Amandine Baillard‡
a,
Paul-Antoine Douissard‡b,
Pavel Loiko*a,
Thierry Martinb,
Eric Mathieub and
Patrice Camya
aCentre de Recherche sur les Ions, les Matériaux et la Photonique (CIMAP), UMR 6252 CEA-CNRS-ENSICAEN, Université de Caen Normandie, 6 Boulevard Maréchal Juin, 14050 Caen Cedex 4, France. E-mail: pavel.loiko@ensicaen.fr; Tel: +33 2 31 45 25 62
bEuropean Synchrotron Radiation Facility (ESRF), 71 Avenue des Martyrs, 38043 Grenoble, France
First published on 4th June 2025
Single-crystal films of terbium-doped gadolinium gallium garnet (Gd3Ga5O12:Tb) were grown by the isothermal dipping liquid phase epitaxy method on undoped (111)-oriented GGG substrates using PbO/B2O3 as a solvent. The effect of the Tb3+ doping level (2 to 10 at%) on the growth parameters, structure, composition, morphology, and emission properties of the films under optical and X-ray excitation was systematically studied. The saturation temperature increased almost linearly with the Tb content. The Tb3+-doped films exhibit a very low lattice mismatch of less than 0.05% with respect to the GGG substrate. The dopant ions are uniformly incorporated in the layers, with a segregation coefficient close to unity. The conversion efficiency of the films is optimized for a doping level of 6 at% Tb3+ in the solution, reaching a maximum light output of 52% with respect to a reference bulk YAG:Ce crystal. The green emission of Tb3+ ions at 543 nm matches with the maximum of sensitivity of CCD/CMOS sensors. The luminescence lifetime of the 5D4 Tb3+ emitting state amounts to ∼2.3 ms and is weakly dependent on the doping level. Minimum afterglow intensities are reached for the GGG:Tb films, as compared to other currently employed scintillators. Gd3Ga5O12:Tb single-crystalline films represent a viable solution for developing novel scintillators providing high efficiency and sub-μm spatial resolution for X-ray imaging.
The spatial resolution R of such detectors can be adjusted through the ratio ∼0.61λ/NA, where λ is the emission wavelength of the detector, and NA is the numerical aperture of the optics.1 Moreover, high quality SCFs with a thickness of a few μm are required for X-ray imaging with sub-μm spatial resolution. Thicker films degrade the contrast for all spatial frequencies contained in the image.5,6
Nowadays, on top of high spatial resolution, detector systems offer the possibility to work at much higher speed with the use of faster CCD/CMOS sensors.7 The basic requirements for developing scintillators for high-resolution imaging have been detailed in the literature, with small variations depending on the targeted application. To combine both sub-μm spatial resolution and high frame rates (>50 frames per s), films with fast scintillation decay (<100 ns at the 1/e level) and especially low afterglow (0.01% after 2 ms) are necessary.6,8
Additionally, the ability to absorb X-rays is also a crucial parameter when selecting a host for a scintillator. This X-ray absorption capability is proportional to ρZeff4, with Zeff and ρ being the effective atomic number and the material density, respectively.6 Consequently, the absorption efficiency is improved by maximizing both parameters.1
Single-crystal film (SCF) scintillators have been elaborated at the European Synchrotron Radiation Facility (ESRF) for sub-μm spatial resolution X-ray imaging. The use of SCFs in X-ray imaging started with cerium-doped Y3Al5O12 (YAG:Ce) garnet films.5 They were soon replaced by Lu3Al5O12 (LuAG) and Gd3Ga5O12 (GGG) films doped with europium ions Eu3+, presenting higher densities and therefore higher efficiencies when integrated in X-ray imaging detectors.6,9 Starting in 2009, the novel Lu2SiO5 (LSO) SCF doped with terbium ions Tb3+ was developed at ESRF in the framework of the European project SCINTAX,10 which provided better results in terms of spatial resolution and efficiency, especially in conjunction with back-illuminated sensors.7,11,12
The SCFs are elaborated using the Liquid Phase Epitaxy (LPE) method, which is well known for the growth of oriented epitaxial films and well adapted to garnets. It is a flux growth process in which the driving force of crystallization on a substrate is provided by the cooling of a supersaturated solution, consisting of a material to be grown (solute) in a suitable solvent. The growth temperatures are usually much lower than those for single crystals (SCs).13–15
The LPE method provides single-crystalline layers of very high optical quality with very few defects (inclusions, cracks). For photonic applications, it is often selected amongst other epitaxial growth techniques, e.g., chemical vapour deposition or molecular beam epitaxy in vapour phase, as it provides nonporous layers at relatively high growth rates (in μm min−1).16 This growth technique presents numerous advantages, such as (i) the large available range of materials and dopants; (ii) the high available doping levels and uniform distribution of dopant ions; (iii) layer thicknesses from a few μm up to tens of μm can be reached on large surfaces of few cm2.
The gadolinium gallium garnet Gd3Ga5O12 (GGG) belongs to the crystal family of body-centred cubic multicomponent garnets with a chemical composition described by the general formula of {A}3[B]2(C)3O12, with {A}, [B], and (C) being the dodecahedral, octahedral, and tetrahedral symmetry cation sites. The GGG formula is represented as {Gd}3[Ga]2(Ga)3O12.17,18 Garnets are very flexible host matrices, with the possibility to extensively vary the chemical composition by substituting A, B and C cations to engineer the material properties. The {A} sites are appropriate for doping with lanthanide ions even with large ionic radii.19,20 Garnets are widely encountered in various applications, such as solid-state lighting and display technologies,21 solid-state lasers,19 scintillators for X-ray imaging,6 and electrolytes for Li-based batteries.22 The large density ρ of 7.1 g cm−3, as well as high effective atomic number Zeff above 50 makes GGG an attractive candidate as host material for high quality thin-film scintillators.1,6,8,23
The SCFs presented in this study were doped with terbium ions. Trivalent terbium ions Tb3+ (electronic configuration: [Xe]4f8) provide multicolour emissions, from blue to deep-red, due to 4f–4f radiative transitions originating from the metastable state 5D4 to a set of lower-lying levels 7FJ (J = 6–0, with 7F6 being the ground state). The most intense emission of Tb3+ ions falls into the green spectral range around 543 nm, corresponding to the 5D4 → 7F5 transition. The large energy gap separating the 5D4 state from the next lower-lying level (about 15000 cm−1) prevents depopulation of the emitting state by non-radiative multiphonon processes. Moreover, the 5D4 state presents a long luminescence lifetime, typically in the range of hundreds of μs to a few ms for garnets and more generally for oxide materials, depending on their crystal-field strength.24–26
Terbium-doped GGG was found to be an efficient green phosphor under both UV and X-ray excitation, thus being very interesting for solid-state lighting. This phosphor provided excellent quantum efficiencies comparable to commercially available ones, with yet a detrimental afterglow of several minutes under UV illumination. However, an afterglow of less than 1 s was observed under X-ray excitation.23,27
There exist studies on the X-ray imaging of the GGG:Tb SCFs produced by LPE.28,29 Jung et al.28 integrated a 6 μm-thick GGG:Tb scintillator in a fast μ-tomography system. This system allows to switch between high-resolution and high-speed acquisition mode. In the first one, the system achieved a breakthrough spatial resolution of 300 nm. In parallel, Douissard et al.29 studied the contrast and spatial resolution capabilities of a similar 6 μm-thick Tb3+-doped SCF in low-dose configuration. They obtained a contrast modulation in the order of 50% and a spatial resolution below 1 μm. Moreover, many synchrotron imaging beamlines use now routinely the GGG:Tb LPE scintillators for X-ray imaging applications.30,31
Nonetheless, no systematic study of the growth nor final properties of GGG:Tb SCFs was performed, unlike for other scintillator hosts such as YAG, LuAG or LSO.
In the present work, we report on the liquid phase epitaxy growth of Tb3+-doped GGG epitaxial layers for various doping levels, ranging from 2 to 10 at% Tb3+. We systematically studied the influence of the Tb3+ doping level on the growth parameters, morphology, composition, structure, and emission properties of the films under optical and X-ray excitation. The goal of this study is to further evidence the relevance of such films for developing novel scintillators providing sub-μm spatial resolution for X-ray imaging.
The growth charge was first melted and homogenized at 1100 °C for several hours between each epitaxy. The molten bath was then stabilized at the growth temperature before dipping the substrate. The substrates were undoped GGG plates of 1 inch diameter with their plane orthogonal to the [111] crystallographic axis and a surface rugosity below 10 Å. The isothermal LPE dipping technique was employed in this study, i.e., the substrates were alternatively rotated at 100 rpm into the solution in a horizontal position during the growth.13 Finally, the substrates were raised just above the solution and rotated at 800 rpm to remove any flux droplets.
The growth parameters are summarized in Table 1, for twelve growth attempts. The growth temperature was within the range of 1013 °C to 1028 °C, in the supersaturation domain of the solution. The growth duration was 10 min for epitaxies no. 1 to 10, 20 min for no. 11, and 4 min for no. 12.
Epitaxy no. | Duration (min) | Temperaturea (°C) | CTbb (at%) | tc (μm) | Growth rate (μm min−1) |
---|---|---|---|---|---|
a Growth temperature.b Nominal doping level of Tb3+ ions in the solution, Tb/(Tb + Gd).c Final thickness of the layer. | |||||
1 | 10 | 1013 | 2 | 9.9 | 1.99 |
2 | 10 | 1015 | 2 | 6.3 | 0.63 |
3 | 10 | 1017 | 2 | 3.3 | 0.33 |
4 | 10 | 1015 | 4 | 12.7 | 1.27 |
5 | 10 | 1020 | 4 | 4.1 | 0.41 |
6 | 10 | 1018 | 6 | 11.3 | 1.13 |
7 | 10 | 1022 | 6 | 4.6 | 0.46 |
8 | 10 | 1020 | 8 | 12 | 1.20 |
9 | 10 | 1023.5 | 8 | 5.2 | 0.52 |
10 | 10 | 1025 | 10 | 7.4 | 0.74 |
11 | 20 | 1028 | 10 | 1.9 | 0.09 |
12 | 4 | 1022.5 | 10 | 6 | 1.50 |
Photographs of the as-grown epitaxies are shown in Fig. 1. They appear transparent under natural light, including the highly doped sample with 10 at% Tb3+, and the intense green emission from the dopant ions is visible under UV illumination. Additionally, a cross-sectional view of an epitaxy was examined using confocal laser microscopy to observe the clean and distinct interface between the substrate and the layer.
The final thickness of the SCFs was determined by weighting the samples before and after the growth. The layers thickness were not exceeding 15 μm during the LPE growth to limit the crystalline defects and especially the formation of cracks due to lattice mismatch tension. The density of the substrate and the layer were assumed to be similar and the growth uniform over the whole surface of the substrate. The growth on the substrate edges was neglected. This method provided a good agreement with the confocal laser microscope observations with a tolerance of ∼5%.
The influence of Tb doping level in the flux on different growth parameters is presented in Fig. 2. For a growth duration of 10 min (epitaxies no. 1 to no. 9), and a given Tb content, the growth is slowed down on increasing the temperature (decreasing the supersaturation), see Fig. 2(a). For a growth duration above 15 min (epitaxy no. 11), the growth rate drops due to the depletion of solute in the vicinity of the substrate.
![]() | ||
Fig. 2 Influence of Tb doping level in the solution on the growth parameters of the GGG:Tb epitaxial layers: (a) variation of the growth rate with the growth temperature for various Tb doping levels; (b) range of the used growth temperatures for employed Tb doping levels. Numbers – epitaxy no., see Table 1. |
Moreover, the saturation temperature increases almost linearly with the Tb content, i.e., between +2 °C and +3 °C every 2 at% of Tb3+, with the solute/(solute + solvent) ratio, see Fig. 2(b). The evaporation of the PbO/B2O3 solvent is also responsible for its increase.
![]() | ||
Fig. 3 Crystalline structure of body-centered cubic Gd3Ga5O12:Tb (sp. gr. Ia![]() |
The composition of the layers was studied using an electron probe microanalysis (EPMA) setup equipped with five wavelength-dispersive X-ray spectrometers (WDS) and an analytical crystal of lithium fluoride LiF. The X-ray tube operated at 25 kV and 1000 nA. To measure the percentages of atoms constituting the SCFs, the epitaxies were coated with a resin and polished on one edge. The crystal was metalized with carbon to ensure the transport of charge through the sample surface, which is necessary for microprobe analysis. The measurements were performed by the French company “Bureau Veritas”. The percentages of atoms were measured with an accuracy of ±0.4%.
The μ-Raman and μ-photoluminescence spectra and the respective maps were recorded using a confocal Raman microscope (InVia Qontor, Renishaw) equipped with an Ar+ ion laser (488/514 nm) and a ×50 Leica objective. A 2400 l mm−1 diffraction grating coupled with a CCD matrix provided a spectral resolution down to 1 cm−1. The μ-Raman spectra were collected after 3 accumulations and 10 s integration.
The in-line transmission spectra were recorded with a spectrophotometer (Cary 5000, Agilent) providing a spectral bandwidth of 0.2 nm in the visible and 0.1 s integration.
The photoluminescence excitation spectrum, as well as the photoluminescence dynamics of Tb3+ ions were studied with a spectrofluorometer (QuantaMaster, Horiba). The excitation spectrum was recorded while monitoring the green luminescence from Tb3+ ions at 543 nm, using a spectral bandwidth of 0.4 nm in the UV-blue spectral range, and 0.8 nm in the visible, with 0.2 s integration. The photoluminescence decay curves were recorded while exciting Tb3+ ions at 488 nm and monitoring their emission at 543 nm.
The light output and the afterglow were recorded at room temperature with an X-ray generator using a copper anode. A 25 μm-thick copper filter was added to select the 8 keV Kα line of Cu. The X-ray tube delivered an X-ray photon flux density of 106 photons per mm2 per s, over a dynamic range of 14 bit. The scintillator was mounted on a high-spatial resolution detector, consisting of microscope optics and a CCD camera. A dark-corrected flat-field image was employed to calculate the mean intensity in the image. The measurements were corrected for the X-ray absorption efficiency (accounting for the thickness of the scintillating layer), so the reported light output reflects the intrinsic conversion efficiency and is independent of the layer thickness. The CCD quantum efficiency was normalized with respect to a “bulk” YAG:Ce single crystal (light output: 30000 ph per MeV, taking its conversion efficiency as 100%) because YAG:Ce scintillators have historically served as the reference for sub-μm spatial resolution imaging systems and are still widely used in X-ray beam monitoring applications at synchrotron sources. By “bulk,” we refer to a single crystal device with a thickness of 500 μm, which has been mechanically thinned down. Such scintillators provide an established performance baseline in terms of light yield and spatial resolution, making them a relevant reference for evaluating the performance of new thin-film scintillators. A comparison with SCF YAG:Ce would not be appropriate in our case, as its undoped YAG substrate also emits light under X-ray excitation. This parasitic luminescence interferes with our measurements and degrades the accuracy of the scintillator light output characterization. The afterglow signal decay was analysed with a photomultiplier (Philips XP2020Q) coupled to a SR400 gated photon counter working in counting mode, sampled at intervals of 8 ms using a SR445 amplifier (Stanford Research Instrument). The exposure time to X-ray was varied in the range of 0.1 s to 10 s, with a temporal resolution of 4 ms.
Measuring the light output of a thin-film scintillator using the pulse height spectrum method with X-rays (commonly used for bulk scintillators) is challenging due to: (i) low stopping power in thin-film scintillators at high energies (tens of keV), as most X-rays pass through without depositing energy, and without sufficient energy deposition, the scintillator does not emit enough light to measure a clear spectrum; (ii) poor energy resolution owing to a low light yield; (iii) poor light collection – most of the light is lost at the optical coupling, and the signal-to-noise ratio becomes too low for pulse height analysis; and (iv) electronic noise – the very low light yield from X-rays in the thin-film scintillator may be close to the noise level of the electronics (photodetectors). Tender X-rays in integration mode (the approach used here) are a better alternative for thin-film scintillators: this detection system is robust, reliable, and similar to the systems employed for X-ray imaging applications at synchrotron facilities. In addition, it provides information regarding imaging quality.
As mentioned above, Tb3+ ions are substituting Gd3+ ones in the dodecahedral sites of the GGG lattice, leading to a decrease of the lattice parameter a due to the difference of their ionic radii. In this case, the tension in the film with respect to the substrate increases with the content of Tb. The lattice mismatch, Δa, between the doped epitaxial layer and the undoped substrate follows the Vegard's law:35 Δa = Δa0 + αTbCTb, where αTb = 0.20 is the hard-sphere diameter ratio of Tb3+, and Δa0 = 4.44 mÅ the lattice mismatch between an undoped layer and the substrate.
For the same material, there is always a slight difference in the crystalline structure induced by the growth method. During the growth of the substrates by the Czochralski method, a small proportion of Gd3+ ions are found in the octahedral sites occupied in theory only by Ga3+ (the so-called antisite defects).36
Thus, the lattice parameter a of the GGG substrate is increased as compared to that of an undoped GGG layer grown by LPE, due to the larger size of Gd3+ ions (rGd = 0.938 Å and rGa = 0.620 Å in VI-fold oxygen coordination32). Therefore, Δa0 is not zero in our case and there is a difference between the lattice of an undoped epitaxial layer and the GGG substrate.
As the Tb content is increased, the lattice mismatch follows the same trend, see Fig. 4(b). For the heavily-doped 10 at% Tb3+ SCFs, Δa is reaching 6.38 mÅ, corresponding to a mismatch of Δa/asubstrate = 0.05%, with the layers presenting higher densities of cracks. High quality garnet epitaxial films with very low surface roughness (down to a few nm) and low tensile stress can only be elaborated if the lattice mismatch with the substrate is very low (<0.1%).14
Furthermore, the diffraction peaks from the GGG:Tb layers are well defined and their full widths at half maximum (FWHM) are comparable to that of the substrate (approximately 0.012° and 0.011°, respectively), which reveals the excellent crystalline quality of the obtained layers.
The WDS spectra for the 4 at% Tb3+-doped GGG film and the GGG substrate are depicted in Fig. 6. The scatter peaks were assigned to the emission lines of Tb, Gd and Ga according to experimental results previously reported in the litterature.37,38 The incorporation of Tb3+ ions in the SCF is evidenced by the appearance of additional peaks assigned to L, Lα2, Lα1, Lβ4, Lβ1 and Lβ2,15 lines, corresponding to photon energies of 5.547 keV, 6.238 keV, 6.273 keV, 6.940 keV, 6.978 keV and 7.367 keV, respectively.37 The potential small contamination by Pb was below the detection limit.
![]() | ||
Fig. 6 Comparison of the WDS spectra of a GGG substrate and a Tb3+-doped layer (epitaxy no. 4 with 4 at% Tb3+ in the flux). Symbols – contributions of Gd, Tb and Ga assigned using experimental data reported in the literature.37,38 |
The homogeneity of the layers was further studied by EPMA by measuring the atomic percentages at ten different positions over the surface (Φ 2.54 cm) of three SCFs doped with 4 at%, 6 at% and 10 at% Tb3+. The root mean square (r.m.s.) values are reported in Table 2. The content of dopant is uniform over the layer surface, as for the other constituents, as expressed by the small r.m.s. errors. The average actual Tb doping levels are 4.3 at%, 6.3 at% and 10.4 at% Tb3+. The segregation coefficient, KTb = Clayer/Csolution, is close to unity in all cases as expected due to the closeness of rGd and rTb, highlighting excellent incorporation of the dopant ions in the GGG lattice, with only a slight reduction on increasing the doping level.
Epitaxy no. | CTba | Gd | Tb | CTb,layerb | KTbc |
---|---|---|---|---|---|
a Nominal doping level of Tb3+ ions in the solution, Tb/(Tb + Gd).b Actual doping level of Tb3+ ions in the layer.c Segregation coefficient of Tb3+ ions, Clayer/Csolution. | |||||
4 | 4 at% | 14.73 ± 0.81 | 0.66 ± 0.03 | 4.3 at% | 1.07 |
7 | 6 at% | 13.31 ± 0.17 | 0.90 ± 0.02 | 6.3 at% | 1.05 |
10 | 10 at% | 12.48 ± 0.22 | 1.45 ± 0.02 | 10.4 at% | 1.04 |
The uniform distribution of the dopant both in the depth of the layer and over the surface ensures a very homogeneous response of the SCF under X-ray illumination.
The down face is directed towards the solution and in contact with the PbO solvent vapours during the LPE growth, creating etch pits on the surface. This face is therefore removed by chemical-mechanical polishing and only the upper face is kept.
The high-energy range of the Raman spectra (ν > 500 cm−1) corresponds to internal vibrations of the [GaO4] and [GaO6] polyhedra, while the low-energy range (ν < 300 cm−1) corresponds to translation motions of the Gd3+ ions. The highest Raman frequency at 741 cm−1 is due to vibrations in antiphase of the [GaO4] and [GaO6] units, associated with a significant stretching of the bonds. Finally, the most intense peak at 353 cm−1 is due to separate rotations of these polyhedra around two different sets of perpendicular axes.17,39,41
The Raman spectra of the low-doped layer and the substrate are almost identical, except for a small decrease of intensity for the layer. The Raman peaks of the film also slightly broaden as compared to those of the substrate, with a FWHM of 17.9 cm−1 and 18.4 cm−1 for the substrate and the SCF at 741 cm−1, respectively. A close look at the highest Raman frequency mode located at 741 cm−1 is shown as an inset in Fig. 8 for a GGG substrate and layers doped with 2 at% and 6 at% Tb3+. A small blue shift of 0.9 cm−1 can be observed for the 6 at% Tb3+ doped layer with respect to the substrate. This phenomenon is explained by the variations of the crystalline structure by incorporation of dopant ions with a different ionic radius. The slight decrease of intensity, as well as the negligible broadening of the peaks for the layers confirm the excellent crystallinity of SCFs, with their crystalline structure almost identical to that of the substrate.
![]() | ||
Fig. 9 Transmission spectra of GGG:Tb epitaxies no. 3, 5 and 7 with 2 at%, 4 at%, and 6 at% Tb3+ in the flux, respectively, in the UV-visible spectral range. Symbols – contributions of Gd3+ and Tb3+ ions, and compensation of the oxygen vacancies VO2−, dashed grey line – theoretical limit T0 set by Fresnel losses after Wood and Nassau.42 |
The presence of ionic impurities has been discussed in the previous works concerning garnet epitaxial films grown by LPE.14,46–51 In particular, Randoshkin et al.48 conducted a systematic study on the LPE growth of undoped GGG layers on (111)-oriented substrates of the same nature using a mixture of PbO/B2O3 as a solvent. They observed broad additional lines in the optical absorption spectra of the epitaxial films, with maxima at 280 nm, 325 nm, and 550 nm, associated with Pb2+ and Pb4+ ions. The presence and intensity of these lines depend on the supercooling, ΔT, of the solution, i.e., the difference between the saturation temperature, Ts, and the growth temperature, Tg. The best quality films are generally grown from low supercooling. There is also a risk of Pt4+ impurities coming from the platinum crucible. These ionic impurities can strongly affect the optical properties.50 Therefore, the broad and weak absorption band between 250 nm and 300 nm could be assigned to a small amount of Pb2+ ions. No additional broad bands associated with Pb4+ ions were observed at the wavelengths reported by Randoshkin et al.48
Additionally, the narrow lines located at 300–315 nm, 273–282 nm, and 244–249 nm correspond to 4f–4f electronic transitions of the host-forming Gd3+ ions in the GGG lattice.23,45,52,53
Finally, the weak absorption band in the 315–395 nm range is connected to the charge compensation of oxygen vacancies.46,53 The parity and spin forbidden 4f–4f transitions of Tb3+ are not visible due to the small thickness of the layers.
The scheme of both 4f and 5d energy levels of Tb3+, as well as 4f ones of Gd3+ ions are shown in Fig. 10(a) after Carnall et al.54,55 It depicts the 4f–4f transitions in absorption and emission from Tb3+ ions.
![]() | ||
Fig. 10 Excitation and luminescence properties of Tb3+ as dopant in GGG layers grown by LPE: (a) partial energy-level scheme of Tb3+ (solid black lines) and Gd3+ ions (dashed orange lines) after Carnall et al.,54,55 arrows – 4f–4f transitions in absorption and emission of Tb3+, NR – multiphonon non-radiative relaxation, grey area – conduction band of Tb3+, blue areas – excited configuration 4f75d1, LS and HS – low- and high-spin states, respectively; (b) photoluminescence excitation spectrum of GGG:Tb epitaxy, λlum = 543 nm, * – Gd3+ 4f–4f excitation lines, diamonds – Tb3+ inter-configurational 4f8 → 4f75d1 transitions; (c) photoluminescence spectrum of Tb3+ ions, λexc = 488 nm, square – Eu3+ impurities in the GGG substrate; (d) μ-luminescence mapping across the end-facet of the GGG:Tb/GGG epitaxy monitoring the peak intensity at 544 nm (the 5D4 → 7F5 transition). The detailed (b) excitation and (c) luminescence spectra are available in the ESI materials.† |
One way to excite a rare-earth-doped photoluminescent material is to target the broad and intense absorption bands of the 4fn−15d1 excited configuration.56 Contrary to the 4fn levels, the positions of the 4fn−15d1 states strongly depend on the crystalline environment.57,58 The barycenter energies of the 4fn−15d1 states are mainly affected by the nature of ligands in the host matrix (the nephelauxetic effect). Blasse and Bril56 have demonstrated that the ligand electronegativity and symmetry influence the amplitude of the 4fn−15d1 level extension. For Tb3+, this excited configuration is located at relatively low energies as compared to other trivalent lanthanides due to the more stable half-full 4f7 shell.56,58,59
During the transition in excitation of a 4fn electron to the excited configuration 4fn−15d1, the f–d electron coupling for the ions in the second half of the lanthanide series 4fn (n > 7) gives rise to 4fn−15d1 high-spin (HS) and low-spin (LS) states. For Tb3+ ions, this transition results in two multiplets 9DJ and 7DJ, exhibiting higher (HS) and lower (LS) spin multiplicity than the 4fn ground-state 7F6, respectively. Consequently, the 7F6 → 9DJ transitions are spin-forbidden and the 7F6 → 7DJ ones are spin-allowed by the selection rules of spin multiplicity (ΔS = 0).57,59 Both LS and HS states of Tb3+ ions are shown in the energy-level scheme. Their energies were determined from the photoluminescence excitation spectrum with an accuracy of ±5 cm−1.
The photoluminescence excitation spectrum of Tb3+ ions as dopant in a GGG epitaxial layer is depicted in Fig. 10(b). The assignment of 4f–4f transitions was done following Carnall et al.54 The measurement was performed while monitoring the green luminescence at 543 nm.
The band falling in the blue spectral range with a maximum at 488.4 nm is due to a transition to the metastable state 7F6 → 5D4. The numerous overlapping bands in the UV-blue spectral range between 330 nm and 390 nm are ascribed to transitions to the higher-lying 5LJ, 5GJ, and 5DJ manifolds.
The inter-configurational 4f8 → 4f75d1 transitions of Tb3+ ions are contributing to intense excitation bands in the UV, see Fig. 10(b). The band at 300–325 nm corresponds to the spin-forbidden 7F6 → 9DJ transitions to HS states, and the most intense one at 255–290 nm – to the spin-allowed 7F6 → 7DJ transitions to LS states. The latter one was also observed in the transmission spectra, Fig. 9. These bands are overlapping with low-intensity 4f–4f lines to higher-lying manifolds of Tb3+. The energy ranges of the HS and LS states of Tb3+ in GGG SCFs are given in Table 3. These experimental values are compared with those previously reported for other Tb3+-doped garnets. The HS–LS splitting corresponds to the energy gap between the onsets, i.e., the lowest energies Ei, of the HS and LS bands.
The more covalent the host matrix is, the larger the 4fn−15d1 extension will be, thus reducing the interactions with the 4fn electrons. This induces a smaller HS–LS splitting.60
Our experimental values are well in line with the ones for Tb3+-doped LuAG, YAG and YGG. Note the significant lowering of the HS–LS splitting when Al3+ ligands are replaced by Ga3+ ones. The electronegativity χ of Ga3+, χ4 = 1.755 and χ6 = 1.579 in IV- and VI-fold coordination, respectively, is superior to that for Al3+, χ4 = 1.691 and χ6 = 1.513.61 Therefore, Ga–O bonds are more covalent than Al–O ones, thus reducing the HS–LS splitting.
Furthermore, Gd3+ cations of the GGG host matrix exhibit 4f–4f excitation lines at 274 nm, 307 nm, 313 nm and 317 nm, Fig. 10(b). These lines are also observed in the transmission spectra, Fig. 9. They are overlapping with the 4f75d1 excitation bands of Tb3+ ions, thus indicating an efficient energy transfer from the host matrix to the excited configuration of dopant ions, followed by a fast relaxation of the latter to the 5D4 metastable state.23,45,52,53
The photoluminescence (PL) spectrum of Tb3+ ions in a GGG layer is shown in Fig. 10(c). The characteristic emission lines of the Tb3+ ions are observed in the visible spectral range, with a maximum in the green at 543.4 nm corresponding to the 5D4 → 7F5 transition. Moreover, the weak background emission between 600 nm and 606 nm is due to Eu3+ contamination of the undoped GGG substrate. However, the content of these Eu3+ impurities is supposed to be low considering no additional absorption lines were observed in the 390–395 nm range, see Fig. 9.44,52
Additionally, μ-luminescence mapping was performed across the end-facet of a GGG:Tb/GGG epitaxy while monitoring variations of luminescence peak intensity at 544 nm, as shown in Fig. 10(d). The results indicate a uniform distribution of Tb3+ ions inside the SCF, with no diffusion into the GGG substrate.
The luminescence kinetics of Tb3+ ions were studied for various doping levels, as shown in Fig. 11. The luminescence lifetime of the 5D4 emitting state of Tb3+ ions was measured while exciting in the blue at 488 nm and monitoring the green emission at 543 nm. The 5D4 lifetime amounts to 2.42 ms for the low-doped (2 at% Tb) layer, being in line with those previously reported for Tb3+ in cubic garnets.26,45 It only slightly decreases down to 2.32 ms on increasing the doping level to 6 at% Tb3+, thus indicating the excellent crystalline quality of the layers with very few quenching centers. The luminescence lifetimes have been measured with an accuracy of ±0.05 ms. The decay curves are single exponential, in agreement with the single type of sites for Tb3+ ions in the GGG lattice (D2 symmetry).
![]() | ||
Fig. 12 Radioluminescence spectra of Tb3+ ions in GGG epitaxial layers doped with 2 at%, 4 at%, and 6 at% Tb3+ in the flux, respectively: (a) UV-blue emission from the 5D3 manifold; (b) visible emission from the 5D4 manifold. X-ray illumination, 8 keV. Squares – Eu3+ impurities in the GGG substrate. The inset in (a) shows the partial energy-level scheme of Tb3+ after Carnall et al.,54 arrows – 4f–4f transitions in emission from the 5D3 manifold, NR – multiphonon non-radiative relaxation. |
The substrate luminescence intensity is less than 1% of that for the layer under X-ray flux at 8 keV. However, the background luminescence of the substrate can become an issue on the synchrotron beamlines at high energies and high X-ray fluxes, as it can decrease the spatial resolution of the system.
As with the PL spectrum, weak parasitic lines associated with Eu3+ impurities in the GGG substrates were observed between 600–614 nm and 704–712 nm.
Afterglow, the delayed luminescence from a scintillator after stopping its X-ray irradiation, is particularly detrimental for fast X-ray imaging applications.11 Since this phenomenon strongly depends on the exposure time to X-rays, its influence on the afterglow properties of GGG:Tb layers was studied. The results are given in Fig. 13(a), with exposure times of 0.1 s, 1 s and 10 s. The time response of the scintillators down to relative amplitudes of 10−4 to 10−5 was measured since a dynamic range up to 14 bit for successive images is required in some X-ray imaging applications. The afterglow of the GGG:Tb SCF was compared to that of its bulk SC counterpart which was grown by the Czochralski method. After 100 ms, the afterglow intensity of the SCF drops down to 6 × 10−5 in average, corresponding to a dynamic range of more than 14 bit. In the case of the GGG:Tb single crystal, the intensity decreases between 5.8 × 10−3 and 0.016 depending on the exposure time, which corresponds to less than 8 bit of dynamic range. The afterglow of the GGG:Tb layer is almost not detected in this explored dynamic range.
These results were compared with the performance of miscellaneous rare-earth-doped SCs and epitaxial layers based on garnets, see Fig. 13(b).
In general, the afterglow intensity for SCFs decreases faster and much further than for SCs, with a minimum value reached with the GGG:Tb layers studied in this work.
The Ce3+-doped LuAG and YAG SCs limit the dynamic range respectively to ∼4–5 bits and ∼7–8 bits after 100 ms. The higher afterglow in the case of SCs grown by the Czochralski method could originate from (i) impurities in the raw powders, and (ii) defects such as vacancies or antisites characteristic of this growth method at high temperatures.36
The afterglow of the Tb3+-doped films decreases much faster than that for other rare-earth-doped garnets already employed as scintillators.
An improvement of the light output was observed as the Tb doping level in the solution was increased, reaching a maximum of 52% at 6 at% Tb3+, see Fig. 14. The same optimal Tb content was determined by Lammers et al. for GGG:Tb powders.23 Above this doping level, the light output stabilizes at ∼51%. Note the exception of the epitaxy no. 11, doped with 10 at% Tb3+, reaching a maximum light output of 57%. This epitaxy was elaborated with the lowest growth rate of 0.09 μm min−1, thus limiting the incorporation of lead from the solvent. In fact, the light output is mainly influenced by the crystalline quality of the films, i.e., by the growth parameters, as the presence of defects and impurities leads to non-radiative processes. Therefore, for a given Tb doping level, the light output slightly increases as the growth rate is lowered and the growth temperature is increased. This conversion efficiency corresponds to approximately 50% of that for a YAG:Ce single crystal.
The GGG:Tb epitaxies present high transmittance in the emission range of Tb3+ ions. The intense excitation bands corresponding to the inter-configurational 4f8 → 4f75d1 transitions of Tb3+ ions are overlapping with low-intensity 4f–4f excitation lines in the UV spectral range. The experimental barycenter energies of the 4f75d1 high-spin and low-spin states of Tb3+ in GGG films were determined by photoluminescence excitation measurements. Moreover, the 4f–4f excitation lines of the Gd3+ host-forming cations are also overlapping with the 4f75d1 bands of Tb3+, revealing an efficient energy transfer from the host matrix to the dopant ions.
Terbium ions as dopant in the GGG layers exhibit their most intense emission in the green spectral range at 543 nm, fitting well with the sensitivity of CCD and CMOS sensors. The luminescence lifetime of the 5D4 Tb3+ emitting state amounted to 2.32 ms for 6 at% Tb3+, and is weakly dependent on the doping level. The conversion efficiency of the GGG:Tb films is optimized for a doping level of 6 at% Tb3+ in the solution, reaching a maximum light output of 52% with respect to a reference bulk YAG:Ce single crystal.
The Tb3+-doped layers do not exhibit any significant afterglow after X-ray irradiation at 8 keV, being absent from the films for a 15 bit dynamic range. Minimum afterglow intensities are reached for GGG:Tb films, as compared to other SCFs currently employed as scintillators. The luminescence intensity of the GGG substrates under X-ray illumination is very low, namely less than 1% of the Tb3+-doped layers.
Following these results, we suggest that the GGG:Tb epitaxial layers are attractive candidates for applications in X-ray imaging with sub-μm spatial resolution. Indeed, the films meet most of the criteria for high-resolution imaging scintillators, namely their excellent optical quality and high density, high transmittance in the spectral range of Tb3+ green emission, and the emission wavelength matching the spectral maximum of sensitivity of CCD/CMOS sensors. The afterglow can be further optimized by eliminating the impurities and crystalline defects introduced during the synthesis of the films.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5ra01784j |
‡ Both are the first authors. |
This journal is © The Royal Society of Chemistry 2025 |