Abstract
In this article, we discuss the crystal chemistry and luminescence properties of 
activated silicate garnets. Our results indicate that the two previously reported end members of this family, 
and 
, form a complete solid solution, and 
primarily substitutes for 
in the octahedral sites with charge compensation by 
substitution in the dodecahedral sites. Within this solid solution, there is a continuous shift in the lowest energy 
excitation band and the 
emission band toward higher energy as the composition is varied between 
and 
.
Export citation and abstract BibTeX RIS
Solid-state lighting based upon a combination of violet/blue light-emitting diodes (LEDs) and downconversion phosphors has created enormous commercial interest because of potentially higher efficacies and long lifetimes for these light sources. Lamps based upon phosphor downconversion of blue LEDs generally use 
garnet 
phosphors. The lowest allowed 
transition strongly absorbs blue radiation and efficiently downconverts this radiation into yellow light through the reverse 
emission transition.1 It is possible to design phosphors that have emission colors that are redshifted vs 
by understanding the relationship between crystal chemistry and luminescence in these phosphors.2 The garnet structure can be represented by the general formula 
, where {A} cations are eight-coordinated in dodecahedra, [B] cations are octahedrally coordinated, and (C) cations are tetrahedrally coordinated. The position of the 
5d levels is strongly influenced by the host lattice selection,3–5 and hence, the 
emission could be controlled through careful substitutions in the dodecahedral, octahedral, and tetrahedral sites. In previous work, we showed that it is possible to synthesize 
garnets, specifically 
, that have redshifted emission 
vs typical 
phosphors; the lower energy for 
emission was attributed to the larger crystal field splitting of the two lowest energy 5d levels.6 
replacement of 
led to a blueshift in the 
emission spectra, but with a lower room and high temperature (HT) efficiency. Other articles on 
luminescence in silicate garnets, such as 
, have a higher energy 
emission band 
with excellent HT efficiencies.7, 8 The differences in the position of the 
emission band between these pure silicate garnets lead to the question of how well 
emission can be tuned in these garnets vs the aluminate garnets, and if there are any apparent trade-offs in the efficiency of these materials, especially at HTs. We address these questions in this article by evaluating nominal compositions within the 
system to determine the range of solid solution between 
and 
and to understand the effects of composition on 
emission, excitation, and efficiency. Our results suggest that there is a full solid solution between the 
and 
silicate garnets with limited 
solubility on the dodecahedral {A} sites. Within these compositions, the emission maximum for 
emission can be fully tuned within the range of 505–605 nm with corresponding shifts in the lowest energy 
excitation band. However, unlike 
additions to 
,6 the blueshift in 
emission in 
compositions does not come with any losses in phosphor efficiency at room or elevated temperatures. Potential reasons for 
luminescence quenching and the changes in the position of the lowest energy 
5d level are also discussed.
Experimental
Powder samples with nominal compositions of 
and 
were synthesized by conventional solid-state reactions starting from high purity 
(PIDC), 
(Aldrich), MgO (Aldrich), 
(PIDC), 
(Aldrich), and fumed 
(Cabosil). The reactants were mixed well and heated in 
mixtures at 
for 5–10 h. The compositions discussed in this article are referred to by their nominal starting compositions. Powder X-ray diffraction (XRD) patterns were recorded for 
using a Philips X'pert diffractometer (
radiation). Lattice parameters were obtained from powder matching of the XRD data using the Fullprof program. The emission and excitation spectra were measured on pressed phosphor powders in an aluminum plaque using an F4500 Hitachi fluorescence spectrophotometer. All measurements were made at room temperature unless otherwise mentioned. The relative quantum efficiency (QE) was measured with respect to a commercial 
phosphor with 
(Kodak) as a reflection standard. HT luminescence measurements were made on pressed powders in an aluminum plaque connected with resistive heaters and a thermocouple that were attached to a standard Watlow temperature controller.
Results and Discussion
Phase characterization and 
site occupation
Silicate garnet synthesis under ambient pressure conditions6–10 is more challenging vs aluminate and germanate garnets with limits on the compositional space. However, we have demonstrated that silicate garnet synthesis at ambient pressure tends to favor compositions with smaller 
ions such as 
, partially occupying the {A} (dodecahedral) sites.6 In addition, larger ions at the [B] octahedral sites also tend to stabilize garnet phase formation.7–10 This interplay between dopant ions and their occupancy in the garnet plays a role in the stability and phase purity of the garnets; it is therefore of interest to determine which site 
primarily occupies in the 
garnet. Typically, 
substitutes into the [B] octahedral site in garnets,10 but hydrothermally synthesized garnet compositions such as 
11 show that 
can occupy the dodecahedral site for certain compositions and synthesis conditions. Therefore, compositions were made where 
is nominally substituted for 
on the dodecahedral site (i.e., 
) and also where 
is nominally substituted for 
on the octahedral site with charge compensation through the 
and 
levels (i.e., 
). Figure 1 shows the comparison of the XRD patterns of (a) 
, (b) 
, and (c) 
. The primary phase in all of these samples are garnets with a small amount of secondary apatite phase present as previously reported.6 In 
samples, apatite is the only secondary phase for 
. However, an additional 
impurity phase is also observed for 
in these samples. This indicates a limited solid solubility of 
in dodecahedral sites. Further, the garnet lattice parameter increases in all compositions where 
nominally replaces 
(Table I). This does not correlate to the substitution of 
by 
, because the 
(
12) ionic radius is significantly smaller compared to that of 
(
12). However, an increase in lattice parameters is expected, if 
(
12) replaces 
(
12) on the octahedral site and 
(
12) replaces 
for charge compensation. In principle, this can lead to additional impurity phases beyond apatite whenever 
nominally replaces 
, in contradiction to our observations of only an apatite secondary phase for 
. However, apatites can have significant nonstoichiometry and compositional flexibility,13 and we surmise that the additional 
that is not incorporated into the garnet phase is incorporated into the apatite at lower levels of 
substitution.
Figure 1. Powder XRD pattern of (a) 
, (b) 
, and (c) 
. The symbol ∗ represents peaks due to a secondary apatite phase.
Table I. Lattice parameters of 
.
 | Lattice parameter (Å) |
|---|---|
| 0 | 11.9743(2) |
| 0.1 | 11.9884(3) |
| 0.2 | 11.9947(3) |
| 0.5 | 12.0062(3) |
| 1 | 12.0706(3) |
Unlike nominal compositions where 
replaces 
in 
, the XRD patterns indicate that 
substitution in the octahedral sites (nominal compositions of 
) does not increase the relative intensity of additional impurity or secondary phase peaks for 
. The expansion in lattice parameter with 
substitution for 
in 
is clearly seen in the 
compositions throughout the entire range of compositions from 
to 
(Table II) with the lattice parameters for the end-member compositions in reasonable agreement with previous results.6, 7 The lattice parameter vs the nominal 
substitution follows Vegard's law (Fig. 2), confirming the complete solid solubility between 
and 
.
Table II. Lattice parameters of 
.
 | Lattice parameter (Å) |
|---|---|
| 0 | 11.9743(2) |
| 0.1 | 11.9890(2) |
| 0.2 | 12.0066(2) |
| 0.5 | 12.0606(2) |
| 1 | 12.1371(3) |
| 1.5 | 12.2070(3) |
| 2 | 12.2521(2) |
Figure 2. Variation in lattice parameter as a function of composition for the samples 
.
luminescence in these silicate garnets
The body color of 
is yellow-orange, due to the lowest energy 
absorption in the blue spectral region, with a broadband emission spectrum 
and an emission maximum of 
.6 For 
compositions, the lowest energy 
absorption is at higher energies with larger values of 
, giving a more yellow-green body color (Fig. 3a). Correspondingly, the emission spectra are at higher energies with higher 
content in 
(Fig. 3b). The 
emission color can be tuned from orange to blue-green emission by systematically replacing 
for 
and 
for 
in 
. For all compositions, the emission spectrum can be deconvoluted into two Gaussians separated by 
. This is typical for the 
emission and is indicative of emission from the lowest energy 5d level to the 
and 
levels. The tunable emission characteristics of these silicate garnet enable flexibility in designing potential LED lamps when using either single phosphors or blends of multiple phosphors. There is no apparent change in the Stokes shift with composition, although an accurate determination of the Stokes shift in these materials is difficult to assess due to severe inhomogeneous broadening that comes from the disorder on the dodecahedral and octahedral sites in these garnets.6–8
Figure 3. (a) Excitation 
and (b) emission 
spectra of 
with increasing Sc content, where (i) 
, (ii) 
, and (iii) 
. The sharp lines in the excitation spectra are experimental artifacts from the Xe lamp source.
We first assess the reasons for the higher energy 
emission and absorption with a higher 
in the 
compositions. The energy of the lowest 5d level for 
in the inorganic hosts is mainly dependent upon two factors: the 5d centroid shift from the 
-free ion level and the crystal field splitting of the 5d levels.3–5 We start by analyzing the 
5d centroid shift in the end members of this system, 
and 
. The 
centroid shift in 
was estimated at 
, 
smaller than the estimates for typical 
garnets.6 Similar calculations using the crystallographic data for 
9 give a centroid shift of 
, primarily from a longer 
bond length. This difference in the centroid shift therefore accounts for at least part of the energy difference between the lowest energy 
transition in 
and 
. Typically, the 
crystal field splitting also has an inverse relationship with the 
–ligand bond length,3–5 which would imply a smaller crystal field splitting as the composition moves toward 
. This is apparently the case in these garnets because the splitting between the lowest two 5d levels in 
is 
and becomes gradually larger for lower 
values in 
(Fig. 3a). Therefore, it is likely that the blueshift of the 
emission and excitation spectra when substituting 
into 
is due to a combination of a smaller 
5d centroid shift and crystal field splitting from the longer 
bond length. The relationship between crystal field splitting and 
–ligand bond length in these silicate garnets is typically observed for 
luminescence,3–5 but differs from the 
garnets where larger {A} cations redshift the 
emission and excitation spectra.2, 14, 15
We finally discuss the QE of these phosphors at room and elevated temperatures. The QE increases with Sc substitution from 70 to 80% of 
. This is a promising value considering that the phosphor synthesis is not optimized for efficiency. It is also important to account for quenching at HTs due to the heat generated by the high power violet/blue LEDs and from the phosphor downconversion. Though parallel efforts are being made to address thermal management issues in LED packages,16 it is essential to have phosphors retain their efficiency at elevated temperatures to address some of these issues. 
has reduced HT quenching compared to 
6–8 and this trend is also reflected within the solid solution compositions (Fig. 3). Hence, increasing 
and 
concentrations in 
not only shifts the emission peak position to higher energy, it also improves phosphor efficiency at elevated temperatures (see Fig. 4). The relationship between absorption/emission position and HT efficiency potentially indicates that 
luminescence quenching occurs by a level crossing mechanism as proposed in the aluminate garnets15, 17 and garnets doped with 
.18 As noted above, the synthesis of these phosphors has not been completely optimized, so further improvements in QE at room temperature and HT could be possible.
Figure 4. Integrated luminescence intensity 
vs temperature for 
, where 
, 1.5, and 2.
Conclusions
In this article, we have investigated the crystal chemistry and 
luminescence properties within silicate garnets that can be synthesized at ambient pressure. There is strong evidence toward a continuous solid solution between the previously reported 
5 and 
.6, 7 Correspondingly, 
substitution in these silicate garnets has a strong preference toward the octahedral site. We have also demonstrated that as the composition moves through this solid solution, the 
emission continuously shifts, allowing for color flexibility in different LED lamp applications and that the HT efficiency of these materials also improves with higher 
levels.
Acknowledgments
The authors thank E. A. Bachniak for the initial sample synthesis and Digamber Porob and Emil Radkov for useful discussions. This work was supported by GE Lumination.
GE Global Research assisted in meeting the publication costs of this article.