were prepared by conventional melting and quenching of a 100 g batch of analytical grade reagents SiO₂, Li₂CO₃, ZnO, Al₂O₃, K₂CO₃, NH₄H₂PO₄ and V₂O₅. Melting was performed in alumina crucibles at 1550 °C for 2 hours. Glass slabs were obtained after pouring the melts into preheated graphite moulds and annealing for 2 h at 450 °C. From these slabs, disks of 10 × 10 × 2 mm³ were cut and polished (SiC/water). Conversion into glass ceramics was done via a single-step heat treatment for 2 h at a temperature between 550 and 700 °C (in steps of 50 °C) in air (applying a heating rate of 3 K/min). Density before and after crystallization was determined by the Archimedes method. The crystallization process was characterized by X-ray diffractometry (XRD Siemens Kristalloflex D500, Bragg-Brentano, 30 kV/30 mA, Cu Kα). UV-NIR absorption spectra were recorded with a double-beam photospectrometer, using a 150 mm integration sphere and a PbS detector (PerkinElmer Lambda 950). Room-temperature photoluminescence and fluorescence decay kinetics were studied with a spectrofluorometer equipped with double monochromators (Czerny-Turner) in excitation and emission (Fluorolog-3, Horiba Jobin Yvon), using a 400 W Xe-lamp for static and a 75 W Xe-flashlamp for dynamic analyses, respectively, as excitation sources. Electron paramagentic resonance spectra (EPR) were taken on an X-band microwave spectrometer operated at a frequency of 9.7 GHz (Bruker ESR 300E). XPS spectra were acquired using a PHI 5600 spectrometer with a 450 W Al Kα X-ray source. Wide scans were made at 300 W source power and a emission angle to the specimen of 45° in a binding energy range of 0 - 1200 eV and pass energy of 187.85 eV. The C1s, O1s and the V2p core levels were recorded with a step of 0.2 eV and pass energy of 23.50 eV, whereby the levels for V₂O₅ were recorded simultaneously. Binding energies of the photoelectron lines of the sample were charge referenced to the C1s line of adventitious hydrocarbon at 284.80 eV. Background subtraction was made using a Shirley function. O1s and V2p signals were fitted to Gaussian peak functions.

3. Results and discussion

Depending on nominal vanadium concentration, as-melted glasses exhibited greenish to brown coloration (the un-doped glass was completely colorless). Corresponding absorption spectra are shown in Fig. 1(A) . The dominating absorption bands are located at 1065, 615 and 450 nm. These are assigned to the presence of V⁴⁺ ions and correspond to the electronic transitions of ²Г_t4→²Г_t5, ²Г_t4→²Г_t3, and ²Г_t4→²Г_t1, respectively [15,16]. The spectra further indicate that no V³⁺ [17] is present in the as-melted glasses. The absorption bands of V⁵⁺ lie in the UV spectral range and can, in a first consideration, not unambiguously be extracted from Fig. 1. With increasing dopant concentration, the UV cut-off is shifted to longer wavelength.

 

Fig. 1 UV-NIR absorption spectra (A) and EPR spectra (B) of SLZAKP for different nominal dopant concentration, expressed as V₂O₅.

Download Full Size | PDF

In silicate glasses, V⁴⁺ ions are typically present in the form of a VO²⁺ complex [18]. Due to the unpaired d-electron of the V⁴⁺ ion, this complex exhibits characteristic electron paramagnetic resonance. Corresponding EPR spectra comprise a well-resolved hyperfine structure (located around g ~2.0) which is associated with the ⁵¹V nucleus (I = 7/2) [19–21]. EPR spectra of as-melted SLZAKP with different nominal vanadium content are shown in Fig. 1(B). In accordance with optical absorption data, the presence of an increasing amount of VO²⁺ with increasing total vanadium concentration is clearly confirmed (noteworthy, resonances were absent in blank reference samples). The resonance intensity is used as an indicator for the relative amount of VO²⁺-species. It increases almost linearly with increasing total vanadium content. Hence, it may be assumed that in the considered compositional range, the ratio of V⁴⁺/V⁵⁺ remains untouched by the total vanadium concentration.

The optical basicity Λ of the glass provides a way to estimate the location of the redox equilibrium between V⁴⁺ and V⁵⁺ [22]. It is obtained via a linear mixing approach, taking into account the molar fraction X _i and the partial molar basicity Λ _i of each component i,

 

Fig. 2 XPS spectra of as-melted SLZAKP with different nominal vanadium concentration (labels). The inset depicts a zoom at the V2p_3/2 signal.

Download Full Size | PDF

Optical excitation and emission spectra are given in Fig. 3(A) . All doped samples show a broad excitation band, ranging from 250 to 400 nm. The excitation band comprises two individual peaks, i.e. at about 275 and 320 nm. These peaks are assigned to the already noted charge transfer reaction in [VO₄]^3- groups [24,25]. Interestingly, the position of the higher-lying excitation band remains unchanged with increasing vanadium concentration whereas the low-lying band appears to red-shift from 313 to 336 nm as the vanadium concentration (V₂O₅) is increased from 0.1 to 4.0 mol%. Emission spectra were recorded by exciting on the lower-lying band. Emission occurs over a broad asymmetric band ranging from 400 to 800 nm with a maximum at 580 nm. Its position and FWHM do not vary with increasing dopant concentration. Emission intensity reaches a maximum for a nominal V₂O₅ concentration of about 1 mol%. For higher dopant concentration, the emission intensity drops sharply, indicating concentration quenching and/or increasing interference of V⁴⁺-species. In the considered spectral range, for excitation at 315 nm, blank samples of SLZAKP did not show any photoluminescence. Figure 3(B) represents the dependence of emission lifetime (time within which the luminescence intensity decays to 1/e times of its initial value) on nominal vanadium concentration (expressed as V₂O₅). Generally, luminescence decay follows a single exponential equation (inset of Fig. 3(B), shown exemplarily for [V₂O₅] = 0.1 mol%), indicating that only a single type of emission centers is present. For a concentration of up to about 0.5 mol%, lifetime remains practically concentration-independent (~34 μs). At higher dopant concentration, it decreases notably, i.e. to 22.4 μs for [V₂O₅] = 4.0 mol%. This decrease is attributed to increasing interaction between neighboring V-species and, hence, increasing probability of fast non-radiative decay. The observed lifetime is comparable to that of crystalline V-MgAl₂O₄ (13.7 μs) [1], much larger than that of V-doped SiO₂/Zn₂SiO₄ composites (13 ns) [2] and significantly below the value observed in C-V_0.05Siβ (88 ms) [4].

 

Fig. 3 Static photoluminescence (A) and luminescence lifetime (B) of SLZAKP for different nominal vanadium concentration (labels). The inset of (B) shows the decay kinetics of an as-melted specimen and a crystallized sample (both doped with 0.1 mol% V₂O₅).

Download Full Size | PDF

Upon annealing at ≥ 550 °C, the samples partially crystallize (Fig. 4(A) ). Crystallization occurs by internal nucleation homogeneously over the bulk of the sample. In accordance with previous observations [10,11,14], XRD data confirm precipitation of a nanocrystalline phase and suggest assignment to the orthorhombic polymorph of Li₂ZnSiO₄ (JCPDS card no. 00-024-0682) [11,26]. However, it must be noted that the phase is presently not indexed. The peak arrangement matches that of orthorhombic Li_3.8Zn_1.2P_1.8Si_0.2O₈, notwithstanding the clear discrepancy in chemistry. Both phases are structurally related to low-Li₃PO₄ (JCPDS card no. 01-084-0003). We therefore interpret the observed patterns as a reflection of the formation of a solid solution of Li_4-2xZn_xSiO₄, eventually incorporating or being nucleated by minor amounts of PO₄ ^3- [11]. In accordance with [11], XRD spectra indicate a phase transition in the temperature range of 600-650 °C. During crystallization, density of the samples does not change significantly (as-made SLZAKP-0.1 mol% V₂O₅: 2,721 ± 0.005 g/cm³; after annealing for 2 h at 700 °C: 2,735 ± 0.005 g/cm³). The crystal volume faction V _c was estimated from the relative area of all diffraction peaks. Calculated values are noted in Fig. 4(A). V _c is limited by the amount of ZnO. A simple consideration of glass composition and stoichiometry indicates a theoretical maximum of V _c of 45-50 vol.%, provided that stoichiometric Li₂ZnSiO₄ is the single crystalline phase. Such a degree of crystallization is achieved after treatment for 2 h at 700 °C. Traces of Zn²⁺-depleted crystallite species, appearing especially in the later stages of crystallization, may be a result of ZnO depletion and higher mobility of Li⁺-ions, and may explain the difficulties in assigning the obtained XRD patterns. The residual glass phase is an aluminosilicate glass with reduced optical basicity. Photoluminescence spectra of exemplary SLZAKP glass ceramics ([V₂O₅] = 0.1 mol%) are shown in Fig. 4(B). Upon crystallization and as a function of annealing temperature (i.e., degree of crystallization), the relative emission intensity increases up to tenfold for the sample which was annealed for 2 h at 700 °C as compared to the as-melted glass. In a first consideration, this increase is at least partly attributed to multiple scattering of incident and luminescent light in the glass ceramic. On the other hand, the position of the emission band blue-shifts with increasing annealing temperature, i.e. from 580 nm (as-melted glass) to 546 nm (glass ceramic after 2 h at 700 °C), what indicates that the active vanadyl groups undergo a change in ligand field strength. This observation is in accordance with the significant increase which is observed in emission lifetime: as shown in Fig. 3(B), upon crystallization, lifetime increases from 34.1 to 45.3 μs, what is a clear indicator for the reduction of non-radiative relaxation and, hence, an increase in quantum efficiency. In the present case, it is also taken as strong evidence for the effective incorporation of V⁵⁺ emission centers into the crystalline phase for the following reason: if vanadyl groups would not partition into the crystalline phase, they would enrich in the residual glass phase. That is, for V _c of ~48 vol.%, V₂O₅ concentration in the residual glass phase would double to about 0.2 mol%. As shown in Fig. 3(B) for the as-melted glasses, at this dopant concentration, lifetime is still about 34 μs. Taking into account the change in chemical composition of the residual glass phase due to selective precipitation of Li₂O, ZnO and SiO₂ in the crystalline phase (resulting in lower optical basicity), one would expect an increase in the amount of V⁴⁺ (Eq. (2)). This would result in additional quenching of photoluminescence due to ²Г_t4→²Г_t3 (what is not the case, here).

 

Fig. 4 (A): Ex situ XRD patterns of as-melted SLZAKP and samples which were annealed for 2 h at different temperatures ([V₂O₅] = 0.1 mol%).The labels indicate the calculated crystal volume fraction. (B): Photoexcitation (right) and emission (left) spectra of crystallized SLZAKP for various annealing temperatures.

Download Full Size | PDF

4. Conclusions

Photoluminescence of vanadium-doped lithium zinc silicate glasses and corresponding nanocrystalline Li₂ZnSiO₄ glass ceramics was studied. Depending on dopant concentration and synthesis conditions, VIS photoemission from [VO₄]^3- centers peaks at 550-590 nm and occurs over a bandwidth (FWHM) of ~250 nm. While within the considered range of dopant concentrations, the emission band position remains unaffected by dopant concentration, a notable blue-shift occurs after crystallization. The corresponding excitation band covers the complete UV-B to UV-A spectral region. Here, the excitation peak at about 315 nm remains unaffected by crystallization, but red-shifts when the dopant concentration is increased. Both observations are attributed to the nature of photoemission in the considered system. That is, photoemission occurs as a result of relaxing charge transfer in vanadyl groups. This process is strongly affected by the ligand field strength, and also by interference of V⁴⁺ species and oxygen vacancies, respectively. Partitioning of vanadium into V⁵⁺ and V⁴⁺ was examined by EPR and XPS. Due to the very high absorption cross section of ²Г_t4→²Г_t3, even small amounts of V⁴⁺ strongly interfere with photoluminescence. In as-melted glasses, the emission lifetime is about 34 μs. After crystallization, it increases to about 45 μs. With respect to concentration quenching, the optimal dopant concentration was found at about 0.5 mol% V₂O₅. Crystallization was observed to results in a tenfold increase of emission intensity, what is attributed to multiple scattering and the incorporation of V⁵⁺ species into the crystal phase.