A Facile Method Using a Flux to Improve Quantum Efficiency of Submicron Particle Sized Phosphors for Solid-State Lighting Applications

This work successfully verified that the addition of a flux (NH4F, NH4Cl, and H3BO3) during synthesis has an impact on the crystallite size and quantum efficiency of submicron-sized particles of CaMgSi2O6:Eu phosphors. The addition of NH4F or NH4Cl increased the crystallite size in the submicron-sized particles, yielding an increase in emission intensity and quantum efficiency. On the other hand, the use of the H3BO3 flux crystallized a secondary phase, SiO2, and changed the lattice parameters, which degraded the luminescent properties. In addition, an excessive amount of NH4Cl was examined, resulting in nucleation of a secondary phase, CaSiO3, which changed the lattice parameters with no improvement in luminescent properties. These results demonstrate that the addition of a flux could be a method to improve the quantum efficiency of submicron-sized particles composed of nanocrystallites; however, a judicious choice of the flux composition and amount has to be carefully considered.


Introduction
Powder phosphors produced by the conventional solid-state reaction method have been widely researched for application in near UV-emitting LEDs (nUV-LEDs) [1,2].This method produces micron-sized powders that have higher quantum efficiencies than smaller-sized powders [3,4], whereas chemical synthesis methods produce submicron-sized powders composed of nanocrystallites.Particle size and crystallite size should be differentiated to understand the quantum efficiency of phosphors.A powder particle can be a single crystal or consist of crystallites, the crystallite size is typically measured by X-ray diffraction or transmission electron microscopy, whereas the particle size is typically measured by dynamic light scattering or scanning electron microscopy.In the remote phosphor configuration, the phosphor particles are on a substrate that is suspended above a nUV-LEDs, as opposed to the conventional configuration where the phosphors are embedded in a polymer around a blue-emitting LED.In addition, in the remote configuration, the packing density of the large particles is low, which generates substantial light scattering [5].To overcome this issue, phosphors with a small, narrow particle size distribution are required.If the particle radii are <~400 nm, these particles will negligibly scatter visible and near UV radiation, because particle size is smaller than the wavelength of the radiation.However, phosphor particles in the submicron-size regime with nano-sized crystallites have poor quantum efficiency compared to micrometer-sized, single crystal phosphor particles [2,6].
A flux is typically mixed with reactants [7,8,[14][15][16]19] or as-synthesized product [20] before an annealing step; thus, is present as the molten phase during the post-synthesis annealing process.Since the flux material should be evaporated after annealing to avoid formation of impurities or second phases in the final product, the melting and boiling temperatures should be considered in relationship to the annealing conditions (temperature, atmosphere), when selecting a flux material.The criteria for selecting a flux are: (1) a low melting temperature, so that it is a liquid during the annealing process [9]; (2) a boiling temperature lower than the annealing temperature, so that the flux can be evaporated to avoid impurity or a second phase formation [13].Fluxes having a higher boiling temperature than the annealing temperature had been used in most of the previous literature (Table 1); and (3) the difference in ionic radii of the flux and the phosphor elements must be more than 30% to avoid doping of the flux elements in the phosphors [22].Chiang et al. [7] reported a formation of a second phase of BaAl 2 O 4 in Y 2.95 Ce 0.05 Al 5 O 12 with a BaF flux.This suggests that some fluxes can remain after the reaction, producing by-products.
Dai et al. [9] discussed the effect of various fluxes (NH 4 Cl, NH 4 F, H 3 BO 3 , LiF, and NaF) on the emission intensity of Y 1.55 Eu(III) 0.45 Ti 2 O 7 phosphors with an orange-red emission under near UV light for display devices such as high-resolution and field emission displays, as well as high-power white light-emitting diodes.It was found that uniform micrometer-sized (~4 µm) particles formed with NaF and LiF fluxes, and the maximum emission intensity was achieved with NaF flux, while a narrow size distribution of the particles was not achieved with NH 4 Cl, NH 4 F, or H 3 BO 3 fluxes.Additionally, Zhang et al. [8] examined the influence of different concentrations of BaF 2 flux on the formation of Ca 0.99 Ce 0.01 Sc 2 O 4 with green emission prepared by a solid-state reaction.After introducing BaF 2 , a higher particle growth rate, larger particle sizes, and more narrow particle size distribution were verified, which resulted in improved emission intensity.The emission intensity increased with the increase of the concentration of BaF 2 and the maximum emission intensity corresponded to 0.5 wt.% of BaF 2 .The emission intensity decreased when the concentration of BaF 2 was higher than 0.5 wt.%, which was attributed to particle agglomeration.Wang et al. [20] examined the effect of Li 2 CO 3 and K 2 CO 3 fluxes on the formation of Ca 0.68 Mg 0.2 Eu 0.12 SiO 3 prepared by a co-precipitation method.The crystallite size increased from ~93 nm to 99 nm (6% of Li 2 CO 3 ) or to 100 nm (5% of K 2 CO 3 ), and the quantum efficiencies were improved (from 12% to 27% with Li 2 CO 3 flux and to 31% with K 2 CO 3 flux).In this work, blue-emitting Ca 0.94 Eu 0.06 MgSi 2 O 6 powders were synthesized through a coprecipitation method [23,24].The powders were annealed with three different fluxes (NH 4 F, NH 4 Cl, or H 3 BO 3 ).In our previous report [24], the Ca 0.94 Eu 0.06 MgSi 2 O 6 powders formed submicrometer-sized particles having blue color with x = 0.14 and y = 0.05 on the CIE diagram, similar to those defined by the National Television System Committee color (0.14, 0.08), but the quantum efficiency was found to be low (Φ, ~5%).A low Φ is a drawback of nanocrystalline-sized phosphors; therefore, the main goal of this study was to improve the Φ using several different types of flux materials.NH 4 F, NH 4 Cl, and H 3 BO 3 were selected as flux materials from the reported flux candidates [7][8][9][13][14][15][16][17][18][19][20] due to their low melting and boiling temperatures (Table 2), which are expected to decompose or evaporate during annealing process.Additionally, the large (>30%) ionic radii difference between the flux and the phosphor components indicates that flux contamination of the phosphor is unlikely to occur.The co-precipitation method was used to synthesize the powders by following previously reported procedures [23,24].The concentration of 6 at.%Eu 2+ activator was chosen as it is reported to have the highest photoluminescence (PL) emission intensity [24].Tetraethyl orthosilicate (2.23 mL) was added to ethanol (20 mL) with several drops of nitric acid and deionized water while stirring for 30 min.Meanwhile, Eu 2 O 3 (0.0015 mol) was dissolved in nitric acid (0.4 mL) solution to form aqueous Eu(NO 3 ) 3 solution.Mg(NO 3 ) 2 •6H 2 O (0.005 mol) and Ca(NO 3 ) 2 •4H 2 O (0.0047 mol) were dissolved in deionized water with stirring.After the two nitrate solutions became transparent, all three solutions (tetraethyl orthosilicate, Eu(NO 3 ) 2 , and Mg(NO 3 ) 2 with Ca(NO 3 ) 2 ) were mixed together and then stirred for 1 h.Subsequently, ammonium hydroxide was added dropwise into the solution to reach a pH of 10 and initiate precipitation.White precipitates were formed, and the suspension was stirred for 8 h.Next, NH 4 F, NH 4 Cl, or H 3 BO 3 was added to the solution at amounts of 2 wt.%, 6 wt.%, or 10 wt.% of Ca 0.94 Eu 0.06 MgSi 2 O 6 .The conversion of wt.% to mol.% for each flux is shown in Table 3.The solutions were then centrifuged and dried at 80 • C for 12 h.Finally, a post-synthesis annealing step was performed at 1100 • C for 2 h in air and then at 1100 • C for 4 h in a 5% H 2 /95% N 2 atmosphere to transform Eu 3+ to Eu 2+ .

Characterization
The powders were analyzed by X-ray diffraction (XRD) on a D2 Phaser (Bruker, Karlsruhe, Germany) using CuKα radiation and a step size of 0.014 • over a 2θ range of 20 to 80 • .The crystallite sizes, lattice parameters, and ratio of the phases presented were calculated by Rietveld refinement using the TOPAS 4.2 software (Bruker).The sizes of the powders were examined by dynamic light scattering (DLS) on a Nanotrac Wave II system (Microtrac Inc., York, PA, USA) [25][26][27].Particles were distributed in an aqueous solution by sonication in a water bath for DLS experiments.A field emission scanning electron microscope (FESEM, XL30, Philips, Amsterdam, Netherlands) at 10 keV was used to image the powders to confirm sizes and determine morphology.Samples were coated with iridium at 85 µA for 10 s before imaging.Energy dispersive spectroscopic analysis (EDS) was performed with a scanning electron microscope (Apreo SEM, FEI, Hillsboro, OR, USA) to analyze the concentration of elements.Absolute quantum efficiency (Φ) measurements were performed using an integrating sphere system, with sodium salicylate (Φ = 44%) as a reference standard.PL emission and excitation spectra were acquired with a fluorescence spectrophotometer (Hitachi F-7000, Hitachi High-Technologies Corporation, Tokyo, Japan) using λ = 350 nm excitation wavelength (pulse = 0.025 s).This excitation wavelength was selected as it produced the highest PL emission intensity for Ca 0.94 Eu 0.06 MgSi 2 O 6 [24].

Crystal Structure and Lattice Parameters
Figure 1 shows the crystal structure of CaMgSi 2 O 6 .Also known as diopside, it has a monoclinic crystal structure with space group C2/c.The lattice parameters are a = 0.9743 nm, b = 0.8879 nm, c = 0.5230 nm, and β = 105.53• [24].The coordination numbers for Ca 2+ , Mg 2+ , and Si 4+ of 8, 6, and 4, respectively.The ionic radii of the ions are listed in Table 4.The radii differences between ions in CaMgSi 2 O 6 and the flux are Mg 2+ -B 3+ = 90%, Si 4+ -B 3+ = 81%, O 2− -F − = 5% (4-coordinated, 6-coordinated), O 2− -Cl − = 26% (6-coordinated).The F − from NH 4 F and Cl − from NH 4 Cl are likely to be composed to HF (g) and HCl (g) during the annealing process (Table 2) although the radii difference between O 2− and F − /or Cl − is less than 30%.H + may occupy interstitial sites in the lattice because of its small size.Eu 3+ (0.107 nm for 8-coordinated) is expected to occupy the Ca 2+ sites due to the radii size similarity (5% difference) before reduction annealing and transforms to Eu 2+ after reduction.XRD patterns of Ca0.94Eu0.06MgSi2O6with and without NH4F and NH4Cl fluxes are shown in Figure 2. The peak widths narrowed with an increase in the concentration of NH4F (Figure 2a), indicating an increase in crystallite size.Several small peaks from CaEu4(SiO4)3O were recorded for 10 wt.% NH4F.CaEu4(SiO4)3O was previously identified in Ca0.94Eu0.06MgSi2O6under annealing temperature of 1247 °C [29], which is a higher temperature than used in the present study, with no flux.As shown previously, a flux can decrease the temperature for the crystallization of phosphor materials [7,13], so that excessive NH4F may lead to the new formation of CaEu4(SiO4)3O at the lower temperature by reducing the corresponding formation temperature.For the diffraction patterns from the powders prepared using NH4Cl flux, shown in Figure 2b, the peak width also slightly narrowed with the increase of concentration of NH4Cl.However, this narrowing is less pronounced in comparison with the NH4F, indicating that the crystallite sizes with NH4F flux were larger.The crystallite sizes for different concentrations of NH4F and NH4Cl fluxes are shown in Figure 2c.For the NH4F flux, the crystallite size increased from roughly 13 nm with no flux to about 31 nm with 10 wt.% NH4F.The maximum crystallite size for NH4Cl flux was about 19 nm with 2 wt.%NH4Cl.Different diffusion rates of the reactants through a flux may explain why the material produced with NH4F flux shows enlarged crystallite sizes compared to that made with NH4Cl flux.Ions move through liquid flux during the annealing process with a diffusion coefficient expressed by D = (1/f)•kT, With the H3BO3 flux, SiO2 impurities were detected, as shown in the XRD patterns in Figure 3a.Even though B 3+ and Si 4+ have quite different ionic radii (0.026 nm and 0.011 nm) [30], a substitution of B 3+ on Si 4+ sites on the tetrahedral site has been previously reported [31,32] although there is a charge difference between B 3+ and Si 4+ .Marler et al. [31] reported synthetic tourmaline (olenite) to replace partial silicon ions with excessive boron ions; and the difference in charge between Si 4+ and B 3+ was compensated by protons leading to the unusually high water content.Xia et al. [32] studied La5(Si2−xB1−x)(O13−xNx):Ce 3+ by replacing partially B 3+ -O 2− by the Si 4+ -N 3− ; and the charge difference from the replacement of B 3+ by Si 4+ was compensated by the substitution of O 2− by N 3− .In the current study, the charge difference from the partial replacement of Si 4+ by B 3+ could be compensated to produce oxygen vacancies due to no other ion replacements.With the H 3 BO 3 flux, SiO 2 impurities were detected, as shown in the XRD patterns in Figure 3a.Even though B 3+ and Si 4+ have quite different ionic radii (0.026 nm and 0.011 nm) [30], a substitution of B 3+ on Si 4+ sites on the tetrahedral site has been previously reported [31,32] although there is a charge difference between B 3+ and Si 4+ .Marler et al. [31] reported synthetic tourmaline (olenite) to replace partial silicon ions with excessive boron ions; and the difference in charge between Si 4+ and B 3+ was compensated by protons leading to the unusually high water content.Xia et al. [32] studied La 5 (Si 2−x B 1−x )(O 13−x N x ):Ce 3+ by replacing partially B 3+ -O 2− by the Si 4+ -N 3− ; and the charge difference from the replacement of B 3+ by Si 4+ was compensated by the substitution of O 2− by N 3− .In the current study, the charge difference from the partial replacement of Si 4+ by B 3+ could be compensated to produce oxygen vacancies due to no other ion replacements.With an assumption that B ions partially replace the Si ions, the corresponding chemical reaction is: As the amount of H3BO3 increased (2-10 wt.%), the corresponding amount of SiO2 increased (53-84 mol.%, Figure 3b).The crystallite size of the phosphor increased from ~13 nm to ~27 nm, while for SiO2 it increased from 20 nm to 80 nm (Figure 3c).
The calculated lattice parameters of Ca0.94Eu0.06MgSi2O6with and without each flux are shown in Figure 4.The lattice parameters of powders prepared with NH4F and NH4Cl (Figure 4a,b) were identical to the phosphors prepared without flux, indicating that these two fluxes did not affect the crystal structure of the resultant materials.In contrast, with H3BO3 flux, the b and c parameters did not show significant change, while a decreased from 0.9743 nm to 0.9670 nm.This is presumably due to the substitution of B 3+ (0.011 nm) with Si 4+ (0.036 nm) sites, causing the lattice parameter to decrease.Vegard's law [33], which is an empirical rule based on a linear relationship found between lattice parameters and the size of the constituent elements.The formation of ½O vacancies from the charge difference between Si 4+ and B 3+ can affect the lattice parameters.Therefore, the relationship between the initial lattice parameter and the one with B 3+ is = − ( − ) − , where a is the lattice parameter (0.9670 nm) with 10 wt.% B 3+ addition, a0 is the initial lattice parameter (0.9743 nm), is the partial substitution of B on Si, , , are the radii of Si 4+ , B 3+ , and O 2− , respectively.The With an assumption that B ions partially replace the Si ions, the corresponding chemical reaction is: As the amount of H 3 BO 3 increased (2-10 wt.%), the corresponding amount of SiO 2 increased (53-84 mol.%, Figure 3b).The crystallite size of the phosphor increased from ~13 nm to ~27 nm, while for SiO 2 it increased from 20 nm to 80 nm (Figure 3c).
The calculated lattice parameters of Ca 0.94 Eu 0.06 MgSi 2 O 6 with and without each flux are shown in Figure 4.The lattice parameters of powders prepared with NH 4 F and NH 4 Cl (Figure 4a,b) were identical to the phosphors prepared without flux, indicating that these two fluxes did not affect the crystal structure of the resultant materials.In contrast, with H 3 BO 3 flux, the b and c parameters did not show significant change, while a decreased from 0.9743 nm to 0.9670 nm.This is presumably due to the substitution of B 3+ (0.011 nm) with Si 4+ (0.036 nm) sites, causing the lattice parameter to decrease.Vegard's law [33], which is an empirical rule based on a linear relationship found between lattice parameters and the size of the constituent elements.The formation of 1  2 O vacancies from the charge difference between Si 4+ and B 3+ can affect the lattice parameters.Therefore, the relationship between the initial lattice parameter and the one with B 3+ is a = a 0 − x(r Si − r B ) − 1 2 xr O , where a is the lattice parameter (0.9670 nm) with 10 wt.% B 3+ addition, a 0 is the initial lattice parameter (0.9743 nm), x is the partial substitution of B on Si, r Si , r B , r O are the radii of Si 4+ , B 3+ , and O 2− , respectively.The x value obtained is 0.09, which is smaller than the molar fraction of H 3 BO 3 added (0.28 converted from 10 wt.% H 3 BO 3 ).The excess B 3+ was not detected in the XRD patterns (detection limit of 3 ~5 wt.%).Therefore, due to the low boiling temperatures, it likely does not remain in the resultant powder.The change of lattice parameter a, while parameters b and c remain unchanged, can be explained by the ion arrangement.Along the [100] and [001] directions (see Figure 1b,d), there are two Si 4+ along the a-axis, four along the b-axis, and three along c-axis.The fractions of one Si 4+ along the a-, b-, and c-axes are 0.50, 0.25, and 0.33, respectively.The fraction of Si 4+ along the a-axis is the largest, so that the replacement of Si 4+ is affected more along this axis, resulting in a decrease of a with an addition of 10 wt.% H 3 BO 3 .Another possible reason for a change in lattice parameter of Ca 0.94 Eu 0.06 MgSi 2 O 6 is H + dissolved in interstitial sites, but this would cause an increase of lattice parameters, which was not observed.

Scanning Electron Microscopy and Dynamic Lighting Scattering Analysis
SEM images of Ca0.94Eu0.06MgSi2O6are shown in Figure 5a for no flux, Figure 5b-d for NH4F flux addition, Figure 5e-g for NH4Cl flux addition, and Figure 5h-j for H3BO3 flux addition.For NH4F or NH4Cl fluxes, the particle sizes and morphologies were similar to those without a flux.However, the particles with the H3BO3 flux were aggregated, and the particle sizes were irregular.From EDS analysis, the ratio of Ca:Mg:Si:O was 1:1:7:13 in the large particles (yellow point in Figure 5h), where the amounts of Si and O are two to three times more than those of CaMgSi2O6 (Ca:Mg:Si:O = 1:1:2:6).In the small particles (red point in Figure 5h), the ratio of Ca:Mg:Si:O was 1:1:3:5, which is closer to that of CaMgSi2O6.Although the results from EDS are considered more accurate for polished surfaces  or NH 4 Cl fluxes, the particle sizes and morphologies were similar to those without a flux.However, the particles with the H 3 BO 3 flux were aggregated, and the particle sizes were irregular.From EDS analysis, the ratio of Ca:Mg:Si:O was 1:1:7:13 in the large particles (yellow point in Figure 5h), where the amounts of Si and O are two to three times more than those of CaMgSi 2 O 6 (Ca:Mg:Si:O = 1:1:2:6).In the small particles (red point in Figure 5h), the ratio of Ca:Mg:Si:O was 1:1:3:5, which is closer to that of CaMgSi 2 O 6 .Although the results from EDS are considered more accurate for polished surfaces than the powders, the ratio of Ca, Mg, Si, O demonstrated differences between small and large particles.Also, given that the EDS analysis penetrates 1-2 µm below the surface [34], CaMgSi 2 O 6 could be detected under the SiO 2 particles, which could be the reason EDS analysis showed a small amount of Ca and Mg, when presumably large SiO 2 particles were analyzed.Therefore, large particles can be considered as SiO 2 and the aggregated, small particles as CaMgSi 2 O 6 .

Scanning Electron Microscopy and Dynamic Lighting Scattering Analysis
Ceramics 2018, 1, x FOR PEER REVIEW 10 of 16 than the powders, the ratio of Ca, Mg, Si, O demonstrated differences between small and large particles.Also, given that the EDS analysis penetrates 1-2 µm below the surface [34], CaMgSi2O6 could be detected under the SiO2 particles, which could be the reason EDS analysis showed a small amount of Ca and Mg, when presumably large SiO2 particles were analyzed.Therefore, large particles can be considered as SiO2 and the aggregated, small particles as CaMgSi2O6.

Photoluminescence Spectra and Quantum Efficiency
The PL excitation (PLE) was monitored at 458 nm, and the PL emission spectra were obtained under λex = 350 nm as an excitation wavelength.Although the Eu ions were introduced from a strong nitric acid solution during the synthetic process, this acid may cause traces of transition metal ions impurities (Cr, Cu, Pb, Ni, Zn, Au, Ti).However, these are negligible due to the very low concentration in the whole solution (<0.01 ppm).If all transition metals are present in the phosphor powders, the concentration would be ~0.1 ppb, which would not influence the luminescence properties.Figure 7a-c shows the PLE spectra (dashed lines) with a broadband absorption in the near UV region from 200 nm to 400 nm, with a maximum at 350 nm, which is attributed to the allowed transition of Eu 2+ .The PL emission (solid lines) shows a spectrum with a maximum at 458 nm, which corresponds to the parity allowed 4f 6 5d 1 → 4f 7 transition of Eu 2+ .In Figure 7a, the absorption and emission intensities, together with the quantum efficiency (Φ), increased with the increase of the amount of NH4F.A maximum Φ of 17% was found at 11 wt.% of NH4F.For the NH4Cl flux (Figure 7b), the quantum efficiency increased from 5% to 11% when 2 wt.% of NH4Cl was added.For the amounts of NH4Cl > 2 wt.%,Φ decreased.As shown in Figure 7c, Φ with H3BO3 was very low, ~1%,

Photoluminescence Spectra and Quantum Efficiency
The PL excitation (PLE) was monitored at 458 nm, and the PL emission spectra were obtained under λ ex = 350 nm as an excitation wavelength.Although the Eu ions were introduced from a strong nitric acid solution during the synthetic process, this acid may cause traces of transition metal ions impurities (Cr, Cu, Pb, Ni, Zn, Au, Ti).However, these are negligible due to the very low concentration in the whole solution (<0.01 ppm).If all transition metals are present in the phosphor powders, the concentration would be ~0.1 ppb, which would not influence the luminescence properties.Figure 7a-c shows the PLE spectra (dashed lines) with a broadband absorption in the near UV region from 200 nm to 400 nm, with a maximum at 350 nm, which is attributed to the allowed transition of Eu 2+ .The PL emission (solid lines) shows a spectrum with a maximum at 458 nm, which corresponds to the parity allowed 4f 6 5d 1 → 4f 7 transition of Eu 2+ .In Figure 7a, the absorption and emission intensities, together with the quantum efficiency (Φ), increased with the increase of the amount of NH 4 F. A maximum Φ of 17% was found at 11 wt.% of NH 4 F. For the NH 4 Cl flux (Figure 7b), the quantum efficiency increased from 5% to 11% when 2 wt.% of NH 4 Cl was added.For the amounts of NH 4 Cl > 2 wt.%,Φ decreased.As shown in Figure 7c, Φ with H 3 BO 3 was very low, ~1%, because of the significant amount of SiO 2 in the sample, as was confirmed by the XRD (Figure 3b). Figure 7d is a plot of Φ as a function of the amount of flux for all three fluxes used in the current study.A cumulative analysis of Figures 2c and 7d demonstrates that the change in the crystallite sizes of the materials produced with NH 4 F and NH 4 Cl fluxes, directly relates to the change in their corresponding quantum efficiencies [35].For 0-10 wt.% of NH 4 F, the average crystallite size increased from 13 nm to 31 nm, and the corresponding quantum efficiencies improved from 5% to 17%.Similarly, for 0 wt.%, 2 wt.%, 6 wt.%, and 10 wt.%, of the NH 4 Cl flux, the average crystallite sizes were 13 nm, 19 nm, 15 nm, 18 nm, respectively, and the corresponding quantum efficiencies showed similar trend.In constant, with 0-10 wt.% of H 3 BO 3 , the crystallite size increased from ~13 nm to ~27 nm, but the quantum efficiency was low due to the presence of a large fraction of the secondary phase, SiO 2 .
of the materials produced with NH4F and NH4Cl fluxes, directly relates to the change in their corresponding quantum efficiencies [35].For 0-10 wt.% of NH4F, the average crystallite size increased from 13 nm to 31 nm, and the corresponding quantum efficiencies improved from 5% to 17%.Similarly, for 0 wt.%, 2 wt.%, 6 wt.%, and 10 wt.%, of the NH4Cl flux, the average crystallite sizes were 13 nm, 19 nm, 15 nm, 18 nm, respectively, and the corresponding quantum efficiencies showed similar trend.In constant, with 0-10 wt.% of H3BO3, the crystallite size increased from ~13 nm to ~27 nm, but the quantum efficiency was low due to the presence of a large fraction of the secondary phase, SiO2.

Effect of High Concentration of NH4Cl
Increased amount of NH4Cl was examined to determine if the high concentration of this particular flux resulted in the formation of second phase, a change in crystal structure, or further improvement of Φ while phosphors with NH4F and H3BO3 already consist of secondary phases below 10 wt.% of the flux.Figure 8a shows the XRD patterns of CaMgSi2O6 produced with 14, 17, and 20 wt.% of NH4Cl flux.There were no secondary phases with 14 and 17 wt.%,but CaSiO3 was detected for 20 wt.% of NH4Cl.The lattice parameters a and c were not affected, but b decreased from 0.8867

Effect of High Concentration of NH 4 Cl
Increased amount of NH 4 Cl was examined to determine if the high concentration of this particular flux resulted in the formation of second phase, a change in crystal structure, or further improvement of Φ while phosphors with NH 4 F and H 3 BO 3 already consist of secondary phases below 10 wt.% of the flux.Figure 8a shows the XRD patterns of CaMgSi 2 O 6 produced with 14, 17, and 20 wt.% of NH 4 Cl flux.There were no secondary phases with 14 and 17 wt.%,but CaSiO 3 was detected for 20 wt.% of NH 4 Cl.The lattice parameters a and c were not affected, but b decreased from 0.8867 nm (for 14 wt.%) to 0.8852 nm (for 20 wt.%) (Figure 8b).Given that the radii difference between Cl − and O 2− is 26%, excessive Cl − may substitute for O 2− and produce a charge imbalance.This could cause cation defects and/or the creation of secondary phases such as CaSiO 3 .MgSiO 3 could also potentially be produced due to the substitution O 2− by Cl − , but it was not detected by XRD.Vegard's law can also be applied to the b-axis change, with an assumption that there are vacancy defects of The PL emission spectra and Φs with additional NH4Cl flux are shown in Figure 8c.The Φs were not significantly changed with an increase in the amount of NH4Cl.A comparison between crystallite size and Φs with the amount of NH4Cl flux is shown in Figure 8d.The maximum Φ corresponded to 2 wt.%; over 2 wt.%Φ initially decreased and eventually leveled off.However, the crystallite sizes increased with an increase in the concentration of the flux.This implies that the flux, although increasing the crystallite size, also changed lattice parameters resulting in no enhancement of Φ due

Figure 2 .
Figure 2. X-ray diffraction patterns of Ca 0.94 Eu 0.06 MgSi 2 O 6 with a flux of (a) NH 4 F and (b) NH 4 Cl.(c) Calculated crystallite sizes of Ca 0.94 Eu 0.06 MgSi 2 O 6 with NH 4 F and NH 4 Cl fluxes.

Figure 3 .
Figure 3. (a) X-ray diffraction patterns of Ca 0.94 Eu 0.06 MgSi 2 O 6 with H 3 BO 3 flux; (b) Calculated molar ratios of the phases present and (c) calculated crystallite sizes.

Ceramics 2018, 1 ,
x FOR PEER REVIEW 9 of 16 value obtained is 0.09, which is smaller than the molar fraction of H3BO3 added (0.28 converted from 10 wt.% H3BO3).The excess B 3+ was not detected in the XRD patterns (detection limit of 3 ~ 5 wt.%).Therefore, due to the low boiling temperatures, it likely does not remain in the resultant powder.The change of lattice parameter a, while parameters b and c remain unchanged, can be explained by the ion arrangement.Along the [100] and [001] directions (see Figure1b,d), there are two Si 4+ along the a-axis, four along the b-axis, and three along c-axis.The fractions of one Si 4+ along the a-, b-, and caxes are 0.50, 0.25, and 0.33, respectively.The fraction of Si 4+ along the a-axis is the largest, so that the replacement of Si 4+ is affected more along this axis, resulting in a decrease of a with an addition of 10 wt.% H3BO3.Another possible reason for a change in lattice parameter of Ca0.94Eu0.06MgSi2O6 is H + dissolved in interstitial sites, but this would cause an increase of lattice parameters, which was not observed.

Figure 4 .
Figure 4. Calculated lattice parameters of Ca 0.94 Eu 0.06 MgSi 2 O 6 from the X-ray diffraction results with a flux of (a) NH 4 F; (b) NH 4 Cl; and (c) H 3 BO 3 .
SEM images of Ca 0.94 Eu 0.06 MgSi 2 O 6 are shown in Figure 5a for no flux, Figure 5b-d for NH 4 F flux addition, Figure 5e-g for NH 4 Cl flux addition, and Figure 5h-j for H 3 BO 3 flux addition.For NH 4 F

Figure 6 .
Figure 6.The distributions of particle sizes of Ca 0.94 Eu 0.06 MgSi 2 O 6 with fluxes (a) NH 4 F; (b) NH 4 Cl and (c) H 3 BO 3 analyzed by dynamic light scattering analysis.

Table 1 .
Reported results of the addition of flux on phosphor preparation.T m = melting temperature, T b = boiling temperature, Φ = quantum efficiency.

Table 3 .
Conversion of wt.% to mol.% for each flux in the solid phosphor powders.