1. Introduction

Divalent manganese ions play an important role as active centers in inorganic phosphors, mainly for lighting applications. They exhibit a 3d⁵ electronic configuration and photoemission usually occurs due to the transition ⁴T₁(G) → ⁶A₁(S). Therefore, the position of the emission band is strongly dependent on the field strength of the surrounding lattice. Typically, the emission peak lies between about 500 and 700 nm with a bandwidth of several tens of nanometers [1–3]. As a consequence, fluorescence from Mn²⁺ dopants can be used, for instance, for probing coordination numbers or other structural data in both glasses and crystalline materials [1,4]. On the other hand, the strong dependence of emission behaviour on host lattice makes the Mn²⁺ center a highly versatile and, thus, attractive component in the development of novel phosphors that emit in the green-to-red spectral range (e.g [5].). However, particularly for applications in solid-state lighting, knowledge on active centers other than rare-earth ions remains comparably limited. Their optical properties, on the other hand, often suggest significant potential as primary phosphor in combination with soft UV and blue excitation sources or as secondary phosphors to improve color rendering index (CRI) and perception of white light emitting devices (WLEDs) [6,7].

Contrary to the often employed solid state reaction for the production of inorganic phosphor materials, in the present study we are focussing on the glass ceramic route. That is, production of a glass by conventional melting and quenching of a mixture of raw materials, and controlled recrystallization of that glass in a secondary heating procedure. In a general way, this procedure offers various advantages over conventional solid state reactions, particularly related to compositional variance, dopant concentration and homogeneity, microstructural design, forming and production of microspheres, and recycling. In this context, as a means to stabilize Mn²⁺ over trivalent manganese, we have chosen a phosphate glass matrix. Phosphate glasses are long known as excellent hosts for high concentrations of transition metal ions as well as rare earth ions, what has made them the most important system for glass lasers, but also high-performance filters. They also exhibit relatively low optical basicity, what enables stabilization of polyvalent ions at lower redox states than are normally found in other glasses. Their major drawback, however, is their chemical stability and resistance to (atmospheric) water, what often renders them useless for any applications where they cannot be protected adequately. In the present case, another problem is presented by the desired controlled recrystallization of the as-melted glass. For that, internal nucleation (versus nucleation and crystal growth only at the surface) must be achieved, what cannot easily be done in phosphate glasses with relatively high kinetic fragility. Furthermore, even if effective heterogeneous nucleation agents were available, as additional components in the potential material, their optical and redox properties might strongly alter those of the final phosphor. Noteworthy, this prevents the use of many of the commonly [8] used nucleant species such as, e. g., TiO₂, metallic nanoparticles, SnO₂ or CeO₂. Therefore, a different strategy was applied in the present case that resulted in a unique material. Two competing network forming units, PO₄ ^3- and SO₄ ^2-, were combined in a glass with relatively low total amount of network former (and high kinetic fragility). It is demonstrated that in this case, phase separation can be induced, resulting in controlled, homogeneous and finely distributed precipitation of an SO₃-rich phase. This then gives rise to laterally confined crystallization of (Zn,Na₂)SO₃:Mn²⁺. The resulting glass ceramic exhibits intense red luminescence, in dependence on Mn²⁺ concentration and composition of the precursor glass, peaking between about 610 and 650 nm with a bandwidth FWHM of about 110 nm.

2. Experimental

Glass samples were prepared by conventional melting and quenching, using analytical grade ZnO, Na₂CO₃, ZnSO₄·7H₂O, NH₄H₂PO₄ and MnCO₃ as raw materials and adding ZnSO₄·7H₂O to obtain SO₃ in excess of 3 mol.% to compensate for volatilization losses. Batches were homogenized by grinding mixtures of appropriate composition in an agate mortar. MnO-containing batches were first pre-calcined for 3 h at 300 °C, and subsequently melted at 750 °C for 1 h in alumina crucibles, using an alumina rod for stirring. For comparison, undoped glasses were melted in Au crucibles. Melting in alumina crucibles resulted in contamination of the glasses with < 0.2 mol.% Al₂O₃ (determined by electron dispersive spectroscopy, EDS). Noteworthy, this contamination had no observable effect on rheology, crystallization or optical properties of the derived glasses. Glass slabs of ~20 g were obtained by quenching on a graphite plate at room temperature. Nominal sample compositions are give in Table 1 . The glass transition temperature T _g of the samples was estimated from differential scanning calorimetry (DSC, Netzsch DSC 404 F1) according to Ref [9]. All glass samples appeared highly transparent, what can be taken as a reliable indicator for 2 + as the dominant valence state of the manganese ions [Fig. 1(a) ]. Analyses of optical emission and excitation behavior were performed with a spectrofluorometer equipped with double monochromators (Czerny-Turner) in excitation and emission (Fluorolog³, Horiba Jobin Yvon, spectral resolution of ~0.1 nm), using a 400 W Xe-lamp for static and a 75 W Xe-flashlamp for dynamic analyses, respectively, as excitation sources. Raman spectra were collected on a Nicolet Almega XR dispersive Raman spectrometer. The crystallization process was studied in situ, using a high-temperature X-ray diffractometer (XRD) at a heating rate of 5 K/min. Finally, electron spin resonance spectra (ESR) of Mn-doped samples were recorded on an X-band microwave spectrometer at a frequency of 9.7 GHz (Bruker ESR 300E). With the exception of XRD, all measurements were carried out at room temperature.

Table 1. Nominal compositions and basic physical data of examined glasses.

View Table

 

Fig. 1 Optical excitation (monitoring wavelength of emission peak) and emission spectra (excitation wavelength of 409 nm), (A), and corresponding electronic band structure (B) of Mn²⁺-doped sulfophosphate glasses of type SP11. Labels in (A) indicate concentration of MnO in mol.%. Inset: Photograph of samples SP11_30 (top), SP11_07 (middle), SP11_01 (bottom).

Download Full Size | PDF

3. Results and discussion

3.1 Photoluminescence in Mn²⁺-doped precursor glasses

The electronic structure of manganese (3d⁵) is relatively well known. It comprises the excited terms ⁴G, ⁴P, ⁴D and ⁴F [10]. For Mn²⁺, the ground state is ⁶A₁(S) and at least five transitions can readily be identified in the optical excitation spectra of Mn-doped sulfophosphate glasses [Fig. 1(a)], peaking at 348.9 nm, 360.7 nm, 409.4 nm, 421.9 and 504.4 nm, respectively. These correspond to the transitions ⁶A₁(S) → ⁴E(D), ⁶A₁(S) → ⁴T₂(D), ⁶A₁(S)→ ⁴A₁(G), ⁴E, ⁶A₁(S) → ⁴T₂(G) and ⁶A₁(S) → ⁴T₁(G), respectively [Fig. 1(b)]. A further excitation band can be detected at ~317.5 nm, corresponding to ⁶A₁(S) → ⁴T₁(P), if the concentration of MnO exceeds ~0.5 mol.%. Noteworthy, for increasing manganese content, samples exhibit increasing violet coloration [inset of Fig. 1(a)]. The origin of this lies in an absorption band at ~490 nm which can readily be assigned to traces of Mn³⁺ (3d⁴; ⁵E →⁵T₂ [11] (note that Mn³⁺ does not exhibit optical emission in the 600-nm spectral range). Due to strong overlap particularly of the lower three bands (excitation to ⁴A₁(G), ⁴T₂(G) and ⁴T₁(G), respectively), excitation can in principle be performed with a relatively broad blue source, i.e. from about 400 to 530 nm. On the other hand, the band around 409 nm (⁴E(G) →⁴A₁(S)) appears most effective. The ratio between blue and UV excitation, I _409nm to I _360nm increases with increasing Mn²⁺ concentration. In the same time, emission intensity increases continuously, surprisingly without notable concentration quenching for up to 3 mol.% of dopant concentration (noteworthy, concentration quenching was observed for SO₃-free zinc phosphate glasses at MnO > 2.5 mol.% upon excitation at 410 nm [4]).

Emission always occurs due to ⁴T₁(G) → ⁶A₁(S), and a significant red-shift as well as increasing emission intensity are observed with increasing concentration of Mn²⁺ [Fig. 1(a)]. Linear regression of the integrated emission intensity I versus manganese concentration c _MnO leads to I(a.u.) = 9.86397 × c _MnO(mol%) with an R-value of 0.98757. The line width at half maximum (FWHM) is ~110 nm for all samples. In a first consideration, the position of the emission band suggests octahedral coordination of the Mn²⁺ ions (while tetrahedrally coordinated Mn²⁺ usually results in green emission [12]). Noteworthy, changing the SO₃ to P₂O₅ ratio at constant MnO concentration (samples SP6_07; SP11_07; SP16_07 and SP19_07) results in a slight blue-shift of emission for increasing amount of SO₃. Considering the d⁵ Tanabe-Sugano diagram, this indicates decreasing crystal field strength with increasing SO₃.

ESR spectra of doped glasses are shown in Fig. 2(a) , revealing the typical [13,14] fingerprint of Mn²⁺ (no resonance was observed in undoped glasses). It comprises a sextet hyperfine line structure. As a result of concentration quenching, this sub-structure disappears at MnO ≥ 2 mol.% [15,16]. Obtained spectra indicate increasing hyperfine interaction with increasing Mn²⁺ concentration and, hence, increasingly ionic character of the bonds between Mn²⁺ and its surrounding ligands [17,18]. The increasing interaction between Mn²⁺-species may further be taken as an explanation for the observed red-shift in optical emission with increasing concentration [19].

 

Fig. 2 ESR (A) and Raman spectra of Mn²⁺-doped sulfophosphate glasses of type SP11. Labels indicate concentration of MnO (mol.%). Inset: Intensity ratio of the Raman bands at 1050 cm⁻¹ and 992 cm⁻¹.

Download Full Size | PDF

Raman spectra further confirm this conclusion [Fig. 2(b)]. They exhibit four major bands: at 1410 cm⁻¹ (symmetric P = O stretch [20], ), 1050 cm⁻¹ (symmetric stretching vibration in PO₃ [20]), 992 cm⁻¹ (υ ₁-vibration in SO₄ ^2- [21,22]) and 750 cm⁻¹ (stretching mode in bridging oxygen [21^]). Relative to the Raman active SO₄ ^2- band, the ν_s-PO₃ band decreases with increasing Mn²⁺-concentration (inset of Fig. 2). Considering charge balance, ionic radii and coordination, it is reasonable to assume that Mn²⁺ ions are incorporated in the lattice on Zn²⁺ sites (note that Zn²⁺ is in octahedral coordination [23], and above luminescence data indicates octahedral coordination for Mn²⁺, too). An increase in (Zn²⁺ + Mn²⁺) then results in further decreasing polymerization of the P₂O₅-network [21]. This is reflected by decreasing intensity of the 1050 cm⁻² and 750 cm⁻² bands.

3.2 Sulfophosphate glass ceramics

In terms of emission efficiency, the preferred environment of active centers should usually be crystalline rather than glassy. To fully benefit from the several advantages of glass forming, but to also obtain high quantum yield, however, controlled recrystallization – the formation of a glass ceramic – may be employed. Thereby, the question whether or not nucleation and crystal growth can be initiated in the bulk material or only at free surfaces is decisive for any potential application. As noted before, bulk (internal) nucleation is usually achieved via the use of chemical agents that promote heterogeneous nucleation, often by epitaxial growth of the desired phase on the nucleant phase. In the present case, as with most phosphate systems, however, such nucleation agents are on the one side not known. On the other side, common choices would have significant impact on the optical properties as well as on the redox distribution of the manganese ions. Therefore, a different approach was pursued that has originally been described by Beall for the controlled precipitation of mullite from an aluminosilicate system [24]: A glass is formed from two glass formers so that its composition lies within the mixability gap of the system, but phase separation is kinetically prevented. Via re-heating this glass, the kinetic barrier is overcome and phase separation occurs, what results in a significant shift of the chemical (and rheological) properties of the demixed phases. Then, crystallization may occur in the demixed phase, either homogenously or at the newly-created interface. Crystal precipitation remains confined to the so-called relict structure [24] of the previous separation process and can thus be controlled to occur finely distributed over the bulk material. This approach was applied in the present case. Upon reheating the glass, the presence of competing network forming structural units PO₄ ^3- and SO₄ ^2- in comparable quantities triggers precipitation of an SO₄ ^2--rich droplet phase. From this phase, crystallization is initiated. This process can be observed in situ by high-temperature XRD (Fig. 3 ). In dependence on SO₃-concentration, crystallization is roughly described as a two-step process. Starting at ~370°C (SP16) and ~390°C (SP19), respectively, crystalline sulfate and sulfophosphate phases are initially precipitated. These comprise dominantly Na₂SO₃ and (Na₂, Zn)SO₃. At this point, crystallization occurs internally, what is clearly evidenced by gradually evolving opalescence of the samples if heat-treated in this temperature regime. Noteworthy, this step cannot be observed in sample SP6. Assumedly, in that case, SO₃-concentration lies within the solubility limit of the phosphate matrix. The second crystallization process is characterized by the precipitation of mixed (Na, Zn) phosphate phases, dominantly via surface crystallization. This second process resembles typical devitrification behaviour of phosphate glasses, occurring in either sample at temperatures between ~470 °C (SP6) and 400 °C (SP19). It is undesired in the sense that it cannot be controlled, does not occur homogeneously over the bulk sample and prevents simple transformation of a glass body into a glass ceramic while maintaining its external geometry (e.g. fiber). Hence, the observed process window for the preparation of glass ceramics lies in the temperature range of ~370°C to ~390°C. If the relatively high heating rates and short observation times, respectively, that were applied in the HT-XRD analyses are taken into account, these temperature limits should be reduced by about 10-20 K.

 

Fig. 3 In situ X-ray diffraction patterns as taken during heating samples SP6, SP16 and SP19 (varying SO₄ ^2-/PO₄ ^3- ratio). Labels indicate respective temperature. Spectra are initially dominated by sulfate phases Na₂SO₄, JCPDS 086-0800 (red arrows); ZnSO₄, JCPDS 070-1255 (green dotted line in SP16), later by various phosphate phases (blue arrows).

Download Full Size | PDF

As a confirmation of the motivating assumption, the resulting glass ceramics exhibit several times higher emission intensity, as compared to the as-melted glass at equivalent composition and MnO-concentration (Fig. 4 ). Besides this significant intensity increase, on a second view, the emission spectrum of the crystallized sample appears slightly blue-shifted and strongly asymmetric. A more detailed view on this is obtained if emission spectra for various excitation wavelengths are considered (Fig. 5 ). Qualitatively, on the crystallized sample, a shift of the peak emission wavelength between three dominant positions can be observed, i.e. ~620 nm, 628 nm and 616 nm for an excitation wavelength of ~350 nm, 409 nm and 500 nm, respectively [Fig. 5(a)]. On the other hand, in the as-made glass, no dependence of emission peak position on excitation wavelength could be detected [Fig. 5(b)]. This indicates that in the glass ceramic, emission centers are present in three different ligand fields. Compared to as-made glasses (0.7 mol.% MnO, emission peak at ~625 nm), two of the three positions result in a slight blue shift of emission, thus indicating lower crystal field strength [25]. The third position practically equals that of a glassy environment with low SO₃ concentration. This is consistent with principle expectations that arise with the synthesis of the glass ceramics. Precipitation of sulfate phases results in a decrease of SO₃ concentration in the residual glass phase. Noteworthy, emission as a function of SO₃-content exhibits a slight red-shift for decreasing SO₃ (Fig. 4). Then, the glass ceramic emission peak at 628 nm is attributed to Mn²⁺-centers in the residual glass phase. Upon crystallization, Mn²⁺ can be incorporated on Zn²⁺-sites in zinc phosphate as well as zinc sulfate lattices, what results, in consistence with XRD data, in two principle emission centers. In both cases, Zn²⁺ and, hence, Mn²⁺ remains in octahedral coordination. Therefore, the extent of the observed blue-shift is relatively small. Considering the ionic radii of S⁶⁺~29 pm and P⁵⁺~38 pm, respectively, a stronger field would be expected to act on Mn²⁺ in a phosphate lattice as compared to a sulfate lattice. Then, emission at 620 nm should be attributed to the phosphate species (higher crystal field strength), while emission at 616 nm originates from Mn²⁺ in sulfate environment. A corresponding picture is revealed by dynamic emission data (Fig. 6 ). Delayed spectroscopy reveals changes in shape and position of the dominant emission peak with increasing delay time. Namely, in accordance with above arguments, the emission spectrum is dominated, in sequence, by peaks at 620 nm, 628 nm and 616 nm, respectively [Fig. 6(a)]. This is taken as another evidence for the presence of at least three emission centers, and is confirmed by the decay kinetic which clearly does not follow a first order exponential equation [Fig. 6(b)], particularly in the beginning period. The effective lifetime (1/e times initial intensity) is 14.0 and 16.8 ms for glass and glass ceramic, respectively.

 

Fig. 4 Optical emission spectra of as-made glass SP16_10 and corresponding glass ceramic (heat treated for 4h at 380 °C). Inset: Photographs of both samples (A) under ambient light and (B) under 40 W UV-A broadband illumination.

Download Full Size | PDF

 

Fig. 5 Optical emission spectra of glass ceramic (A) and glass (B) SP11_07 for varying excitation wavelength (labels).

Download Full Size | PDF

 

Fig. 6 Time-resolved delay curves of emission from glass ceramic (A) and luminescence decay curves (B) of glass and glass ceramic.

Download Full Size | PDF

4. Conclusions

In summary, we report on intense red fluorescence from Mn²⁺-doped sulfophosphate glasses and glass ceramics of the type ZnO-Na₂O-SO₃-P₂O₅. Due to the low optical basicity of the phosphate matrix, stabilization of Mn²⁺ over Mn³⁺ can readily be achieved in this system, even for high dopant concentrations, when preparing the glasses in air. Through thermally induced bimodal separation of a sulphate-rich phase, finely distributed precipitation of (Na, Zn) sulfate phases by internal crystallization is achieved at temperatures around 350-390 °C, and crystal growth is confined to the relict structure of phase separation. At higher temperatures, a secondary crystallization process may be initiated. In that case, dominantly zinc and (Zn,Na) phosphates are precipitated by surface crystallization. In both glasses and glass ceramics, photoemission occurs at around 620-650 nm with a bandwidth of ~110 nm. Emission is red-shifted for increasing Mn²⁺ concentration, and blue-shifted for increasing ratio of SO₃:P₂O₅. Emission intensity is linearly increasing with increasing dopant concentration up to 3.0 mol.%, and no quenching effects could be observed for this concentration range. In the glassy state, only one type of emission centers could be detected, i.e. Mn²⁺ on octahedral Zn²⁺ sites. The position of the emission peak and shifts associated with changes in matrix composition and dopant concentration can clearly be attributed to changes in ligand field strength. As indicated by ESR and Raman spectroscopy, increasing Mn²⁺ content leads to further depolymerization of the glass lattice, and the character of bonds between Mn²⁺ and the lattice becomes increasingly ionic. Recrystallization of the glass and, hence, formation of a glass ceramic results in a significant increase of emission intensity. Then, dynamic emission spectroscopy reveals three different emission centers. On the basis of crystal field splitting, these are attributed to octahedrally coordinated Mn²⁺ in the residual glass phase and in crystalline phosphate and sulfate lattices, respectively. Excitation bands lie in the blue-to-green spectral range and exhibit strong overlap, and the optimum excitation range matches the emission properties of GaN- and InGaN-based LEDs. Noteworthy, in the same time as emission intensity increases with dopant concentration, the ratio between blue and UV excitation increases as well, what presents another significant advantage for application in WLED devices.

Compared to conventional solid-state synthesis of alternative materials, the glass ceramic route offers unique advantages with respect to composition, dopant concentration, formation of microspheres and recycling.