1. Introduction

Nitride/oxynitride phosphors have been investigated extensively for use in white light emitting diodes (LEDs), because these materials exhibit high quantum efficiencies, high thermal and chemical stabilities, and low thermal quenching properties. For example, Eu²⁺-doped M₂Si₅N₈ (M: Ba, Sr) [1–6] and CaAlSiN₃ [7–9] are known to be red-emitting phosphors, while Eu²⁺-doped β-sialon [10] and (Sr,Ca,Ba)Si₂O₂N₂ [11] are used as green phosphors. It was recently reported that Sr₂Si(O_1-xN_x)₄ powders doped with Eu²⁺ ions exhibit red emissions [12, 13].

In addition to the above-mentioned nitride phosphors, BaSi₇N₁₀:Eu²⁺ has also been suggested as a cyan-emitting phosphor [14–17]. BaSi₇N₁₀ is a highly condensed ternary silicon nitride. It has a network structure composed of corner- and edge-sharing SiN₄ tetrahedra; the Ba²⁺ ions occupy the cavities within the network structure of [Si₇N₁₀]^2-. A Ba atom is coordinated by 13 nitrogen atoms, with the Ba–N distances being 2.91 – 3.53 Å, while the coordination number (CN) of Ba can be 12, if only the nitrogen atoms corresponding to Ba–N distances of approximately 3.50 Å are considered [14,18,19].

Earlier studies have reported that the photoluminescence (PL) spectrum of BaSi₇N₁₀:Eu²⁺ contains excitation and emission bands that peak at approximately 300 – 330 nm and 475–500 nm, respectively [14–17]. Further, both band types are located at wavelengths shorter (i.e., at higher energies) than those in the case of red-emitting Ba₂Si₅N₈:Eu²⁺ powders [1,2]. This difference in wavelengths is attributable to the fact that the crystal field splitting of the Eu²⁺ ions in BaSi₇N₁₀ is weaker. The CN of Ba (13 N atoms) in BaSi₇N₁₀ is approximately twice as large as those for the Ba(I) and Ba(II) sites in Ba₂Si₅N₈, which were 6 and 7, respectively, resulting in the volume of the polyhedron in BaSi₇N₁₀ being much larger and the average bond length being longer. BaSi₇N₁₀:Eu²⁺ powders have been prepared by various methods: the nitridation of each metal source and the solid-state reaction process [14], the sol-gel process [15], the nitridation of metal alloys [16], and the conventional solid-state reaction process using barium nitrates, Eu₂O₃, and Si₃N₄ [17].

In spite of the high potential of BaSi₇N₁₀:Eu²⁺ for use in white LEDs, there have been few reports on its luminescent properties, in contrast to the case for other nitride phosphors. In this study, we prepared BaSi₇N₁₀:Eu²⁺ powders through a simple and low-cost process that involved the use of a high-temperature graphite furnace. We also investigated the effects of the Ba content in the starting mixtures and the substitution of Ca²⁺ ions at the Ba²⁺ sites on the luminescent spectrum and intensity exhibited by the synthesized powders.

2. Experimental details

The (Ba,Ca)Si₇N₁₀:Eu²⁺ powders were synthesized using a high-temperature graphite furnace. BaCO₃ (High Purity Chemicals, 99.95%), CaCO₃ (High Purity Chemicals, 99.95%), Si₃N₄ (Ube Industry, Ltd., E10, 99.9%), and Eu₂O₃ (Grand Chemicals & Materials, 99.99%) were used as the starting materials. The mixtures of (x-y-z)BaCO₃–yCaCO₃–7/3Si₃N₄–0.5zEu₂O₃ were ball milled for 24 h, calcined, and fired at 1600 °C for 5 h in a nitrogen (99.999%) atmosphere using graphite crucibles. The crystalline phases of the synthesized powders were determined with an X-ray diffractometer (XRD, Rigaku MiniFlex II) using CuK_α radiation (λ = 1.5406 Å); Si powders were used as the reference. Rietveld refinement was performed using X'pert HighScore Plus (PANalytical B. V.). The particle morphologies and composition of the powders were analyzed using a field emission scanning electron microscopy (FE-SEM, JEOL JSM-6500F) with an energy dispersive spectrometry (EDS) attachment. The average atomic ratios were estimated by calibrating the measured ones using a reference powder consisting of BaCO₃, CaCO₃, Si₃N₄, Eu₂O₃. The nitrogen and oxygen contents (wt%) of the powders were measured using a nitrogen/oxygen analyzer (LECO, TC600). The PL spectra, quantum efficiencies (QEs), and thermal quenching were determined using by a PL system (PSI, Darsa 5000) with a 500 W xenon lamp as an excitation source. The external QE is the ratio of the number of photons of the emission to that of the excitation light irradiated on a phosphor, while the internal QE is the ratio of the number of photons of the emission to that of the excitation light absorbed by a phosphor. QEs were measured using an integrating sphere within the PL system, and BaSO₄ was used as a reference. The decay curves were measured with a time-resolved PL (TRPL) streak-scope system (Hamamatsu), using a 374 nm pulsed laser diode with a pulse duration of 65 ps. All the spectral measurements were performed at room temperature.

3. Results and discussion

3.1. Crystal structure

A number of BaSi₇N₁₀:0.1Eu²⁺ powders were synthesized using different amounts of BaCO₃ in the starting mixtures of (x-0.1)BaCO₃–7/3Si₃N₄–0.05Eu₂O₃. The XRD patterns of the synthesized powders are shown in Fig. 1(a). For x = 0.6, the XRD pattern was predominantly composed of peaks corresponding to BaSi₇N₁₀ (JCPDS #89-6751). A number of weak unknown peaks (◆), which were attributed to BaCO₃ being present in the starting mixtures in an insufficient amount, were also observed. For x = 0.8 – 1.2, a pure BaSi₇N₁₀ phase was observed; its XRD patterns did not changed much over the range of x values. In addition, the unknown peaks noticed in the case of x = 0.6 disappeared completely.

 

Fig. 1 (a) XRD patterns of the samples prepared with (x-0.1)BaCO_3, (b) experimental (red dots) and calculated (blue solid line) XRD profiles of Rietveld refinement for BaSi₇N₁₀:0.1Eu²⁺ powders corresponding to x = 0.8, and (c) a difference between the experimental and calculated data. Blue bars represent Bragg reflections.

Download Full Size | PDF

The variation in the Ba/Si atomic ratio (R_Ba/Si) of the prepared powders is shown in Fig. 2; the estimated R_Ba/Si values were higher than the theoretical ones, and the powders prepared with x = 0.8 had the same R_Ba/Si value (~0.13 ± 3%) as the theoretical one (~0.13) with the stoichiometric composition. The difference between the estimated and theoretical R_Ba/Si values increased gradually with an increase in x. This phenomenon can be ascribed to the formation of a volatile SiO phase in the reductive atmosphere resulting from the use of the graphite furnace and crucibles. The measured nitrogen content of the BaSi₇N₁₀:0.1Eu²⁺ powders corresponding to x = 0.8 was approximately 30.5 wt%; this value was almost the same as the one calculated (29.5 wt%), if one takes into consideration errors in measurement. On the other hand, the measured oxygen content was 0.26 wt%. These results demonstrate that the simple and low-cost process employed in the present study perfectly suited for the synthesis of BaSi₇N₁₀ powders. Possible reactions for the nitridation are speculated to be as follows: BaCO₃ decomposed into BaO and CO₂ (g) during calcination, and then, BaO reacted with C to afford Ca and CO (g); in this case, the C and CO partial pressures in the reacting chamber were assumed to be appropriately controlled. However, oxygen production from Eu₂O₃ was probably inevitable to a certain extent, resulting in the local incorporation of oxygen ions in the BaSi₇N₁₀ lattice, but it was negligible owing to a small amount of Eu₂O₃ in the starting mixtures. G. Anoop et al. prepared BaSi₇N₁₀:Eu²⁺ powders by the conventional solid-state reaction process using BaCO₃ at 1600 °C under the reduction atmosphere, but the detailed preparation conditions, the atomic composition, and the nitrogen content were not reported [17].

 

Fig. 2 Variation in the Ba/Si atomic ratios (R_Ba/Si) of the samples prepared with (x-0.1)BaCO₃.

Download Full Size | PDF

Rietveld refinement was performed for BaSi₇N₁₀:0.1Eu²⁺ powders corresponding to x = 0.8; XRD profiles are shown in Figs. 1(b) and 1(c), and crystallographic data are summarized in Table 1 compared with those of a reference (BaSi₇N₁₀, JCPDS #89-6751). A refinement of the site occupancy factor (SOF) confirmed that the refined composition of this sample was Ba_0.9Eu_0.1Si₇N₁₀, and that each crystallographic site was fully occupied. These findings indicated that 10 mol% Eu²⁺ ions were successfully substituted for the Ba²⁺ sites, and the BaSi₇N₁₀:0.1Eu²⁺ pure nitride with the stoichiometric composition could be obtained by the synthesis process of this study. The refined composition was just in accordance with that estimated by the results of EDS and N/O analysis. Ba_0.9Eu_0.1Si₇N₁₀ has a monoclinic system with a space group of Pc, while its lattice parameters a, b, and c slightly decreased when compared with those of BaSi₇N₁₀, because the ionic radii for Eu²⁺ and Ba²⁺ are evidently different: 1.35 Å (CN = 10) and 1.52 Å (CN = 10) for Eu²⁺ and Ba²⁺, respectively. However, the difference in a unit cell volume (V) and an axis angle β is insignificant. Selected interatomic distances (Eu–N) of Ba_0.9Eu_0.1Si₇N₁₀ are listed in Table 2.The shortest and longest distances of Eu–N are 2.910 Å and 3.490 Å, respectively, which are very nearly equal to or slightly shorter than those of Ba-N for BaSi₇N₁₀ (2.913 Å and 3.529 Å, respectively). In summary, the findings indicated that the changes in the lattice parameters and interatomic distances were insignificant, even though the smaller Eu²⁺ ions were substituted for the Ba²⁺ ions. This was due to the rigidity of the networks of BaSi₇N₁₀. Pilet al. suggested that, when compared with those of BaSi₇N₁₀, the high rigidity of the highly condensed [Si₇N₁₀]^2- networks of SrSi₇N₁₀ resulted in the slight change in the crystal parameters in spite of the smaller ionic size of Sr²⁺ (1.61 Å for Ba²⁺ and 1.44 Å for Sr²⁺, CN = 12) [20].

Table 1. Crystallographic data for BaSi₇N₁₀ and BaSi₇N₁₀:0.1Eu²⁺ (x = 0.8).

View Table | View all tables in this article

Table 2. Interatomic distances (Å) of Eu-N for Ba_0.9Eu_0.1Si₇N₁₀.

View Table | View all tables in this article

3.2. Photoluminescence of (Ba,Ca)Si₇N₁₀:Eu²⁺

The PL excitation (PLE) and PL spectra of the various powders are shown in Fig. 3(a) and 3(b). The PLE spectra ranged from 250 nm to 430 nm, and their intensities changed with x. Li et al. have suggested that the asymmetric PLE band noticed in the spectrum of BaSi₇N₁₀:Eu²⁺ is attributable to the overlapping Gaussian sub-bands at approximately 296, 332, and 363 nm [14]. The PLE spectra observed in the present study could also be decomposed into three sub-bands (A, B, and C), which are represented by dotted lines in Fig. 3(a); either the A-band or the B-band was the strongest one, depending on the value of x. Compared to those in the case of x = 0.6 – 0.8, the B-bands for x = 1.0 – 1.2 were significantly blue-shifted, while the A-bands were shifted slightly toward shorter wavelengths. These blue-shifts were related to the compositions of the powders.

 

Fig. 3 (a) PLE and (b) emission spectra of the samples prepared with (x-0.1)BaCO₃.

Download Full Size | PDF

The ratios of the intensities of the A- and B-bands (I_R = I_A-band/I_B-band) are shown in Fig. 4. For x = 0.6–0.8, the I_R values were greater than one, but decreased significantly for x = 1.0 or greater, becoming less than one. When compared with the variation in the Ba/Si ratio (Fig. 2), it was evident that the A-band was stronger than the B-band in the Ba-deficient and the stoichiometric regions, and vice versa in the Si-deficient region. This result demonstrated that the energy transition of Eu²⁺ ions from the ground state to the excited energy levels E₁, E₂, and E₃, which corresponded to the A-, B-, and C-bands, respectively, was greatly affected by the crystal structure of BaSi₇N₁₀. In summary, the BaSi₇N₁₀:0.1Eu²⁺ powder with the stoichiometric composition (i.e., the one corresponding to x = 0.8) exhibited the strongest excitation intensity at approximately 368 nm, indicating that it has potential for use in white LEDs pumped by near-ultraviolet (nUV) sources.

 

Fig. 4 Variation in the I_R values of the samples prepared with (x-0.1)BaCO₃.

Download Full Size | PDF

The PLE spectra obtained in the present study were completely different from those listed in earlier studies [14–17], which reported that the strongest excitation bands occurred at approximately 280–330 nm; these corresponded to the C- or B-band in Fig. 3(a). The Eu²⁺ ions substituted for the Ba²⁺ ions occupy the cavities within the network structure of [Si₇N₁₀]^2- and are coordinated by 13 nitrogen atoms. Therefore, the crystal-field splitting of the 5d level of the Eu²⁺ ions depends strongly on the nitrogen and oxygen contents. Fang et al. calculated the electronic structure of BaSi₇N₁₀ using first-principles calculations [18]. They confirmed that the top of the valence band of the compound was dominated by N ^[2^] atoms (coordinated by two Si atoms), while the conduction band was determined mainly by the Ba 6s states. The experimentally determined optical band gap of BaSi₇N₁₀ is roughly 4 – 4.8 eV [14,18]. Thus, the highest excited energy level of the Eu²⁺ ions (C-band at 296 nm, ~4.2 eV) lie fully or partly within the conduction band. This suggests that the energy-level scheme of the Eu²⁺ ion is closely related to the crystal structure of BaSi₇N₁₀. Thus, it is likely that the differences in the PLE spectra reported in previous studies and those determined in the present study are attributable to the differences in the compositions, crystal defects, and nitrogen (oxygen) contents of the investigated compounds. The effects of the crystal defects are explained in detail in the discussion on the decay time (Fig. 8).

The dominant-peak wavelengths (DPWs) of the PL emission spectra (Fig. 3(b)) were 478 nm, irrespective of the value of x and the excitation wavelength used (A-, B-, and C-bands). This indicates that there is only one activation enter (Eu²⁺ sites) surrounded by the 13 nitrogen atoms. The intensity of the emission was dependent on the composition (Fig. 2) and the phase purity of the compounds (Fig. 1) and was the strongest for x = 0.8.

The QE and thermal quenching of BaSi₇N₁₀:0.1Eu²⁺ (x = 0.8) were investigated; commercialized (Sr,Ba)-orthosilicate green phosphor was used as the reference. The internal and external QEs were approximately 107% and 79%, respectively, of those of the reference. This demonstrates that BaSi₇N₁₀:Eu²⁺ has high internal QE, while its external QE can be improved further; this can be accomplished by optimizing the synthesis process and the post-synthesis treatments. The effects of the temperature on the emission intensity (i.e., the degree of thermal quenching) and the corresponding spectra of BaSi₇N₁₀:0.1Eu²⁺ (x = 0.8) are shown in Figs. 5(a) and 5(b), respectively. The emission intensity decreased gradually over temperatures of 20–180 °C and remained at approximately 80% of the initial intensity, whereas that of the reference phosphor at 180 °C dropped abruptly to 65% of its initial intensity. Thus, the compound exhibited low thermal quenching, being comparable to those of other nitride phosphors. As can be seen from the Fig. 5(b), the DPWs remained almost constant, and the full width at half maximum (FWHM) of the emission band broadened slightly, by approximately 3.56 nm, when the temperature was increased from 20 °C to 180 °C. These results indicate that BaSi₇N₁₀:Eu²⁺ exhibits highly stable chromaticity and brightness even at the high temperatures. On the other hand, earlier papers reported that the increase in the temperature resulted in a slight blue-shift of the DPWs [15–17].

 

Fig. 5 (a) Thermal quenching and (b) spectral variation of BaSi₇N₁₀:0.1Eu²⁺ (x = 0.8) powders.

Download Full Size | PDF

The effects of the substitution of Ca²⁺ ions for Ba²⁺ ions on the luminescence were also investigated. The variations in the emission intensity and DPWs of (Ba,Ca)Si₇N₁₀:0.1Eu²⁺ powders, which were prepared from a starting mixture consisting of (0.7–y)BaCO₃–yCaCO₃–7/3Si₃N₄–0.05Eu₂O₃, are shown in Figs. 6(a) and 6(b), respectively. The XRD patterns of the powders confirmed that no secondary phases were produced for y ≤ 0.2. With an increase in the Ca content, that is, for an increase in y from 0 to 0.1, the emission intensity increased by approximately 130%, and then sharply decreased at y = 0.2. The DPW shifted continuously to a higher value (i.e., it underwent a red-shift) for y ≤ 0.1, but a sudden blue shift was observed at y = 0.2. The Commission International de L'Eclairage (CIE) chromaticity coordinates of the powders are shown in the inset of Fig. 6(b). They were all in the greenish-blue region (cyan) and varied with the amount of Ca²⁺, which led to changes in the DPW.

 

Fig. 6 Variation in (a) the emission intensity and (b) DPWs of (Ba,Ca)Si₇N₁₀:0.1Eu²⁺ powders with Ca (y). (Inset: CIE chromaticity coordinates).

Download Full Size | PDF

These phenomena can be explained by the changes induced in the crystal parameters of the compounds. The radius of the Ca²⁺ ion (r = 1.34 Å for CN = 12) is much smaller than that of the Ba²⁺ ion (r = 1.61 Å for CN = 12). In addition, both ions exhibit different chemical properties (i.e., they have different electronegativities and ionic bond energies). These differences result in structural modification when Ca²⁺ ions replace Ba²⁺ ions. The crystal parameters of (Ba,Ca)Si₇N₁₀:0.1Eu²⁺ are shown in Fig. 7. When compared to the refined parameters of BaSi₇N₁₀:0.1Eu²⁺, the changes of the lattice constants of (Ba,Ca)Si₇N₁₀:0.1Eu²⁺ for y = 0.05 were insignificant, β was lower, and the unit cell volume was higher. Further, for y = 0.05 – 0.20, a, b, and the unit cell volume decreased continuously; however, c and β increased abruptly at y = 0.20. These finding demonstrate that the substitution of Ca²⁺ ions for Ba²⁺ ions causes local lattice distortion. As a result, the crystal field surrounding the Eu²⁺ ions is the optimal one, with the strongest emission intensity being observed at y = 0.1, as shown in Fig. 6(a). The relationship between the crystal field strength, Dq, and ionic size is Dq ∝ R⁻⁵, where R is the length of the bond between the central ion (Eu²⁺) and a ligand ion (N^3-). Accordingly, the substitution of the smaller Ca²⁺ ions for Ba²⁺ ions increases the strength of the crystal field, resulting in the red-shift observed for y = 0 – 0.1, as shown in Fig. 6(b). However, the unexpected blue-shift at y = 0.2 might be attributable to the marked increase in c and β (see Fig. 7).

 

Fig. 7 Crystal parameters of the (Ba,Ca)Si₇N₁₀:0.1Eu²⁺ powders prepared with different amounts of Ca (y). (a) Lattice constants a and b, (b) lattice constant c, (c) axial angle β, and (d) unit cell volume.

Download Full Size | PDF

The decay time of I_o/e (I_o: initial emission intensity at 481 nm) of (Ba,Ca)Si₇N₁₀:0.1Eu²⁺ (x = 0.8, y = 0.1) was approximately 0.7 μs, as can be seen from Fig. 8. This value is in the range that is typical for the 5d → 4f energy transition of the Eu²⁺ ion in nitride phosphors, while the single exponential curve suggests the presence of a single activation center (see Fig. 3). Earlier studies report significantly longer decay times [15,16], but do not describe the underlying mechanism. For example, Qin et al. have suggested that barium vacancies (V_Ba”) and positive defects are created during the synthesis process owing to local oxidation and that these act as hole- and electron-trapping centers, respectively, leading to longer-lasting phosphorescence [16]. Consequently, it can be speculated that the differences in the decay time and the PLE spectra obtained in the present study and those reported in previous works [14–16] are probably ascribable to the differences in the crystal compositions, defects, and nitrogen contents of the synthesized compounds.

 

Fig. 8 Decay curve of (Ba,Ca)Si₇N₁₀:0.1Eu²⁺ (x = 0.8, y = 0.1).

Download Full Size | PDF

The PL emission spectra of the (Ba,Ca)Si₇N₁₀:zEu²⁺ (x = 0.8, y = 0.1) samples with different Eu²⁺ concentrations (z) are shown in Fig. 9. The emission intensity increases continuously for z ≤ 0.15 and then drops sharply at z = 0.25, owing to a concentration-quenching effect. The change in the DPW with an increase in z was significant, because the large Stokes shift (~6,340 cm⁻¹) does not allow the shift of the emission band, which is attributable to the re-absorption mechanism caused by overlapping absorption and emission bands [15]. On the other hand, at z = 0.25, a weak red band was observed at approximately 650 nm, in addition to the main emission band. The results of the XRD analyses indicated that no impurity phase was present in the compound corresponding to z = 0.25. Nevertheless, the most probable reason for the red emission is the presence of a Ba₂Si₅N₈:Eu²⁺ phase, which might exist in amounts smaller than the detection limit of the XRD analysis system. A similar phenomenon has been reported by Li et al., who found that Ba₂Si₅N₈:Eu²⁺ as well as BaSi₇N₁₀:Eu²⁺ appeared as an impurity phase at high Eu²⁺ concentration (20 at%) and that this phase resulted in orange emission [15].

 

Fig. 9 PL emission spectra of the (Ba,Ca)Si₇N₁₀:zEu²⁺ powders (x = 0.8, y = 0.1) prepared with different Eu²⁺ concentrations (z).

Download Full Size | PDF

4. Summary

(Ba,Ca)Si₇N₁₀:Eu²⁺ powders were synthesized from mixtures consisting of (x-y-z)BaCO₃–yCaCO₃–7/3Si₃N₄–0.5zEu₂O₃ using a high-temperature graphite furnace at normal nitrogen atmosphere. Rietveld refinement confirmed that stoichiometric and pure BaSi₇N₁₀:Eu²⁺ powders were obtained at x = 0.8. The excitation spectra of the powders were composed of three sub-bands, whose intensities varied with the BaCO₃ content; the strongest 368 nm excitation band was observed at x = 0.8, resulting in the strongest 478 nm emission. There were distinct differences in the PLE spectra observed in the present study and those reported previously. These differences are probably attributable to the differences in the compositions, crystal defects, and nitrogen (oxygen) contents of the synthesized compounds. The substitution of Ca²⁺ ion for Ba²⁺ ions increased the intensity of the emission by approximately 130%, owing to local lattice modification. Finally, the powders exhibited low thermal quenching, high QEs, and stable chromaticities, indicating that (Ba,Ca)Si₇N₁₀:Eu²⁺ phosphors show promise for use in white LEDs pumped by nUV sources.