Introduction

Cerium doped glasses are attractive materials for a broad range of applications. Cerium ions can act as sensitizer in luminescent glasses as e.g. scintillator or laser glasses, they are widely used as sensitizer in photo-thermo-refractive glasses, they are often used to decrease photo-darkening in optical fibers or as decolorizing agent for container glass. Furthermore, most white emitting LEDs are based on Ce³⁺-doped yttrium-aluminum-garnet (YAG) phosphors which act as wavelength converter.

Recently Ce³⁺ was proposed as sensitizer ion for quantum cutting materials due to its high absorption cross section and strong fluorescence emission [1, 2]. Quantum Cutting (QC) is the emission of two or more low energy photons upon the absorption of one high energy photon. This effect attained great attention due to its potential utilization in the improvement of photonic materials. Efficient NIR QC in an UV excited photonic material will result in quantum efficiencies higher than unity and therefore increase the quantum yield of these materials. Possible applications are e.g. the improvement of solar cells or wavelength converters for optical sensors. Up to now mostly crystalline quantum cutting materials are described in literature, but lately S. F. Zou et al. have shown that QC is also possible in glasses [3]. To achieve QC, Ce³⁺-Yb³⁺ codoping in oxyfluoride glasses has been used in this case. For energy transfer between fluorophores, the transfer efficiency mainly depends on the spectral overlap of the emission spectrum of the sensitizer ion and the absorption spectrum of the acceptor ion [4, 5]. In principle this is also valid for energy transfer in QC materials [6], but here the self-convoluted absorption spectrum of the acceptor ion has to be used [7]. However, the spectral overlap between sensitizer and acceptor ion was not high in Ref. [3]. A further optimization of the spectral overlap should increase the efficiency of the quantum cutting material. This can be achieved by the variation of the chemical composition of the host glass since the transitions involved in the Ce³⁺ fluorescence are parity allowed 4f-5d electric dipole transitions which are strongly dependent on the local surrounding of the Ce³⁺ sites. This is not only the reason of the strong absorption and emission of Ce³⁺ but also allows the shift of the Ce³⁺ absorption and emission spectra over a very broad spectral range. An optimal QC energy transfer from Ce³⁺ to Yb³⁺ should be obtained if the Ce³⁺ emission wavelength matches with twice the energy of the Yb³⁺ absorption in the same host [6, 7], which is about 465 nm in aluminosilicate glasses [8]. Aluminosilicate glasses are very promising host matrices for utilization of this effect since they offer chemical stability over an exceptionally broad compositional range [e.g 9, 10]. Furthermore they offer a high chemical durability, high mechanical strength, high rare earth solubility, relatively small coefficients of thermal expansion and low production costs [e.g 8–12].

Besides the spectroscopic properties of the Ce³⁺ doped host material also the red-ox state of the cerium ions is crucial for its utilization as QC sensitizer. As a polyvalent ion cerium might occur in glasses as Ce³⁺ and Ce⁴⁺. The ratio of the concentrations of these two oxidation states mainly depends on the composition of the glass [13, 14], the oxygen activity of the melt which might also be affected by the furnace atmosphere [14–16] and the temperature of melting [14, 16, 18]. As in the case of all other polyvalent ions, also Ce³⁺ and Ce⁴⁺ form an equilibrium with the physically dissolved oxygen of the melt according to 4 Ce³⁺ + O₂ ↔ 4 Ce⁴⁺ + 2 O^2-. The equilibrium is shifted to Ce³⁺ with increasing temperature, decreasing basicity and the application of reducing melting conditions. In principle, the equilibrium constants of redox reactions can be determined by equilibration experiments [18] or, however, using electrochemical measurements. Smythe et al. presented a calibration method for the quantitative determination of Ce³⁺/Ce⁴⁺-ratios in alkali-aluminosilicate glasses using Ce M_4,5-edge X-ray absorption near edge structure (XANES) spectroscopy [19]. It was further tried to separate the Ce³⁺ and Ce⁴⁺ absorption bands by spectroscopic methods [17, 20–24]. This is difficult since Ce³⁺ and Ce⁴⁺ ions both show broad-band and largely overlapping near UV absorption. Furthermore, up to now a clear evidence of any Ce⁴⁺ fluorescence emission has not been found, which could help to separate the absorption bands via fluorescence excitation measurements. However, only one of these workgroups applied reducing and oxidizing melting conditions to enable a clear separation of the Ce³⁺ and Ce⁴⁺ absorption bands [21]. Since this work was done in phosphate and fluoride phosphate glasses of relatively low optical basicity, and low melting temperature only small amounts of Ce⁴⁺ could be obtained.

For effective Ce³⁺-Yb³⁺ quantum cutting, glasses of relatively high optical basicity have to be used. In these materials, the co-existence of Ce⁴⁺ which has an extremely strong and broad charge transfer transition would dramatically decrease the absorption of UV photons by Ce³⁺ ions. This would result in very low Ce³⁺ emission intensities and in low quantum yields even at high energy transfer rates to the acceptor ion. Therefore a detailed knowledge of the influence of the host matrix, the melting conditions and the overall cerium concentration on the spectral properties of Ce³⁺ and the Ce³⁺/Ce⁴⁺ ratio is needed. This paper presents a systematical study on the spectral properties of Ce³⁺ and Ce⁴⁺ in aluminosilicate glasses, which is of particular importance for the development of QC materials as well as for the research on photo-thermo-refractive glasses.

Experimental procedure

All glasses were prepared from high purity raw materials (Fe < 10 ppm, other contaminating metals < 0.5 ppm) SiO₂ (Sipur A1, Schott, Germany), Al₂O₃ (Ceralox, Condea Chemie, Germany), MgO (Merck, Germany), CaCO₃ (Merck, Germany), Na₂CO₃ (Merck, Germany), and BaCO₃ (VK Labor- und Feinchemikalien, Germany). The batches were melted in covered platinum crucibles (non-reducing melts) or corundum crucibles (reducing melts) at temperatures in the range from 1550 to 1650°C depending on the individual glass composition. After melting for 2 h, the samples were cast into brass moulds and transferred into a muffle furnace, preheated to temperatures in the range from 720 to 920°C. Subsequently, the cooling furnace was switched off and the samples were allowed to cool (cooling rate: approximately 3 K/min).

Starting with a magnesium aluminosilicate glass, different glass compositions of increasing optical basicities have been melted. Table 1 summarizes the chemical compositions, densities and calculated optical basicities of all prepared glasses. Two series of glass samples were prepared; in the first series the cerium concentration was kept constant at 1∙10¹⁹ ions per cm³ (about 0.04 mol% CeO₂) while varying the glass composition; for the second glass series the glass composition was kept constant (BaAS3510) while the cerium concentration was varied between 5∙10¹⁸ and 5∙10²⁰ ions per cm³. The samples were denoted according to their molar chemical composition. Glass samples were doped with cerium ions using CeO₂ as an additive to the raw materials.

Table 1. The compositions of all studied samples, their densities and calculated optical basicities (after: [28 and references therein]).

View Table | View all tables in this article

The density of each glass sample was measured using an AccPyc 1330 pycnometer (Micromeritics, Germany). Fluorescence emission and excitation spectra were measured using a fluorescence spectrometer RF-5301PC (SHIMADZU, Japan). For these investigations, the samples were cut and polished (thickness 10 mm). Cerium excitation spectra were recorded at the wavelength of the emission peak maximum while emission spectra were obtained using the wavelength of the excitation peak maximum (spectral resolution: 0.2 nm). Optical absorption spectra were recorded using a commercial double beam spectrometer UV-3102PC (SHIMADZU, Japan) in the wavelength range of 190–3,200 nm with an error of about 1%. For these measurements, polished samples with a thickness of 0.1 to 1 mm were used.

Results and discussion

The effect of glass composition

Figure 1 shows the fluorescence emission and excitation spectra recorded from various glass compositions all doped with a cerium concentration of 1∙10¹⁹ ions per cm³ which corresponds to about 0.04 mol% CeO₂ depending on the density of the glass. All curves show a pronounced maximum. The peak emission and excitation of the magnesium containing glass MgAS2020 occur at the smallest wavelengths (385 and 340 nm, respectively) while those of the barium containing glass BaAS3510 are observed at the largest wavelengths (445 and 365 nm, respectively). The intensity of the emission spectrum is largest in the magnesium containing sample MgAS2020, and decreases within the series CaAS3020, CaBaAS251013, CaNaAS251015. Within this series, a continuous decrease of the peak intensity with increasing wavelength of the peak maximum is observed. The ternary barium aluminosilicate glass is observed at the largest wavelength, however, it shows a clearly more intense fluorescence emission than the glasses CaBaAS251013 and CaNaAS251015. Obviously, glasses of larger optical basicity show emission at lower wavenumbers / larger wavelengths (see Table 1). The optical basicity is a phenomenological measure of the electron donor power of an oxidic compound. It is closely related to the polarizability of the material and can be calculated from experimentally determined or theoretically derived basicity values of the individual oxides [27]. For the calculation of the theoretical basicities of the glass compositions the averaged partial basicity values of Duffy and Lebouteiller/Courtine given in Ref. [28]. have been used. The observed dependency on the chemical composition of the glasses can be explained as follows: An increasing optical basicity of the glass composition is related to an increasing polarizability at the local rare earth sites. According to Dorenbos [25, 26] an increasing polarizability results in an increasing shift of the 5d¹ energy level of Ce³⁺ to lower energies which is equivalent to a red-shift of the absorption and emission spectra. In particular the polarizability is reported to increase in the order Mg²⁺ < Ca²⁺ < Na⁺ < Ba²⁺ [25] which exactly resembles the order of the optical basicities of the respective oxides [27, 28]. The splitting of the 5d¹ energy level has a further influence on the spectral position of the absorption/emission peaks. Stronger splitting results in an additional red shift of the spectra. The splitting is reported to be mainly dependent on the Ce³⁺ coordination number and Ce-O bond length but independent of the shift of the 5d¹ level [25]. Hence, varying the glass composition in order to increase the optical basicity results in the desired shift to longer wavelengths of the Ce³⁺ emission as seen in Fig. 1. Furthermore, it can be seen that the Ce³⁺ emission spectra consist of two components (non-symmetrical shape of the emission spectra). This effect is due to the two 4f¹ energy levels (²F_5/2 and ²F_7/2) which are the final states of the fluorescence transition 5d¹→ 4f¹. ²F_5/2 is the ground state while the ²F_7/2 level is located at around 2,000 cm⁻¹ [29] which corresponds very well to the spectra.

 

Fig. 1 Ce³⁺ excitation (left) and emission spectra (right) of various cerium doped aluminosilicate glasses of relatively high optical basicity. The grey short-dashed curve is the Yb³⁺ absorption spectrum in the MgAS2020 glass at doubled energy.

Download Full Size | PDF

Figure 1 additionally includes an energy-doubled absorption spectrum of Yb³⁺ doped MgAS2020 glass (grey short-dashed line, doping concentration 1∙10²⁰ ions per cm³). The peak is due to the ground state (²F_7/2) absorption of the 3-fold splitted ²F_5/2 level of the Yb³⁺ ion. Since this transition is an f-f transition it is – in contrast to the f-d transitions of Ce³⁺ – hardly affected by the glass composition [8]. Its distinctive tall peak is located at about 980 nm in aluminosilicate glasses, the broader subpeak is observed at a smaller wavelength of about 930 nm [8] (in Fig. 1 displayed at 490 and 465 nm, respectively). The largest overlap between the Ce³⁺ emission and the energy-doubled Yb³⁺ absorption in principle may be obtained by the high basicity BaAS3510 glass which shows the Ce³⁺ emission peak at around 445 nm. However, by melting under air only a relatively small fluorescence emission intensity is obtained with this glass composition.

The notable decrease in Ce³⁺ fluorescence intensity of glasses with higher optical basicity is in agreement with the literature where a shift of red-ox equilibria to the higher oxidation state with increasing basicity has often been described [13, 14] (see also introduction section). Therefore under constant melting conditions, the Ce⁴⁺ concentration should be higher in glasses of higher optical basicity. Due to the large spectral overlap of the absorption spectra of Ce³⁺ and Ce⁴⁺ and the strong charge transfer absorption of Ce⁴⁺ more radiation is absorbed by Ce⁴⁺ which does not result in fluorescence emission in these glasses.

The effect of reducing melting conditions

In order to further evaluate the effect of Ce⁴⁺ and to increase the emission intensity of Ce³⁺ in high basicity glasses, three different reduction techniques were applied in this work: melting with argon bubbling for 1 h (curve b in Fig. 2), melting with argon bubbling for 1 h under argon atmosphere (curve c in Fig. 2) and replacing a small part of the aluminum oxide by metallic aluminum powder in the batch (curve d in Fig. 2). For the latter, 0.5 mol% of Al₂O₃ were replaced by 1 mol% Al-powder. The glass composition of the highest optical basicity BaAS3510 and a cerium doping concentration of 1∙10¹⁹ cm⁻³ were used for these experiments. In Fig. 2 the fluorescence excitation and emission spectra of the reduced samples (curves b, c, d) are compared to the unreduced sample (curve a). As expected, reduction of the melt results in an increased Ce³⁺ fluorescence intensity. Obviously, argon bubbling under air only slightly decreased the Ce⁴⁺ concentration. Curves c and d (bubbling with argon under argon atmosphere and reducing by Al-powder respectively) show maxima which are in an almost perfect agreement. This indicates an almost complete reduction of Ce⁴⁺ to Ce³⁺ in both cases. Overall, the Ce³⁺ fluorescence intensity increased by a factor of about 2.4 in these two samples in comparison to the unreduced sample.

 

Fig. 2 Excitation (left) and emission spectra (right) of barium aluminosilicate glasses doped with 1·10¹⁹ cerium ions/cm³ prepared under different melting conditions a) unreduced, b) reduced by Ar bubbling, c) reduced by Ar bubbling in Ar atmosphere, and d) reduced by adding aluminum powder to the batch.

Download Full Size | PDF

Figure 3 shows the absorption spectra of BaAS3510 glass samples doped with 1∙10¹⁹ cerium ions per cm³, melted under reducing (long-dashed line), not reducing (full line) and oxidizing conditions (short-dashed line). For oxidation, the melt was bubbled with oxygen for 2 hours and subsequently casted without refining. An oxidized sample (oxygen bubbling) and a reduced sample (argon bubbling) of this series are shown in Fig. 4(sample thickness 1 cm, sample size about 3 × 2 cm). Although the doping concentration is very low, the oxidized sample shows a clear yellow/brownish coloring while the reduced sample is clear and colorless.

 

Fig. 3 UV-VIS absorption spectra of cerium doped BaAS3510 glasses (1·10¹⁹ ions per cm³) prepared using reducing, non-reducing and oxidizing melting conditions. The spectra of the undoped samples (not shown) have been subtracted from the respective spectra of the cerium doped samples.

Download Full Size | PDF

 

Fig. 4 An oxidized (oxygen bubbling, left) and a reduced BaAS3510 sample (argon bubbling, right). Sample thickness 1 cm, sample size about 2 × 3 cm. Overall cerium doping concentration 1·10¹⁹ ions per cm³. The oxidized sample clearly shows a yellow/brownish coloring.

Download Full Size | PDF

To separate the optical absorption due to Ce³⁺ and Ce⁴⁺ ions from the absorption of impurities, the spectra of reduced, oxidized and unreduced blank glass samples have been measured and subtracted from the respective spectra of doped samples in Fig. 3. At a wavenumber of about 31,000 cm⁻¹, the most intense Ce³⁺ absorption peak (4f¹ → 5d¹) is observed which is clearly visible for the reduced and not reduced samples. For the oxidized sample, this peak is much smaller. However, since the 5d¹ energy level of Ce³⁺ is split into 5 sublevels [21, 29–31], four more Ce³⁺ absorption bands should be visible. On the other hand, it is well known that the charge transfer transition of Ce⁴⁺ is also located in this spectral range (see spectra of not reduced and oxidized samples). Therefore, it is difficult to decide whether the spectrum of the reduced sample solely represents the absorption of Ce³⁺ or a superposition of transitions of both Ce³⁺ and Ce⁴⁺ ions in the range from 47,000 to 31,000 cm⁻¹. In Fig. 2 Ce³⁺ fluorescence spectra of two strongly reduced samples are shown which are in very good agreement, although they had been produced by completely different methods. This is a hint that the maximum possible amount of Ce³⁺ has already been achieved in theses samples. To further evaluate this assumption, samples of different cerium concentrations have been prepared with the aluminum powder reduction method. Up to a concentration of 2∙10²⁰ cm⁻³ all these spectra have almost the same shape (not shown). Samples of higher cerium concentration could unfortunately not be measured, due to the too strong absorption of Ce³⁺, even at a sample thickness of only 0.1 mm. From these measurements it can be concluded that the reduction potential of 1 mol% aluminum powder is sufficient to reduce the entire cerium concentration to Ce³⁺. In fact 1 mol% CeO₂ corresponds to a concentration of about 2.5∙10²⁰ cerium ions per cm³ in this glass. It should further be mentioned that one mol% metallic aluminum should be sufficient to reduce 3 mol% CeO₂ (2 Al + 6 CeO₂ → Al₂O₃ + 3 Ce₂O₃), however, a part of the metal might also be oxidized e.g. by CO₂ from the used raw materials.

Ebendorff-Heidepriem and Ehrt [21] also reported a complete reduction of cerium to Ce³⁺ (1.5 and 2∙10¹⁹ cm⁻³) in phosphate and fluoride phosphate glasses by melting with the addition of sugar to the batch or by remelting the glasses in carbon crucibles under argon atmosphere. In both cases, melting temperatures of 1100 °C have been used. Johnston [16] reports that the percentage of Ce⁴⁺ in a sodium silicate glass could be reduced from 68 to 22% only by applying CO₂ atmosphere instead of melting in air at 1085 °C. A further hint that all of the cerium could be reduced by using the presented methods at low cerium concentrations can be found in Ref. [14]: Schreiber et al. measured the red-ox equilibria of different ions as e.g. Eu^2+/3+, Fe^2+/3+ and Ce^3+/4+ in aluminosilicate glasses at 1500 °C. It is stated that cerium has the highest reduction potential of these ions and therefore is much easier to reduce than e.g. europium. In an earlier work on the fluorescence properties of europium, we reported an almost complete reduction of Eu³⁺ to Eu²⁺ in borosilicate glasses by using similar, strongly reducing melting conditions [32]. Although the glass composition had a lower optical basicity in that case it might be a further indirect hint that all of the cerium could be reduced to Ce³⁺ in the present experiments. However, in order to further evaluate the measured absorption spectra, they were deconvoluted using curves of Gaussian shape.

Figures 5, 6 and 7 represent three examples of deconvolution of absorption spectra of a reduced, a not reduced and an oxidized cerium doped sample respectively (BaAS3510, 1∙10¹⁹ Ce^3+/4+ cm⁻³, also displayed in Fig. 3). Since it can be assumed that most of the cerium is reduced to Ce³⁺ in the reduced sample (Fig. 5) it has been tried to fit 5 Gaussian peaks to the spectrum since the 5d¹ level is split up into 5 components as explained earlier. The positions of the peaks are about 30,000, 30,700, 32,000, 37,200 and 45,700 cm⁻¹. Especially the relative spectral position of the three peaks of lower energies corresponds well to literature data, e.g. 33,000, 34,500 and 36,400 cm⁻¹ in YAlO₃ [29]. The peak positions of the two peaks observed at higher energies of about 37,200 and 45,700 cm⁻¹ still fit relatively well (41,900 and 45,500 cm⁻¹ [29]). Similar Ce³⁺ f-d band positions are also reported for borax glasses [30]. In phosphate and fluoride phosphate glasses, the band positions are reported to be more equally distributed between 34,000 and 49,000 cm⁻¹ [21]. However, Ce³⁺ band positions and line widths can be much different for different host materials [21, 25, 26, 29, 30, 33]. Furthermore, especially for the high energy band, the reliability of the measurements and fits is relatively low since in this spectral range also the absorption of the glass matrix is observed. In addition, here the charge transfer absorption of Ce⁴⁺ is located, which consists of 2 bands [21, 23]. Therefore, also a superposition of Ce⁴⁺ and Ce³⁺ absorption bands could be observed here, which cannot be separated clearly.

 

Fig. 5 Example of deconvolution of the optical absorption spectrum of a cerium doped BaAS3510 sample (1·10¹⁹ ions per cm³) melted under reducing conditions. All five peaks are attributed to Ce³⁺.

Download Full Size | PDF

 

Fig. 6 Example of deconvolution of the optical absorption spectrum of a cerium doped BaAS3510 sample (1·10¹⁹ ions per cm³) melted under not reducing conditions. The five short-dashed grey peaks are attributed to Ce³⁺ (see Fig. 5), while the two long-dashed black peaks represent the two-band Ce⁴⁺ absorption.

Download Full Size | PDF

 

Fig. 7 Example of deconvolution of the optical absorption spectrum of a cerium doped BaAS3510 sample (1·10¹⁹ ions per cm³) melted under oxidizing conditions. The five short-dashed grey peaks are attributed to Ce³⁺ (see Fig. 5), while the two long-dashed black peaks represent the two-band Ce⁴⁺ absorption.

Download Full Size | PDF

Figure 6 shows the deconvolution of the absorption spectrum of the unreduced sample. As can be seen the absorption at wavenumbers above 35,000 cm⁻¹ strongly increased in intensity. To separate the absorption bands of Ce⁴⁺ and Ce³⁺ in this spectrum, the bands of Ce³⁺ obtained from the deconvolution in Fig. 5 have been added at the same spectral positions. Then two more peaks at higher energies have been added which resemble the 2 band absorption of Ce⁴⁺. While keeping the positions, relative intensities and shape of the five assumed Ce³⁺ peaks constant, it has been tried to fit the two additional Ce⁴⁺ peaks to the spectrum. This approach worked reasonably well. Under the assumption that the reduced sample (Fig. 5) contains solely Ce³⁺, a molar extinction coefficient of 807 l∙mol⁻¹∙cm⁻¹ for Ce³⁺ at a wavelength of 324 nm was calculated. Using this result, the deconvolution shown in Fig. 6 furthermore enables to calculate the extinction coefficient of Ce⁴⁺. According to the fit, the intensity of the Ce³⁺ peak decreased to 67% in comparison to the reduced sample. In turn this means that the Ce⁴⁺ concentration in the not reduced sample is 0.33∙10¹⁹ cm⁻³ and a molar extinction coefficient of 4,627 l∙mol⁻¹∙cm⁻¹ can be calculated for Ce⁴⁺ at 250 nm in this sample. In the same way the extinction coefficient of Ce⁴⁺ can also be calculated from the absorption spectrum of the oxidized sample (Fig. 7). Here, a Ce⁴⁺ concentration of 0.53∙10¹⁹ cm⁻³ and an extinction coefficient of 4,843 l∙mol⁻¹∙cm⁻¹ at 250 nm was calculated. These results are summarized in Table 2.The relative difference of less than 5% for these two calculations proves, that the assumptions that the reduced sample (Fig. 5) only contains Ce³⁺ and the Ce³⁺ extinction coefficient is about 807 l∙mol⁻¹∙cm⁻¹ are justified. The average value of these two molar extinction coefficient values is 4,735 l∙mol⁻¹∙cm⁻¹. However, it must be pointed out that the relative intensities of the two Ce⁴⁺ peaks are obviously not the same in the two deconvolutions shown in Figs. 6 and 7. In analogy, a slight difference in the peak positions of these two peaks in both deconvolutions is observed. From the absorption spectra reported by Ebendorff-Heidepriem and Ehrt [21], a molar extinction coefficient of about 710 l∙mol⁻¹∙cm⁻¹ at 290 nm can be deduced for Ce³⁺ in a phosphate glass, and a value of about 470 l∙mol⁻¹∙cm⁻¹ at 265 nm for a fluoride phosphate glass. Especially the value for the phosphate glass is fairly close to our result. However, the reported extinction coefficients for Ce⁴⁺ are much lower than in the glasses presented here (2,230 and 720 l∙mol⁻¹∙cm⁻¹ for phosphate and fluoride phosphate glass respectively). This might be a hint that not all cerium occurred as Ce³⁺ in those samples; most likely due to the much lower melting temperature of these glass compositions.

Table 2. Calculated molar extinction coefficients of Ce^3+/4+ at the respective peak absorption wavelengths in barium aluminosilicate glasses melted under different conditions

View Table | View all tables in this article

From Fig. 7, it is also obvious that the low energy absorption peak of Ce⁴⁺ extends far into the visible range resulting in the yellow/brownish coloring of this glass (left sample in Fig. 4). As a result, the following conclusions can be drawn: The extinction coefficient of Ce⁴⁺ can roughly be estimated to be about 6 times larger than the extinction coefficient of Ce³⁺ at their respective peak wavelengths. Due to the extremely broad Ce⁴⁺ absorption peak at around 37,000 cm⁻¹, the absorption spectrum of Ce⁴⁺ extends far into the visible range of the spectrum. Therefore the yellow/brownish coloring of the samples is due to the Ce⁴⁺ absorption although the peak positions of the Ce⁴⁺ bands are located at much higher energies than the Ce³⁺ absorption.

The effect of cerium concentration

Excitation and emission spectra of samples with different cerium concentrations are shown in Fig. 8.All of these samples are of the glass composition BaAS3510 and have been melted using 1 mol% metallic aluminum powder as reducing agent replacing 0.5 mol% of Al₂O₃ in the batch. With increasing cerium concentration, the intensities in the spectra increase while the peak position is slightly shifted towards lower wavenumbers / higher wavelengths. The spectral shift indicates an increase of the optical basicity with increasing cerium concentration.

 

Fig. 8 Ce³⁺ excitation (left) and emission spectra (right) of BaAS3510 glasses with different cerium concentrations between 5·10¹⁸ and 5·10²⁰ cm⁻³.

Download Full Size | PDF

The intensity increase up to a concentration of 2∙10²⁰ ions per cm³ can be attributed to the increasing Ce³⁺ concentration and therefore to a higher quantity of radiation centers. Nevertheless, the intensity increase is much lower than the respective Ce³⁺ concentration increase. Furthermore the above mentioned shift of the spectra due to the increasing optical basicity is much too large to be solely attributed to the relatively small increase in cerium concentration. Possibly these effects are due to clustering of the cerium ions in these glasses which would result in an increased local basicity (spectral shift) due to the high optical basicity of cerium oxide (0.65 [28]) and increased concentration quenching (non-proportional intensity increase) at the cerium sites. However, further work is needed to fully understand this effect. A further increase in the cerium concentration to 5∙10²⁰ ions per cm³ (about 2 mol% CeO₂) results in a drastic decrease of the emission intensity. At this point, the concentration of cerium exceeds the concentration of metallic aluminum in the batch necessary to reduce all Ce⁴⁺ and hence a fairly high concentration of Ce⁴⁺ is still present. This is supported by the color of the glass samples. While the samples with low cerium concentrations are colorless, samples of higher cerium concentration tend to be yellow/brownish. The sample with 5∙10²⁰ cerium ions per cm³ even appears light brown. The coloring is due to the broad charge transfer absorption band of Ce⁴⁺ which extends far into the visible part of the spectrum as shown earlier. Concentration quenching might play an additional role in the intensity decrease at cerium concentrations above 2∙10²⁰ ions per cm³.

Conclusion

The effect of the glass composition, optical basicity and melting conditions on the Ce³⁺/Ce⁴⁺ ratio and the spectral properties and intensity of the Ce³⁺ fluorescence was studied in aluminosilicate glasses. In samples of high optical basicity, the Ce³⁺ excitation and emission peak positions shift towards longer wavelengths while the Ce³⁺/Ce⁴⁺ ratio decreases. Increasing amounts of Ce⁴⁺ can efficiently hinder the excitation of Ce³⁺ due to their broad and very intensive charge transfer absorption. The application of strongly reducing melting conditions can counteract this effect. High intensity blue emission matching the spectroscopic requirements for potential quantum cutting in Ce³⁺/Yb³⁺ codoped glasses could be achieved with a barium aluminosilicate glass. The molar extinction coefficients of Ce³⁺ and Ce⁴⁺ at their respective absorption peaks were calculated to be around 807 and 4,735 l∙mol⁻¹∙cm⁻¹, respectively.