Harry
Ramanantoanina
*a,
Mohammed
Sahnoun
b,
Andrea
Barbiero
a,
Marilena
Ferbinteanu
c and
Fanica
Cimpoesu
*d
aDepartment of Chemistry of the University of Fribourg (Switzerland), Chemin du Musée 9, 1700 Fribourg, Switzerland. E-mail: harry.ra@hotmail.com; Fax: +41 26 300 9738; Tel: +41 26 300 8700
bLaboratoire de Physique de la Matière et Modélisation Mathématique, LPQ3M, Université de Mascara, Algeria
cFaculty of Chemistry, Inorganic Chemistry Department, University of Bucharest, Dumbrava Rosie 23, Bucharest 0206462, Romania
dInstitute of Physical Chemistry, Splaiul Independentei 202, Bucharest 060021, Romania. E-mail: cfanica@yahoo.com
First published on 15th June 2015
Ligand field density functional theory (LFDFT) is a methodology consisting of non-standard handling of DFT calculations and post-computation analysis, emulating the ligand field parameters in a non-empirical way. Recently, the procedure was extended for two-open-shell systems, with relevance for inter-shell transitions in lanthanides, of utmost importance in understanding the optical and magnetic properties of rare-earth materials. Here, we expand the model to the calculation of intensities of f → d transitions, enabling the simulation of spectral profiles. We focus on Eu2+-based systems: this lanthanide ion undergoes many dipole-allowed transitions from the initial 4f7(8S7/2) state to the final 4f65d1 ones, considering the free ion and doped materials. The relativistic calculations showed a good agreement with experimental data for a gaseous Eu2+ ion, producing reliable Slater–Condon and spin–orbit coupling parameters. The Eu2+ ion-doped fluorite-type lattices, CaF2:Eu2+ and SrCl2:Eu2+, in sites with octahedral symmetry, are studied in detail. The related Slater–Condon and spin–orbit coupling parameters from the doped materials are compared to those for the free ion, revealing small changes for the 4f shell side and relatively important shifts for those associated with the 5d shell. The ligand field scheme, in Wybourne parameterization, shows a good agreement with the phenomenological interpretation of the experiment. The non-empirical computed parameters are used to calculate the energy and intensity of the 4f7–4f65d1 transitions, rendering a realistic convoluted spectrum.
Born more than eighty years ago, from the work of H. Bethe1 and J. H. van Vleck2 it still keeps the position of the most transparent way to describe the optical and magnetic properties of metal ion-based systems (lattices or molecular complexes). As long as quantum chemical methods can compute reliable energy level schemes, the subsequent ligand field analysis of the raw results is the way to illuminate in depth the underlying mechanism.3–5Stricto sensu, the ligand field refers to effective one-electron parameters accounting for the effect of the environment on a metal ion, but the complete frame includes the inter-electron effects, describing the electronic correlation in the active space of dn or fn configurations, and also the spin–orbit coupling, namely the relativistic effects. Besides the standard theory, one must note the paradigm shift due to C. E. Schäffer and C. K. Jørgensen, who revisited the ligand field theory to ensure more chemical insight within their Angular Overlap Model (AOM), initially devoted to the d-type transition metal systems.6 W. Urland pioneered this model for the f-type ligand field, in lanthanide compounds, with convincing applications in spectroscopy and magnetism.7
About two decades ago, given the important growth of computational techniques, the demand for a predictive theory compatible with the classical formalism of the ligand field theory emerged. In particular, this is not a trivial task in the frame of density functional theory (DFT), limited to non-degenerate ground states, while ligand field concerns the full multiplets originating from dn or fn configurations. In the consistent solving of this problem, C. Daul erat primus. He and co-workers (noting the contribution of M. Atanasov) designed a pioneering approach by non-routine handling of DFT numeric experiments, to extract ligand field parameters, in a post-computational algorithm named LFDFT.8–10 The procedure treats the near degeneracy correlation explicitly within the model space of the Kohn–Sham orbitals possessing dominant d and f characters.
In LFDFT, the basic start is a DFT calculation performed in average of configuration (AOC) conditions. Namely, for a given dn (or fn) configuration of the metal ion in the complex, the occupation of five (or, respectively, seven) Kohn–Sham orbitals carrying main d (or f) character is fixed to the general fractional n/5 (or n/7, respectively) numbers. This corresponds to the barycentre conceived in formal ligand field theories. Subsequently, with the converged AOC orbitals, a series of numeric experiments are done, producing the configurations related to the distribution of n electrons in the five (or seven) orbitals identified as the ligand field sequence (this time with corresponding integer populations). These determinant configurations are not real states, but useful computational experiments, able to render ligand field parameters. The situation is somewhat similar to broken symmetry treatments,11–14 where the spin-polarized configurations cannot be claimed as physical states, but artificial constructions relevant for the emulation of the exchange coupling parameters.15 Then, the LFDFT run of different configurations based on AOC orbitals yields ligand field parameters, altogether with inter-electron Coulomb and exchange effective integrals. Thus, the Slater determinants are used as the basis in the computational model. In the advanced background of the theory, a canonical number of configurations needed to reproduce the desired parameters can be defined as a function of the symmetry of the problem (Slater determinant wavefunctions of spin–orbitals weighted by symmetry coefficients).10 In practice, the full set of configurations can be generated, performing the least square fit relating the computed energy expectation values against the ligand field model formulas. The obtained parameters are further used in setting configuration interaction (CI) matrices, in the spirit of the ligand field formalism, sustained in a non-empirical manner. Therefore C. Daul et al. have realized the parameter-free ligand field theory, which became a valuable tool for any consideration of multiplet states in DFT.
We recognize herein the impact of the LFDFT in solving various electronic structure problems. This computational gadget has revolutionized many fields of chemical science, being applied in theoretical investigations16–20 as well as in experimental works.21,22
A priori, LFDFT has determined the multiplet energy levels within an accuracy of a few hundred wavenumbers.23 The model has given satisfactory results for the molecular properties arising from a single-open-shell system, such as zero-field splitting (ZFS),24,25 magnetic exchange coupling,26–29 Zeeman interaction,30 hyper-fine splitting,30 shielding constants,31,32 d–d and f–f transitions.10,17,33,34
Recently, the LFDFT algorithm has been updated to handle the electronic structure of two-open-shell systems, as it is important in the understanding of the optical manifestation of lanthanide phosphors.35,36 Lanthanide compounds are agents in light-emitting diode (LED) technology, used in domestic lighting.37 In the case of a two-open-shell inter-configuration of f and d electrons, the size of the ligand field CI matrices is collected in Table 1, calculated with the following combinatorial formulas:
![]() | (1) |
![]() | (2) |
n | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
a ∑ represents the cumulative sum of N(4fn) and N(4fn−15d1). | ||||||||||||||
N(4fn) | 14 | 91 | 364 | 1001 | 2002 | 3003 | 3432 | 3003 | 2002 | 1001 | 364 | 91 | 14 | 1 |
N(4fn−15d1) | 10 | 140 | 910 | 3640 | 10![]() |
20![]() |
30![]() |
34![]() |
30![]() |
20![]() |
10![]() |
3640 | 910 | 140 |
∑a | 24 | 231 | 1274 | 4641 | 12![]() |
23![]() |
33![]() |
37![]() |
32![]() |
21![]() |
10![]() |
3731 | 924 | 141 |
In this paper, we present new development and applications of the LFDFT algorithm, previously validated for the two-open-shell 4f15d1 electronic structure of Pr3+.35,38–40 Special attention will be paid to Eu2+ systems, i.e. for n = 7 (Table 1), taking as examples divalent europium doped in the fluorite-type lattices CaF2 and SrCl2, comparing the first principles results with the available experimental data.41,42
H = H0 + HEE + HSO + HLF, | (3) |
![]() | (4) |
![]() | (5) |
The matrix elements of HEE are constructed from the two-electron integrals:
![]() | (6) |
ψ(r) = Rnl(r)Ylm(θ,ϕ), | (7) |
Within mathematical operations, eqn (6) is reducible into the product of two integrals of angular and radial components. Once the angular part is explicitly resolved, the whole variety of the eqn (6) integrals can be represented by a few radial Slater–Condon parameters, Fk (eqn (8) and (9)) and Gk (eqn (10)), with intra- or inter-shell nature. In the two-open-shell problem of 4f and 5d electrons, one obtains:
![]() | (8) |
![]() | (9) |
![]() | (10) |
〈nlsmlms|HSO|nls′ml′ms′〉 = ζnl〈lsmlms|![]() | (11) |
![]() | (12) |
![]() | (13) |
![]() | (14) |
Besides the Hamiltonian setting, other specific construction regards the matrix element of the dipole moment operator, important to the computation of the intensity of transitions:
![]() | (15) |
In summary, several series of parameters have to be determined non-empirically in order to perform LFDFT calculations of two-open-shell f and d electrons:
(1) Δ(fd), which represents the energy shift of the multiplets of the 4fn−15d1 configuration with respect to those of the 4fn configuration.
(2) Fk(ff), Fk(fd) and Gk(fd), which represent the static electron correlation within the 4fn and 4fn−15d1 configurations.
(3) ζnl, which represents the relativistic spin–orbit interaction in the 4f and 5d shells.
(4) Bkq(f,f), Bkq(d,d) and Bkq(f,d), which describe the interaction due to the presence of the ligands onto the electrons of the metal centre.
The DFT calculations have been carried out by means of the Amsterdam density functional (ADF) program package (ADF2013.01).48–50 We must point out that the ADF is one of the few DFT codes that has the set of keywords facilitating the AOC calculations and Slater determinant emulation, needed by the LFDFT procedure.35,36 The hybrid B3LYP functional51 was used to compute the electronic structure and the related optical properties, in line with previous works.35,36,39 The molecular orbitals were expanded using triple-zeta plus two polarization Slater-type orbital (STO) functions (TZ2P+) for the Eu atom and triple-zeta plus one polarization STO function (TZP) for the Ca, Sr, F and Cl atoms.
The geometrical structures due to the doping of the Eu2+ ion into CaF2 and SrCl2 lattices were approached via periodical calculations by means of the VASP program package.52 The local density approximation (LDA) defined in the Vosko-Wilk-Nusair (VWN)53 and the generalized gradient approximation (GGA) outlined in the Perdew-Burke-Ernzerhof (PBE)54 were used for the exchange–correlation functional. The interaction between valence and core electrons was emulated with the projected augmented wave method.55,56 External as well as semi-core states were included in the valence. A plane-waves basis set with a cut-off energy of 400 eV was used. Super-cells representing a 2 by 2 by 2 expansion of the unit cells of CaF2 and SrCl2 were simulated, which were found to be large enough to lead to negligible interactions between the periodic images of the Eu2+ impurity. 4 k-points were included in each direction of the lattice. The atomic positions were allowed to relax until all forces were smaller than 0.005 eV Å−1.
T. Ziegler et al. clarified early that the occupation-averaged configurations, called transition states, carry in DFT the meaning of statistically-averaged spectral terms.57 We prepare the wavefunctions ψ4f and ψ5d by AOC where six and one electrons are evenly distributed in the 4f and 5d orbitals of Eu2+, respectively (Fig. 1). This will generate the reference totally symmetric density, which will be used to compute the DFT energy associated with the series of Slater determinants. Thus all the Slater determinant energies are successively computed permuting seven electrons in the 4f wavefunction (Fig. 1) for the 4f7 manifold, and permuting six electrons in the 4f wavefunction plus one electron in the 5d for the 4f65d1 manifold. The results obtained at the B3LYP level of theory are graphically represented in Fig. 2 showing the Δ(fd) gap. Note that Δ(fd) can occasionally have a negative value, indicating that the ground electron configuration of the lanthanide ion is 4fn−15d1 instead of 4fn. Such a situation may appear in the case of lanthanide Gd2+ (n = 8, see Table 1) and La2+ (n = 1, see Table 1) ions.
The lowest energies corresponding to the 4f7 manifold (Fig. 2) are associated with the Slater determinants:
|a+b+c+d+e+f+g+| and |a−b−c−d−e−f−g−|, |
|c±d±e±θ+| and |c±d±e±θ−|, |
The DFT Slater determinant energies (Fig. 2) can also provide information about the two-electron Fk(ff), Fk(fd) and Gk(fd) parameters using Slater's rule3 and least mean square fitting.10 However, this procedure might undergo uncertainty caused by the important number of linear equations versus variables. In the case of two-open-shell 4f7 and 4f65d1 of Eu2+, for instance, it returns to solve 33462 linear equations with nine variables, leading to some misrepresentations of the parameters.58 Therefore, we calculate the Fk(ff), Fk(fd) and Gk(fd) parameters from the radial wavefunctions Rnl of the 4f and 5d Kohn–Sham orbitals of the lanthanide ions following eqn (8)–(10), which is the subject of the next section.
three Fk(ff) parameters: F2(ff), F4(ff) and F6(ff); plus |
two Fk(fd) parameters: F2(fd) and F4(fd); plus |
three Gk(fd) parameters: G1(fd), G3(fd) and G5(fd); plus |
two spin–orbit coupling constants: ζ4f and ζ5d.35 |
Slater–Condon parameters and spin–orbit coupling constants | |||
---|---|---|---|
(a) | (b) | (c) | |
F 2(ff) | 500.19 | 475.60 | 388.47 |
F 4(ff) | 64.66 | 61.32 | 49.92 |
F 6(ff) | 6.87 | 6.51 | 5.30 |
F 2(fd) | 245.32 | 245.36 | 244.72 |
F 4(fd) | 17.86 | 18.12 | 18.82 |
G 1(fd) | 338.38 | 369.81 | 431.92 |
G 3(fd) | 29.97 | 31.66 | 35.34 |
G 5(fd) | 4.70 | 4.91 | 5.40 |
ζ 4f | 2133.90 | 1980.90 | 1246.50 |
ζ 5d | 1279.31 | 1245.93 | 987.25 |
Note that the parameters F2(ff), F4(ff) and F6(ff) are acting principally on the single-open-shell 4fn configuration, but they are also present in the diagonal block of the 4fn−15d1 interaction matrix. Experimentally known spectral terms of the 4f7 configuration of Eu2+ concern only the ground state 8S and the two excited states 6P and 6I,76 although there are 119 levels arising from the multi-electron configuration.77 The calculated energy values of these 8S, 6P and 6I spectral terms are given in Table 3, obtained using the parameters in Table 2. They are also compared with the available experimental data taken from the framework of the NIST atomic spectra database.76
Calc. | Exp.a | |||
---|---|---|---|---|
(a) | (b) | (c) | ||
a Taken from ref. 76 where the energy value of 6P3/2 is not known. | ||||
8S7/2 | 0.00 | 0.00 | 0.00 | 0.00 |
6P7/2 | 36379.05 | 34596.14 | 28854.94 | 28200.06 |
6P5/2 | 37400.88 | 35526.15 | 29317.04 | 28628.54 |
6P3/2 | 38339.68 | 36381.98 | 29758.39 | —a |
6I7/2 | 40277.33 | 38282.44 | 31591.89 | 31745.99 |
6I9/2 | 40978.41 | 38917.65 | 31888.23 | 31954.21 |
6I17/2 | 41370.71 | 39274.11 | 32060.93 | 32073.30 |
6I11/2 | 41542.51 | 39430.07 | 32135.09 | 32179.55 |
6I15/2 | 41901.08 | 39756.07 | 32293.32 | 32307.78 |
6I13/2 | 41881.85 | 39739.21 | 32287.83 | 32314.14 |
We determined the deviations between the calculated and the experimental spectral terms (Table 3) using eqn (16):
![]() | (16) |
The calculated deviations ε (eqn (16)) from the experimentally-known spectral terms76 are represented in Fig. 4 for the three theoretical methods under consideration. Here also the Pauli-relativistic calculation leads to the best reproduction of the experimental data, its mean deviation being 6.17% (Fig. 4), which is far smaller than those obtained at the non-relativistic and ZORA-relativistic levels of theory.
![]() | ||
Fig. 4 Representation of the error distribution ε (in %) with respect to the experimental data76 of the calculated multiplet energy levels corresponding to the 4f65d1 configuration of a gaseous Eu2+ ion, at the non-relativistic (in red, (a)), ZORA-relativistic (in green, (b)) and Pauli-relativistic (in blue, (c)) levels of theory. The calculated mean deviations from the experimental data are also given. |
In this section, the impact of the relativistic correction on the spectroscopy of lanthanide ions is clearly justified; an appropriate description of the radial R4f and R5d wavefunctions is a prerequisite, enabling a good reproduction of the experimental data.
For the pristine CaF2 and SrCl2 systems (Fig. 5), the calculated lattice parameters are given in Table 4 in terms of the DFT functional used in the band structure algorithm. It is found that both GGA and LDA calculations yield different lattice equilibrium constants (Table 4), i.e. different local relaxations. In terms of a direct comparison, we consider the GGA calculation most appropriate to simulate the experimental data, although the cells are slightly larger than the experimental ones.
For the CaF2:Eu2+ and SrCl2:Eu2+ systems, we constructed super-cells which double the number of the unit cells of CaF2 and SrCl2 in the a, b and c directions. The Eu2+ ion was placed in the position (0,0,0). In these cases, the super-cells are big enough inasmuch as the interactions between two Eu2+ ions are minimized. We relaxed the positions of the atoms, fixing the lattice parameters to the theoretical values obtained for the pure systems. This mimics the resistance of the whole lattice against defect-induced distortions, under the conditions of a lower doping concentration than the 2 × 2 × 2 super-cells actually worked upon have. The optimized Eu–F and Eu–Cl bond lengths are 2.4732 Å and 3.0774 Å, respectively, which represent an elongation with respect to the Ca–F and Sr–Cl bond lengths obtained for the pure systems: 2.3893 Å and 3.0515 Å. The description of the local structure of doped materials is important in the further evaluation of the ligand field Hamiltonian (eqn (13)), the presence of the impurity in the host materials producing distortions due to differences in the ionic radii or electronic structure. We favoured here the band structure algorithms for geometrical purposes, although we can certainly conceive of a cluster geometry optimization approach, which is already popular in computational chemistry, especially while dealing with excited states geometry.36,39
B40(f,f), B44(f,f), B4−4(f,f), B60(f,f), B64(f,f) and B6−4(f,f), |
B40(d,d), B44(d,d) and B4−4(d,d), |
The inversion center in the Oh point group allows vanishing of the elements of the sub-matrix 〈f|HLF|d〉.47
The ligand field energy schemes of the 4f and the 5d orbitals of Eu2+ in the CaF2:Eu2+ and SrCl2:Eu2+ systems were calculated taking the cubic clusters (EuF8)6− and (EuCl8)6−, respectively, which have the optimized geometries obtained in the previous section. Point charges were placed at the coordinates of the next-neighbouring Ca2+ and Sr2+ ions, which are also shown as ball-and-sticks in the super-cell in Fig. 5. These were used in order to mimic the long-range interaction of the crystal hosts.
The ligand field energies and wavefunctions were obtained from Kohn–Sham orbitals of restricted DFT calculations within the AOC reference, by evenly placing six electrons in the 4f orbitals and one electron in the 5d. We previously presented the analysis of the ligand field interaction with respect to the change of the DFT functional for the two-open-shell 4f and 5d problem in Pr3+.35 It was found that, in the 5d ligand field, the DFT functional does not play an important role, whereas in the 4f, the hybrid B3LYP functional is required in order to obtain realistic ligand field parameters.35 Therefore we have used B3LYP for the computation of the electronic structure of Eu2+.
The 4f orbitals form the basis of t1u, t2u and a2u irreducible representations (irreps) of the Oh point group. The 5d orbitals are in the basis of the eg and the t2g irreps. The values of the ligand field Bkq parameters were determined by linear equation fitting using eqn (13), knowing the following ratios for the octahedral symmetry constraint:
![]() | (17) |
![]() | (18) |
t1u < t2u < a2u, |
t1u < a2u < t2u, |
The change in the orbital ordering may be attributed to the impact of the neighbouring cations, where the symmetry-adapted linear combination of their virtual orbitals may stabilize the a2u irrep. This is not achieved here in the small cluster models of (EuCl8)6−. Nevertheless, a direct comparison between Bkq(f,f) and Bkq(d,d) indicates that the effect of the 4f parameters will be completely superseded by the 5d ones.
![]() | (19) |
![]() | (20) |
A general problem in establishing the parametric conversion is the fact that the AOM matrix is not traceless, the sum of the diagonal elements for a homoleptic [MLn] complex with linearly ligating ligands (isotropic π effects) being n(eσ + 2eπ), instead of zero, like in the standard ligand field model. In the case of the 4f shell, in octahedral symmetry, the situation does not impinge upon the parametric conversion since we have two independent parameters, B40(f,f) and B60(f,f) in the Wybourne scheme (Table 5), versus two AOM parameters eσ(4f) and eπ(4f), uniquely related to the two relative gaps in the ligand field splitting in Oh symmetry.
The mutual conversion is done by the following formulas:
![]() | (21) |
![]() | (22) |
For the 5d shell, the single gap between eg and t2g does not need the two AOM parameters, so that must impose certain conventions, like the eσ(5d)/eπ(5d) = 3 ratio.35 However, we have not advanced in this direction, given the good match of the computed and fitted 5d-type Bkq parameters, which do not demand the call of AOM as a further moderator in the comparative discussion.
The ligand field interaction, besides lifting the degeneracy of the 4f and 5d orbitals, also has a side effect expanding the radial wavefunctions towards the ligands positions. This is commonly known as the nephelauxetic effect, a concept coined by C. K. Jørgensen84 which is the subject of the next section.
![]() | ||
Fig. 6 Representation of R5d of Eu2+ in the free ion (in blue), in (EuF8)6− (in pink) and (EuCl8)6− (in violet), obtained at the Pauli-relativistic level of theory. |
Recalling eqn (9) and (10), we calculated the Fk(fd) and Gk(fd) parameters in the complex, based on the radial shapes shown in Fig. 6. Compared with Fig. 3, one notes that R4f remains almost the same, while R5d were shifted by the nephelauxetic effect (see also ref. 47).
The results are given in Table 6, together with the calculated spin–orbit coupling constant ζ5d, using eqn (12) and the Δ(fd) gap. All the parameters (Table 6) are reduced when compared to the Pauli-relativistic quantities in Table 2. The nephelauxetic ratio β is defined as the fraction made from the inter-electron parameters obtained in the complex and in the free ion, for instance:
![]() | (23) |
CaF2:Eu2+ | SrCl2:Eu2+ | |||||
---|---|---|---|---|---|---|
Calc. | β | Exp.a | Calc. | β | Exp.a | |
a The Fk(fd) and Gk(fd) are taken from ref. 41 and 42. They are converted to the corresponding Fk(fd) and Gk(fd) parameters using the conversion factor in ref. 43. | ||||||
F 2(fd) | 138.42 | 0.57 | 133.33 | 100.56 | 0.41 | 117.43 |
F 4(fd) | 9.88 | 0.53 | 10.25 | 6.79 | 0.36 | 8.54 |
G 1(fd) | 232.08 | 0.54 | 192.29 | 160.56 | 0.37 | 162.06 |
G 3(fd) | 18.22 | 0.52 | 17.30 | 12.31 | 0.35 | 14.41 |
G 5(fd) | 2.74 | 0.51 | 2.72 | 1.84 | 0.34 | 2.26 |
ζ 5d | 505.76 | 0.51 | 760 | 371.14 | 0.38 | 844 |
Δ(fd) | 18![]() |
— | 23![]() |
12![]() |
— | — |
We can obtain the Δ(fd) gap for the CaF2:Eu2+ system from ref. 41 which we compare with our calculated value (Table 6). Unfortunately, the experimental value for the same parameter was not specified for the SrCl2:Eu2+ system.42 The difference between the calculated Δ(fd) gap and that obtained in ref. 41 is directly related to the Fk(ff) parameters (Table 2), which is also present in the diagonal elements of the CI matrix of the 4f65d1 configuration of Eu2+. Since the values of our calculated Fk(ff) parameters are larger than that given in ref. 41 our Δ(fd) is accordingly smaller.
The transitions from the initial 4f7(8S7/2) state to the final 4f65d1 are electric dipole-allowed, with the calculation of the electric dipole transition moments obtained from eqn (15). The oscillator strength for the zero phonon lines between the ground state 8S7/2 of 4f7 and the final states of 4f65d1 are calculated and represented in Fig. 7. The most intense transitions are given with respect to the irreps of the octahedral double group. In the circumstances of a non-degenerate 8S7/2 state of the 4f7 subsystem, the energies of the 4f7–4f65d1 transitions are practically the same as the position of 4f65d1 spectral terms. The intensities were computed by corresponding handling of the dipole moment represented in the ligand field CI basis, depending all on a single reduced matrix element, ultimately irrelevant as an absolute value, if we consider an arbitrary scale of spectral rendering. The zero-field splitting, which transforms the 8S7/2 state of 4f7 to Γ6 + Γ7 + Γ8 in the actual octahedral symmetry, is in the magnitude of tenths of cm−1. The 4f65d1 transitions are characterized by two dominant bands (Fig. 7), in line with the excitation spectrum seen in ref. 41 and 42 for CaF2:Eu2+ and SrCl2:Eu2+. The correspondence between the theoretical results and the excitation spectrum is seen in the ESI‡, where the excitation spectra of CaF2:Eu2+ (Fig. S1, ESI‡) and SrCl2:Eu2+ (Fig. S2, ESI‡) are reproduced from ref. 41 and 42.
There are several advantageous characteristics that this fully non-empirical LFDFT method possesses which should be noted and remembered, besides the predictive capability, very important today for the vast number and kind of rare-earth-based technological materials. The method can be applied to any lanthanide ions, for general 4fn–4fn−15d1 transitions with different coordination symmetries. The LFDFT approach has other advantages against widespread semi-empirical and full ab initio methods, not least the fact that it can be applied to larger-size systems in a relatively short computational time.
Footnotes |
† Dedicated to Professor Claude Daul and Professor Werner Urland in the celebration of their seventieth and seventy-first anniversaries. |
‡ Electronic supplementary information (ESI) available: The whole range of the spectral energy obtained for CaF2:Eu2+ and SrCl2:Eu2+ together with a direct comparison between the theoretical spectra and the excitation spectra in ref. 41 and 42. See DOI: 10.1039/c5cp02349a |
This journal is © the Owner Societies 2015 |