Correlation of EPR and Photoluminescence Analysis for Crystalline Defects in Eu³⁺/Yb³⁺-Doped Lutetium Silicate Sol–Gel Powders

Abstract

Crystalline defects such as oxygen vacancies have been studied little by electron paramagnetic resonance (EPR) spectroscopy for silicate-based luminescent materials. In this study, lutetium oxyorthosilicate powders were prepared by the sol–gel method, using TEOS (silicon source) and rare earth salts as precursors. The cross-linking agent, Glymo, contributed silicon atoms to the precursor solution in all systems. The addition of Glymo to Lu₂SiO₅, Lu₂SiO₅:Eu and Lu₂SiO₅:Eu/Yb influenced the morphology and chemical structure of the powders, leading to Lu₂Si₂O₇ formation. The crystalline defects in the lutetium silicate systems were investigated by EPR spectroscopy, and several defects related to oxygen were identified, as well as impurities from the precursors. Photoluminescence emission spectra revealed Eu³⁺ transitions between ⁵D₀ → ⁷F₀, ⁵D₀ → ⁷F₁ and ⁵D₀ → ⁷F₂ under 258 nm excitation, in addition to oxygen vacancy emissions between 500 and 550 nm. Oxygen vacancies were identified and confirmed by correlating EPR and photoluminescence studies.

1. Introduction

It is known that crystals have defects in their crystalline structure. Oxygen vacancies (V_O) are an example of this type of defect. The experimental characterization of V_O has been conducted using X-ray photoelectron spectroscopy (XPS), Raman spectroscopy complemented by UV–Vis spectroscopy [1,2,3] and EPR spectroscopy, primarily for other applications such as photocatalytic activity [4]. In structural studies related to crystalline defects in silicate-based luminescent materials, little has been studied by correlating EPR spectroscopy and photoluminescence (PL) [5,6]. Previous studies have experimentally and theoretically shown the presence of oxygen vacancies in lutetium silicate systems doped with Ce³⁺ [7,8]. Lutetium oxyorthosilicate is a rare earth silicate with similar characteristics to commercial phosphors, such as high density (7.4 g/cm³), and for scintillation application, a high light output yield (~27,500 photons/MeV), and a fast response (~40 ns), enabling the system to be able to be doped with different rare earth ions [9,10,11,12,13,14]. Structurally, the fundamental silicate unit consists of a [SiO₄]⁴⁻ tetrahedron; the Si-O bond tends to be of a more covalent character than ionic, resulting in a stable bond [15]. The moderate repulsive forces between silicon atoms allow the linking of [SiO₄]⁴⁻ via common oxygen atoms to form many polymorphs; thus, silicate can be organized in different ways, like tetrahedron rings, chains, sheets or isolated. Generally, silicate compounds contain Si–O–M bonds, where the M–O bond is weaker than the Si–O bond due to the metal ions having a lower valence than silicon. This results in the formation of [SiO₄]⁴⁻ tetrahedra where silicon attracts oxygen ions more strongly than the metal ions [15]. Rare earth silicate compounds consist of isolated (SiO₄) tetrahedra or groups of (Si₂O₇) and (Si₃O₁₀), where the rare earth cations are coordinated from six to ten oxygen atoms. The binary system RE₂O₃–SiO₂ with a composition of 1:1 has two different monoclinic structural types with the space group P2₁/c for type A and the space group C2/c for type B. In rare earth (RE) silicates, the RE–O bond is essentially electrostatic because of the strongly electropositive nature of RE³⁺. Lu₂SiO₅ is an oxyorthosilicate compound containing two types of anions: the (SiO₄) tetrahedra and non-silicon-bonded oxygen ions. Lutetium oxyorthosilicate crystallizes in a monoclinic structure containing two crystallographic sites for lutetium in the space group C2/c (type B) [16]. The first lutetium site (Lu1) is coordinated with six oxygen atoms with Lu–O bond distances in the range of 2.13–2.25 Å, and the second lutetium site (Lu2) is coordinated with seven oxygen atoms with Lu–O bond distances in the range of 2.12–2.62 Å [17]. The aim of this investigation is to identify the crystalline defects present in lutetium silicate phosphors. This will contribute to the structural analysis of luminescent materials by considering intrinsic defects and correlating EPR spectroscopy analysis with photoluminescence properties.

2. Results and Discussions

The XRD patterns of the Lu₂Si₂O₇/Lu₂SiO₅/Glymo (LSG), Lu₂Si₂O₇/Lu₂SiO₅/Glymo:Eu³⁺ (LSG:Eu) and Lu₂Si₂O₇/Lu₂SiO₅/Glymo:Eu³⁺/Yb³⁺ (LSG:Eu/Yb) powders calcined at 1100 °C are shown in Figure 1A. The stable phase (type B) of Lu₂SiO₅ with a monoclinic structure and space group C2/c (ICSD 98-008-9624) was identified in all samples. Type B Lu₂Si₂O₇ (ICSD 98-041-2249) was also present due to the addition of Glymo as the cross-linking agent, which acts as a silicon precursor like TEOS, contributing silicon atoms to the reaction. To obtain the Lu₂Si₂O₇ system by the Sol–Gel method, it is necessary to use 1 mol of Lu₂O₃ sol and 2 mol of SiO₂ sol. Even if only 1 mol of SiO₂ sol was used, the amount of Glymo added to the precursor solution contributes 1 mol of silicon to form lutetium pyrosilicate (Lu₂Si₂O₇).

The hydrolyzable methoxy groups in Glymo can undergo nucleophilic attack by the solvents present in the lutetium silicate precursor solution, such as ethanol and water, where the OH⁻ groups displace a hydrogen atom from the methoxy group to interact with the lutetium atoms. In this way, Glymo contributes a larger amount of silicon atoms, thus forming Lu₂Si₂O₇. The crystallographic planes were identified for the main crystalline signals of Lu₂SiO₅ (denoted with “*”) and Lu₂Si₂O₇ (denoted with “●”). As demonstrated in the diffractogram, a greater number of signals match for lutetium pyrosilicate ICSD no. 98-041-2249, thus indicating that the addition of Glymo favored the formation of Lu₂Si₂O₇ and as a secondary phase, lutetium oxyorthosilicate. The crystallographic signals for the three samples were identified at the same diffraction angles, indicating no displacement, which suggests the incorporation of Eu³⁺ and Yb³⁺ as substitute atoms replacing the Lu site in the crystalline structure. The average crystallite size was calculated using Scherrer’s formula $D=k\lambda /\beta \cos \theta$ [18], where $D$ is the average crystallite size, $k$ is the shape factor, $\lambda$ is the X-ray wavelength, $\beta$ is the peak width at FWHM and $\theta$ is the Bragg angle at 16, 22 and 19 nm for the LSG, LSG:Eu and LSG:Eu/Yb samples, respectively.

Porosity was evident in Figure 2 due to the elimination of vapors, forming cavities or empty spaces within the matrix. Initially, the addition of Glymo favored the interactions with the precursor solution, increasing the formation of a 3D network during the Sol–Gel process because of the ability of Glymo molecules to cross-link [19,20]. Subsequently, the heat treatments eliminated the organic Glymo components, generating the observed porosity. The larger surface area, attributed to porosity, facilitates interaction with the environment; for example, it enables oxygen loss during heat treatments. Defects have been observed on surfaces, where they concentrate and stabilize, thus forming oxygen vacancies in the system [21,22,23].

Figure 3 shows the EPR spectra for the lutetium silicate powders with the addition of Glymo. For each sample (LSG, LSG:Eu and LSG:Eu/Yb), a very broad signal appears to extend from 0 to 600 mT or more and increases its intensity in relation to the amount of lanthanide ions; the signal width is due to the short relaxation time that these ions have [24]. Also, these signals present a maximum at approximately g ~ 16.344 (40 mT). This is important because it allows us to infer the possible oxidation states for the rare earth ions. For Lu ([Xe]4f¹⁴5d¹6s²), the common oxidation state is 3+, and although less frequent, it can be 2+, 1+ and 0. For EPR signals to exist, there must be at least one unpaired electron; this implies that the EPR signals due to lutetium could originate from some Lu²⁺ ions present in the samples, since the other oxidation states do not provide unpaired electrons that could contribute to the EPR signal. The radii of Lu²⁺ are considerably smaller than that of Lu³⁺ because the loss of electrons, particularly from the 6s orbitals, results in greater nuclear attraction for the remaining electrons and a reduction in the size of the ion. From the above, it can be deduced that the possible existence of Lu²⁺ ions and the decrease in their atomic radius with respect to Lu³⁺ would break the crystallinity of the lattice, creating defects or vacancies, which were detected by EPR around g~2 (H~333 mT), as shown in Figure 3.

For europium ([Xe]4f⁷6s²), the most common oxidation state is 3+; but it is also common to find the 2+ state in a stable manner [25]. However, Eu³⁺ is not paramagnetic, implying that there may also be a certain amount of Eu²⁺ ions that could contribute to the increased EPR signal intensity. As shown in Figure 3, the signal intensity at g~16.344 (40 mT) for LSG and LSG:Eu is not significant, implying that the oxidation state of Eu is 3+. Nevertheless, the appearance of new vacancies or defects is evident by EPR, as can be seen in the g~2 region (H~333 mT). This indicates that Eu entered the lattice substitutionally, and due to the different atomic radii of the Lu²⁺ and Eu³⁺ ions, a greater detriment in the lattice crystallinity of the LSG:Eu sample is induced compared to LSG. Analogous to Eu, ytterbium ([Xe]4f¹⁴6s²) also exhibits stable oxidation states 3+ and 2+; however, the only oxidation state detectable by EPR is 3+. Two related effects are observed in the EPR spectrum (Figure 3) for the LSG:Eu/Yb sample: the first effect is the increase in signal intensity at g~16.344; while the second is the significant reduction in defects (signals at g~2). This implies that Yb enters the lattice substitutionally, most likely at Eu ion sites with oxidation states of 3+, which increases the intensity and crystallinity of the g~16 signal by significantly reducing the defects in the lattice as observed in the g~2 region. Around 158 mT, an absorption signal is identified with a value of g~4.25, which is related to impurities of Fe³⁺ ions present in the precursors [26]. Such impurity of Fe³⁺ in EPR analysis is of little importance in the discussion of defects or vacancies.

In the 300 to 400 mT range, oxygen-related absorption signals are identified. All samples were calcined at 1100 °C for 2 h and then irradiated at 258 nm for photoluminescence measurements. Lutetium silicate is a ceramic material that has a theoretical bandgap of 4.68 eV [27]. The excitation source energy supplied to the system was 4.8 eV, which can promote the passage of electrons from the valence band to the conduction band, thus forming electron–hole pairs. The charge carriers generated can stabilize at defect sites, forming paramagnetic species detectable by EPR, as observed in the spectrum. On the other hand, the holes generated by exposure to ultraviolet light are stabilized at O²⁻ sites mainly on the sample surface, forming O⁻ ions. All samples show intense EPR signals in the region of g = 2 (Figure 3 and Figure 4), which is typical for free radicals and paramagnetic structural defects.

Figure 4 shows the absorption signals related to defects assigned to oxygen. From the three lutetium silicate systems (LSG, LSG:Eu and LSG:Eu/Yb), signal B is identified, and it is of anisotropic-axial type spectrum with values of g// = 2.021 and g⊥ = 2.002. This signal is found in different proportions due to the formation of holes in the system. That is, all the samples have a different concentration of defects, with the LSG:Eu sample being the one with the highest concentration of O⁻ type species, in addition to the A signal, which is isotropic and corresponds to an F⁺ color center where an electron is trapped in an oxygen vacancy [28]. Previous studies reported that the signal corresponding to ionized oxygen vacancies (V’_O) is obtained in an interval with g values between 1.9560 and 2.0030 [29], obtaining in this case signal A with a value of g = 2.0014. On the other hand, sample LSG:Eu/Yb is the one with the lowest concentration of defects, with a relative amount of O⁻ type species of 1:2.5:4.8 for samples LSG:Eu/Yb:LSG:LSG:Eu, respectively.

The excitation spectra for lutetium silicate systems are shown in Figure 5A when monitored at 612 nm, which corresponds to the characteristic emission (⁵D₀ → ⁷F₂) of Eu³⁺ ion. Absorption signals related to Eu³⁺ ion are identified at 360, 380, 391, 400 and 413 nm which are attributed to ⁵D₄ ← ⁷F₀, ⁵L₇ ← ⁷F₀, ⁵L₆ ← ⁷F₀, ⁵L₆ ← ⁷F₁ and ⁵D₃ ← ⁷F₁ respectively [30,23]. Defects like oxygen vacancies can absorb excitation energy, giving rise to absorption bands different to the dopant ion in the excitation spectrum, and thus the absorption signals at 298 and 320 nm are attributed to defects [31,32]. The charge transfer band (CTB) between the oxygen to europium ions (O-Eu) is identified as a broad absorption signal centered at ~250 nm. The emission spectra for the three lutetium silicate samples (LSG, LSG:Eu and LSG:Eu/Yb) in Figure 5B were obtained at an excitation wavelength of 258 nm. It has been reported that defects can introduce energy levels for electrons in the bandgap of the material [33]. As observed in the EPR spectra, all samples exhibit defects associated with O⁻, like oxygen vacancies.

The electrons that are not trapped by intrinsic defects in the system recombine with the photogenerated holes in the valence band. The excitation wavelength used for the measurement corresponds to the charge transfer band between the O–Eu (λ_exc = 258 nm), and emissions related to oxygen vacancies at approximately 513 and 540 nm were observed for all the systems [34,35]. Previous studies have discussed the role of oxygen vacancies, which can act as intrinsic luminescent centers [36], and are also able to function as sensitizers, thereby transferring energy to activators [37]. It is possible that an energy transfer occurs between oxygen vacancies and Eu³⁺ ions, as absorption signals associated with oxygen vacancies were identified in the excitation spectra. In the case of LSG:Eu and LSG:Eu/Yb, several emissions corresponding to Eu³⁺ ions were identified. The presence of a ⁵D₀ → ⁷F₀ transition (λ_em = 581 nm) indicates that the Eu³⁺ ion is inserted into the same crystallographic site for the Lu₂Si₂O₇ system. Since this transition (⁵D₀ → ⁷F₀) is known as a forbidden transition according to the selection rules, its observation is related to the J-mixing, where it can be assumed that a strong crystal-field effect occurs [38,39,40]. The emissions located at 588, 590 and 599 nm correspond to a magnetic dipole type transition (⁵D₀ → ⁷F₁) that is independent of the chemical environment surrounding the dopant ions; in this case, europium, unlike the emissions at 612 and 620 nm which correspond to an electric dipole type transition that is sensitive to the chemical environment surrounding the europium ion. It is important to mention that the characteristic emission of the europium ion is usually the one with the highest intensity corresponding to the electric dipole type (⁵D₀ → ⁷F₂); however as mentioned above, this transition is sensitive to the chemical environment, which is modified by adding Glymo as cross-linking agent, thus limiting the radiative transitions between the ⁵D₀ excited state and the ⁷F₂ ground state.

This may benefit the magnetic dipole type transition (⁵D₀ → ⁷F₁), so it is possible that the modification of the chemical environment, in addition to the generation of holes due to the excitation energy of the system, favor the effective radiative transitions between the ⁵D₀ excited state and the ⁷F₁ ground state, considering that there is no influence from the addition of Glymo on the emission at 590 nm [41]. The decrease in luminescence intensity in the LSG:Eu/Yb system can be attributed to the incorporation of the Yb³⁺ ion. The co-doping of the system with Yb³⁺ results in a competitive environment amongst the dopant ions, leading to the occupation of the preferential Lu site in the Lu₂SiO₅ phase, which is coordinated with seven oxygen atoms [17]. A schematic representation of the luminescence process is shown in Figure 6A. Electrons are excited from the valence band to the conduction band by the CTB (λ_exc = 258 nm), and the excited electrons in the conduction band have two possible paths: (1) De-excitation through non-radiative transitions to the level introduced by V_O, followed by emission (radiative transitions) at 513 and 540 nm. (2) Electrons decay to Eu³⁺ levels and through non-radiative relaxation to the bottom of the excited states (⁵D₀), to decay by radiative transitions to the ground states ⁷F_J (J = 0, 1, 2) while emitting at different wavelengths. Figure 6B corresponds to the chromaticity diagram obtained using the CIE 1931 standard which provides a visual understanding of color properties. The color coordinates for the LSG system are (0.3049, 0.6450), and this corresponds to the pure system which indicates only the oxygen vacancy emissions in the green region of the visible spectrum. For LSG:Eu and LSG:Eu/Yb systems, the color coordinates are (0.5345, 0.4518) and (0.5006, 0.4804), respectively, which are in the orange–red region of the diagram, and are typical of emissions for Eu³⁺ ions with similar intensities at ⁵D₀ → ⁷F₁ and ⁵D₀ → ⁷F₂ transitions. In summary, all lutetium silicate samples are active under irradiation and generate excitons which can stabilize in the intrinsic defects of the material. As observed in the SEM results (Figure 2), the porous morphology favors environmental interaction during heat treatments, generating intrinsic defects, such as oxygen vacancies, which were identified by EPR spectroscopy and correlated with the luminescent properties for the lutetium silicate systems. The presence of O⁻ type defects creates new energy levels in the bandgap of lutetium silicate, and the electrons that are stabilized in an oxygen vacancy form the color center.

3. Materials and Methods

3.1. Preparation of Powders

Lutetium nitrate [Lu(NO₃)₃, 99.9%], tetraethyl orthosilicate [TEOS Si(OC₂H₅)₄, 99%], europium nitrate [Eu(NO₃)₃, 99.9%], ytterbium chloride [YbCl₃, 99.9%], ethanol [C₂H₅OH, 99%], acetylacetone [AcAc C₅H₈O₂, 99%] and glycidyloxypropyl trimethoxysilane [Glymo C₉H₂₀O₅Si, 98%] were used as the starting materials. The lutetium silicate powders were obtained by the Sol–Gel method. Firstly, Lu(NO₃)₃ was mixed with ethanol and the chelating agent AcAc in a 1:1 molar ratio of Lu:AcAc under vigorous stirring for an hour to obtain the Lu₂O₃ sol [42,43,44]. Two dissolutions were prepared to obtain the SiO₂ sol: (1) TEOS was added to ethanol in a 1:4 molar ratio of TEOS:ethanol, and (2) HCl was added to deionized water in a 530:1 molar ratio of H₂O:HCl and stirred for one hour. Then, the solutions were mixed and stirred for 21 h [45]. The Lu₂O₃ and SiO₂ solutions were mixed in a 1:1 molar ratio and stirred for 2 h to obtain the Lu₂SiO₅ sol [46]. Glymo was added as a cross-linking agent to all samples in a molar ratio of 1.3 of Glymo:Lu₂SiO₅. The samples were named LSG, corresponding to the pure sample, LSG:Eu (lutetium silicate doped with Eu³⁺, 5% mol) and LSG:Eu/Yb (lutetium silicate co-doped with Eu³⁺, 5% mol and Yb³⁺, 5% mol). A detailed description of the main reactions and the proposed reaction mechanism for the Lu–Si–Glymo complex formation is reported in our previous paper [47].

3.2. Characterization

The crystalline structures were analyzed by X-ray diffraction (XRD) using a powder diffractometer Bruker D8 Advance with Cu target and kα radiation (1.5406 Å), at a step of 0.02° and a θ/2θ equipment configuration in the range of 10 to 70°. The powder morphology was assessed using a Quanta FEG 250 scanning electron microscope operated at 15 kV. EPR spectroscopy was performed on a Bruker ELEXSYS II E 500 spectrometer with X-band (~9.5 GHz) at 77 K. Photoluminescence studies were performed on an F-7000 FL spectrophotometer (Hitachi High-Technologies Corporation, Tokyo, Japan).

4. Conclusions

Lutetium oxyorthosilicate (Lu₂SiO₅) type B and lutetium pyrosilicate (Lu₂Si₂O₇) type B powders were obtained by the Sol–Gel method and calcined at 1100 °C. The addition of Glymo modified the crystalline structure and morphology to form Lu₂Si₂O₇, and the incorporation of Eu³⁺ and Yb³⁺ ions was confirmed by XRD. Oxygen vacancies were identified in all the samples besides the F⁺ center for LSG:Eu. The formation of defects, like oxygen vacancies, was confirmed in the lutetium silicate systems by photoluminescence studies showing emission signals between 500 and 550 nm. These results provide the basis for the development of luminescent materials in the study of crystalline defects by EPR spectroscopy and their relationship with luminescent properties.