1. Introduction

Broadly speaking, the interest in synthesizing in different shapes rare-earth sesquioxides of the cubic phase is related to the diverse range of applications which can be envisioned from these materials: phosphor powders [1–4], scintillating thin films or powder materials [5–8], photonic nanomaterials, high-power and/or sub-100 fs lasers, eyesafe telecommunications lasers, yellow-orange-red solid state lasers, upconverter materials for the 3rd generation of photovoltaïc cell, and the list is not exhaustive [9–11]. The growth of rare-earth sesquioxides bulk single crystals is recognized as one of the most challenging endeavours of today’s crystal growth, in the sense defined by these two science drivers: to produce well-established crystalline materials with improved quality and larger size at a lower cost ; to grow in bulk new categories of materials with extreme thermodynamic characteristics. Owing to both the extreme thermodynamic characteristics in which their growth takes place and a complex crystallographic polymorphism upon cooling, cubic rare-earth sesquioxides single crystals with improved quality and large size are difficult –if not still impossible– to obtain, especially at a lower cost [12]. In recent years, we have developed a cheap and user friendly flux growth process that works at half the melting temperature of rare-earth sesquioxides (between 1225 and 1130°C), and established that the new solvent it uses favours the crystallization of the cubic Gd₂O₃, Y₂O₃ and Lu₂O₃ phases with unprecedented Yb³⁺-substitution levels without resorting to any post-growth thermal treatment [13]. The spectroscopic properties and laser operation based on these crystals have been investigated and found to be encouraging from the viewpoints of high-power and/or sub-100 fs pulse laser applications [14–16]. In addition to the first cubic Gd₂O₃:Yb³⁺ single crystals ever grown to date, pure Tb₂O₃ single crystals have been grown with promising Faraday rotation properties. Indeed, these heretofore unknown crystals displayed a Verdet constant ≈3.35 times higher than that of well-known commercial TGG crystals at 1064 nm, and an absorption coefficient (corrected for Fresnel reflexion losses) at the same wavelength ≈0.28 cm⁻¹, still too high to comply with industrial specifications for Faraday rotators but nevertheless quite encouraging [17]. They also proved to be fully Tb³⁺ reduced crystals by thermodynamic response functions (specific heat and magnetic susceptibility) measurements and analysis.

In this work we follow our exploration of the discovery potential of our growth process, by addressing the growth of Eu³⁺-doped rare-earth sesquioxides. This oxidation state turns out to be even more reducible to Eu²⁺ than Yb³⁺ to Yb²⁺ in most oxides, chlorides and bromides [18–22]. We focused our efforts on the Y₂O₃:Eu³⁺ and (Lu,Gd)₂O₃:Eu³⁺ solid solutions for several reasons. The former one is the only one known to date and “historical” crystal that proved laser operation based on Eu³⁺ cations [23]. Moreover, its hexagonal P6₃/mmc to cubic Ia-3 phase transition upon cooling makes it difficult to obtain optical quality and laser grade materials by classical high-temperature (≈2400-2500°C) solidification methods (heat-exchanger method, laser-heated pedestal growth or micro-pulling down), that is, directly from the molten state of the same composition. Indeed, such high-temperature methods use Re crucibles the stability of which requires to flow a mixture of gases containing explosive and highly reductive H₂. Our flux growth process was found to be operative in air, with Pt crucibles and a solvent completely devoid of OH^--groups, heavy metals (Pb, Bi, etc.) or any corrosive species. The interest in (Lu,Gd)₂O₃:Eu³⁺ lies in its potential X-ray imaging applications which are being developed at the ESRF, because of its high volumic mass (≈9 g/cm³) and high average effective Z [24–26]. Eu³⁺ was chosen as an optically active cation because its well-known luminescent and scintillating properties [1–8,27] occur mainly in the orange-red spectral range, where cheap and almost 100% efficient detectors exist.

2. Experimental procedures

All the raw materials (Lu₂O₃, Gd₂O₃ and Eu₂O₃) were commercial powders of 5N purity, dried in air at 500°C during 12h, and 1000°C during 12h. The crystal growth of (Lu_0.5Gd_0.5)₂O₃:Eu³⁺ using a Li₆(Lu_0.475Gd_0.475Eu_0.05)(BO₃)₃ solvent was performed in several stages. At first, a solute of formula (Lu_0.475Gd_0.475Eu_0.05)₂O₃ was prepared by reacting together a stoichiometric and mechanically mixed amount of Lu₂O₃, Gd₂O₃ and Eu₂O₃ powders. Then, the Li₆(Lu_0.475Gd_0.475Eu_0.05)(BO₃)₃ solvent was synthesized by mechanically mixing and grinding stoichiometric amounts of Li₂CO₃, H₃BO₃, Lu₂O₃, Gd₂O₃ and Eu₂O₃ powders, which were subsequently heated in air in a Pt crucible at 450°C during 12h, and at 750°C during 12h. The resulting mixture was then ground again in an agate mortar. Finally, a growth load composed of 124.9781 g (80 mol. %) of solvent Li₆(Lu_0.475Gd_0.475Eu_0.05)(BO₃)₃ and 30.7919 g (20 mol. %) of solute (Lu_0.475Gd_0.475Eu_0.05)₂O₃ was mechanically mixed and ground together. Melting of this mixture in air in a Pt crucible was performed at 1225°C. By means of a Pt and alumina double reflector (thickness higher than 1 mm for Pt and 2 mm for alumina), typical thermal gradients in the furnace can be reduced to 1°C.cm⁻¹, so that the temperature distribution inside the molten charge remains as uniform as possible. Growth was initiated on a Pt spatula immersed into the melt. The molten bath stirring was undertaken by rotating the alumina rod at 30 rpm during 24h at 1225°C in order to dissolve completely the solute and homogenize the melt composition. Then, the Pt spatula was lifted in such a way as to maintain it a few millimeters below the melt surface, right at the center of the crucible section. In order to avoid sedimentation of the possibly remaining undissolved solute particles, a 20 rpm rotation was kept during the growth stage itself. The slow cooling of the flux was performed with a “shark teeth” profile at −0.2°C.h⁻¹ between 1225°C and 1130°C. At this temperature, the spatula was removed from the molten bath, and then the whole setup was allowed to cool down to 850°C at 6.5°C.h⁻¹, and then at 30°C.h⁻¹ down to room temperature. The alumina rod resisted quite well and just required cleansing in a diluted HCl solution to be fully operational for the next experiment.

22 facets could be identified on the crystal shown in Fig. 1, among which two hexagonal ones, a majority of distorted diamond ones, and a small square one. 21 of the 22 facets could be oriented by the Laue method on a Delta Technologies International GM WS Series X-Ray goniometer head and found to be (211) planes. Interestingly, the small square facet was found to be (100). As the four largest d_hkl-spacings in this crystal are d₁₀₀>d₂₀₀>d₂₁₁>d₂₂₀, we conclude that the crystal grew close to equilibrium but that it was removed from the molten bath too early to observe a full equilibrium habit made of (100) facets. These observations are similar to those already made in Yb³⁺-doped Gd₂O₃, Y₂O₃ and Lu₂O₃ crystals and in pure Tb₂O₃ crystals [13,17].

 

Fig. 1 As-grown (Lu,Gd,Eu)₂O₃ bulk single crystal (~0.5 cm³) obtained by the flux method.

Download Full Size | PDF

Gd- and Eu-concentration were measured by scanning (211) facets, at a depth down to 2 µm, by means of EPMA (Cameca SX 100, 20 keV electron beam), and found to be independent on the spatial coordinates. In Lu_2-x-yGd_xEu_yO₃ crystals, we found x = 0.41 (≈5.69 10²¹ Gd.cm⁻³) and y = 0.03 (≈4.16 10²⁰ Eu.cm⁻³), whereas for the Y_2-xEu_xO₃ crystal, we got x = 0.134 (≈1.79 10²¹ Eu.cm⁻³). These amounts of substituents dissolved in their respective crystalline structures permit to estimate some effective and average distribution coefficient for each of them: k_av(Gd³⁺)≈0.43 and k_av(Eu³⁺)≈0.3 in Lu₂O₃, and k_av(Eu³⁺)≈0.67 in Y₂O₃. This trend in effective and average distribution coefficients can be correlated to the differences in ionic radii. Indeed, the difference in ionic radii between Eu³⁺ and Y³⁺ is smaller than that between Gd³⁺ and Lu³⁺, which in turn is smaller than that between Eu³⁺ and Lu³⁺.

Optical transmittance of the Lu_1.56Gd_0.41Eu_0.03O₃ (t = 360 µm) and Y_1.866Eu_0.134O₃ (t = 350 µm) unoriented single crystals from 200 to 3000 nm was recorded with a Cary 5000-UV-vis-NIR spectrometer and a resolution of 1 nm. The baseline difference between the two samples is likely to be due to the fact that owing to smaller dimensions of the Y_1.866Eu_0.134O₃ crystal (2x2 mm² section), the aperture angle was smaller. The emission spectrum at room temperature was measured with a Triax 320 spectrophotometer from Horiba-Jobin Yvon equipped with a PM Hamamatsu R928 detector and a resolution of 0.2 nm. The light excitation was delivered by an intense Xe lamp connected to a monochromator for spectral selection.

Magnetic susceptibility was measured in field-cooled mode using a Quantum Design SQUID MPMS XL magnetometer operated in the 4.2−354 K temperature range under an applied magnetic field of 1 mT for the flux-grown Lu_1.56Gd_0.41Eu_0.03O₃ crystal of mass 45.1 mg, and 8 mT for the flux-grown Y_1.866Eu_0.134O₃ crystal of mass 17.8 mg, both mounted in a capsule placed in a straw. The negligibly small diamagnetic contribution of the capsule was not subtracted from our data.

The X-ray conversion efficiency was measured with a monochromatic 8 keV energy X-ray beam produced by a lab X-ray generator (copper anode with 25 µm copper filter). Bulk YAG:Ce was used as a reference for the conversion efficiency measurement (conversion efficiency of bulk YAG:Ce = 100%). The scintillators were coupled to an X-ray microscope equipped with an Olympus 2x magnification microscope objective and a PCO Sensicam SVGA camera. The conversion efficiency was calculated from the flat field image (dark image corrected) obtained under 30 s exposure. The mean ADU signal extracted from a Region Of Interest (ROI) of the image was measured and corrected by the crystal absorption efficiency (crystal thickness) and the CCD camera spectral matching factor (convolution of scintillator emission spectrum and CCD sensor quantum efficiency).

The scintillation spectrum was measured under X-ray excitation energy of 8 keV with a copper anode. The spectrophotometer consisted of a CP140 Jobin Yvon monochromator coupled to a PCO2000 camera. The spectrum was corrected by the monochromator grating efficiency and the CCD camera Quantum Efficiency (QE).

The time response of the scintillators down to relative amplitudes of 10⁻⁵–10⁻⁴ was studied since a dynamic range of up to 14 bits for successive images is required in some X-ray imaging applications. The afterglow decay depends on the exposure time and more slightly on the X-ray photon flux. Typical exposure times on synchrotron sources for X-ray tomography are 0.1–10 s (monochromatic X-ray beam up to 10¹³ photons/(s.mm²)). The afterglow was measured at 8 keV with an X-ray tube equipped with a copper anode and copper filter. This corresponds to an X-ray photon flux of 10⁶ photons/(s.mm²), which is considerably lower than the synchrotron beam intensity. The signal decay was measured with a PMT 2020Q coupled to a SR445 Stanford Research amplifier and a SR400 gated photon counter from Stanford Research Instrument working in counting mode and sampled at intervals of 4 ms.

3. Emission spectra and Judd-Ofelt analysis

The average refractive index was assumed to be 1.9 for both crystals, and was used for all the spectral range investigated to calibrate the spectra by the Fuchtbauer-Ladenburg method. The stimulated emission cross section for a given transition from state J to state J₀ can be evaluated by the following expression:

 

Fig. 2 FL-calibrated stimulated emission cross-section of the Y_1.866Eu_0.134O₃ and Lu_1.56Gd_0.41Eu_0.03O₃ crystals at room temperature under excitation at 393 nm.

Download Full Size | PDF

The number of crystal-field sublevels shown in the emission spectra indicates a complete degeneracy lifting consistent with the C₂ site symmetry adopted by ~¾ of the Eu³⁺ ions. We performed the Judd-Ofelt analysis to estimate the intensity parameters Ω_λ (λ = 2,4) exactly as in [19] and [20], by using the ratio of the integrated intensity of the ${}^{5}D{}_0\to {}^{7}F{}_{2,4}$ electric dipole transitions, ${\displaystyle \int I_J\left(\nu \right)}d\nu$, to the integrated intensity of the ${}^{5}D{}_0\to {}^{7}F{}_1$ purely magnetic dipole transition, ${\displaystyle \int I_1\left(\nu \right)}d\nu$. The Judd-Ofelt parameters, oscillator strengths, spontaneous emission probabilities, branching ratios, ratio “R” of integrated emission intensity of the hypersensitive transition ${}^{5}D{}_0\to {}^{7}F{}_2$ to that of the ${}^{5}D{}_0\to {}^{7}F{}_1$ magnetic transition, and radiative lifetimes are gathered in Table 1.

Table 1. Results of the Judd-Ofelt analysis carried out in the Y_1.866Eu_0.134O₃ (A) and Lu_1.56Gd_0.41Eu_0.03O₃ (B) single crystals: energy barycenter ν, Ω_2,4 JO parameters, electric and magnetic oscillator strength (f), radiative transition rates (A_R), branching ratios (β_R), radiative lifetimes (τ_R).

View Table | View all tables in this article

The emission band for the ⁵D₀-⁷F₆ transition, centered around 810 nm, could not be measured accurately because of limitations of our detector in this spectral range. We found that Ω₂>Ω₄ for both crystals, exactly as in Li₆Eu_1-xGd_x(BO₃)₃ solid solution crystals [19,20]. The striking feature about Ω₂ lies in its high sensitivity to the covalence degree of the chemical bond: the more covalent the Eu-O chemical bond, the higher the Ω₂ value, the higher the contribution of the electric dipole mechanism, but also the lower the symmetry. Since it is very sensitive to the crystal field, the Ω₂ value changes greatly from compound to compound. The ⁷F₁ and ⁷F₂ multiplet crystal field splittings in the two crystals are consistent with previous energy level diagram determination and crystal field calculations [27–29]. The energy level positions of the lowest spin-orbit multiplets are given in Table 2

Table 2. Energy level positions determined at room temperature in the Y_1.866Eu_0.134O₃ (A) and Lu_1.56Gd_0.41Eu_0.03O₃ (B) single crystals.

View Table | View all tables in this article

4. Magnetic susceptibility behaviour

In cubic Eu₂O₃, Caro and Porcher firmly established the important contribution from the Eu³⁺ cations located in the C_3i-site to the total magnetic susceptibility [28]. The fact that in this crystallographic site the lowest of the crystal field sublevels of the ⁷F₁ multiplet is noticeably low in energy (≈185 K) leads to a typical low temperature plateau of the Van Vleck paramagnetic behaviour which starts at lower temperature (≈32 K) and higher susceptibility value. Hence, the resulting average magnetic susceptibility displays a low temperature plateau which starts at lower temperature than if it was due only to the Eu³⁺ cations located in the C₂-site.

In Fig. 3, it is seen that Y_1.866Eu_0.134O₃ exhibits the typical Van Vleck paramagnetic behaviour (see [19] for a detailed phenomenological description) with the notable exception of the Curie-type tail at temperatures below ~16 K, which is likely to be due to some Gd³⁺ contamination and/or negligible Eu³⁺ reduction to Eu²⁺. One interesting thing to observe is that the magnetic susceptibility plateau starts at a higher temperature upon cooling than the one of cubic Eu₂O₃ shown in [28]. Even if the overall crystal field strength is different in both compounds, the crystal field sublevels energy positions of the ⁷F₁ multiplet remains virtually the same in both cubic Eu₂O₃ and Y₂O₃:Eu³⁺ [27,28], and so our magnetic susceptibility data suggests that in the flux-grown Y_1.866Eu_0.134O₃ crystals the Eu³⁺ cations distribution, which is expected to follow the ¾-C₂/¼-C_3i crystallographic landscape offered by the cubic structure, is shifted in favour of the C₂-symmetry site by a substantial amount, roughly ≈84%-C₂/16%-C_3i. When a ¾-¼ distribution is included in the magnetic susceptibility calculation, the low temperature plateau (fixed by the Eu³⁺ concentration, ≈1.79 10²¹ Eu.cm⁻³) is clearly overestimated by an amount which compares to the error bar of the measurement. Neutron or XRD data refinement, or ¹⁵¹Eu Mössbauer spectroscopy, would certainly help cross-checking this trend. Consequently, it seems that the liquid-solid partition coefficient is higher for the Eu³⁺ cations dissolved on the C₂-smmetry site in Y₂O₃. In Fig. 4, the Lu_1.56Gd_0.41Eu_0.03O₃ flux-grown crystal magnetic susceptibility behaviour could not be fitted to a Curie-Weiss law, a Van Vleck paramagnet law, nor a linear combination of both over the whole temperature range investigated. From 4.2 K to ~65 K, a reasonable Curie-Weiss fit could be obtained, leading to a Curie constant C≈0.983 K, a paramagnetic Curie temperature θ_p≈-3.2 K and an effective magnetic moment μ_eff≈8.13 μ_B/Gd³⁺. The latter value corresponds to 102.4% of the theoretical Gd³⁺ free ion value and suggests that at low temperatures, where the Van Vleck paramagnetism of the Eu³⁺ cations is not expected to affect the temperature dependence of the overall magnetic susceptibility, the paramagnetism is dominated by the Gd³⁺ magnetic moments. At higher temperatures, the (tiny) deviation from the Curie-Weiss behaviour is likely to be due to the Van Vleck paramagnetism of Eu³⁺ cations. As a matter of fact, the three crystal field sublevels of the C₂-site Eu³⁺ cations ⁷F₁ multiplet were observed at 275, 538 and 824 K, and the first ⁷F₂ crystal field sublevel in the same site was observed at 1232 K [29]. The fact that no linear combination of Van Vleck and Curie-Weiss paramagnetisms could fit the data stresses the complexity of this paramagnetism. Indeed, given the high substitution levels, Gd³⁺ and Eu³⁺ magnetic moments are probably not independent from each other. In principle, there can be Gd(C₂)-Gd(C₂), Gd(C₂)-Gd(C_3i), Gd(C_3i)-Gd(C_3i), Eu(C₂)-Gd(C₂), Eu(C₂)-Gd(C_3i), Eu(C_3i)-Gd(C₂) and Eu(C_3i)-Gd(C_3i) magnetic interactions, the Eu-Eu interactions being certainly negligible. In the Li₆Eu_1-xGd_x(BO₃)₃ solid solution crystals, drastically different paramagnetic behaviours were also observed between the x = 0 and the x = 0.25 members [19] and such discrepancies (Van Vleck versus Curie-Weiss) are similar to those observed in the comparison between the Lu_1.56Gd_0.41Eu_0.03O₃ magnetic susceptibility and that of Y₂O₃:Eu³⁺ in Figs. 3 and 4.

 

Fig. 3 MKSA magnetic susceptibility of an Y_1.866Eu_0.134O₃ flux-grown crystal. The energy level positions in the calculations were: C₂ symmetry, E₁ = 267.3 K, E₂ = 510 K, E₃ = 800 K and E₄ = 1750 K (adapted from Table 2) ; C_3i symmetry, E₁ = 183.3 K, E₂ = 714.4 K, E₃ = 714.4 K and E₄ = 1750 K (adapted from [28]).

Download Full Size | PDF

 

Fig. 4 MKSA magnetic susceptibility and inverse magnetic susceptibility of a Lu_1.56Gd_0.41Eu_0.03O₃ flux-grown crystal.

Download Full Size | PDF

5. Characterization of the Lu_1.56Gd_0.41Eu_0.03O₃ single crystal optical/scintillation properties

The Lu_1.56Gd_0.41Eu_0.03O₃ single crystal scintillation properties were characterized with a view to synchrotron X-ray imaging applications, with spatial resolutions in the ~2-25 µm range [30–32]. In this realm of applications, bulk Y₃Al₅O₁₂:Ce (YAG:Ce) and Lu₃Al₅O₁₂:Ce (LuAG:Ce) single crystals are currently reference materials [30–32]. These crystals exhibit a good tolerance to radiations, a good chemical (not hygroscopic) and mechanical stability, are easy to process with a good optical quality and can be polished down to thicknesses of 5 µm. However, YAG:Ce suffers from relatively low X-ray absorption efficiency. On the other side, LuAG:Ce displays higher stopping power and conversion efficiency, although it generally shows a high level of afterglow [33]. This is a problem when acquiring tomographic slices with high dynamic range.

In order to evaluate the scintillation properties of the Lu_1.56Gd_0.41Eu_0.03O₃ single crystal, the conversion efficiency, scintillation spectrum, UV-Visible transmission, afterglow and X-ray images were measured.

The (Lu,Gd)₂O₃:Eu³⁺ conversion efficiency is very high. Some comparisons are drawn in Table 3, especially with LuAG:Ce which is a reference crystal used for X-ray imaging with medium spatial resolution in synchrotron facilities.

Table 3. Conversion efficiency of Lu_1.56Gd_0.41Eu_0.03O₃ crystal compared with YAG:Ce and LuAG:Ce reference crystals.

View Table | View all tables in this article

The scintillation spectrum of the Lu_1.56Gd_0.41Eu_0.03O₃ crystal shown in Fig. 5 displays several bands which, by comparison with Fig. 2, can be assigned to the Eu³⁺ cations intraconfigurational transitions ⁵D₀→⁷F_0-5. The relative peak intensities seem to compare with those of Fig. 2, except for the ⁵D₀→⁷F₄ emission, which is slightly decreased. The scintillation spectrum of Lu_1.56Gd_0.41Eu_0.03O₃ single crystals matches well the quantum efficiency wavelength dispersion of the Frelon 2k (equipped with the chip TH7899 atmel), and more broadly speaking is well adapted to front illuminated sensors.

 

Fig. 5 Lu_1.56Gd_0.41Eu_0.03O₃ single crystal scintillation spectrum under X-ray excitation at 8 keV and matching with the atmel TH7899 Quantum Efficiency.

Download Full Size | PDF

The transmission of the Lu_1.56Gd_0.41Eu_0.03O₃ crystal was found to be 0.65-0.79 and devoid of sharp absorption edge over the 300-800 nm spectral range (Fig. 6). By performing Fresnel reflection losses correction, with n = 1.9, we could roughly estimate that the absorption coefficient in the red spectral range is ~1.3 cm⁻¹. Moreover, using the energy levels position given in [29] to calculate the ⁷F₂ multiplet partition function (Z≈2.51, at room temperature), we can estimate by the reciprocity method that the absorption coefficient at 610.4 nm is ~1.8 cm⁻¹. Hence, the crystal does not reabsorb significantly the scintillation light (l_ABS~5500 μm) where it is the most intense and the detector proves to be the most efficient.

 

Fig. 6 Transmission spectra of Lu_1.56Gd_0.41Eu_0.03O₃ and Y_1.866Eu_0.134O₃ single crystals between 300 and 800 nm.

Download Full Size | PDF

The afterglow at room temperature of the Lu_1.56Gd_0.41Eu_0.03O₃ crystal is low (Fig. 7). After ~100 ms, a dynamic range of 13 bits can be solved. The afterglow of the Lu_1.56Gd_0.41Eu_0.03O₃ crystal is lower than its ceramics analog [26] and also much lower than the LuAG:Ce single crystal. The afterglow in LuAG:Ce is attributed to vacancies, Lu-Al antisite defects or dislocations, which trap electron and holes, and then recombine non-radiatively [32]. This afterglow depends on the Ce concentration and that is why some discrepancies can arise from sample to sample in LuAG:Ce. An X-ray image with the Lu_1.56Gd_0.41Eu_0.03O₃ single crystal is shown in Fig. 8. It was difficult to reach a good focus with the Lu_1.56Gd_0.41Eu_0.03O₃ crystal. In comparison with standard YAG:Ce crystal, our Lu_1.56Gd_0.41Eu_0.03O₃ crystal clearly still have many defects with a size ranging from 1 to 20 μm.

 

Fig. 7 Afterglow of Lu_1.56Gd_0.41Eu_0.03O₃ single crystal at 8 keV for different exposure times (0.1 s, 1 s and 10 s), and comparison with a bulk LuAG:Ce crystal exposed during 1 s to X-rays.

Download Full Size | PDF

 

Fig. 8 X-ray resolution target image of the Lu_1.56Gd_0.41Eu_0.03O₃ crystal. The field view is 1.3x1.3 mm² and the grid lines thickness is ~25 μm.

Download Full Size | PDF

The high melting temperature of rare-earth sesquioxides raises the question of the crucible and refractory materials to be used: Pt, Mo, W, Ir, alumina and magnesia have to be avoided. So far, the best results have been achieved with expensive Re crucibles, pedestal, nozzles and top cap reflectors, which exhibit a propensity to oxidize in contact with the oxide-based refractory material and with the few ppm of oxygen flowing in the furnace, and also dissolve in molten Lu₂O₃ bath, forming ~20 μm-sized particles embedded in the resulting crystals. Re is too much expensive because its high mechanical hardness makes it difficult to shape in a specific form. The Re powder compaction process involved to obtain specifically shaped Re parts of the furnace, is also likely to introduce Fe and other undesired metallic optical contaminants. The chemical (and subsequently mechanical) stability of the Re parts requires a flux gas containing H₂ at ~2400°C, which entails health and safety requirements unlikely to favour large scale industrial production.

Moreover, for the rare-earth sesquioxydes preceding Yb₂O₃ in the periodic table, there exists structural phase transition, and even series of structural phase transitions, which make it impossible to obtain directly the cubic phase, or to obtain it with good (laser grade) crystal quality [34], as is the case for rare-earth doped Y₂O₃ crystals. If we calculate the average ionic radius on the rare-earth crystallographic sites in Lu_1.56Gd_0.41Eu_0.03O₃ and Y_1.866Eu_0.134O₃ crystals, we see that in Adachi and Imanaka’s diagram, the former is in some sense « equivalent » to growing Ho₂O₃, and the latter « equivalent » to growing Dy₂O₃, so that there must be hexagonal P6₃/mmc → cubic Ia-3 and hexagonal P6₃/mmc → monoclinic C2/m → cubic Ia-3 phase transitions, especially in Lu_1.56Gd_0.41Eu_0.03O₃ [35]. The chemical, spectroscopic and magnetic characterizations presented in this work firmly establish that, for the first time, cubic Lu_1.56Gd_0.41Eu_0.03O₃ and Y_1.866Eu_0.134O₃ crystals could be grown at half their melting point temperature. At this temperature, the cubic phase stability domain is directly reached from the liquid state, and it is possible to take advantage of isotropic thermal, thermomechanical and thermo-optical behaviors of these crystals upon cooling. In addition, the growth process is favored by the fact that the crystals have higher thermal conductivity upon cooling, the liquid phase exhibits higher liquid surface tension and lower vapor pressure.

The crystal growth takes place in air so that Eu³⁺ oxidation state can be stabilized without resorting to any post-growth thermal treatment annealing. As a matter of fact, the thermochemical data tables indicate [36,37] that the reduction temperature of Eu₂O₃ to EuO under an oxygen partial pressure of 10⁻⁷ atm is ~2400 K. Hence, melting Eu₂O₃ under a flow of gas containing hydrogen (P_O2~10⁻¹¹ atm) will completely reduce it with formation of many gas bubbles in the melt. From an industrial viewpoint, one should also note that the solvent used contains neither heavy metal, hydroxyl groups nor corrosive species, and that the furnace scheme is the most simple one can conceive for flux growth, with easily found high purity Pt crucibles and spatula, cheap (Kanthal) resistive heating and alumina refractory elements. In spite of the huge body of literature dealing with Eu³⁺-doped rare-earth sesquioxides and their spectroscopic properties, the availability of single crystals under “mild” conditions permitted us to obtain emission spectra calibrated in cross section units for the first time. Single crystals are the only objects which allow for the determination of intrinsic properties and of their anisotropy, even in cubic phases, when the property under investigation depends on the direction or implies the orientation of some field with respect to the structural polyhedra relevant to it. In comparison with complex preparation processes [1,2,4,8,11], note that in our growth process, the microstructure of the starting powder is not a key parameter.

6. Conclusions

We have successfully grown a ~0.5 cm³ Lu_1.56Gd_0.41Eu_0.03O₃ as well as smaller Y_1.866Eu_0.134O₃ sesquioxide single crystals of the cubic phase. The Li₆(Lu,Gd,Eu)(BO₃)₃ and Li₆(Y,Eu)(BO₃)₃ solvents that we used for the high-temperature solution growth makes it possible to obtain these crystals at half their melting temperature, in air, without resorting to any post-growth thermal annealing to stabilize the Eu³⁺ oxidation state. According to the Van Vleck paramagnetism observed in Y₂O₃:Eu³⁺ crystals, the Li₆(Y,Eu)(BO₃)₃ solvent leads to a preferential dissolution of the Eu³⁺ cations on the C₂-site of the cubic Y₂O₃ crystal structure. Transmission measurements established the promising optical quality of the samples, especially for Lu_1.56Gd_0.41Eu_0.03O₃ in the red spectral range where most of the scintillation light is emitted. The Lu_1.56Gd_0.41Eu_0.03O₃ single crystal, grown and tested for the first time, is an efficient X-ray scintillator and its characteristics seem promising for X-ray imaging with medium spatial resolution, especially when it is coupled to such a front-illuminated CCD chip as atmel TH7899 2k with a maximum quantum efficiency between 620 nm and 800 nm. The Lu_1.56Gd_0.41Eu_0.03O₃ emission spectrum exhibits in the 575-750 nm spectral range, series to 4f-4f transitions arising from the Eu³⁺ cations, particularly well-suited to most CCD/CMOS cameras.