Luminescence Mechanistic Study of BaLaGa3O7:Nd Using Density

Mar 8, 2016 - Synopsis. The luminescence mechanism of BLGO:Nd is clarified by density functional theory. Luminescence can be realized by the introduct...
0 downloads 3 Views 4MB Size
Download PDF
Article pubs.acs.org/IC

Luminescence Mechanistic Study of BaLaGa3O7:Nd Using Density Functional Theory Calculations Junling Meng,†,‡ Xiaojuan Liu,*,† Congting Sun,† Chuangang Yao,†,‡ Lifang Zhang,†,‡ Fen Yao,† Dongfeng Xue,† Jian Meng,† and Hongjie Zhang† †

State Key Laboratory of Rare Earth Resource Utilization, Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, Changchun 130022, P. R. China ‡ University of Chinese Academy of Sciences, Beijing 100049, P. R. China S Supporting Information *

ABSTRACT: BaLaGa3O7:Nd (BLGO:Nd) has been investigated as a laser crystal material for about three decades. In the present work, the luminescence mechanism of BLGO:Nd is clarified by density functional theory (DFT) calculations. Structural optimization was first performed on the constructed supercell to obtain the equilibrium geometry. On the basis of the optimized crystal, the electronic structures of the BLGO host (without and with single defects) and the BLGO:Nd phosphor (without and with neighboring defects) were comprehensively investigated. Three important features are revealed by theoretical analyses. First, single defects in BLGO have little effect on the light emission, although the impurity levels appeared within the band gap. Second, luminescence can be realized by the introduction of Nd ions. Calculations of optical properties demonstrated that parity-forbidden transitions among the 4f levels are partially allowed because the mixing of 4f and 5d configurations occurs at higher empty 4f levels. It is thus clear that the electronic transitions between occupied 4f and empty 4f−5d states are electric-dipole-allowed. Therefore, light emission in BLGO:Nd can be achieved in the electronic transition process of Nd 4f electrons → empty 4f−5d levels → empty 5d levels → Nd 4f levels. The neighboring intrinsic defects play only an auxiliary role in prolonging the decay time. Third, co-doping of Tb in BLGO:Nd is considered to be beneficial to luminescence in theory because of its shallow to deep distribution of impurity orbitals in the band gap. Therefore, BLGO:Nd co-doped with other lanthanide ions will offer guidelines in the search for the best luminescent materials.

1. INTRODUCTION Disordered laser crystals have attracted much attention because they can combine excellent thermomechanical features and relatively broad emission lines. The rare-earth gallates, belonging to the large group of crystals with the general formula ABC3O7 (A = Ca, Sr, Ba; B = Y, La−Gd; C = Al, Ga), have been widely investigated because of their lasing possibilities.1−5 Polycrystalline BaLaGa3O7 (BLGO) was prepared for the first time by Ismatov and co-workers by sintering a stoichiometric mixture of the components at high temperature.6 Afterward, large single crystals of BLGO were produced by Piekarczyk et al. through the Czochralski method.7 Subsequently, researchers have paid much attention to the measurement of various parameters of BLGO single crystals to explore its potential properties, such as refractive indices and elastic, piezoelectric, and dielectric constants.8,9 Experimentally, inorganic compounds doped with trivalent lanthanide ions are widely used as luminescence materials. The luminescence of these ions originates from transitions within the partially filled 4f shell. Because single crystals of BLGO are transparent in the 1500−41000 cm−1 range,10 they are of interest as potential host lattices for trivalent lanthanide-ion dopants to form luminescent and laser materials. The © 2016 American Chemical Society

luminescence of Nd-doped BLGO (BLGO:Nd) is the most interesting in academic respects, because the ionic radii of La and Nd do not differ considerably and Nd ions have partially filled 4f orbitals, so a high solubility limit of Nd3+ and 4f → 5d transition can be expected.5,11,12 Meanwhile, the luminescence properties of BLGO:Nd are the most significant compared to those of other lanthanides.13−15 To date, the luminescence mechanism of BLGO:Nd has still not yet been determined even though a large number of experimental studies have been carried out. In general, rare-earth ions always serve as phosphors in the host lattice, because of their 4f → 5d electronic transition after being excited. Meanwhile, the intrinsic defects, such as oxygen, gallium, lanthanum, and barium vacancies (VO, VGa, VLa, and VBa, respectively) in BLGO, which can be generated under the annealing conditions, are expected to be crucial to the luminescence. A previous experimental study assumed that the blue emission in BLGO occurs when an electron from a VO or Ga center recombines with a trapped hole.13 However, scientific evidence to support this statement is still absent. Therefore, a study of the electronic Received: November 22, 2015 Published: March 8, 2016 2855

DOI: 10.1021/acs.inorgchem.5b02714 Inorg. Chem. 2016, 55, 2855−2863

Article

Inorganic Chemistry

description of correlation effects in rare-earth oxides. In this work, we used Dudarev et al.’s method29 in which only an effective Hubbard parameter Ueff = U − J appears in the Hamiltonian. The magnitude of Ueff was varied between 0.0 and 2.0 eV for the Nd and Tb f orbitals. The plane-wave expansion cutoffs were 7.0 for expanding the wave function (RKMAX) and 12 for expanding the densities and potential (GMAX) in all compounds. A 4 × 9 × 2 Monkhorst−Pack grid in the complete Brillouin zone was employed, and Brillouin zone integration was carried out with a modified tetrahedron method.30 The selfconsistent calculations were considered to be converged when the charge convergence was less than 10−4 e. The defective structural models, based on a 2 × 2 × 1 supercell with a stoichiometry of La8Ba8Ga24O56 (BLGO), were constructed by removing one of the O, Ga, Ba, or La atoms in a perfect La8Ba8Ga24O56 structure or substituting one of the La atoms with a Nd, Tb, or Eu atom (NdLa, TbLa, or EuLa, respectively). The BLGO host without defects and with single defects of an oxygen vacancy (VO‑I, VO‑II, and VO‑III), a gallium vacancy (VGa‑I and VGa‑II), or a lanthanum vacancy (VLa) was first investigated. Subsequently, the doping of Nd3+ in BLGO without and neighboring defects and with a neighboring oxygen vacancy (NdLa + VO‑I, NdLa + VO‑II, and NdLa + VO‑III), gallium vacancy (NdLa + VGa‑II), or lanthanum vacancy (NdLa + VLa) was also studied in this work. Finally, other rare-earth ions, such as Tb and Eu ions, were employed as activators, and their luminescence possibilities were explored.

structures of the dopants and defects in BLGO should be able to reveal the luminescence mechanism. In recent years, many researchers have applied theoretical calculations to investigate the luminescence rationale. For example, in 2011, Muñoz-Garciá and Seijo studied the atomistic structure and electronic structure of singly substituted (Ce or La) and co-doped (Ce and La) yttrium aluminum garnet (Y3Al5O12, YAG) by means of first-principles periodicboundary-condition DFT calculations.16 Detailed discussions were presented to reveal the significant roles of the crystal structure and 4f electrons in Ce,La−YAG. Recently, Qu and coworkers used first-principles calculations to investigate the luminescence mechanism of co-doped CaAl2O4:Eu,Nd, which is a class of persistent luminescence materials.17 They reported that the 4f and 5d levels of the luminescent center Eu are located within the band gap, the O vacancies serve as electron traps, and the Ca vacancies play an assistant role to the oxygen defects. Furthermore, Wen et al.18 and Ramanantoanina et al.19 also investigated 4f → 5d transition of γ-Ca2SiO4:Ce3+ and CsMgBr3:Eu2+ phosphors, respectively, using first-principles calculations. However, theoretical accounts of the electronic structure and properties of Nd3+ in a luminescent host considering its 4f orbital configurations are relatively rare. Herein, we present a fully theoretical work based on DFT enabling the study of the luminescence mechanism of lanthanide-doped BLGO. This study was undertaken with three goals: First, it was intended to explore whether the BLGO host without and with single defects could emit light. The second aspect was to analyze the effects of Nd ions on the luminescence of BLGO:Nd, as well as the effects of neighboring defects. Finally, we also calculated the properties of other phosphor components, namely, BLGO:Tb and BLGO:Eu, to explore the reason why Nd ion is the best phosphor with the BLGO matrix. It is worthwhile to note that, on the basis of DFT calculations, co-doping of BLGO with Nd and Tb is anticipated to provide a better luminescent material than BLGO:Nd in terms of the luminescence intensity and lifetime.

3. RESULTS AND DISCUSSION 3.1. Perfect BLGO. Barium lanthanum gallate (BaLaGa3O7) crystallizes in the tetragonal P4̅21m space group (No. 113) within the D2d3 point group with two formula units in the primitive cell and the Ba and La ions randomly distributed in a 1:1 ratio in one position.2,31 The crystal structure is built up from CO4 layers in the crystallographic ab plane based on the experimental structure, and the divalent Ba and trivalent La cations between the layers are distributed randomly in 8-foldcoordinated sites. The trivalent lanthanide ions are able to enter into the sites formally occupied by La3+. The structure model is shown in Figure 1. This structural framework is composed by two types of vertex-sharing GaIO4 and GaIIO4 tetrahedrals with La and Ba atoms inserted in the interstitials. On the basis of the different coordination environments (see Figure 1), the O atoms occupy three different crystallographic positions

2. COMPUTATIONAL DETAILS AND MODELING In present work, the equilibrium geometries and formation energies of defects were calculated using the Vienna ab initio simulation package (VASP),20,21 which is based on DFT and implemented with the projector-augmented wave (PAW) pseudopotential.22 The exchange and correlation functional was given by the generalized gradient approximation (GGA) as formulated by Perdew, Burke, and Ernzerhof (PBE).23 Each crystal structure was fully relaxed with a plane-wave energy cutoff of 500 eV under a residual force of ≤0.05 eV/Å. The electronic structure properties were further obtained from DFT calculations with the program package WIEN2K,24,25 by applying the full-potential linearized augmented-plane-wave (FL−LAPW) method,26,27 which is among the most accurate methods for performing electronic structure calculations for crystals and has been used for a wide spectrum of materials, including insulators, semiconductors, metals up to f-electron systems, and intermetallic compounds. The band structure can be directly compared to experiment in weakly correlated cases. The aim of employing this calculation program was to precisely analyze the role of the 4f orbitals of lanthanide ions in the luminescence. In the LAPW method, space is divided into atomic spheres and the interstitial region, and then the electronic states are classified as core states and valence states. The valence states are expanded using the plane-wave basis functions. The GGA−PBE method was also used for the exchange correlation potential. To account for the strong Coulomb repulsion between the localized 4f states of lanthanide ions, appropriate Hubbard-like terms (U) were introduced in the effective potential,28 leading to an improved

Figure 1. Schematic views of the perfect BLGO structure along the c direction. Color scheme: red, orange, and magenta for O-I, O-II, and O-III, respectively; aqua and olive green for Ga-I and Ga-II, respectively; blue and dark yellow for Ba and La, respecticely. 2856

DOI: 10.1021/acs.inorgchem.5b02714 Inorg. Chem. 2016, 55, 2855−2863

Article

Inorganic Chemistry

perfect BLGO is an insulator with a calculated band gap of 3.33 eV. To the best of our knowledge, the experimental band gap of crystalline BLGO has not been reported in the literature. However, the measured band gap should be slightly smaller than that calculated for perfect BLGO, because experimental samples will contain some interfaces, defects, and disordered domains. In addition, the single-crystalline structure of BLGO is mainly composed of GaO4 tetrahedra. Therefore, the bonding orbitals of Ga−O bonds contribute mainly to the valence bands (VBs). As can be seen in Figure 2, the VB is indeed mainly occupied by O 2p states. From the PDOS, one can also see that the 5d orbitals of Ba and La make the main contributions to the conduction band. A previous experimental study assumed that the blue emission in BLGO occurs when an electron from an oxygen vacancy or Ga center recombines with a trapped hole.13 Can VO levels offer electrons to other captured holes? Does such a trapped hole level exist in BLGO? The following sections provide a detailed discussion of these topics. To determine the stabilities of various defects, we first computed their formation energies EF. According to ref 17, EF for a single defect in BLGO is calculated as

(denoted O-I, O-II, and O-III). By means of equilibrium geometry calculations, the optimized lattice parameters of the pristine BLGO system were calculated by both the GGA and LDA methods, as summarized in Table 1, in which the Table 1. Lattice Parameters of the Primitive Unit Cell Based on the Experimental BLGO Structure Calculated with Both the GGA and LDA methods, along with the Experimental Data as a Reference a (Å) b (Å) c (Å) ⟨GaI−OI⟩ (Å) ⟨GaII−OI⟩ (Å) ⟨GaII−OII⟩ (Å) ⟨GaII−OIII⟩ (Å) ⟨GaI−OI−GaII⟩ (deg) ⟨GaII−OIII−GaII⟩ (deg)

GGA

LDA

experimental2,24

8.246 8.246 5.478 1.870 1.889 1.825 1.847 115.4 127.9

8.073 8.073 5.345 1.836 1.854 1.795 1.829 114.7 128.7

8.145 8.145 5.382 1.837 1.859 1.791 1.833 116.2 127.4

experimental data are also listed as a reference. It seems that the LDA-optimized parameters are more consistent with the experimental results than the GGA-obtained parameters. This is a prevalent phenomenon in geometry optimization. However, in this case, we still would like to select the GGA approach for calculating electronic structures and optical properties because, from a general perspective, the GGA exchange-correlation functional performs much better than the LDA exchange-correlation functional in describing the transition-metal oxide bonds and their electronic structures. To ensure a low concentration of doping and defects and simultaneously reduce the CPU time, we built the 2 × 2 × 1 supercell with 96 atoms in BLGO on the basis of the experimental tetragonal structure. After the supercell had been built, a monoclinic crystal structure with the P1 space group was presented. Even though the space group had changed, the reasonable properties of the BLGO primitive cell were the same as those of its supercell. The computed band structures and projected densities of states (PDOS) of BLGO are shown in Figure 2. Obviously,

E F = E(Ba8 − xLa8 − yGa 24 − zO56 − δ X n) − E(La8Ba8Ga 24O56 ) − nμ X + xμBa + yμLa + zμGa + δμO

where E(Ba8−xLa8−yGa24−zO56−δXn) (X = Nd, Tb, or Eu) is the total energy of the supercell containing the substituted defect XLa; E(La8Ba8Ga24O56) represents the total energy in the same supercell without defects; n, x, y, z, and δ are the numbers of atoms of the X, Ba, La, Ga, and O constituents, respectively; and μX, μBa, μLa, μGa, and μO are the chemical potentials of bulk X, Ba, La, Ga, and O2 molecules, respectively. The calculated results are listed in Table 2, in which the three types of oxygen Table 2. Formation Energies of the Single Defects defect

formation energy (eV)

defect

formation energy (eV)

NdLa VO‑I VO‑II VO‑III VGa‑I

0.364 4.195 4.279 4.076 9.604

VGa‑II VLa VBa TbLa EuLa

6.963 13.359 7.539 0.347 1.164

defects have similar formation energies, whereas the formation energy of VGa‑II is ∼2.641 eV lower than that of VGa‑I; such a large energy difference demonstrates that the Ga-II site is more prone to generate defects the Ga-I site. Detailed discussions about single defects in BLGO are presented in the following three sections. 3.2. Single Defect in BLGO. To determine the role of VO in BLGO, the electronic properties of VO‑I, VO‑II, and VO‑III were calculated. The band structures of VO‑I, VO‑II, and VO‑III are plotted in Figure 3, in which the defect levels are situated below and very close to the Fermi level. According to our calculations, there is no spin polarization for VO‑I, VO‑II, and VO‑III, so their band structures for only the spin up states are exhibited. Clearly, the defect levels, occupied by electrons, are ∼2.58, ∼1.88, and ∼2.50 eV below the conduction-band minimum (CBM) for VO‑I, VO‑II, and VO‑III, respectively. However, these electrons in the defect levels are difficult to excite directly into the conduction band because of the large energy gap. Therefore, no room is left in the VO levels for capturing

Figure 2. Computed band structure and PDOS of the perfect BLGO structure. The Fermi level is set to 0 eV and indicated by a red dashed line. 2857

DOI: 10.1021/acs.inorgchem.5b02714 Inorg. Chem. 2016, 55, 2855−2863

Article

Inorganic Chemistry

maximum (VBM), which is similar to those of VCa in the CaAlO4 compound.17 Removing a La atom would leave three unpaired electrons, which prefer to associate with O and Ga atoms. Incorporating the former discussions, the presence of VLa is beneficial for transporting electrons to O atoms and higher empty levels (such as VGa levels). We also calculated the band structure and PDOS of VBa in BLGO (see Figure S7). Only weak impurity levels lie above the Fermi level, demonstrating that the formation of VBa has a minimal impact on the luminescence. According to these results, combined with the VO, VGa, VLa, and VBa analysis, even though electrons can indeed travel between the defect levels, their effect on the luminescence of BLGO is faint. This is consistent with the experimental finding that the BLGO host is only a very weak or even not a lighting material. According to the experimental investigation, the luminescence of BLGO:Nd is crucially dependent on the transition between the 4f and 5d levels of the Nd dopant. A corresponding discussion of the luminescence mechanism of BLGO:Nd is provided in the next section. 3.3. Nd Doping in BLGO. Lanthanum compounds are suitable hosts, because their 4f shells are empty and, therefore, no 4f electronic transitions are possible unless other lanthanide ions are incorporated into the materials. As the luminescent center, the transition of electrons between the 4f and 5d orbitals of Nd ions must play a vital role in luminescence. It is well-known that correlation effects of localized 4f electrons always exist in lanthanide ions, and one of the most popular approaches to solving this problem is DFT + U, where U is a semi-empirical energy and different systems have different U values based on the available experimental determinations. For rare-earth ions, for the calculation of magnetic properties, the U value can be as high as 8 eV,33,34 whereas for the investigation of luminescent families, it can be as low as about 1 eV.17 In the present work, different U values (0, 0.5, 1, 1.5, and 2 eV) were tested for the 4f orbitals of Nd ions. From the calculated PDOS patterns (see Figure S8), one can see that both of the occupied Nd 4f levels and Fermi levels move toward the VB after the addition of U to the Nd 4f orbitals, whereas the empty Nd 4f levels almost do not migrate compared to the CB. This would result in an enlargement of the band gap between the occupied and empty Nd 4f levels. According to a previous report on luminescence calculations,17 we found here that, for the calculation of the luminescence mechanism of BLGO:Nd, the value of U = 1 eV on the Nd 4f orbitals have already given a good agreement with the experimental results. Therefore, we selected U = 1 eV for the 4f orbitals of Nd ions in the present work. The electronic structure of BLGO:Nd was computed, and the band structure and DOS of NdLa are displayed in Figure S9. It is obvious that some impurity levels were introduced into the band gap after the substitution of the Nd complex compared to the perfect BLGO. The PDOS, plotted in Figure 4, reveals that the impurity levels originate from Nd 4f orbitals, and the 4f orbitals are partially occupied because the Fermi level lies between these impurity levels. In theory, the Nd3+ ion has three occupied majority-spin f electrons; from Figure 4, one can see that the splitting of the ground state 4I9/2 of Nd3+ into ligand field levels is very small. The 4f orbitals of the Nd ions were split using the QTL and TETRA programs of WIEN2K software, and the splitting of the 4f levels of Nd ions are presented in Figure 5. In Cartesian coordinates, there are seven nonequivalent f orbitals, namely, f x(x2 − 3y2), f y(3x2 − y2), fz(x2 − y2), f xyz, f x(x2 − 3y2), f xz2, fyz2, and fz3, which is

Figure 3. Computed band structures of BLGO with an oxygen vacancy. The impurity levels of the oxygen vacancies are denoted by the lines of pink, black, and olive green spheres for VO‑I, VO‑II, and VO‑III, respectively. The Fermi level is set to 0 eV and indicated by a red dashed line.

electrons. Consequently, VO is not an effective defect level in BLGO. Nevertheless, the electrons in the VO levels can be trapped by other hole levels that are located above the Fermi level, playing an auxiliary role in luminescence. This is consistent with the experimental analysis that VO indeed can offer electrons to other trapped holes. It was previously reported that Ga3+ ions can capture electrons that are knocked from the O2− ion by γ-irradiation;32 therefore, the Ga3+ ion should be crucial to the luminescence of BLGO. We constructed VGa‑I and VGa‑II in BLGO, and the computed results are shown in Figures S1−S4. For VGa‑I (see Figures S1 and S2), an impurity level is located above the Fermi level; thus, the defect level is unoccupied and can accept electrons. From the above VO analysis, we know that the VO defect levels are situated just below the Fermi level and can release electrons to a slightly higher hole level, which is consistent with the experimental conclusion: O2− + γ → O− + e− and Ga3+ + e− → Ga2+.32 However, the electrons trapped in these empty levels will be hardly released to conduction band, because of the large energy gap (∼2.77 eV) between the impurity level and the CBM. Therefore, VGa‑I in BLGO cannot be very good for promoting luminescence. For VGa‑II, there are two types of impurity levels presented in Figures S3 and S4, one located just below the Fermi level, and the other spin-down part situated above the Fermi level. Thus, the electronic transition can be easily realized between the two impurity levels. Certainly, these empty impurity levels can also accommodate electrons excited from VO and leaving hole carriers behind. However, the large energy gap (∼2.34 eV) also leads to the electrons in the empty states being difficult to release to the CBM. Nevertheless, the energy gap of VGa‑II is ∼0.43 eV lower than that of VGa‑I, implying that the electrons in the defect level of VGa‑II is more easily released than that in the VGa‑I level, which is in agreement with the fact that the formation energy of VGa‑II is lower than that of VGa‑I. We removed one of the La atoms from the BLGO model, and the resulting band structure and PDOS are shown in Figures S5 and S6. As can be seen in Figure S5, there are spindown impurity levels located just above the valence-band 2858

DOI: 10.1021/acs.inorgchem.5b02714 Inorg. Chem. 2016, 55, 2855−2863

Article

Inorganic Chemistry

Figure 4. Calculated PDOS of the doped compound BLGO:Nd. The Fermi level is set to 0 eV and indicated by a red dashed line.

Figure 5. Split Nd 4f orbital in BLGO:Nd. The Fermi level is set to 0 eV and indicated by a red dashed line.

shown in the inset of Figure 5. It is obvious that some of the 4f levels are degenerate and that the three 4f electrons are distributed on the Nd 4f orbitals except for 4fx(x2 − 3y2). The corresponding luminescence mechanism is discussed in sections 3.6 and 3.7. 3.4. Complex Defects in BLGO:Nd. If a luminescent Nd center already exists, is it a benefit to luminescence? To answer this question, we computed the formation energies and band structures of the complex defects NdLa + VO, NdLa + VGa, and NdLa + VLa. First, for NdLa + VO, we calculated the formation energies when the distance between NdLa and VO was lowest and greatest. The result was that VO prefers being far from NdLa, as the formation energies of NdLa + VO‑II are 4.312 and 4.194 eV for neighboring and distant positions with respect to NdLa, respectively. However, to check the effects of VO on NdLa, we still removed one oxygen atom that was neighboring to Nd. The computed band structures and DOS are displayed in Figure 6. The defect levels of VO‑I and VO‑III in BLGO:Nd are ∼0.5 eV lower than that of VO in BLGO, whereas VO‑II seems to exhibit no change under the two conditions, so its influence is negligible. Therefore, VO‑I and VO‑III in NdLa + VO could be deeper than the luminescent center and be beneficial for its

Figure 6. Calculated band structures and DOS of the complex defect NdLa + VO with the shortest distance between them. In the band structure, the impurity levels of Nd 4f are plotted by lines of dark yellow spheres. The line of pink spheres in panel a is for VO‑I, the line of black spheres in panel b is for VO‑II, and the line of olive green spheres in panel c is for VO‑II. The Fermi level is set to 0 eV and indicated by a red dashed line.

longevity. In addition, there are only two types of effective oxygen bridge bonds, namely, GaI−OI−GaII and GaII−OIII− 2859

DOI: 10.1021/acs.inorgchem.5b02714 Inorg. Chem. 2016, 55, 2855−2863

Article

Inorganic Chemistry

introduced. From Figures S14 and S15, one can see that these impurity levels stem from Tb 4f states. The PDOS patterns also showed that the occupied 4f states of the dopant Tb3+ lie deep in the VB (from about −3 to −5 eV), so the 4f → 5d electronic transition is difficult because of the energy gap between the VB and the impurity levels. In addition, there is only an occupied Tb 4f level lying just below the Fermi level. Therefore, the luminescence of Tb-doped BLGO is weak. It is worthwhile to note that the empty spin-down Tb 4f impurity levels are situated very close to the CBM and could interact with the conduction band by radiation. Therefore, we suppose that the luminescence lifetime of co-doped Tb ions in BLGO:Nd should be longer than that of BLGO:Nd. Accordingly, we performed calculations on Nd,Tb-co-doped BLGO to test our supposition, and the computed results are shown in Figure 8. It is obvious that the empty defect levels are

GaII, between two types of tetrahedral Ga atoms (GaIO4 and GaIIO4) in the BLGO crystal structure. Thus, VO‑I and VO‑III should play more important roles in luminescence than VO‑II. Second, for NdLa + VGa, according to the computed formation energies in Table 2, Ga-II forms a vacancy much more easily than Ga-I; thus, we considered only VGa‑II lying nearest to NdLa. The calculated electronic structure is plotted in Figures S10 and S11, whereas the VGa‑II defect level is located just above the Fermi level and the impurity levels of both Nd 4f and VGa‑II move closer to the VB compared to the single defects of NdLa (see Figure S5) and VGa‑II (see Figure S3), such that the occupied Nd 4f levels move toward the VBM. However, because the empty Nd 4f levels also move toward the low energy range, the band gap between the highest empty Nd 4f level and the CBM is too large to allow an electronic transition. Finally, for NdLa + VLa, the electronic structure is very similar to that of the complex NdLa + VGa‑II defect. These results are shown in Figures S12 and S13 and also make little contribution to the luminescence in BLGO:Nd. 3.5. Other Lanthanide Ions Doped in BLGO. Experimental results have shown that other rare-earth ions, such as Tb, Ho, Er, and Tm, can also act as luminescent centers in BLGO.13−15 Experimentally, the luminescence of Tb 3+activated BLGO is very weak at room temperature.13 Lammers and Blasse observed that only the 4f8 lines and a weak broad band appeared in the excitation spectrum. They ascribed this low intensity of the excitation band to the Tb3+ 4f → 5d transition. In the present work, the band structures and DOS of BLGO:Tb were calculated with Tb occupying one La site to determine the effects of Tb 4f electrons on the luminescence properties. The formation energy of TbLa is slightly lower than that of NdLa (see Table 2), demonstrating that a Tb ion can be more easily doped in the BLGO lattice compared to a Nd ion. In the case of the Hubbard U correction for the 4f orbitals of Tb ion, values of U = 0.5, 1, 1.5, and 2 eV were simulated (Figure S14). Because the situation was similar to that of Nd ions, through a careful comparison, a value of 1 eV for U was added to the 4f orbitals of the Tb ion. The electronic structure is shown in Figure 7, in which spin-down impurity levels are

Figure 8. Computed band structures, DOS, and PDOS of BLGO:Nd,Tb. In the band structure, the lines of dark yellow and magenta spheres represent the NdLa and TbLa impurity defects, respectively.

distributed over a wide energy range above the Fermi level and are located from shallow to deep positions in the band gap. Hence, the doped Tb ions not only can provide extra electrons but also introduce a deeper electron trap center. It is noticeable that the occupied 4f orbitals of the Nd ions located below the Fermi level in BLGO:Nd,Tb (see Figure 8) are lower than the same 4f orbitals of Nd in BLGO:Nd (see Figure 4); thus, the electrons can move up and down between 4f levels of Nd and Tb for a very long time before having a chance to move back to the conduction band. Therefore, in theory, the co-doping of Tb and Nd in BLGO indeed can prolong the decay time. The above investigation indicated that co-doped lanthanide ions in BLGO:Nd should be good candidates for applications in light emissions, although singly doped BLGO:X (where X represents rare-earth ions other than Nd) might not be good phosphors for luminescence. This issue deserves further investigation both experimentally and theoretically for lanthanide-ion-doped BLGO phosphors. To the best of our knowledge, a Eu-doped BLGO system has not yet been reported, even though the Eu ion is a good luminescent center in Y2WO6,35 YVO4,36 Ba2LiSi7AlN12,37 and Ba3P5N10X (X = Cl, I),38 an so on. In the present work, we investigated Eu substitution for one La atom in BLGO to

Figure 7. Computed band structures and DOS of Tb-doped compound BLGO:Tb with Tb occupying the La site. The impurity levels of TbLa are represented by the lines of magenta spheres. The Fermi level is set to 0 eV and indicated by a red dashed line. 2860

DOI: 10.1021/acs.inorgchem.5b02714 Inorg. Chem. 2016, 55, 2855−2863

Article

Inorganic Chemistry

Figure 9. Calculated optical properties of BLGO and rare-earth-substituted BLGO. (a) Real part of the dielectric function, (b) imaginary part of the dielectric function, (c) absorption coefficients as a function of wavelength.

compounds can be ascribed to the low symmetry of Nd3+ and Tb3+ coordination in BLGO. Therefore, the appearance of electric dipole transitions can be attributed to some 5d state (antiparity state) mixed with 4f states, so that the parityforbidden transition in the 4f state is partially allowed. The absorption coefficients as a function of wavelength (λ) are displayed in Figure 9c. It is well-known that the ionic radii of Nd3+ and Tb3+ are smaller than that of La3+, so their doping introduces additional structural disorder. It can be seen that rare-earth doping in BLGO results in the displacement of the absorption edge toward longer wavelengths, which is consistent with previous experimental data.1 3.7. Luminescence Mechanism of BLGO:Nd and BLGO:Nd,Tb. On the basis of the preceding analysis and discussion, the luminescence mechanism in BLGO:Nd and BLGO:Nd,Tb can be elaborated as follows. Note that the single defects are not considered because the single defects in BLGO play little role in light emission. The mechanism can be divided into two groups with respect to BLGO:Nd and Tb-co-doped BLGO:Nd. First, as shown in Figure 10, the Nd ion acts as a luminescent center, and its empty 5d levels are located in the CBM of the host. The empty Nd 4f impurity levels lie close to the Nd 5d levels, so mixing of the 4f and small amounts of the 5d orbitals can occur. As a result, the electric dipole transition can proceed in the 4f state because the parity-forbidden transition of the 4f configuration is partially allowed. Therefore,

explore whether the Eu ion is also a good dopant in BLGO. The formation energy of EuLa is much higher than those of Nd and Tb, so BLGO:Eu is more difficult to prepare than BLGO:Nd and BLGO:Tb. The band structure and DOS of BLGO:Eu are plotted in Figures S16−S18, in which the spin-up impurity band is situated just above the Fermi level and very close to the VBM. In addition, the band gap is too large to allow for an electronic transition. However, the Hubbard correction simply pushes the occupied 4f levels downward, making the electronic transition more difficult to achieve. Therefore, unlike for Nd- and Tb-doped BLGO, BLGO:Eu does not fulfill the requirements for light emission. Therefore, in theory, Eu is not an effective luminescent center in BLGO. 3.6. Optical Properties. The nature of a beam of light acting on matter is that it can generate polarization effects on the valence electrons of the material. Therefore, the macroscopic optical response of a material can normally be represented by a complex dielectric function (ε), which consists of a real part (ε1) and an imaginary part (ε2). Resonance appears between the photon frequency and the electronic transition of the material when the real part of the dielectric function decreases sharply, which corresponds to an absorption peak in the imaginary part. The OPTIC module was employed to calculate the optical properties of the studied materials with consideration of spin−orbit coupling effects. Figure 9 shows plots of the dielectric function and absorption coefficients. Two important features are revealed in this figure. On one hand, the decrease of the real part corresponds to the absorption peaks of the imaginary part very well through the Kramers−Kronig relation.39,40 On the other hand, the weak absorption visible in the 1−2.5 eV energy range corresponds to the 4f orbitals of the lanthanide ions doped into BLGO (see Figures 4, 7, and 8). In principle, the 4f → 4f transitions do not fulfill the electric dipole moment selection rule. However, from Figure 4 and Figure S14b, one can see that the 4f levels actually mix with a very small amount of 5d orbitals. A previous study reported that low-symmetry lanthanide coordination (D4d) can lead to the mixing of 4f and 5d wave functions.41 It can thus be deduced that the mixing of 4f and 5d levels in our studied

Figure 10. Schematic of the luminescence machanism of BLGO:Nd and BLGO:Nd,Tb. 2861

DOI: 10.1021/acs.inorgchem.5b02714 Inorg. Chem. 2016, 55, 2855−2863

Inorganic Chemistry

■

under applied irradiation, the Nd 4f3 electrons can be excited into the higher empty 4f−5d levels, and then these excited 4f− 5d electrons can jump into the 5d levels on the CBM under continuous irradiation. When these excited electrons in the 5d levels move back to the 4f bands, luminescence will occur. In BLGO:Nd, there might also coexist a number of intrinsic defects, such as VO, VGa, and VLa, that are usually generated in the process of preparation. However, such complex defects (NdLa + VO, NdLa + VGa, and NdLa + VLa) have little effect on luminescence. The dominant role in luminescence is still played by the transitions of 4f electrons into empty 4f−5d levels, then to empty 5d levels, and then returning back to 4f levels. Second, BLGO:Nd,Tb co-doping is expected to extend the luminescence lifetime. In BLGO:Nd,Tb, Nd serves as the luminescent center, and Tb acts as the electron trap center, where electrons can stay for a while in shallow capture levels and then be released back into the conduction band, thereby prolonging the decay time. These co-doped impurity-induced states were accurately predicted to lie in the host band gap of the phosphors, which is important in the design of phosphors with co-doping of other lanthanide ions in BLGO:Nd in experiments.

ACKNOWLEDGMENTS This work was supported by the National Natural Science Foundation of China under Grants 21571174, 51372244, and 21221061; the Major Program of National Natural Science Foundation of China under Grant 21590794; and the Natural Scientific Foundation of Jilin Province (20130101016JC).

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.5b02714. Band structures and DOS of VGa‑I, VGa‑II, VLa, VBa, NdLa, NdLa + VGa‑II, NdLa + VLa, and EuLa; PDOS of VGa‑I, VGa‑II, VLa, NdLa + VGa‑II, NdLa + VLa, EuLa, Tb 4f, and Eu 4f with different values of U on the 4f orbitals of BLGO:Nd and BLGO:Tb (PDF)

■

REFERENCES

(1) Ryba-Romanowski, W.; Gołąb, S.; Dominiak-Dzik, G.; Berkowski, M. Mater. Sci. Eng., B 1992, 15, 217−221. (2) Hanuza, J.; Andruszkiewicz, M. Spectrochim. Acta, Part A 1995, 51, 869−881. (3) Kaczkan, M.; Pracka, I.; Malinowski, M. Opt. Mater. 2004, 25, 345−352. (4) Karbowiak, M.; Gnutek, P.; Rudowicz, C.; Ryba-Romanowski, W. Chem. Phys. 2011, 387, 69−78. (5) Agnesi, A.; Pirzio, F.; Tartara, L.; Ugolotti, E.; Zhang, H.; Wang, J.; Yu, H.; Petrov, V. Laser Phys. Lett. 2014, 11, 035802−035805. (6) Ismatov, A. A.; Kolesova, V. A.; Piryutko, N. M. Izu. Akad. Nauk SSSR Neorn. Mater. 1970, 6, 1361. (7) Piekarczyk, W.; Berkowski, M.; Jasiolek, D. J. Cryst. Growth 1985, 71, 395−398. (8) Berkowski, M.; Borowiec, M. T.; Pataj, K.; Piekarczyk, W.; Wardzynski, W. Physica B+C 1984, 123, 215−219. (9) Soluch, W.; Ksiezopolski, R.; Piekarczyk, W.; Berkowski, M.; Goodberlet, M. A.; Vetelino, J. F. J. Appl. Phys. 1985, 58, 2285−2287. (10) Ryba-Romanowski, W.; Jeżowska-Trzebiatowska, B.; Piekarczyk, W.; Berkowski, M. J. Phys. Chem. Solids 1988, 49, 199−203. (11) Ryba-Romanowski, W.; Gutowska, M. U.; Piekarczyk, W.; Berkowski, M.; Mazurak, Z.; Jeżowska-Trzebiatowska, B. J. Lumin. 1987, 36, 369−372. (12) Ryba-Romanowski, W.; Sokòlska, I.; Gołąb, S.; Berkowski, M.; Kück, S.; Meyn, J. P. Phys. stat. sol. 1994, 142, K51−K53. (13) Lammers, M. J. J.; Blasse, G. Mater. Chem. Phys. 1986, 15, 537− 544. (14) Ryba-Romanowski, W. J. Phys. Chem. Solids 1990, 51, 229−237. (15) Gao, S. F.; Zhu, Z. J.; Wang, Y.; You, Z. Y.; Li, J. F.; Xu, J. L.; Tu, C. Y. Laser Phys. 2014, 24, 085004−085010. (16) Muñoz-García, A. B.; Seijo, L. J. Phys. Chem. A 2011, 115, 815− 823. (17) Qu, B.; Zhang, B.; Wang, L.; Zhou, R.; Zeng, X. C. Chem. Mater. 2015, 27, 2195−2202. (18) Wen, J.; Ning, L.; Duan, C. K.; Zhan, S.; Huang, Y.; Zhang, J.; Yin, M. J. Phys. Chem. A 2015, 119, 8031−8039. (19) Ramanantoanina, H.; Cimpoesu, F.; Gottel, C.; Sahnoun, M.; Herden, B.; Suta, M.; Wickleder, C.; Urland, W.; Daul, C. Inorg. Chem. 2015, 54, 8319−8326. (20) Kresse, G.; Hafner, J. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 47, 558−561. (21) Kresse, G.; Furthmüller, J. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 11169−11186. (22) Blochl, P. E. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50, 17953−17979. (23) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865−3868. (24) Schwarz, K.; Blaha, P. Comput. Mater. Sci. 2003, 28, 259−273. (25) Blaha, P.; Schwarz, K. ; Madsen, G. K. H.; Kvasnicka, D.; Luitz, J. Computer Code WIEN2K, An Augmented Plane Waves + Local Orbitals Program for Calculating Crystal Properties, revised ed. 2001, Vienna University of Technology: Vienna, Austria, 2003. (26) Andersen, O. K. Phys. Rev. B 1975, 12, 3060−3083. (27) Mattheiss, L. F.; Hamann, D. R. Phys. Rev. B: Condens. Matter Mater. Phys. 1986, 33, 823−840. (28) Anisimov, V. I.; Aryasetiawan, F.; Lichtenstein, A. I. J. Phys.: Condens. Matter 1997, 9, 767−808. (29) Dudarev, S. L.; Botton, G. A.; Savrasov, S. Y.; Humphreys, C. J.; Sutton, A. P. Phys. Rev. B: Condens. Matter Mater. Phys. 1998, 57, 1505−1509.

4. CONCLUSIONS In the present work, the luminescence mechanism of BLGO:Nd was studied using first-principles calculations. The BLGO host was first investigated without and with single defects of VO, VGa, VLa, and VBa. Then, the luminescence mechanism of BLGO:Nd phosphor was comprehensively studied. It was found that the dominant role in luminescence is the series of electronic transitions 4f → 4f−5d → 5d → 4f . Complex defects in BLGO:Nd play only weak assistant roles in light emission. Certainly, other lanthanide ions can also be employed as luminescent centers in BLGO; however, their luminescence is weak (e.g., Tb) or even lacking (e.g., Eu). In addition, although the luminescence of singly doped Tb ions is not strong, when co-doped in BLGO:Nd, Tb serves as an electron trap center to prolong the luminescence lifetime. The present microscopic mechanistic study can provide valuable guidance for future experiments in lanthanide-ion-co-doped BLGO:Nd in the search for the best luminescent materials.

■

Article

AUTHOR INFORMATION

Corresponding Author

*Tel.: +86-431-85262415. Fax: +86-431-85698041. E-mail: [email protected]. Notes

The authors declare no competing financial interest. 2862

DOI: 10.1021/acs.inorgchem.5b02714 Inorg. Chem. 2016, 55, 2855−2863

Article

Inorganic Chemistry (30) Blöchl, P. E. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50, 17953−17979. (31) Hanuza, J.; Hermanowicz, K.; Maczka, M.; Ryba-Romanowski, W.; Lis, T.; Berkowski, M. J. Raman Spectrosc. 1995, 26, 255−263. (32) Jabłoński, R.; Kaczmarek, S.; Berkowski, M. Spectrochim. Acta, Part A 1998, 54, 2057−2063. (33) Ning, L.; Zhang, Y.; Cui, Z. J. Phys.: Condens. Matter 2009, 21, 455601−455607. (34) Ke, J.; Xiao, J. W.; Zhu, W.; Liu, H.; Si, R.; Zhang, Y. W.; Yan, C. H. J. Am. Chem. Soc. 2013, 135, 15191−15200. (35) Kaczmarek, A. M.; Van Hecke, K.; Van Deun, R. Inorg. Chem. 2014, 53, 9498−9508. (36) Duée, N.; Ambard, C.; Pereira, F.; Portehault, D.; Viana, B.; Vallé, K.; Autissier, D.; Sanchez, C. Chem. Mater. 2015, 27, 5198− 5205. (37) Takeda, T.; Hirosaki, N.; Funahshi, S.; Xie, R. J. Chem. Mater. 2015, 27, 5892−5898. (38) Marchuk, A.; Wendl, S.; Imamovic, N.; Tambornino, F.; Wiechert, D.; Schmidt, P. J.; Schnick, W. Chem. Mater. 2015, 27, 6432−6441. (39) de Laer Kronig, R. J. Opt. Soc. Am. 1926, 12, 547−557. (40) Kramer, H. A. Atti del Congresso Internazionale dei Fisici 1927, 2, 545−557. (41) Ramanantoanina, H.; Urland, W.; Cimpoesu, F.; Daul, C. Phys. Chem. Chem. Phys. 2014, 16, 12282−12290.

2863

DOI: 10.1021/acs.inorgchem.5b02714 Inorg. Chem. 2016, 55, 2855−2863