Recent progress in advanced optical materials based on gadolinium aluminate garnet (Gd₃Al₅O₁₂)

Abstract

This review article summarizes the recent achievements in stabilization of the metastable lattice of gadolinium aluminate garnet (Gd₃Al₅O₁₂, GAG) and the related developments of advanced optical materials, including down-conversion phosphors, up-conversion phosphors, transparent ceramics, and single crystals. Whenever possible, the materials are compared with their better known YAG and LuAG counterparts to demonstrate the merits of the GAG host. It is shown that novel emission features and significantly improved luminescence can be attained for a number of phosphor systems with the more covalent GAG lattice and the efficient energy transfer from Gd³⁺ to the activator. Ce³⁺ doped GAG-based single crystals and transparent ceramics are also shown to simultaneously possess the advantages of high theoretical density, fast scintillation decay, and high light yields, and hold great potential as scintillators for a wide range of applications. The unresolved issues are also pointed out.

• gadolinium aluminate garnet
• lattice stabilization
• down-/up-conversion phosphors
• single crystal
• transparent ceramic
• scintillator
• energy transfer

1. Introduction

Rare-earth aluminate garnets, having a general formula of Ln₃Al₅O₁₂ (LnAG, Ln: lanthanide and Y), are an important family of multi-functional ceramic materials. The compounds crystallize in a bcc structure (space group: with 160 (80) atoms in the cubic (primitive) cell, where the Ln occupies the 24c sites (D₂ point symmetry, CN = 8; CN: coordination number) and the oxygen atoms take the 96h sites. The Al atoms have two positions to reside on: the 16a sites with an octahedral point symmetry (C_3i, 40%; CN = 6) and the 24d sites with a tetragonal point symmetry (S₄, 60%; CN = 4) [1]. The garnet structure can be viewed as a framework built up via corner sharing of the Al–O polyhedra, with the Ln residing in dodecahedral interstices [1]. A schematic diagram of the crystal structure is shown in figure 1.

Figure 1. A schematic illustration of the crystal structure of LnAG, where Al_oct and Al_tet represent the Al atoms taking octahedral and tetrahedral lattice sites, respectively. Adapted with permission from [1], copyright 1999 by the American Physical Society.

YAG might be the best-known garnet compound owing to its excellent chemical stability, high creep resistance, optical isotropy, and particularly the ability to accept substantial trivalent Ln³⁺ for diverse optical functionalities. In the bulk form, the YAG:Nd single crystal is one of the most widely used solid laser materials since its discovery in the 1960s [2]. Transparent YAG:Nd ceramics that are equal to or superior to single crystals in optical transmittance and laser performance have also been successfully developed since the 1990s via advanced powder processing and sintering technologies [3, 4]. YAG:Ho³⁺ and YAG:Er³⁺ are important infrared (IR) laser materials for medical surgery, since their fluorescence lines (∼2 μm for Ho³⁺ and ∼3 μm for Er³⁺) match well with the water absorptions of the human body [5]. Transparent YAG:Ce³⁺ ceramic is nowadays being developed as an encapsulant for white-light emitting diodes (LEDs) to replace the widely used but readily degradable resin sealant [6]. Since the parity-law allowed 4f⁰5d¹ → 4f¹5d⁰ transition of Ce³⁺ has a very short fluorescence lifetime of ∼10–100 ns and the intrinsic quenching temperature of Ce³⁺ in YAG is very high (>700 K) [7], YAG:Ce³⁺ transparent ceramic has been considered as a scintillator [8], but does not seem to have a satisfactory stopping power for the incident radiations (x-, α- or γ-rays) owing to the relatively low theoretical density of YAG (∼4.55 g cm⁻³) and the small atomic weight of Y (∼89). In the powder form, rare-earth (Eu³⁺, Tb³⁺, Ce³⁺ etc) activated YAG is being widely studied and used as phosphors for fluorescent lamps, field emission displays (FEDs), and white LEDs.

The occurrence and thermal stability of compounds in the Ln₂O₃–Al₂O₃ binary system heavily depend on the ionic radius of Ln³⁺, conforming to lanthanide contraction. Earlier studies by Mizuno et al [9–13] on the phase diagram of Ln₂O₃–Al₂O₃ found the two intermediate compounds of Ln₂O₃·11Al₂O₃ (β-Al₂O₃ type) and LnAlO₃ (commonly referred to as aluminate perovskite or LnAP, an orthorhombic modification of ideal perovskite) for the large ions of La³⁺–Nd³⁺, the two compounds of LnAP and monoclinic Ln₄Al₂O₉ (commonly called LnAM) for the intermediately sized Sm³⁺–Gd³⁺, and the three compounds of LnAM, LnAP, and LnAG for the small Ln³⁺ ions of Tb³⁺–Er³⁺. Though GdAG (further abridged as gadolinium aluminate garnet (GAG) hereafter) was not identified in the work of Mizuno et al, it was successfully synthesized via flux growth by Van Uitert et al [14] and Manabe and Egashira [15] for potential optical applications. Later on, Shishido et al [16] found via annealing Gd₂O₃·5/3Al₂O₃ amorphous glass that GAG is metastable and would completely decompose to α-Al₂O₃ and GdAP (Gd₃Al₅O₁₂ → Al₂O₃ + 3GdAlO₃) upon prolonged annealing at 1500 °C. A recent work by Li et al [17] further showed that the stoichiometric GAG synthesized via low temperature combustion starts to decompose at ∼1300 °C. All these studies point to the fact that thermodynamically stable LnAG only exists for the Ln³⁺smaller than Gd³⁺ and Gd³⁺ is the boundary for a LnAG to be formed. This is understandable from the crystal structure shown in figure 1. That is, the dodecahedral interstice has a certain geometric shape and dimension, and thus a size limit exists for Ln³⁺ to enter the space without disintegrating the Al–O framework. Mainly due to its structural metastability, GAG has been much less explored than YAG for its properties and applications, though its specific heat and thermal expansion coefficient were experimentally determined by Chaudhury et al [18]. Compared with YAG, however, GAG may hold a number of merits for optical applications: (1) the intrinsic ⁸S_7/2 → ⁶I_J transition of Gd³⁺ (usually centered at ∼275 nm) can be utilized as a new excitation source for some types of rare-earth activators, and enhanced luminescence may also be attained via an efficient energy transfer from Gd³⁺ to the activator [19–22], (2) the GAG lattice is more covalent than YAG due to the lower electronegativity of Gd³⁺ (χ = 1.20) than Y³⁺ (χ = 1.22), which may produce new emission features and result in improved emission intensity, and (3) GAG has a significantly higher theoretical density (5.97 g cm⁻³) than YAG (4.55 g cm⁻³) and the atomic weight of Gd (157, close to the 175 of Lu) is much higher than Y, and thus GAG is more desirable for scintillation applications. Similar to the growth of single crystals and sintering of transparent ceramics, a reasonably high processing temperature is usually needed to produce high quality phosphors through crystal perfection. In this context, lattice stabilization becomes a prerequisite for any practical application of GAG in advanced optical materials. This review article summarizes the recent achievements in GAG, including lattice stabilization via doping and its application in down-/up-conversion (UC) phosphors and transparent ceramic/single crystal scintillators.

2. Lattice stabilization of GAG by modifying the Gd/Al sites

There are two primary ways to stabilize the garnet lattice of GAG, as can be perceived from the crystal structure shown in figure 1, with the first one partially replacing the Al sites with suitably larger trivalent ions to enlarge the dodecahedral interstices via forming Gd₃(Al_1−xM_x)₅O₁₂ solid solution and the second one being partially replacing Gd³⁺ with a smaller Ln³⁺ to form (Gd_1−xLn_x)₃Al₅O₁₂. Ga³⁺ is the main choice in the former case, and Gd₃Ga₅O₁₂ (GGG), known as a thermodynamically stable garnet host for phosphors and solid lasers [2], is an extreme example. The effectiveness of Ga³⁺ doping was experimentally demonstrated by Chiang et al [23], who found that phase-pure garnet can be crystallized from chemically precipitated precursors at ∼1400 °C in the presence of 10 at% of Ga³⁺ and the crystallization temperature decreases to 1300 °C with 20 at% of Ga³⁺ addition. Without Ga³⁺ doping, only a phase mixture of LnAP, LnAG and amorphous alumina was formed. By applying the same stabilization strategy, Kamada et al were able to grow two-inch-diameter Gd₃(Al₂Ga₃)O₁₂:Ce³⁺ single crystals by the Czochralski (Cz) method using [100] oriented seeds [24] and Gd₃(Ga,Al)₅O₁₂:Pr³⁺ single crystals by a micro-pulling down (μ-PD) technique [25] (figure 2). Though Ga³⁺ was thought to exclusively replace Al³⁺ in these studies, atomistic modeling using the static lattice computational approach and pairwise (Buckingham) interatomic potentials by Maglia et al [26] revealed that Ga³⁺, though it prefers to take the octahedral Al³⁺ site, can also be inserted into the dodecahedral position of Gd³⁺ with the generation of anti-site defects owing to its relatively large ionic radius. In addition, suppressing activator oxidation (such as Pr³⁺, Ce³⁺, and Tb³⁺) and Ga³⁺ reduction should be made at the same time to avoid lattice defects and deterioration of optical performance.

Figure 2. Appearance of the Gd₃(Al₂Ga₃)O₁₂ single crystals doped with 1 at% of Ce³⁺ (a) and 1 at% of Pr³⁺ (b). Part (a) reproduced with permission from [24] and part (b) reproduced with permission from [25], copyright 2012 by Elsevier.

Lu³⁺ (0.0977 nm for CN = 8) is the tiniest Ln³⁺ ion and would thus be the most effective to stabilize GAG via replacing the Gd³⁺ site to form a (Gd_1−xLu_x)AG solid solution. With coprecipitated carbonate precursors, Li et al [17] thoroughly studied the effects of Lu content on phase evolution and also properties of the resultant (Gd_1−xLu_x)AG (x = 0–0.5). It was shown that the garnet phase generally crystallizes via LnAM and LnAP intermediates, as is commonly observed for YAG, but the crystallization temperature substantially decreases towards a higher Lu content. With x = 0.3–0.5, phase-pure garnet can even be crystallized at a temperature as low as 1000 °C (figure 3(a)), revealing the significant effectiveness of Lu³⁺ doping. Again, only a phase mixture of LnAG, LnAP and amorphous Al₂O₃ was produced in the absence of Lu³⁺ (figure 3(b)).

Figure 3. The effects of Lu content (x value) on phase evolution of (Gd_1−xLu_x)AG solid solution. Parts (a) and (b) are for calcination temperatures of 1000 and 1500 °C, respectively. Reproduced with permission from [17], copyright 2012 by the American Ceramic Society.

A simultaneous advantage of Lu doping is that it improves the already high theoretical density of GAG (5.97 g cm⁻³). The (Gd_0.5Lu_0.5)AG solid solution, for example, reaches the high value of ∼6.44 g cm⁻³ (figure 4(a)), being close to that of the heavy LuAG (6.73 g cm⁻³). Since Gd is commercially much cheaper than Lu, the (Gd_1−xLu_x)AG solid solutions may replace LuAG to be used as cost effective and high density scintillation materials. Assayed from UV/vis absorption, the (Gd_1−xLu_x)AG solid solutions were found to have increasing optical bandgaps of ∼5.87, 5.97, 6.07, 6.17, 6.27, and 6.37 eV with increasing x from 0 to 0.5 (0.1 interval, figure 4(b)) [17], and the bandgap of (Gd_0.5Lu_0.5)AG has been close to that (∼6.40 eV) of a YAG single crystal [27]. The results may also imply that the luminescence property of a (Gd, Lu)AG based phosphor can be finely tuned by varying the Lu content and the onset of optical transmittance of a transparent (Gd, Lu)AG bulk (single crystal or transparent ceramic) would shift towards a shorter wavelength with increasing Lu incorporation.

Figure 4. Lattice constant and theoretical density (a) and bandgap (b) of the (Gd_1−xLu_x)AG solid solution, as a function of the Lu content. Part (a) reproduced with permission from [17], copyright 2012 by the American Ceramic Society.

Tb³⁺ is the largest single Ln³⁺ for a stable LnAG to be formed, and thus its size can be taken as a reference for a lattice stabilization study of GAG. The minimum amount of Lu³⁺ (∼17 at%) calculated from the ionic size of Tb³⁺ (0.1040 nm for CN = 8), however, is significantly larger than the ∼10 at% found in practice (figure 3(b)). This indicates that stable garnet solid solutions exist if the average ionic size of (Ln₁,Ln₂)³⁺ pair lies in between those of Gd³⁺ (0.1053 nm for CN = 8) and Tb³⁺, in agreement with the fact that TbAG [28–30] and even (Gd_0.9Lu_0.1)AG [31, 32] can be further doped with larger Eu³⁺ (0.1066 nm, CN = 8) and Ce³⁺ (0.1143 nm, CN = 8) for luminescence. Taking the average ionic size of (Gd_0.9Lu_0.1)³⁺ (~0.1045 nm) as a standard, Li et al [33] analyzed the minimum amounts of various small Ln³⁺ that are needed for GAG stabilization, and the x value was predicted to be ∼0.5 for Tb³⁺, 0.3 for Dy³⁺ (0.1027 nm), 0.22 for Y³⁺ (0.1019 nm), 0.2 for Ho³⁺ (0.1015 nm), 0.15 for Er³⁺ (0.1004 nm), 0.13 for Tm³⁺ (0.0994 nm), and 0.11 for Yb³⁺ (0.0985 nm). Practical powder synthesis indeed shows that (Gd_1−xLn_x)AG garnet can be obtained in high phase purity with incorporation of the calculated amount of dopant in each case (figure 5). The results may thus lay a base for flexible materials design by properly combining different types of stabilizers to achieve diverse optical functionalities. The characteristic emission of Ln³⁺ in (Gd_0.5Ln_0.5)AG was observed by the authors for Tb (green), Dy (similarly strong blue and yellow), Ho (green), and Tm (blue), despite the high Ln³⁺ concentration (figure 6).

Figure 5. Powder XRD patterns of the phase pure (Gd_1−xLn_x)AG garnets obtained by doping GAG with the calculated amount (x value) of different Ln³⁺. Reproduced with permission from [33], copyright 2013 by Trans Tech Publications.

Figure 6. Emission spectra for the (Gd_0.5Ln_0.5)AG solid solution, with the excitation wavelength (E_x) and the origin of luminescence indicated. Reproduced with permission from [33], copyright 2013 by Trans Tech Publications.

3. Down-conversion (DC) phosphors based on GAG

DC phosphors are generally referred to those that absorb high-energy photons and re-emit them at longer wavelengths in the visible range. There are many host lattices for DC luminescence, and the most extensively employed ones may include single-/multi-cation oxides, oxysulfides, phosphates, vanadates, borates, and molybdates/tungstates. Different hosts are used in practice to meet different application needs. Though almost all of the optically active Ln³⁺ can be used as the activator for DC luminescence, the four ions of Ce³⁺, Eu³⁺, Tb³⁺, and Dy³⁺ are the most efficient since their energy gaps between the lowest emission state and ground state are sufficiently wide to avoid significant non-radiative cross relaxations. The emission behaviors of these activators are generally governed by their site symmetry in the host lattice, lattice covalency, lattice defects/distortion, and particle size/shape (surface effects).

3.1. Eu³⁺ doping for red luminescence

Eu³⁺ is well known for its orange red/red emission arising from the ⁵D₀ → ⁷F_1,2 electronic transitions. The Eu³⁺ ions doped in LnAG are assumed to replace the Ln³⁺ sites and thus inherit a D₂ point symmetry, which is only slightly perturbed from the highly symmetric D_2h point group [34]. For this, the emission of YAG:Eu and LuAG:Eu is dominated by the parity-law allowed ⁵D₀ → ⁷F₁ magnetic dipole transition at ∼590 nm rather than the forced ⁵D₀ → ⁷F₂ electric dipole transition at ∼610 nm as observed from the well-known Y₂O₃:Eu red phosphor. A [(Gd_1−xLu_x)_1−yEu_y]AG solid solution has recently been developed as efficient red phosphor with Lu³⁺ as the lattice stabilizer, and the effects of various factors on optical properties were thoroughly investigated [31]. Taking [(Gd_0.7Lu_0.3)_1−yEu_y]AG for example, the material was shown to be efficiently excitable with the charge transfer band (CTB) at ∼239 nm to produce a sharp orange–red emission at 591 nm (figure 7), with CIE chromaticity coordinates of (0.62, 0.38) and a full width at half maximum of only ∼6 nm for the emission peak. The optimal Eu³⁺ content was experimentally determined to be ∼5 at% (y = 0.05), and the quenching mainly resulted from exchange interactions, possibly via phonon assisted three Eu³⁺ ion nonresonant interactions. Greatly improved emission intensity and quantum yield, shortened fluorescence lifetime, and increased asymmetry factor of luminescence (the I₅₉₁/I₆₁₀ intensity ratio) were observed along with increasing synthesis temperature up to 1500 °C [31], primarily owing to lattice perfection, defect elimination, and particle growth. The [(Gd_0.7Lu_0.3)_0.95Eu_0.05]AG phosphor synthesized at 1500 °C has internal/external quantum efficiencies (%) of 83.2/56.1, an asymmetry factor of ∼2.85, and a fluorescence lifetime of ∼4.1 ms for the 591 nm emission [31]. The lifetime is close to that (4.66 ms) reported for YAG:Eu [35] but is significantly longer than those (generally 0.5–2.5 ms) of the well-known red phosphors of Y₂O₃:Eu [36, 37], (Gd_1−xLn_x)₂O₃:Eu (Ln = Y, Lu) [22, 38], and La₂O₂S:Eu [39], since the Eu³⁺ activator takes the highly symmetric D₂ lattice site in garnet. Increasing Lu incorporation up to x = 0.5 would lower excitation/emission and also blue-shift the CTB due to gradually decreased covalency of the host lattice (χ = 1.27 for Lu³⁺), for which a minimized Lu content was recommended as long as the garnet lattice can be effectively stabilized [31]. Similar phenomena were observed in the development of (Gd_1−xLn_x)₂O₃:Eu red phosphors (Ln = Y, Lu) [38]. Compared with YAG:Eu, the GAG-based phosphor (figure 7) has an additional excitation band arising from the ⁸S_7/2 → ⁶I_J Gd³⁺transition at ∼275 nm (significantly stronger than the strongest ⁷F_0,1 → ⁵L₆ intra-4f⁶ transition of Eu³⁺ at ∼395 nm), suggesting substantial energy migration from Gd³⁺ to Eu³⁺. The advantage of GAG over YAG as a host lattice was demonstrated in another study by the authors [40]. For example, the internal quantum yield (∼76%) of [(Gd_0.9Lu_0.1)_0.95Eu_0.05]AG is appreciably higher than that (∼71%) of (Y_0.95Eu_0.05)AG at the same temperature of phosphor synthesis. This is primarily owing to the higher lattice covalency of the former, which allows improved excitation absorption and higher probability of electronic transitions. Though [(Gd_1−xLu_x)_0.95Eu_0.05]AG has had a sufficiently high theoretical density and emission intensity, its fluorescence lifetime is too long for scintillation. Shortening the lifetime to below ∼1.0 ms via codoping (such as Pr³⁺) seems necessary for it to compete with the commercialized (Y,Gd)₂O₃:Eu scintillator [41].

Figure 7. Excitation and emission behaviors of the [(Gd_0.7Lu_0.3)_1−yEu_y]AG red phosphors. Reproduced with permission from [31], copyright 2012 by the National Institute for Materials Science.

The carbonate coprecipitation technique, with ammonium bicarbonate as the precipitant, has been able to produce low-aggregation garnet powders, but the primary particles are not separable from each other and are not in a spherical shape (figure 8(a)). Current advances in high-resolution displays not only need finer phosphor particles to improve the resolution by decreasing pixel size but also prefer a spherical particle shape to build a uniform luminescence screen and to improve the brightness of luminescence by minimizing the light scattering on particle surfaces. Urea-based homogeneous precipitation (UBHP) frequently finds success in synthesizing well-dispersed spherical particles of uniform size (monosheres) for single-/multi-cation oxides of the lanthanides [19, 42–48], but failed for YAG in most of the previous studies [49, 50] owing to substantially different solution chemistries of the constituent Y³⁺ and Al³⁺ ions. YAG:Ce phosphor microspheres are thus alternatively made via crystallizing the glassy microbeads quenched from melt droplets produced with laser heating [51]. Xu et al [52] identified that, with nitrate as the rare-earth source in UBHP, the aluminum source plays an essential role in the formation of precursor microspheres for YAG:Nd. They found that ammonium aluminum sulfate (alum) is indispensable and the optimal alum/Al(NO₃)₃ molar ratio is 1/1. Mechanistic study further revealed that microspheres of the Al component are formed at the early stage of precipitation, followed by Y³⁺ precipitation as basic carbonate. Annealing the sulfate-containing precursor at 1100 °C produced YAG:Nd microspheres that can be densified to a translucent state via vacuum sintering of the dry-compacted green body at 1650 °C for 3.5 h. Such a strategy proved similarly successful for [(Lu_1−xGd_x)_0.95Eu_0.05]AG red phosphor microspheres (x ≤ 0.4, figure 8(b)) [53], though Gd³⁺ and Lu³⁺ are different from Y³⁺ in solution chemistry owing to lanthanide contraction. Again, the best results were obtained with alum/Al(NO₃)₃ = 1:1 molar ratio. When Al(NO₃)₃ is the sole Al source, only a gelatinous precursor that would aggregate into a glasslike hard mass upon drying was produced, implying that the sulfate anions from alum have significantly modified the solution chemistry of cations, particularly that of the significantly smaller Al³⁺, and have taken part in precipitation. When the alum/Al(NO₃)₃ ratio is over 1, the primary spheres tend to glue together to form porous precipitates as observed for YAG [50], suggesting that superfluous SO₄²⁻ may serve as a flocculant. The diameter of [(Lu_1−xGd_x)_0.95Eu_0.05]AG microspheres can be finely tuned from ∼500 to 150 nm by increasing the urea/(Al+Ln) molar ratio from 20 to 100, showing the flexibility of the UBHP technique. A photoluminescence study [53] found successively stronger ⁵D₀ → ⁷F₁ emission (591 nm) with increasing x (the Gd content), owing to increased lattice covalency (figure 8(c)), and gradually weaker emission at a decreasing particle size owing to surface effects [47, 54].

Figure 8. Typical TEM (a) and SEM (b) morphologies of the [(Lu_1−xGd_x)_0.95Eu_0.05]AG red phosphor particles obtained via carbonate coprecipitation and urea-based homogeneous precipitation, respectively. Part (c) shows that the emission intensity of the phosphor spheres improves with increasing Gd content. Parts (a) and (b) reproduced from [31], copyright 2012 by the National Institute for Materials Science.

3.2. Ce³⁺ doping for yellow luminescence

The yellow emission of Ce³⁺ arises from the 4f⁰5d¹ → 4f¹5d⁰ (²F_5/2 and ²F_7/2 ground states) inter-configurational electronic transition. As the exposed 5d electron readily interacts with the surrounding anion ligands, Ce³⁺ emission is strongly influenced by centroid shifting and crystal field splitting of the 5d energy level (figure 9) [55]. YAG:Ce³⁺ has been the most prominent and widely used yellow phosphor in LED lighting, since it can be efficiently excited with (Ga, In)N blue LED chips (∼450 nm) and exhibits a high quantum yield of ∼90% for its ∼540 nm emission [7]. One shortcoming is that, for warm-white lighting, YAG:Ce³⁺ has low color rendering and high correlated color temperature due to its lack of a sufficient red portion in the emission spectrum. To overcome this, partially replacing the Y³⁺ sites with less electronegative La³⁺ or Gd³⁺ [6, 56] and more recently doping YAG with Si₃N₄ to form the oxynitride solid solution of Y_2.925Ce_0.075Al_5−xSi_xO_12−xN_x (x < 0.4 for phase pure garnet) [57] were shown to be able to push down the lowest crystal-splitting component (²D_3/2) of the 5d¹ excited level to yield red-shifted emissions. Red-shifting can also be achieved by partially substituting Al³⁺ with Mg²⁺–Si⁴⁺ pairs on the octahedral and tetrahedral sites, respectively, to enhance lattice covalency [58, 59]. Alternatively, red-shifted Ce³⁺ emission can be directly attained with a more covalent host lattice, such as TbAG and GAG-based garnets. With single crystal films, Zorenko et al [29] found that TbAG:Ce exhibits broad band emission peaking at 550 nm under 470 nm excitation [4f¹(²F_5/2) → 5d¹(E_2g) Ce³⁺ transition], with light yields of ∼62–71% depending on the Ce³⁺ content. An efficient Tb³⁺ → Ce³⁺ energy transfer was identified through directly exciting the Tb³⁺ ions in the host lattice at 325 nm. Chiang et al [23] found that the emission wavelength of Ga³⁺-stabilized (Gd_0.97Ce_0.03)₃(Al_1−xGa_x)₅O₁₂ yellow phosphors gradually shortens from ∼565 to 552 nm (λ_ex = 470 nm) with increasing Ga³⁺ substitution from x = 0.1 to 0.3, yet substantially longer than the ∼540 nm emission of YAG:Ce, and the shortening was ascribed to the higher electronegativity of Ga³⁺ (χ = 1.81) than Al³⁺ (χ = 1.61). Li et al [32] studied in detail the synthesis and optical properties of [(Gd_1−xLu_x)_1−yCe_y]AG yellow phosphors. It was found that 1 at% (y = 0.01) of much larger Ce³⁺ (0.1143 nm) can be doped into the garnet lattice in the presence of 10 at% of Lu (x = 0.1) and more Ce³⁺ needs more Lu³⁺. The optimal Ce³⁺ concentration was experimentally determined to be ∼1 at%, and luminescence quenching mainly resulted from exchange interactions. Intensity ratio (I_b/I_a) of the 460 nm [4f¹(²F_5/2) → 5d¹(E_2g)] to 340 nm [4f¹(²F_5/2) → 5d¹(T_2g)] excitations was observed to significantly increase from ∼4.8 at y = 0.01 to ∼9.0 at y = 0.02 and then to ∼12 at y = 0.03, due to successively stronger non-radiative absorptions. Energy transfer from Gd³⁺ to Ce³⁺ was identified from the appearance of ⁸S_7/2 → ⁶I_J Gd³⁺ transition at ∼275 nm. Figure 10 compares the emission spectra of [(Gd_1−xLu_x)_0.99Ce_0.01]AG (λ_ex = 455 nm), (Y_0.99Ce_0.01)AG (λ_ex = 454 nm), and (Lu_0.99Ce_0.01)AG (λ_ex = 448 nm), from which it is seen that the emission covers the broad range of ∼475–750 nm in each case and the peak wavelength of [(Gd_1−xLu_x)_0.99Ce_0.01]AG red-shifts relative to those of YAG:Ce and LuAG:Ce even at the high Lu content of 50 at% (x = 0.5). Increasing Lu incorporation steadily shortens the emission wavelength due to decreased lattice covalency by the high electronegativity of Lu³⁺ (χ = 1.27) and monotonically lowers the emission intensity possibly owing to lattice distortion and defect introduction. The best luminescent [(Gd_0.9Lu_0.1)_0.99Ce_0.01]AG has an integrated emission intensity ∼97% of (Y_0.99Ce_0.01)AG and ∼128% of (Lu_0.99Ce_0.01)AG at the same temperature of powder synthesis. The excellent emission, high theoretical density, and relatively low cost of [(Gd_0.9Lu_0.1)_0.99Ce_0.01]AG may allow it to compete with YAG:Ce and particularly LuAG:Ce for scintillation applications. CIE chromaticity coordinates (figure 11) of the three phosphors are around (0.48, 0.51), (0.39, 0.57), and (0.31, 0.58), corresponding to color temperatures of ∼3044, 4612 and 6010 K, respectively. The chromaticity data again confirm that the GAG-based yellow phosphor has a stronger red component in its emission and is more desirable for warm-white LED lighting.

Figure 9. A schematic energy diagram showing the effects of host lattice A on centroid shift and crystal field splitting of the Ce³⁺ 5d energy level. Such effects shorten the energy difference between the 5d excited state and 4f ground state, known as red-shift D(A). Reproduced with permission from [55], copyright 2013 by Annual Reviews Inc.

Figure 10. Emission spectra for the [(Gd_1−xLu_x)_0.99Ce_0.01]AG, (Lu_0.99Ce_0.01)AG, and (Y_0.99Ce_0.01)AG yellow phosphors. Reproduced with permission from [32], copyright 2013 by the National Institute for Materials Science.

Figure 11. Emission color coordinates for the (Gd, Lu)AG:Ce, YAG:Ce, and LuAG:Ce yellow phosphors.

3.3. Tb³⁺ doping for green luminescence

When a Tb³⁺ activator is excited with light of sufficient energy, such as UV light, its 4f⁸ electrons would be raised to the higher 4f⁷5d¹ level and then fed to the ⁵D₃,₄ excited states, from which fluorescence is produced by transitions to the ⁷F_J (J = 1–6) ground states. Though the excited 5d electron is exposed, Tb³⁺ transitions involve only a redistribution of electrons within the inner 4f sub-shell [60], and thus similar emissions are usually observed from different types of host lattices, with the ⁵D₄ → ⁷F₅ green emission at ∼545 nm being dominant.

YAG:Tb green phosphors are widely studied for applications in cathode ray tubes and flat panel displays such as FED and electroluminescent display, since it is thermally stable and resists saturation under high-current excitations [61, 62]. The GAG-based green phosphor of (Gd, Lu)AG:Tb could be an alternative choice for these purposes and improved performance might also be expected from the possible Gd³⁺ → Tb³⁺ energy transfer. The effects of Tb³⁺ content on photoluminescence of [(Gd_0.8Lu_0.2)_1−xTb_x]AG were studied in figure 12, from which the quenching concentration was determined to be ∼10 at% (x = 0.1), almost identical to that of YAG:Tb [63], and luminescence quenching was suggested to occur via exchange interactions [53]. The excitation spectrum consists of three 4f⁸ → 4f⁷5d¹ transition bands at ∼227 nm (E^1–2₃ level, spin allowed), 276 nm (E^1–3₂ level, spin allowed), and 323 nm (E₁ level, spin forbidden), with the 276 nm one being dominant as widely observed [29]. It should be noted that the ⁸S_7/2 → ⁶I_J Gd³⁺ transition (∼275 nm) well overlaps the 276 nm excitation, suggesting the likelihood of Gd³⁺ → Tb³⁺ energy transfer, since the ⁶I_J state of Gd³⁺ lies higher than the ⁵D_3,4 emission states of Tb³⁺ in the energy diagram of excited states for Ln³⁺ [64–66]. The emission spectrum obtained under 276 nm excitation has four groups of bands at ∼490 (blue), 545 (green, the strongest), 589 (yellow), and 623 nm (red), corresponding to the ⁵D₄ → ⁷F_6,5,4,3 transitions, respectively. Emission from the higher ⁵D₃ excited level, usually in the 450–490 nm region, is hardly observable, which can be explained by cross-relaxation via resonance between the excited and ground states of two Tb³⁺ ions, that is, populating the ⁵D₄ level by quenching the ⁵D₃ level via Tb³⁺ (⁵D₃) + Tb³⁺ (⁷F₀) → Tb³⁺ (⁵D₄) + Tb³⁺ (⁷F₆) [67]. Emission from the ⁵D₃ state was experimentally found for YAG doped with 1 at% [61] but not with 5 at% of Tb³⁺ [62], and 1 at% is generally accepted as the up-limit for the ⁵D₃ emission to appear in many hosts. It is noteworthy that the charge transfer state (CTS) of the host lattice also determines the occurrence of ⁵D₃ emission [68]. In La₂O₂S:Tb³⁺, for example, it is completely quenched even at very low Tb³⁺ concentrations, not owing to cross relaxation but by thermal excitation of the ⁵D₃ electrons into CTS since the two states have similar energies [68]. Comparative studies showed that the [(Gd_1−xLu_x)_0.9Tb_0.1]AG phosphors with x = 0.1 and 0.2 have emission intensities close to (Y_0.9Tb_0.1)AG and (Lu_0.9Tb_0.1)AG, though the latter two have better crystallinity owing to their ease of crystallization, and have fluorescence lifetimes of ∼3.31 ms (3.18 ms for YAG:Tb) and CIE color coordinates of (0.35, 0.57) [53]. Electron-beam excited luminescence of (Gd,Lu)AG:Tb³⁺ is yet needed to study for the aforesaid applications.

Figure 12. Excitation (λ_em = 545 nm) and emission (λ_ex = 276 nm) spectra for the [(Gd_0.8Lu_0.2)_1−yTb_y]AG green phosphors.

3.4. Dy³⁺ doping for white luminescence

The primary interest in using Dy³⁺ as an activator is that it simultaneously emits blue (∼483 nm, ⁴F_9/2→⁶H_15/2 transition) and yellow (∼584 nm, ⁴F_9/2 → ⁶H_13/2 transition) lights, which are needed to develop white light in LEDs and optical display systems [69]. Dy³⁺-containing compounds are also used as thermographic phosphors to measure surface temperature by applying a thin coating of the phosphor to the substrate [70]. The luminescence behavior of Dy³⁺ is governed by parity law, and in the cubic lattice of Ln₂O₃ sesquioxide the emission spectrum is dominated by the yellow band at ∼584 nm [71, 72]. This is because the blue (parity allowed) and yellow (parity forbidden) emissions come from the Dy³⁺ ions taking symmetric and non-symmetric (or low symmetric) lattice sites, respectively, while in this type of oxide the centrosymmetric S₆ site has a much lower occupancy (25%) than the non-centrosymmetric C₂ site (75%) [73]. Relative intensity of the blue emission can be improved in YAG lattice owing to higher site symmetry, but the overall emission intensity is rather limited since within the 4f⁹ configuration of Dy³⁺ the excited electrons have high probabilities of non-radiative cross relaxation owing to the relatively limited energy gap between the excited and ground states and also the abundant energy multiplets for both the states [64, 66]. (Gd,Lu)AG was recently demonstrated to be significantly superior to YAG as the host for Dy³⁺ emission [74]. With the more covalent lattice and particularly via an efficient Gd³⁺ → Dy³⁺ energy transfer, greatly enhanced blue and yellow emissions were simultaneously attained. The optimal Dy³⁺ concentration was found to be ∼2.5 at%, close to the ∼2.0 at% reported for YAG [69], above which luminescence quenching occurs via dipole-dipole interactions. The excitation behaviors of [(Gd_1-xLu_x)_0.975Dy_0.025]AG, (Y_0.975Dy_0.025)AG, and (Lu_0.975Dy_0.025)AG (x = 1.0) are compared in figure 13, where the intra-4f⁹ excitations of Dy³⁺ are similarly found at ∼326, 352, 366, and 386 nm for the ⁶H_15/2 to ⁶P_3/2, ⁴I_11/2+⁴M_15/2+⁶P_7/2, ⁴P_3/2+⁶P_3/2,_5/2, and ⁴I_13/2+⁴F_7/2+⁴K_17/2+⁴M_19/2,21/2 transitions, respectively. It is also seen that the main excitation at 352 nm is generally stronger for [(Gd_1−xLu_x)_0.975Dy_0.025]AG than (Y_0.975Dy_0.025)AG and particularly (Lu_0.975Dy_0.025)AG owing to the lower electronegativity of the (Gd_1−xLu_x)³⁺ pair. A significant difference is that [(Gd_1-xLu_x)_0.975Dy_0.025]AG has an additional excitation band at 275 nm, being the strongest in the whole excitation spectrum, that corresponds to the ⁸S_7/2 → ⁶I_J Gd³⁺ transition, indicating the happening of efficient Gd³⁺ → Dy³⁺ energy transfer. The ⁸S_7/2→ ⁶P_J Gd³⁺ transition appears at ∼312 nm. Figure 14 compares luminescence spectra of the three types of phosphors, from which it is seen that neither the Lu content nor excitation wavelength (275 or 352 nm) brings about appreciable change to the peak position. Emission intensity of the Gd-containing phosphor under 275 nm excitation is roughly two times that under 352 nm excitation, implying that the energy transfer is of high efficiency. Exciting the most luminescent [(Gd_0.8Lu_0.2)_0.975Dy_0.025]AG phosphor under 275 nm produced an emission intensity roughly six and three times those of (Lu_0.975Dy_0.025)AG and (Y_0.975Dy_0.025)AG under 352 nm excitation, respectively (figure 14(a)). Even under identical excitation at 352 nm, the emission intensity of [(Gd_0.8Lu_0.2)_0.975Dy_0.025]AG is about 3.1 and 1.5 times those of (Lu_0.975Dy_0.025)AG and (Y_0.975Dy_0.025)AG, respectively (figure 14(b)). Furthermore, the [(Gd_1−xLu_x)_0.975Dy_0.025]AG phosphor has color coordinates of (0.33, 0.35), very close to the ideal white point of (0.33, 0.33) in the CIE chromaticity diagram, with a color temperature of ∼5609 K [74].

Figure 13. Excitation spectra for the [(Gd_1−xLu_x)_0.975Dy_0.025]AG, (Lu_0.975Dy_0.025)AG, and (Y_0.975Dy_0.025)AG white phosphors (λ_em = 483 nm). Reproduced with permission from [74], copyright 2013 by the Royal Society of Chemistry.

Figure 14. Emission spectra for the [(Gd_1−xLu_x)_0.975Dy_0.025]AG, (Lu_0.975Dy_0.025)AG, and (Y_0.975Dy_0.025)AG white phosphors, taken under excitations with the ⁸S_7/2 → ⁶I_J Gd³⁺ transition at 275 nm (part (a)) and the intra-4f⁹ Dy³⁺ transition at 352 nm (part (b)). Reproduced with permission from [74], copyright 2013 by the Royal Society of Chemistry.

3.5. Eu³⁺/Tb³⁺ codoping for color tunable luminescence

Energy transfer between two types of activators is widely utilized in the phosphor field to tune the emission color, to produce a specific color that cannot be attained with one single type of activator, and to enhance the desired emission. The Ce³⁺/Tb³⁺ and Tb³⁺/Eu³⁺ combinations are among the most frequently adopted activator pairs. In the former, the direction of energy transfer largely depends on the 5d¹ energy level of Ce³⁺, which is as aforesaid readily subjected to centroid shift and crystal field splitting [55]. For example, Ce³⁺ → Tb³⁺ energy transfer is found in CePO₄:Tb [75] while Tb³⁺ → Ce³⁺ in TbAG:Ce [28, 29]. Dorenbos [76] determined that crystal field splitting of the Ce³⁺ 5d level is affected by coordination geometry, and tends to decrease following the order: octahedral > cubic > dodecahedral > tricapped trigonal prisms and cuboctahedral. Only Tb³⁺ → Eu³⁺ transfer can be observed for the Tb³⁺/Eu³⁺ pair, since the ⁵D_3,4 excited states of Tb³⁺ lie higher than the ⁵D_0,1 emission states of Eu³⁺ and both the ions have relatively fixed energy levels for the 4f electrons [64–66]. The Tb³⁺ → Eu³⁺ energy transfer is of high efficiency ( can be ∼90%, for example), because of significant overlapping of the emission spectrum of Tb³⁺ with the excitation spectrum of Eu³⁺ [77, 78]. With such an energy transfer, occurring via electric multipole interactions [78], the emission color of Tb³⁺/Eu³⁺ codoped Y₂O₃ can be finely tuned between red and green by varying the atomic ratio of the two activators [78]. Energy transfer and emission control were recently studied for the GAG-based phosphor of [(Gd_0.8Lu_0.2)_0.9−xTb_0.1Eu_x]AG [53], where the Eu content was varied from x = 0 to 0.1. The excitation spectra taken for the Tb³⁺ green emission at ∼545 nm and the Eu³⁺ red emission at ∼592 nm are shown in figure 15. For Tb³⁺ emission (figure 15(a)), only the characteristic excitation bands of Tb³⁺ are resolved, with the inter-configurational 4f⁸ → 4f⁷5d¹ transition at ∼276 nm being dominant as found for (Gd,Lu)AG:Tb³⁺. Intensity of the excitation significantly decreases with increasing Eu³⁺ addition and finally becomes negligible at x = 0.1, primarily owing to Tb³⁺ → Eu³⁺ energy transfer and also concentration quenching at high total contents of the two activators. The excitation spectra taken for Eu³⁺ emission are, however, dominated by Tb³⁺ transitions, and only very weak CTB and intra-4f⁶ transitions originated from Eu³⁺ are found (figure 15(b)). This indicates that, in the codoped system, exciting Tb³⁺ is the only efficient way to produce Eu³⁺ luminescence through energy transfer. Intensity of the 276 nm excitation reaches its maximum at x = 0.03, followed by a steady decrease at higher Eu contents owing to concentration quenching. Figure 16(a) analyzes intensities of the 592 nm Eu³⁺ and 545 nm Tb³⁺ emissions (λ_ex = 276 nm), where the strongest emission is normalized to 10 for both the activators. It is seen that the Tb³⁺ emission is monotonically weakened at a higher Eu content while the Eu³⁺ emission gradually gains intensity up to x = 0.03 and then decreases, following the tendency found from the excitation spectra. The I₅₉₂/I₅₄₅ intensity ratio steadily increases with increasing Eu³⁺ incorporation, which may suggest a persistent energy transfer from Tb³⁺ to Eu³⁺ or the quenching of Eu³⁺ emission is less than that of Tb³⁺. The CIE color coordinates shown in figure 16(b) indicate that the emission can be well tuned from green to orange red via yellow (figure 17). Further analysis indicated that energy transfer may have occurred via electric dipole-quadrupole interactions [53]. It should be noted that the energy process is more complicated for (Gd, Lu)AG than Y₂O₃ owing to the presence of optically active Gd³⁺. Since the ⁸S_7/2 → ⁶I_J Gd³⁺ transition well overlaps the 4f⁸ → 4f⁷5d¹ Tb³⁺ transition at ∼276 nm, multichannel energy transfer is highly possible, including Gd³⁺ → Tb³⁺ → Eu³⁺, Tb³⁺ → Eu³⁺ and Gd³⁺ → Eu³⁺ (figure 17), though further studies are needed to clarify the exact routes. The excitation behavior of Eu³⁺ and the significantly lowered Tb³⁺ while improved Eu³⁺ emissions up to x = 0.03, however, unambiguously reveal the presence of Tb³⁺ → Eu³⁺ transfer path.

Figure 15. Excitation spectra for the [(Gd_0.8Lu_0.2)_0.9−xTb_0.1Eu_x]AG phosphors, taken by monitoring the green Tb³⁺ emission at 545 nm (a) and the red Eu³⁺ emission at 592 nm (b).

Figure 16. Relative intensities (a) of the 545 nm Tb³⁺ and 592 nm Eu³⁺ emissions and color coordinates (b) of the Tb³⁺/Eu³⁺ co-activated [(Gd_0.8Lu_0.2)_0.9−xTb_0.1Eu_x]AG phosphors.

Figure 17. A scheme showing possible pathways of energy transfer (left) in the [(Gd_0.8Lu_0.2)_0.9−xTb_0.1Eu_x]AG phosphor and digital pictures (right) showing color-tunable emission through the energy transfer (excitation: 275 nm).

4. UC phosphors based on GAG

UC luminescence is an anti-Stokes process in which a longer wavelength radiation, usually near infrared (NIR) or IR light, is converted to a shorter wavelength such as UV or visible light via a two-photon or multi-photon mechanism [79]. The materials are drawing wide interest since they can be used as biological labels for medical diagnosis and therapy, in photovoltaic cells to efficiently harvest solar energy, and also in laser and anti-counterfeit fields [79]. The activators used for UC are those that usually do not exhibit efficient DC luminescence, including Pr³⁺, Sm³⁺, Ho³⁺, Er³⁺, and Tm³⁺. Though when properly doped in a host the above activators themselves have been able to produce UC emission, the efficiency is usually rather limited owing to their unsatisfactory NIR/IR excitations. Yb³⁺ is thus widely employed as a codopant to improve NIR absorption (at ∼980 nm, the ²F_7/2 → ²F_5/2 Yb³⁺ transition) and to sensitize UC emission via nonradiative energy transfer from Yb³⁺ to the activator. The most preferred and widely used hosts for UC luminescence are fluorides owing to their low phonon energy, though other material types, such as Y₂O₃ transparent ceramics [80, 81] and Gd₂O₂S powder [82], were also explored. YAG was suggested to possess a large ground-state Stark splitting and has a quasi three-level energy structure, which may enable a broad and intense absorption of Yb³⁺ in it [83]. The energy transferred from Yb³⁺ may effectively populate the upper level of Tm³⁺ in YAG [83]. UC performances were recently studied for stabilized GAG with the compositions of [(Gd_1-xLu_x)_0.948Yb_0.05Ln_0.002]AG (x = 0.1–0.5, Ln = Er, Ho, and Tm) [53, 84, 85]. Despite the dilute Yb and Ln concentrations, strong UC luminescence was clearly observed in each case upon laser exciting Yb³⁺ at 978 nm, as shown in figure 18 together with the mechanism of UC. The UC luminescence presents as an intense blue band at ∼487 nm (¹G₄ → ³H₆ transition) and a weaker red one at ∼650 nm (¹G₄ → ³F₄) for Tm³⁺, as a fairly strong green band at 543 nm (⁵F₄,⁵S₂ → ⁵I₈) and a strong red band at 668 nm (⁵F₅ → ⁵I₈) for Ho³⁺, and as three bands at 525 nm (green, ²H_11/2 → ⁴I_15/2), 556 nm (yellow, ⁴S_3/2 → ⁴I_15/2) and 655 nm (red, ⁴F_9/2 → ⁴I_15/2) for Er³⁺. CIE chromaticity coordinates of the UC luminescence were found to be around (0.14, 0.19) for Tm³⁺, (0.38, 0.58) for Ho³⁺, and (0.34, 0.64) for Er³⁺, corresponding to blue, greenish yellow, and green colors, respectively (figure 19). Lowered luminescence with increasing Lu incorporation and enhanced emission with increasing temperature of phosphor synthesis were found for the above UC systems. Analyzing the emission intensity against excitation power indicated that the UC luminescence may have occurred via a three-photon process for Tm³⁺ and a two-photon process for both Ho³⁺ and Er³⁺, which are schematically shown in the right part of figure 18 [53, 84, 85].

Figure 18. Up-conversion (UC) luminescence spectra of the Yb/Ho, Yb/Er, and Yb/Tm codoped (Gd_1−xLu_x)AG solid solutions. Mechanisms of the UC processes are presented in the right-hand schemes. Reproduced with permission from [85], copyright 2014 by Trans Tech Publications, and [84], copyright 2014 by Elsevier.

Figure 19. Color coordinates for the UC luminescence of (Gd, Lu)AG:Yb/Ln (Ln = Er, Ho, and Tm).

5. Single-crystal and transparent-ceramic scintillators based on GAG

A scintillator is essentially a luminescent material that absorbs high-energy photons and then emits visible light, for which efficient absorption of the excitation source is a fundamental requirement [41, 86]. Since the relation among absorption coefficient (η_abs), theoretical density (ρ), and effective atomic number (Z_eff) can be expressed as η_abs = ρZ_eff⁴ [87], high theoretical density (generally >6 g cm⁻³) and particularly high effective atomic number are thus needed for an excellent scintillator, though other characteristics such as high light yield and fast response (10–100 ns, dominant decay generally <3 μs to avoid signal pile-up with standard shaping electronics) are also essential [86]. Scintillators combined with photodetectors are widely used in various medical imaging technologies, such as x-ray computed tomography, positron emission tomography (PET), and single-photon emission computed tomography (SPECT), and also in high energy and nuclear physics. The most common scintillators up to date are CsI:Tl, CdWO₄, and Bi₄Ge₃O₁₂ (BGO) single crystals, but they have their respective shortcomings such as hygroscopicity, poor machinability, insufficient light output, and slow blinking [86]. For these reasons, Ce-doped silicates, such as Gd₂SiO₅ (GSO), Lu₂SiO₅ (LSO), (Lu_1−xY_x)₂SiO₅ (LYSO), and LaBr₃ are being developed as alternatives [88–94].

GAG-based single crystal scintillators are mostly studied by Kamada et al [24, 25, 95–97] through crystal growth by the μ-PD and CZ techniques. The CZ-grown Gd₃(Al₂Ga₃)O₁₂:1 at%Ce single crystal, where Ga³⁺ is a lattice stabilizer, was reported to have a high light yield of 46 000 photons/MeV for the Ce³⁺ emission (31 000 photons/MeV for LYSO:Ce), an energy resolution of 4.9% at 662 keV, a primary decay time of 88 ns (91%), and a high theoretical density of 6.63 g cm⁻³ [24]. The crystals are thus regarded as promising scintillators for PET, SPECT, and gamma camera. The μ-PD grown Gd₃(Al_5−xGa_x)O₁₂:1 at%Pr single crystals (x = 1–5) were found to exhibit the 5d → 4f and intra-4f² transition emissions of Pr³⁺ in the 300–400 and 480–600 nm (dominant) regions, respectively, together with the intra-4f⁷ Gd³⁺ emission at 310 nm (figure 20) [25]. A higher Ga³⁺ content would suppress the 5d → 4f Pr³⁺ emission while enhance the f–f transitions of both Gd³⁺ and Pr³⁺. A low light output of ∼4500 photons/MeV, only about 1/5 of the CZ-grown LuAG:Pr standard, and a relatively long primary decay time of 214 ns (98.8%) were reported for the Gd₃(Al₂Ga₃)O₁₂:1 at%Pr single crystal [25]. The poor light output was suggested to be associated with an energy transfer from the 5d state of Pr³⁺ to the 4f state of Gd³⁺ and non-radiative relaxations from the 5d to 4f states of Pr³⁺, with the former being dominant, as schematically shown in figure 21 [98]. Pr³⁺ was thus suggested not to be a proper activator for Gd³⁺-containing scintillators. By simultaneously modifying GAG with Lu³⁺ for Gd³⁺ and Ga³⁺ for Al³⁺, (Gd₂Lu₁)(Al_5−xGa_x)O₁₂:Ce single crystals (x = 1–5; Ce³⁺ content: 0.2, 1.0, and 3 at%) were grown by the μ-PD technique and thoroughly investigated for their luminescence properties [95]. It was shown that decay of the 5d → 4f Ce³⁺ emission at ∼520 nm accelerates with increasing Ga or Ce concentration, and the best composition of (Gd₂Lu₁)(Al₂Ga₃)O₁₂:1 at%Ce has a light yield of ∼22 000 photons/MeV, about 70% of LYSO:Ce (31 000 photons/MeV), a theoretical density of 6.88 g cm⁻³, and a decay time of 76.5 ns (83%) and 282 ns (17%). With the combinatorial approach, Kamada et al made comprehensive composition optimization for 0.2 at% Ce³⁺ doped (Gd_3−xLu_x)(Al_5−yGa_y)O₁₂ [96] and (Gd_3−xY_x)(Al_5−yGa_y)O₁₂ [97] single crystals. The light output of Ce³⁺ in the best hosts of (Gd₂Y₁)(Al₂Ga₃)O₁₂ and Gd₃(Al₂Ga₃)O₁₂ reached ∼42 000–44 000 photons/MeV, being ∼150% of the value of LYSO:Ce and 730% of that of BGO (5700 photons/MeV), with the scintillation decay time dominated by 50–80 ns. The energy resolution at 662 kV was determined to be 8.3% [96], which, though inferior to the 6.7% for high-quality CZ-grown LuAG:Ce, is comparable to the 8.7% measured for LYSO:Ce. Bandgap engineering was pointed out to be crucial in developing high quality scintillators, as also suggested by previous studies. For example, the 5d → 4f luminescence of Ce³⁺ is quenched in the high density garnets of Lu₃Ga₅O₁₂ (7.4 g cm⁻³) and Gd₃Ga₅O₁₂ (7.04 g cm⁻³) owing to positioning of the Ce³⁺ 5d states in the conduction band of the host [99], and the performance of LuAG:Ce is strongly degraded by shallow electron traps (Lu_Al anti-site defects) via delaying energy transfer to Ce³⁺ and giving rise to quite slow components in the scintillation response [100, 101]. A balanced Ga and Gd admixture may eliminate the trapping effects by burying the shallow traps in the bottom of the conduction band and at the same time avoid ionization of the Ce³⁺ activators by separating the 5d excited level from the conduction band, as schematically shown in figure 22 [96, 97].

Figure 20. Radioluminescence spectra of the Gd₃(Al_5−xGa_x)O₁₂:1 at%Pr single crystals under γ-ray excitations. Reproduced with permission from [25], copyright 2012 by Elsevier.

Figure 21. Energy diagram for the Gd³⁺ and Pr³⁺ centers in (Gd_xLu_3−x)(Ga₃Al₂)O₁₂ (x < 0.2), with the energy transfer channel from Pr³⁺ to Gd³⁺ indicated. The right-hand scheme depicts non-radiative relaxation from the lowest 5d¹ to low-lying ³P₀ and ¹D₂ levels of Pr³⁺. Reproduced with permission from [98], copyright 2013 by Elsevier B V.

Figure 22. Energy level scheme showing bandgap and 5d¹ level engineering for the (Gd, Lu)₃(Ga, Al)₅O₁₂:Ce scintillation crystals. The Ga component helps to lower the conduction band (CB) to bury the shallow traps while the Gd component pushes away the 5d¹ level of Ce³⁺ from the bottom of CB to avoid Ce³⁺ ionization. Reproduced with permission from [96], copyright 2011 by the American Chemical Society.

Aside from single crystals, transparent ceramics are under active development for scintillation applications, since the sintering technique generally has advantages in product size and fabrication cost. The (Y,Gd)₂O₃:Eu transparent ceramic (inline transmittance ∼73% at the 610 nm emission of Eu³⁺ or ∼90% of the theoretical value, figure 23), fabricated via pressureless sintering, pressureless sintering plus hot isostatic pressing, and vacuum hot pressing, has been the first commercialized polycrystalline scintillator used in medical x-ray detectors [41, 86]. The (Y_0.67Gd_0.30Eu_0.03)₂O₃ composition (5.92 g cm⁻³) was reported to have a relative light output 2.5 times higher and ∼30% lower than those of CdWO₃ and CsI:Tl single crystals, respectively [41]. Li et al [102] fabricated transparent (Y_0.3Gd_0.67Eu_0.03)₂O₃ ceramic with a higher Gd content, an inline transmittance of ~68% at the 610 nm Eu³⁺ emission and a higher theoretical density of 6.87 g cm⁻³, by vacuum sintering at 1670 °C for 2 h of the oxide powders calcined from coprecipitated carbonate precursors. The main problem associated with Eu³⁺ emission is the long fluorescence decay, which is usually ∼1 ms [41, 86]. Yanagida et al [8, 103] made transparent YAG:Ce and (Gd,Y)AG:Ce ceramics for possible scintillation applications (the Gd content and optical transmittance not specified), and the dominant decay time was found to be ∼90 ns under γ-ray excitations. Light yield of the latter was reported to be 30% of the former, without specifying the reason, and the energy resolution was found to be ∼8% at 662 keV when coupled with an avalanche photodiode [103]. The stopping power of (Gd, Y)AG:Ce was reported to be five times as high as that of YAG:Ce [103]. Li et al [104] studied sintering of (Y_1−xGd_x)AG:Ce (x = 0.01, 0.5, 0.75, and 1; Ce content: 2 at%) transparent ceramics with commercially available nano-sized powders. It was shown that the GAG composition (x = 1.0) cracks owing to the stress (volume change) arising from thermal decomposition while all the other compositions can be densified to transparency by vacuum sintering at 1600–1700 °C for 5 h. The (Y_1.48Gd_1.5Ce_0.02)AG ceramic, with an inline transmittance of ∼65% at 800 nm, exhibits Ce³⁺ emission at ∼570 nm under 340 nm UV excitation, which is red-shifted ∼40 nm relative to YAG:Ce (∼530 nm) owing to the high content of less electronegative Gd³⁺ (figure 24). Cherepy et al [105] studied radioluminescence properties of translucent (Tb_3−xCe_x)AG ceramics (figure 25), and much higher light yields (80 000 photons/MeV) than LuAG:Ce (∼30 000 photons/MeV) were found under β-ray excitations. The decay, however, is slow owing to energy migration on the Tb³⁺ sites, and the principal decay time reaches <1 μs only at the high Ce³⁺ content of x = 0.12. By applying vacuum sintering plus hot isostatic pressing, Cherepy et al [105] also fabricated a series of GAG-based scintillation ceramics (Ce³⁺ content: 3 at%), including (Gd_1.5Y_1.5)₃Al₅O₁₂ (GYAG), (Gd_1.5Y_1.5)₃(Al_5−xSc_x)O₁₂ (GYSAG), and Gd₃(Al₃Sc₂)O₁₂ (GSAG), using Y³⁺/Sc³⁺ as dopant for lattice stabilization (figure 26). GYAG:Ce was reported to have a very high light yield of ∼100 000 photons/MeV under β-ray excitation, due to an efficient energy transfer from Gd³⁺, and have decay time in the 100–200 ns range. Transparency, however, was not achieved for this composition due to the presence of a small amount of GdAlO₃ secondary phase. In contrast, GSAG and GYSAG only produced the moderate light yields of ∼20 000 photons/MeV, with the primary decay time <200 ns, but formed phase-pure garnet with excellent transparency for the former and acceptable transparency for the latter (x ≥ 0.12, transmittance data not shown). In general, the Ce³⁺ emission in Gd-containing scintillators simultaneously has fast primary decay (<200 ns) and high light yield, as compared from table 1 for the typical garnet compounds discussed in this work, showing substantial advantages of the GAG host lattice.

Figure 23. Transmittance and luminescence (excitation: 254 nm UV light) spectra of a 1.5 mm thick (Y_0.67Gd_0.30Eu_0.03)₂O₃ ceramic scintillator. Reproduced with permission from [86], copyright 1997 by Annual Reviews Inc.

Figure 24. In-line transmittance and luminescence (excitation: 340 nm UV light) spectra of a 1.3 mm thick (Y_1.48Gd_1.5Ce_0.02)AG scintillation ceramic. Reproduced with permission from [104], copyright 2010 by the American Ceramic Society.

Figure 25. Beta-excited radioluminescence spectra and appearances of the Ce³⁺ doped TAG and LuAG ceramics with scatter mean free path >1 cm. Reproduced with permission from [105], copyright 2009 by IEEE.

Figure 26. Beta-excited radioluminescence spectra and appearances of the Ce³⁺ doped gadolinium-based garnet ceramics. Reproduced with permission from [105], copyright 2009 by IEEE.

Table 1. A summary of the scintillation properties of the promising single-crystal/ceramic scintillators discussed in this work.

Host lattice	Ce content (at%)	Material form	Density (g cm^−3)	Excitation source	Light yield (photons/MeV)	Resolution at 662 kV (%)	Primary decay (ns)	Reference
LuAG	1.0	Crystal	6.73	γ-ray	7500	66.2 (88%)	—	[95]
LuAG	3.0	Crystal	6.73	γ-ray	28 000	8.6	—	[105]
				β-ray	30 000
LuAG	2.0	Ceramic	6.73	γ-ray	20 000	11.4	—	[105]
				β-ray	30 000
YAG	1.0	Crystal	4.55	γ-ray	22 700	—	83.1 (71%)	[97]
YAG	—	Ceramic	4.55	γ-ray	30 000	7.3	90	[103, 105]
				β-ray	40 000
TbAG	2-16	Ceramic	—	γ-ray	19 500	10.6	4500–600	[105]
				β-ray	80 000
(Gd1.5Y1.5)Al5O12	3.0	Ceramic	—	γ-ray	16 500	11.2	100–200	[105]
				β-ray	>80 000
Gd3(Sc1.89Al3.11)O12	3.0	Ceramic	—	γ-ray	7500	10.8	100–200	[105]
				β-ray	20 000
Gd3(Al2Ga3)O12	1.0	Crystal	6.63	γ-ray	46 000	4.9	88 (91%)	[24]
Gd3(Al2Ga3)O12	0.2	Crystal	6.63	γ-ray	42 217	8.3	52.8 (73%)	[96]
(Gd2Lu1)(Al2Ga3)O12	1.0	Crystal	6.88	γ-ray	22 000	11.2	76.5 (83%)	[95]
(Gd2Y1)(Al2Ga3)O12	1.0	Crystal	—	γ-ray	44 000	8.2	56.9 (66%)	[97]

6. Summary and outlook

Lattice stabilization of GAG (Gd₃Al₅O₁₂, GAG) and the related development of advanced optical materials are summarized in this article, including down-/up-conversion phosphors, transparent ceramics, and single crystals. It is shown that novel emission features and significantly improved luminescence can be achieved for a number of phosphor systems with the more covalent GAG lattice and the efficient energy transfer from Gd³⁺ to the activator. GAG-based single crystals and transparent ceramics with Ce³⁺ as the activator are shown to have fast scintillation decay and high light yields, and thus hold great potential as scintillators for a wide range of applications. Anti-site defects commonly exist in aluminate garnets [106–108], and their occurrence and energy level have profound influences on various aspects of emission through interacting with the excited electrons. These issues, however, have rarely been tackled for the GAG-based garnets, either when Lu³⁺ or Ga³⁺ is used as lattice stabilizer, and thus need to clarify in future studies for a better control of optical properties. In addition, GAG-based transparent ceramics have been much less developed than their YAG and LuAG counterparts, and need significantly more studies on powder processing and sintering technologies to improve their overall transmittance and other optical performances.

Acknowledgment

Experimental contribution to GAG-based phosphors by Dr Jinkai Li, through his PhD thesis work under the supervision of Dr Ji-Guang Li, is acknowledged. Sincere thanks are due to the editorial board of STAM for inviting this review.