1. Introduction

Glass-ceramics (GCs) [1] is a polycrystalline material that can be prepared by controlled crystallization of glass. It is the material that consists of one or more amorphous and crystalline phases [2,3]. The new nanocrystal structures grow directly in the original glass phase and at the same time change the composition and properties of the optical material. It makes it possible to combine the properties of the amorphous and crystalline phases [4]. GCs have a major benefit over their amorphous counterparts in several respects. For example, these materials have high electrical and thermal resistance, and high mechanical strength. The development of micro/nanosized crystallites with the chemical composition of the host material and the dopant is quite possible during heat treatment. The crystallization process of glasses does not affect the material transparency for UV-visible spectral range, provided that the size of the crystallites formed in the material is properly controlled. As a rule, their size should be less than the wavelength of the incident electromagnetic radiation [5,6].

The main factors in the development of GCs are the glass composition. The composition of glass-ceramic phases (crystal types and crystal-to-glass ratio) is responsible for many physical and chemical properties, including thermal and electrical characteristics, chemical resistance and hardness. In many cases, the properties of the initial glass can be tailored for ease of manufacture while modifying the properties of the glass-ceramic for a particular application.

Transition metal ions are of great interest due to the need to develop new compositions of GCs that can be used in technological applications. A characteristic feature of transition metals is due to their ability to form compounds, in which a partially filled d-shell is retained [7]. Moreover, the phenomenon of variable valence, which is inherent in these elements to a greater extent than others (except for some actinides), is due to unusual relationships that exist between successive ionization potentials of d-electrons.

Among the various transition metal ions, the Mn ion greatly influences the physical characteristics, including the spectral and luminescent properties of the material. Under normal conditions, these ions exist in glasses and glass-ceramic materials in oxidation states +2 and +3 [8,9]. The +4 and +5 oxidation states of the manganese ion are also fairly common [10]. The simultaneous presence of the Mn ion in different valence states in both crystalline and amorphous materials leads to nonstandard changes in optical characteristics.

So far, variable composition phosphors activated with Mn⁴⁺ ions have been reported [11–16]. Interest in such systems is due to the fact that, occupying an octahedral or distorted octahedral environment Mn⁴⁺ ions with the 3d³ electronic configuration have wide absorption bands in the near UV and blue spectral region due to ⁴A_2g → ⁴T_1g and ⁴A_2g → ⁴T_2g transitions and luminescence (620–750 nm) due to the ²E_g → ⁴A_2g transition. Radiation in the region from 620 to 750 nm coincides with the absorption spectra of plant pigments chlorophylls a and b, phytochromes P_R and P_FR and can be adapted in combination with LEDs for supplementary illumination and improvement of plant growth [14,17]. The particle size and concentration of the Mn⁴⁺ dopant also have a great influence on the value of the external luminescence quantum yield. The efficiency of radiation conversion can be improved by optimizing the synthesis methods and by varying the process conditions.

However, the most effective red phosphors doped with tetravalent manganese are currently considered oxygen-free [18]. The only oxygen-containing crystalline phosphors with manganese that can compete with them in terms of luminescence parameters are those based on germanium oxide [19]. Since germanium oxide is a well-known glass former, which makes it possible to obtain glasses of good quality, it was interesting and promising in terms of forming optical elements to obtain luminescent germanate glass-ceramics, activated with Mn⁴⁺.

At present, there is a couple of works devoted to the luminescent properties of glass-ceramics obtained by bulk crystallization of the alkali-germanate glass with Mn⁴⁺ [20,21]. Also there are several studies on the luminescence of Mn⁴⁺ in crystalline germanate phosphors with various alkali ions [22–25]. These systems had in common that germanium ions form GeO₆ and GeO₄ structures [22]. When embedded in the structure, manganese ions were situated in Ge(Mn)O₆ octahedrons connected via alkali metal. The main difference between the luminescence of glass-ceramics and crystalline phosphors was in the number of luminescence bands (the presence of inhomogeneous broadening was obvious for glass-ceramics) and in the quantum yield (in crystalline phosphors of a similar composition, it was higher: 80% versus 44%) [21,25].

A comparison of the materials showed that the Mn⁴⁺ luminescence was almost the same due to the similar crystal symmetry of the environment. In all articles, the luminescence and photoexcitation of manganese corresponded to classical transitions of the 3d³ ion in an octahedral environment [22–25]. Adachi referred such phosphors to the type O-A with clearly resolved Stokes and anti-Stokes counterparts on the lower and higher photon [18]. The main difference in the experimental results was the luminescence quantum yield, the maximum values were shown by materials with mixed Li-Na alkali ions. In general, appropriate ions such as Ca²⁺, Rb⁺, Li⁺, or B³⁺ can be co-doped with Mn⁴⁺ ions to improve the photoluminescent properties by improving micromorphology, reducing the number of impurity phases, or modulating the matched structure etc. When synthesizing glass-ceramics with two or more alkali ions, it is not always possible to predict which crystalline phase will prevail in the composition. Therefore, in this work, we set the goal of comprehensive study the effect of the alkali ion type on the luminescent properties of manganese ions in germanates glass-ceramics.

2. Experimental section

Glass/glass-ceramics preparation procedure: A total of 21 glasses were synthesized with the following composition of the batch: xRO₂– (10-x) Li₂O – 89.75 GeO₂ – 0.25 MnO₂ mol.% (where R = Na, K, Rb, Cs, x = 0; 2.5; 5; 7.5) (for more details please refer to Table S1). Reagents such as Li₂CO₃, K₂CO₃, Na₂CO₃, Rb₂CO₃, Cs₂CO₃, GeO₂ and MnO₂ (manganese was introduced already in the desired valence form) with a purity of reagent grade and higher. A glass batch was weighed with 0.001% accuracy and then thoroughly mixed. The synthesis of the initial glass was held by a standard melt-quenching technique in alumina crucibles for 30 min at 1170°С. After that, the melt was poured on a steel mold at room temperature and inertially annealed from 490°C in the muffle furnace. Manganese oxide was introduced into the glass by equimolecular substitution of germanium oxide. All glass compositions contained 10 mol.% of alkali oxides in various ratios. The glass-ceramics was synthesized using bulk crystallization method of a glass host via single-stage heat treatment in a 560-730°C temperature range for 2-5 hours.

Characterizations: Glass transition and crystallization temperatures were determined using an STA 449F1 Jupiter (Netzsch) differential scanning calorimeter (DSC) [26] with an accuracy of ±10°C in a temperature range of 200-800°C, and a scanning rate of 10°/min.

The studies of glass chemical composition were performed by X-ray photoelectron spectroscopy (XPS). The XPS analysis was done using a Combined Auger, X-ray and Ultraviolet Photoelectron spectrometer Thermo Fisher Scientific ESCAlab 250Xi with monochromatic AlKα radiation (photon energy 1486.6 eV). System of electron-ion charge compensation was used when recording the photoelectron spectra. Spectra were recorded in the constant pass energy mode at 100 eV for survey spectrum and 50 eV for element core level spectrum, using XPS spot size of 650 µm. A total energy resolution was no worse than 0.8 eV. The study was done at ambient temperature in UHV with pressure of the order of 1×10⁻⁹ mbar.

The polythermal method for studying the crystallization ability was carried out in an air atmosphere in a tubular horizontal furnace with a temperature gradient from 350 to 640°C.

UV-Vis-NIR absorption spectra were recorded using a Lambda 650 two-beam spectrophotometer (Perkin Elmer) in the 300-900 nm range, with a step of 0.2 nm and an integration time of 1 s.

X-ray diffraction patterns were obtained using a Rigaku Ultima IV X-ray diffractometer (Japan). The radiation from a copper anode with λ (CuK_α) = 1.5418 Å was used. The radius of the goniometer was 285 mm. The X-ray image was taken in the 2θ/θ angle range from 14° to 135° in the Bragg-Brentano geometry. The measurements were carried out using a CuK_β filter. In the experiment, the voltage across the tube was 40 kV, the current was 40 mA, and the output power was 1.6 kW. The scanning speed along 2θ was 0.5°-1°/min. The diffraction reflections were interpreted using the ICDD PDF-2 diffraction database. For the calculation of the X-ray diffraction patterns, the positions of the diffraction peaks were determined, and the relative integral intensity was calculated. The interplanar distances were calculated using the Wolfe-Bragg formula. Crystal sizes were calculated by the Williamson-Hall method, which is an improved classical Scherrer formula corrected for microstrain based on position and intensity at half-width (FHWM) diffraction maxima.

An electron accelerator of the GIN-400 type was used for CL measurements. Cathodoluminescent (CL) properties were investigated using the electron pulses with duration at half-width about 12 ns with average energy of accelerated electrons is 250 keV generated by electron accelerator. The luminescence decay kinetics was recorded using an FEU-106 photomultiplier tube via an MDR-12 monochromator (spectral range 200–2000 nm, linear dispersion 2.4 nm/mm) and a LeCROY digital oscilloscope (350 MHz). The CL integrated spectra were measured with an AvaSpec-2048 optical fiber spectrometer (spectral range 340–1100 nm).

Absolute quantum yield was measured on an Absolute PL Quantum Yield Measurement System С9920-02G, -03G (Hamamatsu) consisting of PMA-12 Photonic multichannel analyzer with InGaAs sensor (200-950 nm range with 2 nm resolution), A10104-01 Integrating sphere unit, a Monochromatic light source L9799-01 with a 150 W Xenon light source and remote-controlled monochromator (250-950 nm range, bandwidth from 2 to 5 nm).

Photoexcitation and photoluminescence spectra were obtained by spectrofluorometer LS-55 (Perkin Elmer). The kinetics of luminescence decay were obtained using a spectrofluorometer; a built-in xenon lamp in an impulse mode was used as an excitation source. The luminescence level was registered by varying the time delay from the exciting pulse with a step of 0.5 µs. Based on this, the dependence of the luminescence intensity decay was plotted. The mathematical processing of the obtained kinetics was carried out in the Origin Pro software.

3. Results and discussion

3.1 Glass characterizations

3.1.1 Structural studies of alkali-germanate glass

The disadvantage of most glasses is that when manganese is introduced into the glass in a tetravalent form, during the synthesis it goes into a bivalent or trivalent state. Germanate glasses are promising for solving this problem, since the ionic radius of Ge⁴⁺ (0.53 Å) is close to the radius of Mn⁴⁺ (0.52 Å) [27]. Thus, the probability of maintaining manganese in a tetravalent state during glass synthesis increased.

The XRD characterization confined the amorphous nature of the initial glass structure. No sharp peak appeared in the spectra of all initial glass samples, showing the only presence of a broad halo around 2$\theta $ ≅20–35°. Figure 1(a) shows the results of DSC studies for compositions of the series 2.5R₂O – 7.5 Li₂O – 89.75 GeO₂ – 0.25 MnO₂ mol.% (where R = Na, K, Rb, Cs).

 

Fig. 1. (a) DSC results for mixed-alkali germanate glasses; (b) photo of the initial glasses after the crystallization ability studies by the polythermal method; (c) phase equilibrium diagram in Li₂O-GeO₂ system; (d) phase equilibrium diagram in Na₂O-GeO₂ system (the red arrows indicate the batch compositions synthesized in the work).

Download Full Size | PDF

For the pure lithium-germanate composition, the DSC data showed two exothermic regions with maxima at 559 and 630°С and one endothermic region with a peak at 1031.5°C. According to the Li₂O-GeO₂ phase equilibrium diagram in the region of high-germanium compositions [28], the formation of the following compounds with the corresponding melting points should be observed: Li₂O*7GeO₂ - 1033°C, 3Li₂O*8GeO₂ - 953°C (Fig. 1(c)). The composition of the batch corresponded to 3Li₂O-97GeO₂ (wt.%) – the composition with congruent melting at a temperature of 1033°C, which coincided with the position of the endothermic peak on the DSC curve.

When two alkalis were introduced into the glass composition, the exothermic and endothermic regions shifted towards lower temperatures compared with mono-alkaline compositions. When lithium was co-doped with another alkaline ion in the Na-K-Rb-Cs series, the exothermic regions shifted towards higher temperatures, and the endothermic regions - towards lower temperatures (Table S2). According to the phase equilibrium diagram of the Na₂O-GeO₂ system (Fig. 1(d)), in the region of high GeO₂ content, the compound Na₂O*4GeO₂ or 2Na₂O*9GeO₂ should crystallize according to Ref. [29]. The eutectic between this compound and GeO₂ accounted for 94.5 wt.% GeO₂ and melted at a temperature of 950°C. Since in our case the concentration of germanium oxide was higher, we should expect the precipitation of a mixture of crystals in the material. In the K₂O-GeO₂ system, in the range of the compositions used, a mixture of GeO₂ and K₂O*4GeO₂ crystals should be precipitated (Fig. S1) [30], in the Rb₂O-GeO₂ system: in addition to GeO₂, Rb₂Ge₇O₁₅ and Rb₂Ge₈O₁₇ should be precipitated, in the Cs₂O-GeO₂ system: Cs₂Ge₆O₁₃ crystals were expected [31].

The analysis of the glass chemical composition by the XPS method showed that, despite the synthesis in a closed crucible, the components with alkaline ions volatilize at least three times compared to the composition of the batch. The concentration of manganese oxide is halved (Table S3, Fig. S2). Thus, we can say that the real composition of the glass matrix was close to 3 R₂O-97 GeO₂ mol. %, and during the crystallization of the glass, a mixture of crystals should precipitate in it, which was confirmed by the XRD method.

In addition to the DSC, an express analysis of the crystallization ability of alkali-germanate compositions was carried out by the polythermal method in a gradient furnace. It did not give exact temperatures of phase transitions; however, it made it possible to quickly assess the prospects of the composition in order to obtain glass-ceramics (Fig. 1(b)).

3.1.2 Absorption studies of alkali-germanate glass

Figure 2 shows the absorption spectra of the initial glasses where dominated the Mn³⁺ bands associated with the transitions from the ground ⁵E_g level to the Jahn-Teller splitted sublevels of ⁵T_2g level in 3d⁴ ion according to Tanabe-Sugano diagram [32,33]. In some works were demonstrated more than two absorption bands in Mn³⁺ - series of ⁵B₁ (⁵E) → ⁵E (⁵T₂), ⁵B₁ (⁵E) → ⁵B₂ (⁵T₂) transitions in green region, ⁵B₁ (⁵E) → ⁵A₁ (⁵E) transition in red and near IR region, and also absorption transitions to ³E (³H), ³T₁ (³H), ³T₂ (³F) in blue and near UV region [34,35].

 

Fig. 2. (a) Absorption spectra of Mn-doped initial glasses with different alkaline ions (inset: photo of samples under a UV lamp); (b) absorption spectra of Mn-doped initial glasses with a decrease in the Li₂O content and an increase in the Rb₂O content; (c) 3d⁴ Tanabe-Sugano diagram; (d) Jahn-Teller splitting of ⁵D level in 3d⁴ ions.

Download Full Size | PDF

Based on the presence of level splitting, the absorption spectra of Mn³⁺ ions were decomposed into components, which showed the presence of two components: one is in the region of 19000–21000 cm^-1, and the second one in the region of 13000-15000 cm^-1. It is known that the splitting between the ⁵E and ⁵B₂ (⁵T₂) levels is within 2000 cm⁻¹ [36]. In our case, the second component is much further away. At the same time, in different previous studies, the location of the band corresponding to the ⁵B₁ (⁵E) → ⁵A₁ (⁵E) transition varied from 7200 to 15500 cm⁻¹ depending on the material structure (crystalline or amorphous) and the degree of the octahedral environment distortion of the ion [36–38]. Thus, the band with a maximum in 13000-15000 cm^-1 region can be attributed to a transition to ⁵A₁ sublevel of the strongly split ⁵E(D) ground state of pseudo-tetragonally Jahn-Teller-distorted sixfold-coordinated Mn³⁺ with d⁴ electron configuration.

Figure 2(a) shows that when replacing the second alkali ion (besides lithium) in the initial glasses, their absorption spectra had no changes. However, when decomposed into components, it was obvious that both absorption bands shifted with increasing radius of the second alkali ion: ⁵B₁ (⁵E) → ⁵E (⁵T₂) band shifted from 482 to 485 nm, ⁵B₁ (⁵E) → ⁵A₁ (⁵E) band shifted from 690 to 674 nm (Table S4). The greater the difference in the ionic radii of alkali ions, the stronger the distortion of the octahedral environment of Mn³⁺ ions, the greater the Jahn-Teller splitting, and the closer the ⁵E (⁵T₂) and ⁵A₁ (⁵E) levels become.

The redshift of the ⁵B₁ (⁵E) → ⁵E (⁵T₂) absorption band with equimolar substitution of an alkali ion in the Li-Na-K-Rb-Cs series was consistent with the Tanabe-Sugano diagram for 3d⁴ ions and suggested that lithium ions created a more highly symmetric environment around manganese ions, and cesium ions created the weakest crystal field. This trend was also maintained in the absence of lithium ions in the glass (in the presence of two other alkali ions) (Fig. S3(a)). Thus, a simple dependence was obtained: with an increase of the alkali ion radius, the environment symmetry of the manganese ion decreased, shifting the main absorption band to the low-energy region (from 478 nm in LiG to 520 nm in CsG, more details in Fig. S3(b) and Table S4).

3.1.3 Luminescent studies of alkali-germanate glass

Experiments to search for luminescence in initial glasses from photoexcitation have not been crowned with success (photo of the samples in the inset in Fig. 2(a)). This was also explained by the strong Jahn–Teller splitting, since the energy of the⁵B₁ (⁵E) → ⁵A₁ (⁵E) absorption transition practically coincided with the energy of the possible luminescent transition ³T₁ (³H) →⁵E (⁵D).

However, electron beam irradiation led to the appearance of a broad cathodoluminescence band with a maximum in the 650–660 nm spectral regions that can be corresponded to Mn²⁺ ions (Fig. 3(b)) [39]. The luminescence band location of Mn²⁺ can be affected by the concentration of ions (the higher the concentration, the longer the wavelength maximum) and by the degree of environment ordering, which can be characterized by the ligand crystal field strength Δ/B. Usually, the luminescence location of Mn²⁺ ions is much more sensitive to the environment ordering than the Mn⁴⁺ luminescence (it is worth comparing the slope of the ⁴T_1g (⁴G) level vs. $\varDelta $/B of Mn²⁺ and ²E_g (²G) level of Mn⁴⁺ on Tanabe-Sugano diagrams). However, in our case, the cathodoluminescence location had little changes while the alkali ion changed. This indicated a little change in the environment of manganese ions for two reasons: (1) in the series shown in Fig. 4(a), the content of only 2.5 mol.% of the glass composition changed, this might not be enough to manifest in luminescent properties; (2) the main contribution to the environment of manganese ions in mixed-alkali germanate glasses was made by lithium ions.

 

Fig. 3. The X-ray patterns of (a) 3Li1NaGC glass-ceramics after 560°C-temperature heat treatment, (b) 3Li1NaGC glass-ceramics after 735°C-temperature heat treatment, (c) 3Li1KGC and (d) 3Li1RbGC glass-ceramics after 560°C-temperature heat treatment for 5 hours.

Download Full Size | PDF

 

Fig. 4. (a) Absorption spectra of initial glasses with different alkali ions after electron beam irradiation; (b) Cathodoluminescence spectra of Mn-doped initial glass with different alkali ions; (c) absorption spectra of Mn-doped glass-ceramics with different alkali ions (inset: difference absorption spectrum of manganese ions without a glass host and its decomposition);(d) absorption spectra of Mn-doped glass-ceramics after electron beam irradiation.

Download Full Size | PDF

The electron beam served as a source of additional electrons, which led to the reduction of manganese ions from the 3 + state to the 2 + state (the same reduction effect was observed in crystalline manganese oxides) [40]. A visual comparison of the initial glasses’ absorption spectra before and after electron beam irradiation resulted in a long-wavelength shift. Decomposition of the spectra confirmed that the bands corresponding to the ⁵B₁ (⁵E) → ⁵E (⁵T₂) and ⁵B₁ (⁵E) → ⁵A₁ (⁵E) transitions in Mn³⁺ ions shifted by 30 nm on average. However, in addition to these absorption bands, a band appeared with a maximum in the region of 425–435 nm (23000-23500 cm^-1) (Fig. S4), which can be attributed to ⁶A_1g(S) → ⁴E_g(⁴G) transition in Mn²⁺ ions [20,38]. The absorption cross-section of trivalent manganese ions exceeded that of divalent manganese by 10 times, which meant that only a small part of the manganese ions had recovered from the 3 + to the 2 + state and therefore the Mn²⁺ cathodoluminescence intensity was low.

3.2 Glass-ceramics characterizations

3.2.1 Structural studies of alkali-germanate glass-ceramics

Of the 21 glass compositions studied, only in 9 glasses crystal phases precipitated, namely, in LiG glasses, the Li-Na series (3 glasses), 3Li1KG, 2Li2KG, 3Li1RbG, 2Li2RbG, 3Li1CsG. That is, in most cases, the nucleation of the crystalline phase occurred in glasses in which the content of lithium oxide was not less than the content of another alkali oxide. In the rest, the diffraction patterns showed the presence of only an amorphous halo, as in the initial glasses.

After isothermal treatment of LiG glasses at temperatures of both exothermic DSC regions, Li₂Ge₇O₁₅ orthorhombic crystals (LGO crystal) with a mean size of about 13 nm were precipitated in the matrix. X-ray diffraction studies could not show whether manganese was included in the crystalline phase. However, this was possible since germanium ions in Li₂Ge₇O₁₅ crystals were in the tetravalent state, and the ionic radius of Ge⁴⁺ coincides with the ionic radius of Mn⁴⁺.

Thus, during the crystallization of the glass matrix, it is quite probable that germanium ions in the crystal structure will be replaced by manganese ions. When tetravalent manganese is in an octahedral environment, it will induce radiation in the red region upon UV excitation, which corresponds to the ²E - ⁴A₂ transition.

In glasses with mixed Li-Na alkalis, during heat treatment in the low-temperature exothermic region (540–580°C), LiNaGe₄O₉ crystals with the mean size of 26.2 nm precipitate (Fig. 3(a)). During heat treatment in the high-temperature exothermic region (730–740°C), a mixture of LiNaGe₄O₉, Li₂Ge₇O₁₅ and GeO₂ crystals precipitated in the matrix with the mean sizes of 39, 26, 43 nm respectively (due to the intense precipitation of the latter, the exothermic region had a very small width and location at high temperature region) (Fig. 3(b)). The ratio of different crystals within the crystalline phase was approximately 1:1:1. With the predominance of sodium over lithium in the composition, the intensity of the reflections of LiNaGe₄O₉ crystals began to dominate over the reflections of Li₂Ge₇O₁₅ until they disappeared. In other glasses with mixed alkalis, the predominant phase was Li₄Ge₅O₁₂ (Fig. 3(c)). The mean size of the crystals precipitated in the 3Li1KGC composition (Fig. 3(c)) was 21 nm. The degree of crystallinity (the volume occupied by the crystalline phase in the material) of glass-ceramics depended on the temperature and duration of heat treatment. With an increase in both, the degree of crystallinity increased. For example, for the samples shown in Fig. 3(a), b and c (heat treatment temperature 560°C, duration 5 hours), the degree of crystallinity was 24, 72 and 15% correspondingly.

Lithium-free glasses demonstrated the absence of crystallization regions at temperatures under study and XRD curves showed only an amorphous halo. However, in this work, the maximum content of alkali oxides was 10 mol.%. When 20 mol.% oxides of alkaline elements were introduced into the glass composition, Na₂Ge₄O₉ and Na₄Ge₉O₂₀, K₂Ge₄O₉ and K₄Ge₉O₂₀, Rb₂Ge₄O₉ crystals precipitated in the case of sodium-germanate, potassium-germanate, and rubidium-germanate glass, respectively. At the same time, these compositions were located at the boundary (or even beyond the boundary) of the glass formation region, and the resulting glass-ceramics can hardly be called optical.

Due to the fact that the presence of a low-temperature crystallization region was characteristic for all glasses, and its temperature range was approximately the same (550–580°C), it was this region that was used for the synthesis of glass-ceramics.

3.2.2 Absorption studies of alkali-germanate glass-ceramics

The absorption spectra of glasses in which no crystals appeared after the heat treatment were subjected to decomposition by analogy with the initial glasses. The decomposition showed that the spectra also consist of two absorption bands of trivalent manganese. In this case, the location of the band maxima coincided with those in the initial glasses with an accuracy of 100 cm^-1 (in that spectral region, this was about 2 nm). Considering the overall large half-width of the bands and a slight absorption shift, we can conclude that in the absence of crystallization in the glass matrix, no other changes in the structure occurred.

In our case, most of the absorption spectrum was occupied by the exponential decay of the fundamental absorption edge of glass host. However, after heat treatment, the fundamental absorption edge has shifted, which indicated the appearance of a new absorption band of high intensity in the 300–350 nm region (curve LiG in Fig. S5(a) and curve 3Li1RbGC in Fig. S5(c)). Similar band can be seen in the excitation spectrum of glass-ceramics (Fig. 5(c)), corresponding to the ⁴A₂ (⁴F) → ⁴T₁(⁴F) transition in tetravalent manganese ions.

 

Fig. 5. (a) Photoluminescence spectra of Mn-doped glass-ceramics with different alkali ions $\lambda $_ex= 335 nm (inset: photo of samples under a UV lamp); (b) cathodoluminescence spectra of Mn-doped glass-ceramics with different alkali ions; (c) photoexcitation spectra of Mn-doped glass-ceramics $\lambda $_lum= 667 nm and its components; (d) practical application in illumination system for plant growth.

Download Full Size | PDF

Before the decomposition, we subtracted the fundamental absorption of the glass and the short-wavelength band and obtained difference spectra, an example of which is shown in the inset in Fig. 4(c). The resulting difference spectra were decomposed in wavenumbers and showed that the absorption of manganese ions consisted of two bands with maxima in the region of 453-480 nm (20000-22000cm-1) and 514-638 nm (15500-19500 cm^-1), in some compositions there was also a small band at 393-408 nm (24500-25500 cm^-1).

The absorption spectra of divalent manganese ions include an intense absorption band, corresponding to the ⁶A_1g (S) →⁴E_g(⁴G) transition, the position of which is highly dependent on the environment and can vary from 400 to 440 nm [34,41,42]. Since the possibility of obtaining manganese in the divalent form has already been shown in this matrix by the electron beam irradiation, the absorption band at 393–408 nm can therefore be attributed to Mn²⁺.

According to the Tanabe-Sugano diagram, Mn⁴⁺ (3d³) ions had two main allowed transitions ⁴A₂→⁴T₁ and ⁴A₂→⁴T₂ in the visible and near UV spectral regions (Fig. S5(d)), which usually correspond to the absorption bands with maximum at 330 nm and 460 nm [43]. It was not possible to resolve the absorption band at 330 nm due to the large absorption cross-section. Accordingly, the band visible in the decomposition with a maximum in the 480–453 nm region can be attributed to the ⁴A₂ (⁴F) → ⁴T₂(⁴F) transition in Mn⁴⁺.

After the nucleation of crystals in the glass matrix, it can be considered that manganese ions entered the intermediate or high crystal field (approximately in the middle of the Tanabe-Sugano diagram). For this reason, we can observe Mn⁴⁺ luminescence, and for the same reason, the absorption bands of the remaining Mn³⁺ ions will shift from their original positions in the initial glass (Table S4). The d⁴ diagram (Fig. 2(c)) shows that the⁵E (D) → ⁵T₂(D) transition increases. Presumably, the band in the region of 514–638 nm can be attributed to the ⁵B₁ (⁵E) → ⁵A₁ (⁵E) transition. However, in articles devoted to crystalline materials, this transition was more strongly shifted to the IR region. Therefore, it can be assumed that this band referred to the transition to another Jahn-Teller sublevel, for example, to ⁵B₂ (⁵T₂) (Fig. 2(d)).

When decomposing the absorption spectra of glass-ceramics with different alkalis (Fig. S5), it can be seen that in the Cs-Rb-K-Na-Li series, the cross-section of the trivalent manganese absorption band decreased due to the appearance of the tetravalent manganese absorption band. The presence of manganese ions in a tetravalent form was also confirmed by luminescent properties studies.

3.2.3 Luminescent studies of alkali-germanate glass-ceramics

The photoexcitation spectrum of alkali-germanate glass-ceramics with manganese consisted of four pronounced bands with maxima at 21692, 25974, 29850, 34250 cm^-1 (Fig. 5(c)), which corresponded to the ⁴A₂(F)→²A₁(G), ⁴A₂(F)→⁴T₁(F), ⁴A₂(F)→²T₂(G) and ⁴A₂(F)→⁴T₂(F) transitions according to Tanabe-Sugano diagram for 3d³ elements in octahedral crystal field. The photoluminescence spectra of glass-ceramics consisted of an intense band with a maximum in the region of 670 nm (Fig. 5(a)), which corresponded to the ²E_g(G)→⁴A_2g(F) transition and was consistent with numerous published data on “red” manganese-containing crystalline phosphors [18,19]. Of the 21 studied compositions, luminescence was observed only in 9 compositions (indicated in Table S5), where the crystalline phase was nucleated. This is consistent with the Tanabe-Sugano diagram: when manganese ions are in a weak crystal field, the ⁴T₂(F) level lies below the ²E(G) level, and upon single-photon excitation to the ⁴T₂(F) level, the electrons relax back to the ground level nonradiatively. When manganese ions are in a strong crystal field, the ⁴T₂(F) level lies above the ²E(G) level, and upon single-photon excitation to the ⁴T₂(F) level, electrons relax nonradiatively to the ²E(G) level, and then from it, radiatively to the ground level.

On the basis of the excitation bands determined during the decomposition process (presented in Table S5), the environmental parameters of manganese ions were determined: the Racah parameter of the electrostatic interaction B, the ligand crystal field strength Dq/B [44,45]. The series of Li-Na glass-ceramics showed the highest values of Dq/B = 3.43 ($\varDelta $=10Dq so on the x-axis in Fig. S5(d) the value will be $\varDelta $/B = 34.3). For other compositions, the ligand crystal field strength fluctuates around Dq/B = 3. This correlated with the values of the quantum yield and the luminescence lifetime.

The Tanabe-Sugano 3d³ diagram shows the degeneracy of electronic levels in a crystal field of octahedral symmetry as a function of the ligand crystal field strength Dq/B. Thus, substituting the obtained parameters into the diagram, we see that, indeed, manganese ions in all the glass-ceramics studied are in a strong crystalline field.

Lithium-sodium compositions had the best luminescence characteristics [21]. The maximum luminescence quantum yield was obtained for the 3Li1NaGC composition and amounted to 37% for HT at 560°C for 5 hours (insert in Fig. 5(a)). According to the results, the luminescence quantum yield was directly related to the Mn⁴⁺/Mn³⁺ ratio. Despite the fact that excitation with a wavelength of 335 nm was specifically used in order not to take into account the absorption of Mn³⁺, it was likely that a nonradiative energy transfer from tetravalent manganese ions to trivalent manganese ions occurred in the material, and the Mn³⁺ ions relaxed nonradiatively. And the higher the number of Mn³⁺ ions, the higher the probability of nonradiative energy transfer. However, this situation was probable only in the case of sufficiently close arrangement of manganese ions of different valences, so that if Mn⁴⁺ ions were located at lattice sites instead of Ge⁴⁺ ions, then Mn³⁺ ions should be embedded in neighboring interstices.

The photoluminescence decay curves for all luminescent glass-ceramics were approximated by two exponentials: fast and slow (the lifetime of both component is given in Table S5, decay curves in Fig. S6), as for most 3d³ ions in an octahedral environment (Mn⁴⁺, Cr³⁺) [43,46–49]. The two components are due to the presence of two closely spaced levels ²Е_g and ²T_2g (G) and an inclusion of the effects of Fano mixing between them [47] (which, yet, has been experimentally confirmed only for Cr³⁺ ions). The radiative ²E to ⁴A₂ transition of Mn⁴⁺ is spin- and parity-forbidden, but both selection rules can be weakened, causing the transition probability to increase and luminescence lifetime to decrease. The parity selection or Laporte selection rule can be attenuated by distortion of the coordination octahedron and mixing of Mn⁴⁺ 3d and O^2- 2p orbitals [49]. For crystalline fluoride phosphors with manganese 4+, the total lifetime can reach 8 ms [43]. For oxide phosphors (including germanate ones) doped with Mn⁴⁺, the lifetime values vary from 0.2 to 6 milliseconds [11,48,50,51] due to the presence of nonradiative energy transfer, with the short component usually being 2.3 3 times shorter than the long one, which is also observed in our case.

The cathodoluminescence of glass-ceramics did not differ much from photoluminescence. A part of the manganese ions could also be restored from the 4 + to the 3 + state, but the shape of the spectrum and its location were little affected, since in glass-ceramics we always had a mixture of 4 + and 3 + manganese ions, which was shown above.

Even though alkali germanate glasses had a strong tendency to crystallize, only lithium ions were the crystallization driver each time, and the crystalline phase consisted mostly of various lithium germanates. Because of this, glass-ceramics were successfully synthesized only in compositions with a lithium oxide content of more or equal to 5 mol.%.

With the introduction of two alkali ions with different ionic radii, the nucleation of crystals with lithium ions is more likely. In the case of a mixed Li-Na system, the radii of these ions differ little, which allows them to create a joint crystal structure of LiNaGe₄O₉. In the case of Li-K, Li-Rb, Li-Cs systems, alkali ions with large radii only introduce additional disorder into the system, thereby reducing the probability of manganese ions entering GeO₆ structures. Therefore, with an increase in the radius of the second alkali ion, the luminescent properties of glass-ceramics deteriorate. For additional verification of this theory, we synthesized the glasses of the K-Rb and Na-Rb systems. However, despite the numerous works on synthesizing crystalline materials of the K₂Ge₄O₉ and Rb₂Ge₄O₉ types [22,52,53], such crystals did not nucleate in a glass matrix. Nevertheless, these results suggest that a lithium-sodium germanate glass-ceramic host is favorable for stabilizing the emission for Mn⁴⁺ in the oxide systems for the further use in red light sources, namely, in phytolamps.

4. Conclusion

Here we presented the investigations of new germanate glass doped with Mn ions containing different alkali ions and their influence on the structural and spectral characteristics of the material. Out of the 21 studied compositions of the initial glass with different combinations of alkali ions pair, only 9 compositions became glass-ceramics. The basis of glass-ceramics was LiNaGe₄O₉ crystals in case of Li-Na germanate glass and Li₂Ge₇O₁₅ crystals for all other glass-ceramics. Crystals were nucleated in a glassy matrix via volume crystallization technique. The initial glass had no photoluminescence, since all manganese ions were in the 3 + state. The electron beam irradiation initiated the appearance of orange emission in the region of 650 nm from initial glass due to the reduction of a part of the manganese ions from the 3 + state to the 2 + state. The glass-ceramics possessed intense emission near 660-670 nm under two-band photoexcitation at 330 and 460 nm. After heat treatment the transition of manganese ions into the 4 + state occurred. The calculation of the ligand crystal field strength showed that Mn⁴⁺ ions were part of the crystalline phase in glass-ceramics. The maximum obtained luminescence quantum yield was 37% for Li-Na germanate glass-ceramics designated by two-exponential decay with lifetime equal to 1.29 msec. Thus, the obtained alkali germanate glass-ceramics with Mn⁴⁺ can be used as effective luminescent materials for the light source of deep-red radiation for plant growth cultivation.